Electrical Instruments in Hazardous Locations PDF

ELECTRICAL
INSTRUMENTS IN
HAZARDOUS
LOCATIONS
4th Edition
By Ernest C. Magison
Copyright © 1998 by International S
ociety of Automation
67 Alexander Drive
P.O. Box 12277
Research Triangle Park, NC 27709
All rights reserved.
Printed in the United States of America.

10 9 8 7 6 5 4 3 2
ISBN 10: 0-9792343-1-X ISBN 13: 978-0-9792343-1-6
No part of this work may be reproduced, stored in a retrieval system, or

transmitted in any form or by any means, electronic, mechanical, photocopying,
recording or otherwise, without the prior written permission of the publisher.
Library of Congress Cataloging-in-Publication Data - In Progress

PREFACE
The first edition of this book went to press in 1966 just as ISA-RP12.2,
“Intrinsically Safe and Non-Incendive Electrical Instruments,” was pub-
lished. ISA-RP12.2 was the product of 15 years of intensive effort by
Instrument Society of America (ISA) Committee RP12 (now SP12). During
that period, the committee, as individuals and as a whole, researched
information pertinent to the engineering of safe electrical systems; formu-
lated, debated and reformulated the requirements of a safe system; and
took the lead to educate the American instrument industry about funda-
mentals of electrical hazards and safe practices.
The prominence of ISA RP12 in matters of electrical safety was evident in

the first edition of this book in the subjects covered, in the philosophy of
safety that was presented, and in the many references to publications by
RP12 members and to papers by other experts delivered at ISA-sponsored
symposia.
Most of the first edition was written as a text for the first ISA Short Course
in Electrical Safety, sponsored by the Wilmington Section in 1964. The first
edition came at the end of a period during which the American instrument
industry and others concerned with safe electrical systems learned that the
safety problems of electrical installations in hazardous atmospheres yield
to the same analytical approaches that produce solutions to other engi-
neering problems.
When the first edition was published, intrinsic safety had been accepted as
a powerful technique for safety. Many people concerned with safety had
accepted the essential concept that safety can be engineered into a system
by applying a relatively few fundamental principles—safety by engineer-
ing and design rather than by emotion and decree.
Between the first and second editions there was a period of building on
the foundation prepared by SP12. This period saw activity in electrical
safety on a much broader scale. Both the National Fire Protection Associa-
tion (NFPA) and the U. S. Coast Guard published requirements for
intrinsically safe systems that incorporated the principles of ISA RP12.2. In
addition, Factory Mutual, Underwriters Laboratories, and Canadian Stan-
dards Association (CSA) listed many intrinsically safe systems. Intrinsic
safety became a common tool of the American instrument industry.
Adopting intrinsic safety as a working tool was only one highly visible
manifestation of increased sophistication about electrical safety. Many
organizations and individuals began to reevaluate past practices, trying to
xii Preface
ascertain the facts and proposing new and different solutions to safety
problems.
With growing sophistication about electrical safety also came increased

awareness that work done in other countries, some of which predated
American activity by decades, is pertinent to our problems. Because the
nature of the problem and the validity of the solution do not depend on
the language in which they are stated, parochial, chauvinistic attitudes
were supplanted by a real desire to learn from the work and experience of
the rest of the world. An ever-growing number of individuals and organi-
zations became committed to establishing standards of safe practice that
are acceptable worldwide.
The second edition recognized the maturation processes of that period.

There was less evangelical fervor, though no less enthusiasm for a subject
that still had many unknowns and many problems to challenge the techni-
cally or politically inclined. Because the basic principles had been sold the
tone of the second edition was, therefore, somewhat more matter of fact
and tutorial.
During the period between the second and third editions important, and
not altogether welcome, changes occurred in the field of electrical safety. It
was a period when safety standards and regulations often became grist for
bureaucratic mills. In all industrialized countries the achievement of safety
became more a legislated, regulated endeavor and less an engineering
undertaking. The influence of the Occupational Safety and Health Act,
and similar legislation, in the United States was matched by parallel regu-
lations in other countries.
The general statements of principle supplied by RP12.2 no longer sufficed

in an environment where third-party certifications became the accepted
norm. Certifications, whether by governmental or non-governmental bod-
ies, demand ever-increasing detail in standards so that compliance or
noncompliance can easily be demonstrated by reference to the written
word, unencumbered by the necessity for making professional judgments.
Many design engineers and inspectors also want the solace of unequivocal
rules that need no interpretation.
The trend toward more comprehensive and detailed standards was exac-
erbated by movements toward harmonization of standards to make safety
standards the same in all countries and marketplaces. Harmonization is
healthy, as anyone employed by a manufacturer or user with plants or
markets in many countries can attest. But, as one attempts to include in a
single document all the requirements of different jurisdictions, it becomes
essential to add text to recognize historical national practices. More detail
is needed also to enhance uniform interpretation among a larger, more
heterogeneous group of design and certifying engineers.
Preface xiii
It was even more necessary to focus on fundamental principles in the third

edition than it had been 15 years earlier. As standards give more attention
to detail it becomes much easier to forget fundamentals and to continually
add more restrictive requirements “for the sake of safety.”
Although the third edition recognized these changes in the environment

in which issues of safety were addressed, like the first and second edi-
tions, it was dedicated primarily to providing a summary of
fundamentals, both theory and data, and secondarily to detailed discus-
sions of standards and regulations. The objective was still to provide
readers with a sound basis for interpretation and use of standards. It was
hoped that some readers would be both prepared and encouraged to con-
tinue the battle—to continually review standards and regulations in the
light of fundamental principles and pertinent technical facts.
The structure of this new edition is somewhat different from that of earlier
ones. The order of the chapters has been changed to what the author now
considers to be a more logical order. But the years have not altered basic
principles, and the emphasis on them in this edition is undiminished.
Many chapters have been rewritten, either to better reflect today's per-
spective or to clarify the presentation. In some places data and
illustrations have been removed if they are of historical value only and the
principles can be better illustrated with more contemporary data. Of
course, some new material has been added. Treatment of theory and prin-
ciples has been expanded in some chapters to provide the foundation for
approaching some of the questions that arise because standards and regu-
lations can't anticipate all situations. In other places, more practical
examples of application of the principles in real situations are given. The
amount of space devoted to work outside North America has again been
increased. The subject matter of this book is truly an international concern
and contributions towards settling the pressing issues come from many
sources.
I have also added and deleted full chapters. Specifically, chapters on

increased safety and ignition by optical sources have been added. Recog-
nition of Zone 1 in the National Electric Code (NEC) will increase the
interest in increased safety apparatus; therefore, it needs to be discussed in
this book. Recent investigations of ignition by optical sources and the very
early efforts of ISA SP12.21 motivated a new chapter that attempts to sum-
marize the author's understanding of this mode of ignition and offer some
opinions about how the safety standards community might react to the
issues raised.
The chapter on flame arresters has been deleted from this edition. Flame
arresters in instrumentation and other small enclosures are relatively rare
and, when used, are typically of the sintered metal variety. The references
xiv Preface
on which the chapter was based are relatively old, and the author has not
undertaken the research necessary to update the presentation.
The chapter on the European Community has also been removed from
this edition. The European Community is now such an influential factor in
the lives of a large percentage of the potential readers of this edition that
the historical background is no longer necessary. The discussion of CEN-
ELEC standards is now distributed throughout the book according to
subject matter.
Those readers who are familiar with earlier editions may notice that this
edition contains more of the author's personal opinions on the issues. It
has been my intent to make it obvious to the reader which are my opinions
and to differentiate them from tutorial material. If I have failed in this
objective, I apologize.
I have included my personal opinions for two reasons:

(1) As standard development within ISA and other organizations has
become more demanding of participants’ time, many participants tend
to assume that what has been written and adopted by others is worthy
of being copied. The drafts of documents passing across my desk are
numerous and demonstrate, in many cases, that the objective is to get a
document into print, whether or not it is needed or founded on a well-
developed philosophy and rationale. In many cases it is obvious that
the drafters are completely unaware of the historical precedents for
their work.
(2) I have now reached the age where the old man's disease makes its
appearance. I have seen it in others. This disease leads people to think
that everything they have to say is worth listening to, and everything
they think is worth saying!
Ernie Magison
TABLE OF CONTENTS
PREFACE xi
CHAPTER 1 HISTORICAL BACKGROUND AND PERSPECTIVE 1
Growing Interest in the Safety of Electrical Instrument Installations, 1
The Role of Underwriting and Standards Developing Organizations, 3
Occupational Safety and Health Administration, 12
Progress in Electrical Safety Standards Development, 14
References, 14
CHAPTER 2 COMBUSTION AND EXPLOSION FUNDAMENTALS 17
Some Underlying Theory, 17
Ignition By a Point Source, 21
Ignition Energy and Flame Velocity, 28
Lower and Upper Explosive Limits, 29
Most Easily Ignited Concentration, 31
Minimum Ignition Energy, 32
Effect of Test Conditions, 32
Effect of Changing Inert Gases, 38
Electrode Geometry, 39
Time Scale of the Ignition Process, 40
References, 41
CHAPTER 3 CLASSIFICATION OF HAZARDOUS LOCATIONS AND
COMBUSTIBLE MATERIALS 43
Why Area Classification?, 43
The North American System, 43
Classes and Groups of Materials, 44
Determination of Area Classification Relative to Class I Hazards, 49
Considerations in Classification of Class I Locations, 51
Some Guides for Class I Area Classification, 55
Special Cases of Area Classification in Instrument Systems, 58
Area and Material Classification Outside the
United States, 60
Material Classification in the United States—Tutorial Discussion and Opinion, 75
References, 94
vi Table of Contents
CHAPTER 4 PRACTICE AND PRINCIPLES OF HAZARD REDUCTION

PRACTICE 99
United States, 99
Canada, 102
Countries Following IEC or CENELEC Standards, 102
The Types of Protection, 103
Principles, 108
References, 134
CHAPTER 5 EXPLOSIONPROOF ENCLOSURES 137
Why an Explosionproof Enclosure Works, 137
Influence of Enclosure Design and Test Procedure, 139
Pressure Piling, 153
Correlation Between MESG and Spark Ignition Energy, 157
Design Criteria—Introduction, 159
References, 176
CHAPTER 6 REDUCTION OF HAZARD BY PRESSURIZATION 183
Recognition in the National Electrical Code, 183
The Standardization of Pressurization, 184
Classification of Pressurization Systems, 186
NFPA 496 Requirements for Pressurization Systems, 187
Additional Requirements for Pressurized Enclosures in Class I Locations, 192
Requirements for Pressurized Enclosures in Class II Locations, 193
Requirements for Pressurized Control Rooms, 193
Pressurization of Enclosures with Internal Source of Release—Continuous
Dilution, 194
CENELEC EN50016, 204
Installation of Pressurization Systems, 211
References, 215
CHAPTER 7 ENCAPSULATION, SEALING, AND IMMERSION 217
Oil and Sand Immersion, 217
Sealing, 219
Defining Safety Requirements for Sealed Devices, 229
Standards for Sealed Devices, 235
Encapsulation or Potting, 244
References, 247
CHAPTER 8 INCREASED SAFETY, TYPE OF PROTECTION e 251
Construction Requirements, 253
Creepage and Clearances, 254
References, 263
Table of Contents vii
CHAPTER 9 IGNITION OF GASES AND VAPORS BY ELECTRICAL

MEANS 265
Characteristics of Electric Arcs, 267
Arcing at Closing Contacts and in Capacitive Circuits, 274
Typical Test Equipment for Capacitive Circuits, 278
Precautions in Testing, 281
Typical Ignition Test Results, 284
Effect of Resistance in Discharge Path, 286
Effects of Electrode Material and Geometry, 294
Opening Contacts in Inductive and Resistive Circuits, 294
Test Equipment for Break-Spark Ignition Measurements, 296
Influence of Contact Material, 299
Effect of Contact Separation Speed, 306
Ignition of Materials Other than Methane, 307
Probability of Ignition—Influence of Current and Voltage Level, 311
Effect of Shunt Elements on Ignition Current, 315
Hot Wire Ignition, 324
References, 335
CHAPTER 10 INTRINSICALLY SAFE AND NONINCENDIVE
SYSTEMS 341
Historical Review, 341
Interpreting the Definition, 345
Specific Requirements for Intrinsically Safe Systems, 349
Requirements for All Intrinsically Safe Apparatus, 354
Requirements for Features and Components on which Intrinsic Safety
Depends, 359
Requirements for Infallible (Protective) Components, Infallible Assemblies, and
Infallible Connections, 367
How Safe is Safe?, 374
Installation of Intrinsically Safe Systems, 394
Inspection of Intrinsically Safe Systems, 400
Maintenance of Intrinsically Safe Systems, 404
Nonincendive Equipment and Wiring, 406
References, 410
viii Table of Contents
CHAPTER 11 DESIGN AND EVALUATION OF INTRINSICALLY SAFE

APPARATUS, INTRINSICALLY SAFE SYSTEMS, AND
NONINCENDIVE SYSTEMS 415
Design and Evaluation of Intrinsically Safe Apparatus, 415
Power Supplies, 428
Inductors with Ferromagnetic Cores, 433
Zener Diode Track Layout, 461
Design of Intrinsically Safe Systems, 469
References, 489
CHAPTER 12 IGNITION BY OPTICAL SOURCES 493
Overview, 493
Ignition by Radiating Sources, 497
The Nature of the Hazard, 507
Standards Actions to Address Hazards of Radiation - Opinions of the Author, 516
Need for Further Study and Research, 517
References, 518
CHAPTER 13 DUST HAZARDS 521
Nature of the Dust Hazard, 521
Influence of Chemical Composition of the Dust, 523
Influence of Shape, Size, and Concentration, 527
Chemical Composition of the Suspending Medium, 529
Distinguishing Features of Dust Hazards, 531
Area Classification, 533
Electrical Apparatus For Dusty Locations, 538
Ways to Reduce Hazard, 541
IEC and CENELEC, 541
References, 544
CHAPTER 14 HUMAN SAFETY 547
Introduction, 547
Effects of Electrical Shock, 547
The Body as a Circuit Element, 550
What is a Safe Level of Circuit Voltage?, 553
Conclusion, 554
References, 554
CHAPTER 15 DEGREE OF PROTECTION BY ENCLOSURES 555
The NEMA System, 555
IEC 529, 558
Correlation of NEMA and IP Codes, 562
References, 562
Table of Contents ix
APPENDIX A DERIVATION OF INITIAL PURGE VOLUME

REQUIREMENT 563
APPENDIX B EXAMPLES OF DERIVATIONS 567
Derivation of an Equation for Loss of Pressure in a Sealed System, 567
Derivation of Expressions for Seal Breathing, 569
Derivation of Expressions for Pressure Testing Sealed Enclosures, 572
Derivation of Flow Test Equation, 574
INDEX 575
CHAPTER 1
HISTORICAL BACKGROUND AND
PERSPECTIVE
Growing Interest in the Safety of Electrical Instrument

Installations
Concern of instrument manufacturers and users for the problems of safety
in electrical instrument installations has, not surprisingly, been closely
related to the number of electrical instruments used in hazardous loca-
tions. In the typical refinery or chemical plant of the 1930s and early 1940s,
most instruments were locally mounted mechanical flowmeters, Bourdon-
actuated pressure gauges, and galvanometer-actuated mechanical potenti-
ometers and millivoltmeters. When safety was of concern, even the chart
drives of these instruments could be hand-wound, spring driven, or air-
operated devices. Such installations posed few problems of electrical
instrument safety. Installation practices for electrical instruments were
derived from power and lighting practices. In hazardous locations explo-
sionproof construction was typical, although pressurization or purging
was not uncommon. These practices might not have been optimal; how-
ever, there was no strong motivation to reduce the cost of such safety
measures or to increase the degree of safety that could be achieved at the
same cost.
During the 1940s, instrument systems grew more complex, requiring large
panel boards in control rooms that were so large that one operator could
not monitor all the instruments. A control room crew typically consisted
of several men. Small-case pneumatic receivers and field-mounted trans-
mitters were introduced to reduce space and manpower requirements.
Safe operation of electrical instruments was still not of major concern,
even though electrical potentiometric recorders were often used in control
rooms. If there was real concern for safety, the most conservative users
still adopted the approach of explosionproofing parts of the instrument or
pressurizing it; less conservative people used the instrument in its normal
operating condition. Because most instrumentation was pneumatic or
mechanical, however, there was no real incentive to develop a sophisti-
cated concern for electrical safety.
During the 1950s, the trend toward faster, more versatile control systems
continued, with the introduction of small case electrical control systems.
These electrical control systems met growing demands for higher speed
2 Historical Background and Perspective
and were more compatible with data loggers and computers. Even in
those installations where the electrical control system did not have signifi-
cantly higher speed of response than the pneumatic system it replaced,
users found that the higher sensitivity of the electrical system gave tighter
control of the process. This, too, encouraged the installation of electrical
systems. Concurrently, an increasing number of quality analyzers were
being applied to supplement conventional flow, pressure, temperature,
and level control loops. Because most quality-measuring instruments —
such as calorimeters, infrared analyzers, chromatographs, pH meters, and
conductivity analyzers — yield electrical signals, compatibility of such
instruments with conventional control loops biases the system designer
towards a completely electrical system. In the 1960s, systems comprising
hundreds of electrical instruments in hazardous locations became
commonplace.
Such installations demanded reconsideration of the methods used to

ensure the safety of electrical instrument installations. The excessive cost
of past practices, when duplicated hundreds of times, could no longer be
ignored. New devices needed new approaches to safety. Many analyzers
were not amenable to brute-force explosionproofing, because they had to
be opened frequently for routine calibration and maintenance. Methods
for ensuring safety were needed that would make routine operations more
convenient and less expensive.
When accessibility was not a prime factor in generating interest in new

ways to handle safety issues in electrical installations, installation cost
was. The additional expense of installing an explosionproof enclosure was
negligible compared to the cost of installing conduit, sealing fittings, and
explosionproof junction boxes to provide the explosionproof wiring sys-
tem that must be used in conjunction with explosionproof enclosures.
In addition to the economic motivation for safer and more efficient instal-
lation and operating practices, there was also a movement toward
standardization of safety requirements. Over the years many users had
developed their own corporate practices and internal standards. There
were no commonly accepted industry-wide standards of safety. The
requirements of the National Electrical Code (NEC) applied in general,
but there were no specific rules for instrument systems. Manufacturers
were, therefore, required to satisfy individual customer requirements and
were not always able to standardize design to achieve manufacturing
economy. As the quantity of electrical instruments increased, so did the
magnitude of the safety problem and the need for standardized require-
ments for equipment in hazardous locations.
Historical Background and Perspective 3
The Role of Underwriting and Standards Developing

Organizations
The National Electrical Code
In the United States most electrical installation practices are based either
directly or indirectly on the NEC. Although States, municipalities, or
insurance companies may have their own codes for electrical installations,
they are usually based on NEC requirements. Many municipalities adopt
the NEC by ordinance. Until recently the NEC, prepared by the NEC
Committee of the National Fire Protection Association (NFPA), was pri-
marily concerned with power and lighting in public and private
occupancies. Instrument requirements emphasized panel board instru-
ments and power control devices, not industrial measuring and control
instrumentation.
Many users of industrial measuring and control instrumentation devel-

oped installation practices that were proven safe by experience, but which
did not adhere to the letter of NEC requirements. With the advent of the
Occupational Safety and Health Administration (OSHA), and the threat of
citations for installations not conforming to the letter of the NEC, organi-
zations such as the American Petroleum Institute, the Manufacturing
Chemists Association, and the Institute of Electrical and Electronic Engi-
neers began to propose amendments and additions to the Code to
recognize safe practices that had not been previously codified. The NEC
has evolved to recognize the needs of instrument systems as more manu-
facturers and users recognize that the NEC is a dynamic document that
can be changed to meet new circumstances or to recognize new technol-
ogy. In addition, the NEC has added requirements for low energy systems
such as cable TV, fiber-optics, and alarm systems, which can sometimes
provide guidance for the installation of instrument systems.
The NFPA has assigned responsibility for maintaining and revising the
NEC to the National Electrical Code Committee. This committee consists
of a correlating committee and a number of code-making panels, each cog-
nizant of one or more sections of the Code. Panel membership is as
broadly based as practical, typically including representatives of insurers,
manufacturers, users, testing authorities, inspectors, government agen-
cies, labor unions, and similar bodies. The correlating committee
establishes the timetable for triennial revision of the NEC and reviews rec-
ommendations for NEC changes from all the panels to ensure that no
changes are made without substantial consensus. The committee also
ensures that the NEC is editorially consistent and, most importantly, that
any panel action that affects more than one portion of the NEC is reviewed
and agreed upon by all the panels whose sections may be affected.
Any interested person or organization may propose a change to the Code.

A proposal must be specific, stating the exact alteration or addition
desired, with substantiating reasons. A desired format for proposals and
the calendar for preparation of the next edition of the NEC is given in the
Appendix to the NEC. Action by the panels in response to proposals is
available for review and comment by the public.
Underwriters Laboratories, Inc., and Factory Mutual

Research Corporation
The NEC frequently refers to “approval” by “the code-enforcing author-

ity,” or in recent years, “the authority having jurisdiction.” This may be a
local inspector, an insurance underwriter's representative, or a municipal
authority such as a fire marshal or electrical inspector. The “code-enforc-
ing authority” is the person responsible for approving a specific
installation. “Approved” is commonly misinterpreted to mean listing or
labeling by Underwriters Laboratories (UL) or approval by Factory
Mutual Research Corporation (FM). In most cases, however, UL or FM
listing or approval of equipment ensures approval by the local inspector
or “code-enforcing authority.” In recent years the NEC has begun to use
the terms “listed” or “labeled” when it is intended that a product must
have been examined and found to be in compliance with a particular
requirement by a third party testing organization such as FM or UL.
UL was founded in 1894 as a testing laboratory. Its purpose was to exam-

ine products for personnel and fire hazards for the benefit of insurance
companies. In 1917 it became an independent, self-supporting, non-profit
safety-testing laboratory. At that time, it was still under the sponsorship of
the National Board of Fire Underwriters (now, the American Insurance
Association). In 1968 the charter of UL was broadened. The board of trust-
ees now includes representatives of consumer interests, government
agencies, public safety bodies, utilities, and standardization bodies, as
well as those from the insurance industry.
UL has prepared standards for a broad range of safety devices, consumer

products, and industrial control devices. Many UL standards are ANSI
(American National Standards Institute) standards; that is, they are recog-
nized as the primary standard in the United States for that product. UL
provides to manufacturers testing, labeling, and listing servicers that sig-
nify adherence to UL standards. In general, UL labeling or listing is
acceptable to all inspection authorities in the U.S. In recent years UL has
begun to offer other services, such as certification to product and safety
standards of other organizations, and auditing and certifying for compli-
ance with the ISO (International Organization for Standardization) quality
assurance standards.
Factory Mutual Research Corporation is a division of the Factory Mutual

System, sponsored by the major mutual companies in the property insur-
ance business. Although not as well known to the average consumer as
UL, FM has provided similar testing, listing, and labeling services for con-
struction materials, industrial equipment, and safety-related products for
many years.
Until about 1960, the industrial instrument industry only infrequently

used UL or FM testing, listing, or labeling services. Approval can only be
granted for a specific design or family of designs. Because much industrial
instrument business at that time was to special order, the cost, both in time
and investment, of obtaining approval was thought to be a major impedi-
ment to use of listing services. Users felt sufficiently capable of designing
installations and selecting equipment that would receive approval by the
local inspector or other enforcement authority that they did not demand
UL or FM listing.
During the late 1960s, many users began to require UL listing or FM

approval of intrinsically safe systems, and manufacturers obtained such
verification of major lines of electric process control instrumentation.
Intrinsic safety was new. Many users who felt competent to pass on the
acceptability of unlisted explosionproof housings did not feel competent
to assess the safety of unlisted intrinsically safe systems. As a practical
matter, though, even those who were competent to judge the safety of a
system could not usually obtain the detailed design information necessary
to make an assessment. Most importantly, a third party agency has the
power to require that no changes be made to an approved design without
prior approval to ensure that safety was not affected. Few customers could
make such a demand or marshal the resources to enforce compliance.
The demand by users for UL or FM labeling of industrial instruments

increased rapidly after 1971 because of the influence of OSHA regulations.
Many users now demanded listing of all apparatus to be installed in loca-
tions classified per Article 500 of the NEC, whether explosionproof or
intrinsically safe. OSHA regulations mandated FM or UL approval, not
only for hazardous location equipment, but for electrical equipment in
ordinary locations as well. This is still true, but OSHA regulations now
recognize the ability of additional organizations to certify compliance to
safety standards, and there are options for the use of unapproved equip-
ment in some circumstances. The organizations empowered by OSHA to
certify compliance to recognized standards are known as NRTLs (Nation-
ally Recognized Testing Laboratories).
Canadian Electrical Code and Canadian Standards

Association (CSA)
In Canada the Canadian Electrical Code, Part 1, similar in format and con-
tent to the National Electrical Code, is the reference for electrical
equipment installations. Specific component and equipment requirements
are spelled out in standards published by CSA, a non-profit non-govern-
ment organization established in 1919. These standards form Part 2 of the
Canadian Electrical Code.
Committees of manufacturers, inspectors, and users prepare CSA stan-

dards. Adherence to standards is determined by the Certification Division
of CSA (until 1969, called the CSA Testing Laboratories), which was inau-
gurated in 1940. Before the 1940s, evaluation and certification of
equipment had been a function of the Ontario Hydroelectric Commission,
whose certification was recognized across Canada. Because Ontario
Hydro was not a national organization, it asked CSA to assume responsi-
bility for the national testing program.
An Approvals Council for all electrical equipment has authority to accept

or reject CSA electrical certifications. Approval is not final until the
Approvals Council has balloted favorably. Because the Approvals Council
membership includes the chief inspectors of all ten provinces and two city
inspectors, Canadian inspection and enforcement authorities recognize
CSA certifications.
CSA is an NRTL and its certification to ANSI standards for electrical

equipment is acceptable to OSHA. Certification to CSA standards is not
acceptable unless they have been harmonized with the ANSI standard. A
product certified by CSA to ANSI standards is marked with the CSA logo
that has an NRTL subscript. Though this certification may be acceptable to
OSHA it may not be acceptable in the U. S. marketplace, depending on the
type of equipment and user.
Although onsite inspection services are available for unique installations

of apparatus not certified by CSA, the service is costly. As a practical mat-
ter all apparatus to be marketed in Canada should be CSA-certified to
ensure acceptance by local inspection authorities. This applies to appara-
tus for both hazardous and nonhazardous locations.
Influence of the International Society for Measurement

and Control
The Instrument Society of America (ISA) Committee SP12 (formerly 8D-

RP12) was established in 1949, recognizing that both users and manufac-
turers need standard practices to promote safety. Early progress of the
committee was slow. In place of accepted standards of safety there existed
a mass of lore, opinion, and prejudice about the hazards of electrical
devices. Respect for the letter of the NEC was not matched by understand-
ing of either its spirit or the underlying principles of safe design and
installation.
The philosophy of ISA SP12 has been to use NEC Article 500 as the start-
ing point for a series of recommended practices to promote the safe,
economical installation of electrical instruments in hazardous locations. It
is not the intention of ISA SP12 to prepare recommendations that are con-
trary to the spirit of the NEC. Its aim is to recommend specific safe and
economical practices for those situations commonly found in the instru-
ment industry that are not dealt with in detail in the NEC. The
recommended practices of SP12 are intended to be as free as possible from
arbitrary rules. They are to be based on consideration of the physical fac-
tors determining safety, and are to represent reasonable solutions to a
problem involving both cost and safety considerations, without sacrificing
either cost or safety.
From 1954 to 1974 ISA SP12 made substantial progress under the leader-
ship of Chairman F. L. Maltby of Drexelbrook Engineering Company.
Subcommittees were established to broaden participation and expedite
committee work in areas such as:
• Class II dust hazards
• Wiring in hazardous locations
• Sealing and encapsulation
• Purging and pressurization
• Area classification
During this period, SP12 produced three recommended practices. RP12.1,

“Electrical Instruments in Hazardous Atmospheres,” published in 1960,
established basic definitions and guidelines. It summarized the relation-
ship between the classification of the hazardous location, the nature of the
electrical equipment, and the degree of protection required. It brought
together in single document the essential principles of safe application of
electrical instruments in hazardous locations. RP12.1 was completely
revised and expanded in 1991. It remains a tutorial and reference docu-
ment that attempts to give an overview of the methods of protection
available to reduce electrical hazards. The revised document also attempts

to make the reader aware of important differences between North Ameri-
can vocabulary, standards, and practices and those used in other
jurisdictions.
ISA issued RP12.4, “Instrument Purging for Reduction of Hazardous Area

Classification,” in 1960 and revised it in 1970. This document provided
much-needed standards for purging and pressurizing instrument enclo-
sures, a commonly accepted technique for reducing hazards. RP12.4 was
used by NFPA as the basis for the first edition of NFPA 496, which was
later expanded to cover other applications of purging and pressurization
(see Chapter 6). It is now intended that RP12.4 will become a guide to the
design of purging and pressurization systems, rather than a standard for
what is required.
RP12.2, “Intrinsically-Safe and Non-Incendive Instruments,” published in

1965, provided the first published requirements in the United States for
intrinsically safe equipment and wiring and for equipment suitable for
installation in Division 2 locations without special protection.
In recent years the subcommittee structure has been expanded to include

subjects such as combustible gas analyzers used to ensure safety and ana-
lyzers for specific hazardous gases.
The aim of each of the five subcommittees is to make available recom-

mended practices or standards in its area of cognizance. At the outset it
was expected that if a standard or recommended practice was not yet
appropriate the subcommittee could publish an information bulletin.
These are now called Technical Reports. ISA TR12.2-1995, “Intrinsically
safe System Assessment Using the Entity Concept” is an example. If
another organization is already actively working on similar documents,
the subcommittee formulates an ISA position and attempts to influence
these activities accordingly. Similarly, SP12 encourages liaison members
from other organizations with like interest in safety practices to eliminate
unnecessary duplication of effort. The American Petroleum Institute, Divi-
sion of Refining, through such liaison membership on SP12, endorsed the
first editions of RP12.1, RP12.2, and RP12.4. In recent years the necessity
for formal liaisons has decreased because many members of SP12 are also
members of the other committees active in the field of electrical safety.
SP12 members have had continuing influence in the standards activities of

the American Petroleum Institute, NFPA, the Chemical Manufacturers
Association (formerly Manufacturing Chemists Association), NEC Panel
14, UL industry advisory groups, and National Academy of Science panels
on hazardous materials and grain dust explosions. Committee members
are also influential in the activities of TC31 of the International Electro-
technical Commission, which is charged with preparing international
standards for electrical devices in hazardous locations. Several serve on

the advisory group to the U.S. National Committee for TC31. Others have
attended IEC TC31 meetings as U.S. delegates. One member of SP12
serves on the Canadian National Committee for IEC TC31, has been chair-
man of IEC SC31A on flameproof (explosionproof) enclosures, and is now
chairman of IEC TC31.
Recognition of intrinsic safety and nonincendive circuits in the NEC, intro-

duction of Article 504 on installation of intrinsically safe circuits in the
1990 NEC, introduction of Zone definitions in Article 505 in the 1996 code,
and several other significant amendments to the Code can be attributed
largely to the efforts of SP12 members.
Two of the Recommended Practices written by SP12 served as the founda-

tion for NFPA standards. During the early years of SP12 activity, NFPA
standards had wider acceptance by inspection authorities and by approval
laboratories than ISA Standards and Recommended Practices. For this rea-
son the philosophy of SP12 in the 1970s was to concentrate on free
discussion of issues and the development of a sound engineering position.
Through liaison memberships, technical papers, and the like, the commit-
tee attempted primarily to influence other standards-writing bodies to
incorporate the ISA position into their documents. For example, rather
than prepare an ISA recommendation on area classification, which,
although it might be technically sound, would not be widely accepted as
authoritative, SP12 concentrated its energies on education and attempting
to have the NEC changed. SP12 still wrote standards which did not dupli-
cate existing activity, or if it could not influence another organization to
undertake the project. The preparation of RP12.6 concerning the installa-
tion of intrinsically safe systems was undertaken for this reason.
Accreditation of ISA as a voluntary consensus standards writing organiza-

tion by ANSI in 1976 was one indication of the increased acceptance of ISA
as an important participant in the electrical safety standards community.
ISA SP12 therefore resumed its standards development work, and it was
no longer necessary to use another organization's standard as a vehicle for
SP12 ideas.
National Fire Protection Association
The NEC is only one of hundreds of standards published by the NFPA in

the interest of reducing fire and personnel hazards. NFPA standards are
readily accepted in the electrical industry. Publication of NFPA standards
on intrinsic safety and pressurization that used ISA-recommended prac-
tices as a starting point did much to disseminate ISA concepts and to give
them credibility.
In general, NFPA standards have tended to emphasize recommendations

for safe use of electrical apparatus, area classification, fire protection, and
hazards of materials rather than standards for equipment. NFPA 493, con-
cerning construction and testing of intrinsically safe systems, was a
notable exception. After UL agreed to develop a revision of UL913, using a
consensus committee approach, it was no longer necessary to maintain
NFPA 493, and it has been withdrawn.
NFPA committees often provide a forum for consideration of important

proposals for amending the NEC. The Sectional Committee on Electrical
Apparatus in Chemical Atmospheres (EECA), for example, was influential
in the adoption of several important changes in the scheme for material
grouping in the NEC and successfully proposed grouping of many new
materials in the NEC. The NFPA Standards Council has now assigned the
responsibility for material classification to the EECA committee. Because
the number of materials now classified is too large to be conveniently
accommodated in the NEC, the materials are now listed in NFPA 497, a
publication of the EECA committee.
National Academies of Science and Engineering
The National Academies of Science and Engineering operate through the

National Research Council to provide advisory services to government
agencies. Although the academies are chartered to provide technical and
scientific advice to agencies of government, they are not government orga-
nizations. They are privately incorporated.
During the past 25 years a number of agencies, including the U. S. Coast

Guard, OSHA, and National Institute for Occupational Safety and Health
(NIOSH) have funded studies on material classification, testing, and pre-
vention of grain elevator explosions. The results of these studies are
published in reports of the National Materials Advisory Board, an operat-
ing arm of the National Research Council. They have, in many instances,
influenced the NEC, NFPA 497M, and NFPA 497, which deals with the
classification of hazardous locations in chemical plants.
International Electrotechnical Commission

The International Electrotechnical Commission (IEC), in which most coun-
tries hold membership, prepares international recommendations and
standards for construction and use of electrical apparatus. IEC standards,
if adopted by the member nations as national standards or by purchasers
of electrical equipment, can directly and substantially affect products
manufactured or designed in the United States for sale abroad. Many
industrialized nations adopt IEC standards, and it is common practice for

developing nations to pattern their standards after IEC standards.
A particularly important influence of IEC standards arises from the com-

mitment of CENELEC, discussed below, to adopt IEC standards whenever
possible, rather than write new ones.
Since 1950 United States manufacturers and some large users have greatly
increased support of delegates to IEC committees in an effort to ensure
that IEC standards recognize U.S. needs and practices. A concomitant ben-
efit of participation is recognition that United States standards must be
consistent with IEC documents, where feasible, so that both users and
manufacturers of electrical apparatus can avoid costly differences in plant
or equipment design in different nations. It is the policy of the IEC that
nations who vote in favor of an IEC standard must, in due course, amend
their national standards to harmonize them with the IEC document.
CENELEC
Since 1978 any electrical apparatus for use in hazardous locations that is to
be marketed in the European Community (EC) and enjoy free passage
through customs of the EC member nations must conform to the relevant
European Electrotechnical Standards Committee (CENELEC) standard
and be certified by a competent laboratory in the EC.
CENELEC is made up of members of the EC and the European Free Trade

Association — essentially all European and Scandinavian countries. Its
standards, when adopted by directive of the Council of the EC, have the
force of law in the EC-member countries. These countries are obliged to
adopt the standard in accordance with the provisions of the treaties that
established the EC. EFTA countries are not bound to adoption of CEN-
ELEC standards; however, as a matter of facilitating trade, there has been
a history of harmonization of requirements even in the absence of such a
mandate.
In general, if an IEC standard is available and is technically acceptable,

CENELEC will try to adopt the IEC standard because an IEC standard
already represents a significant harmonization of differing national stan-
dards. If acceptable IEC standards are not available (as was the case when
standards for hazardous location equipment were needed in the late
1970s), CENELEC will write its own documents. CENELEC will also devi-
ate to some extent from an IEC standard if the IEC standard is considered
to be too vague. Often ambiguous or vaguely worded clauses are adopted
in an IEC standard in order to achieve consensus despite widely divergent
views or longstanding practices. This practice is justified on the basis that
a standard that can serve as an objective for national standards to emulate
is better than no standard at all. CENELEC, a regional organization with

less heterogeneous membership, can often resolve such ambiguities in the
preparation of its regional standard, and so its document will differ from
the parent IEC document.
The United States has no representation in CENELEC committees. Its

interests may be reflected only in the content of IEC standards, if they are
adopted, or through influencing the thinking of members of IEC commit-
tees who are also members of the corresponding CENELEC committees,
which is a common situation.
To further the objective of a single European economic entity by 1992

CENELEC was assigned the task of harmonizing all standards that would
otherwise be barriers to trade among the EC countries. The specter of an
Economic Fortress Europe with regulations and standards biased against
North American and Far Eastern manufacturers caused great concern. To
ameliorate that concern CENELEC agreed to allow Secretaries of IEC com-
mittees to sit with corresponding CENELEC committees.
The influence of CENELEC standards was further raised in importance by

the EC Council mandate that, effective January 1, 1996, all products must
bear the CE Mark. The CE mark indicates conformance to all Council
directives whose scope includes that product. At present, manufacturers
of electrical equipment are required to conform to CENELEC EMC stan-
dards and a general purpose electrical safety standard such as EN61010,
the CENELEC version of IEC1010. Hazardous location certification contin-
ues as in the past while the details for mandatory application of the
Explosives Atmospheres Directive in 2002 are being settled. This directive
covers all equipment that might be used in a hazardous location, not only
electrical equipment, and prescribes acceptable patterns of internal quality
control and certification. In general, equipment that poses the most severe
hazards in use, such as that for use in Zone 0 or 1, requires third party cer-
tification. Equipment that poses lesser hazards may be subject only to self-
certification and quality control by the manufacturer. The various mod-
ules provide some flexibility for the manufacturer to choose the
combination of certification, internal quality control, and followup verifi-
cation to suit the nature of his product and its manufacturing volume.
Occupational Safety and Health Administration

The enactment of the Occupational Safety and Health Act on December 29,
1970, ushered in a new era relative to standards for electrical apparatus
and listing of electrical equipment. Part 1910 of the OSHA regulations
(29CFR), published on May 29, 1971, adopted the 1968 NEC and defined
“approved” to mean “listed by UL or FM.” On February 15, 1972,
“approved” was redefined, and the 1971 NEC was adopted by reference.
The new definition seemed to provide certain exceptions to FM or UL list-

ing, but in practice the emphasis on listing was not changed. The listing
requirements of Part 1910 greatly increased interest in developing stan-
dards for categories of apparatus such as process control instrumentation
for which none had ever been written, other than those for hazardous
locations. More importantly, the regulations sparked interest in harmoniz-
ing diverse standards so that FM and UL, or other agencies that might be
accredited by OSHA, would all judge apparatus by the same criteria.
Because of the emphasis on listing, it became necessary to write standards
with third party certification in mind, that is, to include more detailed con-
structional and performance requirements. As a standard becomes widely
used for certification, both the submitter of equipment and the examining
engineer tend to want detailed guidelines in order to avoid issues arising
from differences in judgment or interpretation.
OSHA has not attempted to write product standards for electrical appa-
ratus. However, on January 16, 1981, rather than continuing to adopt the
NEC by reference, OSHA published its own electrical standards. These
standards agreed in most respects with the NEC, but those portions of the
NEC that were not directly pertinent to health and safety were eliminated.
Publication of the electrical standards, instead of adopting the NEC by ref-
erence, eliminated a potential administrative problem of gigantic
proportions that would have resulted every time a new edition of the NEC
was adopted by NFPA. OSHA would have had to propose adoption of the
new edition and solicit public comment, incurring the risk that issues that
had already been dealt with by the NEC Committee would be reactivated
during the comment period. In the Electrical Standards, Part 1910.307, elec-
trical installations in hazardous locations are required to be intrinsically
safe, approved for the hazardous (classified) location, or safe for the haz-
ardous (classified) location. This latter equipment is equipment that the
employer demonstrates will provide protection from the hazards arising
from the combustibility and flammability of vapors, liquids, gases, dusts,
or fibers.
By funding the operations of the National Research Council Committee on

Evaluation of Industrial Hazards, OSHA encouraged the classification of
many new gases, vapors, and dusts and supported studies of possible
improvements in the way materials are grouped in the NEC. The list of
materials in NFPA 497M-1983 was adapted from the report of this com-
mittee, Classification of Gases, Liquids, and Volatile Solids Relative to
Explosionproof Electrical Equipment, publication NMAB 353-5.
Progress in Electrical Safety Standards Development

The organizations mentioned in this chapter are not the only ones that
influence standards for electrical apparatus in hazardous locations in the
United States, but they are the most influential. In recent years, especially
since the advent of OSHA, all these organizations have recognized that an
effective standard must attempt to reflect the interests of all concerned
parties, that is, they must be of stature suitable for adoption as an Ameri-
can National Standard. Only in this way can the standard be a strong
working document in international negotiations or in dealing with domes-
tic regulatory authorities. These organizations have also learned to
cooperate to achieve consensus in a single standard rather than prepare
competing documents. Considerable impetus for such cooperation came
from the recognition that there are not enough financial and personal
resources available to prepare the standards that are needed, and, there-
fore, certainly not enough to squander on parochial competing standards
designed to serve narrow interests.
During the past few years efforts to harmonize CSA, FM, and UL stan-
dards have strengthened, but progress is slow. These laboratories have
also instituted programs for obtaining certification to the standards used
by other laboratories. It is now possible to obtain certification from several
laboratories by making a submittal to only one.
References
Johnston, J., Jr., “Organization and Operation of the Electrical Code Com-
mittee,” Proceedings of 1960 Symposium on Safety for Electrical
Instrumentation in Hazardous Areas, Instrument Society of America,
Pittsburgh.
Maltby, F. L., “History of ISA Committee on Hazardous Area Instrumen-

tation,” Proceedings of 1960 Symposium on Safety for Electrical
Pittsburgh.
Magison, E. C., “Engineering Instruments for Safety,” Instrumentation

Technology (February 1969), pp. 41-46.
Schall, G. E., Jr., “Underwriters' Laboratories: Testing for Public Safety,”

IEEE Transactions on Industry and General Applications, Vol. IGA-6, No. 5
(September/October 1970), pp. 425-429.
Burklin, C. R., “Safety Standards, Codes and Practices for Plant Designs,”
Chemical Engineering (September, October, November 1972).
Magison, E. C., “OSHA Subpart S: Dilemma for the Electrical Industry,”

Electrical Consultant (November 1974), pp. 29-32.
Magison, E. C., “The National Electrical Code, Safety, and OSHA,” Pro-
ceedings of Symposium on Instrumentation in the Process Industries,
Texas A&M University, College Station, Texas, January 1975.
Magison, E. C., “Electrical Safety Standards in the United States—A Status

Report,” IEE Conference Publication No. 134, London, 1975, pp. 11–15.
Riddlestone, H. G., “1971–1975—A Review of Developments and Prob-

lems Outstanding,” IEE Conference Publication No. 134, London, 1975,
pp. 1-5.
Committee on Evaluation of Industrial Hazards, “Classification of Gases,

Liquids, and Volatile Solids Relative to Explosionproof Electrical Equip-
ment,” NMAB 353-5, National Academy Press, Washington, D.C., 1982.
Magison, E. C., “A North American Looks at Electrical Apparatus Stan-

dards,” IEE Conference Publication 218, London, 1982.
CHAPTER 2
COMBUSTION AND EXPLOSION
FUNDAMENTALS
Some Underlying Theory

This chapter presents a simplified view of the development of a combus-
tion wave. It provides the basis for a qualitative understanding of the
effects of changing the properties of the combustible material or the char-
acteristics of the ignition source. The inclusion of the term “explosion” in
the title is deliberate to emphasize that “combustion” and “explosion” are
simply two faces of the same phenomenon. The term “explosion” is used
in this book to denote any uncontrolled and undesired combustion. Our
definition of the term includes unconfined phenomena, which others pre-
fer to call flash fires or deflagrations and high-speed detonations.
Although some authors associate some degree of confinement of combus-
tion to make an explosion, this distinction is not, in the context of this
book, essential.
An explosion is not a unique phenomenon. It is only a self-propagating

combustion wave that is not kept under control. There is no fundamental
difference between an industrial fire or explosion that destroys property
and combustion in a domestic gas stove, except that the former is not kept
under control. This basic similarity is well understood by anyone who has
first turned on the gas in the oven and then had difficulty lighting the
match.
A theoretical treatment of ignition phenomena and combustion theory is

beyond both the scope of this book and the author's ability. However, a
detailed treatment of chemical kinetics and thermodynamics is unneces-
sary for profitable consideration of the relationship of electrical apparatus
to ignition of flammable gases, vapors, and dusts. (This chapter will focus
on gases and vapors. Dusts will be discussed in Chapter 13.) It is possible
to view the ignition process in an approximate phenomenological manner
and gain useful understanding of it. The material that follows treats the
ignition process in a highly oversimplified manner, using simple concepts
and models that the author and engineers of his acquaintance have found
useful. The material is presented as an aid to visualization of the ignition
process. For a consistent, theoretically grounded treatment of ignition and
combustion phenomena the reader is referred to the writings of specialists
such as Lewis and von Elbe and Khitrin.
18 Combustion and Explosion Fundamentals
Combustion is not at all simple. The reaction that we simplistically sum-

marize as: 2H2 + O2 → 2H2O, with a release of 115.6 kilocalories of energy,
is in reality a collection of simultaneous and consecutive reactions, each
with characteristic thermochemical, thermodynamic, and kinetic proper-
ties. M indicates the presence of a solid body, such as the wall of the
vessel. H2O indicates the presence of water.
M-
H2 – → 2H
M
O2 – → 2O
M-
H2 + O2 – → H2 O2
M
H2 O2 – → 2OH
OH + H 2 → H 2 O + H
M
H + O2 – → HO 2
H2 O
H + O2 – → HO 2
H + O 2 → OH + O
O + 2H 2 → H 2 O + 2H
H + OH → H 2 O
Some of the reactions are exothermic. Some are endothermic. The several
species are present in vastly differing concentrations. Some can be
involved in catalytic effects at the walls of the containing vessel.
The combustion of hydrocarbons is even more complex. In addition to the

species noted above there are C/C, C/H, C/O and even C/N fragments in
all conceivable configurations and combinations. Most of the species are in
very low, transient concentrations.
A summary overall reaction can give a first approximation of flame tem-

perature and the explosion's potential, particularly when combustion
takes place in air. The heat capacity of the nitrogen in air opposes the
attainment of extremely high flame temperatures. Were the nitrogen not
present the calculation of flame temperature, taking into account the many
reactions which occur, is extremely tedious and error prone.
In most cases energy must be provided to produce significant quantities of

species formed by endothermic reactions. These species then initiate exo-
thermic reactions, releasing energy for the formation of more endothermic
species. The integration of all these reactions results in a high temperature
within the flame zone and for some distance behind it. This is illustrated in
Figure 2-1.
Combustion and Explosion Fundamentals 19
COMBUSTION ZONE
TEMPERATURE
BURNED GAS UNBURNED GAS
DISTANCE FROM POINT OF IGNITION
Figure 2-1 Temperature Profile in Flame
The time during which the initiating energy is supplied is critical. If the
energy is not supplied fast enough, or if there isn't enough energy sup-
plied, concentration of endothermically generated species may be too low,
and these will react exothermally too slowly to provide the required
energy feedback to cause a self-sustaining reaction. All this verbiage can
be summarized by referring to the classical Arrhenius equation:
dx – E ⁄ RT
------ = Ae ( b – x ) ( c – x )…
dt
where
x = concentration of product species
t = time
A = a constant dependent on the frequency of occurrence of the
species
E = activation energy
R = the gas constant
T = absolute temperature
b, c, … = initial concentration of reacting species
With known values of A and E, this equation can be used to calculate the
rate of the component reactions, or, if desired, the rate of the overall reac-
tion. In this latter case the constants are close to those of the slowest
reaction, the rate-controlling reaction.
Even though we will not use the Arrhenius function for quantitative pur-
poses, it is useful to look at the function from a qualitative standpoint.
Increasing temperature increases the speed of a reaction. Except for con-
centration of reactants, we have no control over the other factors in the
equation. Concentration is expressed in absolute terms, moles/liter, not in
percent. The effects of diluents and pressure, as well as initial tempera-
ture, on the rate equation can be assessed.
The autoignition temperature (AIT), also called the spontaneous ignition

temperature (SIT) is a value determined by a standard method such as that
in ASTM E659-79 or IEC Publication 79-4. In the IEC method, liquid or gas
is injected into a 200 ml Erlenmeyer flask that has been heated to a prede-
termined temperature. Great pains are taken in the design of the apparatus
to ensure that the entire flask is at the same temperature. When a sample is
injected, a timer is started. The timer is allowed to run until a flame is
observed or until 5 minutes has passed. The test is repeated with different
sample volumes and different flask temperatures until no ignition occurs.
Finally, the test is repeated five times at the lowest temperature at which
ignition was observed, rounded down to the nearest lower Celsius degree.
If no ignition is observed that value of temperature is reported as the igni-
tion temperature. The time lag before ignition is recorded, along with the
barometric pressure. The entire process takes place in a darkened room.
The IEC standard states that duplicate tests by the same operator are sus-
pect if they differ by more than 2%; differences between laboratories
should not exceed 5%. The ASTM method is similar and is stated to have
similar reproducibility. The method differs from the IEC method in that (1)
the flask is a 500-ml spherical-body laboratory flask, (2) the maximum
waiting time for ignition delay is 10 minutes, and (3) a new flask is to be
used when testing a new material. The IEC standard specifies a chemically
clean flask.
Because wall effects on the combustion process are reduced in larger vol-
ume apparatus and because the allowable ignition delay is 10 minutes
instead of 5 minutes, the ASTM method should lead to somewhat lower
values of AIT than the IEC method. However, for the purposes of safety
standards and practices, the differences are not important. The ASTM
method referenced here replaces an older standard method which, except
for a 250-ml flask volume, parallels the IEC method. Much of the AIT data
in the U.S. literature is based on the earlier method of measurement.
To electrical engineers and instrument engineers a reproducibility of 5%

among laboratories may seem to be a rough measurement indeed, but few
measurements of combustion parameters are as reproducible—another

reason that neat, clean correlation functions between parameters continue
to elude the hopeful.
Consideration of the Arrhenius equation helps us understand that the

ignition lag is observed to be longer at lower temperatures. The exother-
mic reaction rate is temperature-dependent. We can also see why some
materials will ignite spontaneously at temperatures below the accepted
AIT. All it takes is the proper combination of accelerative endothermic
reactions and poor heat transfer to the outside, as in the case of the oily
rag, which doesn't know that it has only 5 minutes to burst into flame.
Much more importantly, understanding the significance of the rate equa-

tion helps us realize that the autoignition temperature has no direct
quantitative correlation with the sensitivity of a compound to ignition by
electrical arcs, friction, static electricity, optical sources, or even by hot sur-
faces. The value of the autoignition temperature is also known to be
susceptible to significant variation if the flask is made of a different mate-
rial or if coupons of some metals are inserted into the flask, presumably
because some of the reactions are subject to catalytic wall effects.
Steady-state flame propagation, then, involves the generation of heat, and

its effective use to promote further reaction. In the combustion zone, on
the order of 0.1 mm thick, heat is generated by the chemical reactions.
Both the temperatures in this zone and the concentration of reactive spe-
cies are higher than would be predicted by steady-state theory, probably
because the equilibria are temporarily unstable. Ahead of the combustion
zone, heat is conducted and radiated into the unburned gases, raising its
temperature to a value that depends on the flame temperature, the dura-
tion of exposure, the distance from the flame front, and its heat capacity.
Exposure duration and distance regulate the propagation rate. The steady
state is not necessarily constant velocity unless the front is planar and of
unchanging geometry. Neither does the flame front always adjust its
velocity to maintain a continuous steady state from ignition by a source to
completion of the combustion process. In the initial stages, the flame may
expand too rapidly and outrun its ability to pass sufficient energy to the
unburned gases to ignite them.
Ignition By a Point Source

Assume that a point source of energy imparts an amount, We, Joules of
energy to a combustible mixture, raising the local temperature at the site
of energy injection until a small kernel of the combustible mixture ignites.
After ignition, the burning material adds energy to the kernel of gas. At
the same time thermal conduction and radiation transfer heat to the sur-
rounding unburned gas. The gas layer immediately surrounding the
initially ignited kernel in turn reaches ignition temperature; the ignition of

this new layer of gas and thermal expansion of the previously burned gas
cause the kernel to grow in size. In the ideal case, it will grow spherically.
The combustion wave can be considered to act very much like the skin of a
bubble. In this case, the bubble starts with almost zero initial volume and
grows spherically. The combustion wave progresses into the unburned
gas, leaving behind burned gas at a higher temperature.
Experimental evidence shows that there is a critical ignition energy, Wc,

that must be injected into any particular flammable mixture to cause the
incipient flame sphere to grow indefinitely. Experimental evidence also
shows that this amount of energy is related to a critical flame sphere diam-
eter, Dq. If the energy We supplied initially is less than the critical ignition
energy Wc the combustion wave will die out before it reaches the critical
flame diameter, Dq. If the energy We imparted to the mixture is equal to or
greater than the critical energy Wc, the combustion wave will continue to
grow. It will reach the critical flame sphere diameter, Dq, and the flame
will continue to propagate—in conventional terms, there is an explosion.
The interrelationship between the critical ignition energy and the critical
flame sphere diameter can be viewed in the following manner. This is an
oversimplification of the conclusions drawn by Lewis and von Elbe in
their detailed theoretical treatment of the ignition process.
In a steady-state plane combustion wave (or in a spherical wave whose

diameter is large compared to the width of the reaction zone) the amount
of energy per unit volume of wavefront, WA, added in the reaction zone
by combustion is just sufficient to raise the adjacent unburned gas to igni-
tion temperature and supply losses to the burned gases behind the
combustion zone. In a small-diameter spherical wave, however, the
energy in the reaction zone must be greater than it is in a plane wave or
the spherical wave will not propagate. This can be seen as follows (refer
also to Figure 2-2).
Assume that the energy density in the reaction zone of a kernel of diameter
D is WA, the critical energy density required for propagation of a plane
wave. When energy is transferred to an adjacent zone of diameter, D + ΔD,
the energy density will be less because the volume of the reaction zone is
larger. This energy density is less than the critical density, WA, so that the
flame dies out. If the incipient wave is to live through the highly divergent
early stages of growth, the energy density must be greater than that
required for plane-wave propagation from inception until the critical diam-
eter is reached. The excess energy must be supplied by the ignition source.
Figure 2-2 Divergence in Expanding Flame
Another view of this same phenomenon, which the author finds aids
understanding when coupled with the previous one, has been stated by
Litchfield. His view is that, if the expanding flame sphere is considered to
be an expanding bubble, the ignition energy must be of the same order as
the mechanical energy, pV, required to cause the expanding sphere to
reach the critical diameter, Dq. p is the pressure within the bubble, nearly
atmospheric, and V is the volume of the bubble. In support of this thesis
he presents the data which have been summarized in Table 2-1.
Components of Quenching Measured Ignition

pV (mJ)
Flammable Mixture Distance (mm) Energy (mJ)
Acetylene-oxygen 0.18 0.0003 0.0002–0.0004
Acetylene-air 0.64 0.014 0.017–0.018
Hydrogen-oxygen 0.25 0.0008 0.0012–0.0014
Hydrogen-air 0.64 0.014 0.017–0.018
Ethylene-oxygen 0.23 0.0006 0.0009–0.001
Ethylene-air 1.22 0.095 0.07–0.08
Methane-oxygen 0.30 0.0014 0.0027
Methane-air 2.03 0.44 0.3
Nitric oxide-hydrogen 6.35 13.47 8.7
Methane-nitric oxide 6.35 13.47 8.7
Table 2-1 Comparison of Mechanical and Ignition Energies
Lewis, in his contribution to ISA Monograph 111, noted that Litchfield’s

data include several cases in which the measured ignition energy is less
than the mechanical energy calculated at the critical diameter. He stated
that this is consistent with an equation he presented that predicts that mix-
tures with a high flame temperature reaction could have ignition energies
lower than predicted from consideration of mechanical energy alone. The
equation he presented is as follows:
( pV – n b RT f )C p
W e = ---------------------------------------
-
R
where
Cp = mean molar heat capacity of burned and unburned reac-
tants (an assumption that they are the same)
R = gas constant
pV = mechanical energy to reach critical bubble diameter
nb = moles of burned gas in the critical bubble
Tf = flame temperature
Lewis did not present the derivation of this equation. The author believes
that the equation can be justified by the following argument:
pV
------- = nT
R
where
n = the number of moles of burned and unburned gas in the
critical sphere volume
T = the average temperature in the sphere
Because combustion has taken place during the time it takes the sphere to
grow to the critical diameter, a quantity of heat, nbCp(Tf − T), should be
subtracted from the mechanical energy pV. This amount of heat is the heat
contributed by the nb moles of burned gas cooling from the flame temper-
ature, Tf, to the average temperature of the gas in the critical sphere, T.
Because Tf is much higher than T, the term nbCpT can be neglected.
This equation's significance is more apparent if the parentheses are

removed:
C p pV
W e = -------------
- – nb Cp Tf
R
= nC p T – n b C p T
The first term is the total heat used to raise a volume, V, of burned and
unburned gas to the average temperature within the critical sphere. The
second term is the heat contributed by the combustion process. Lewis
states that generally the ratio CP/R is about 3 to 4 so that the ignition
energy would be greater than the mechanical energy, pV. However, if the
flame temperature is high, as it is in reactions involving nitric oxide or
undiluted oxygen, the second term may be of significant size, and the igni-
tion energy may be lower than the mechanical energy, pV. Lewis noted
that the equation he presented is not of value for computation; it is useful
only for helping understanding of the ignition phenomenon, because the
value to be assigned to nb is difficult to assess.
If one examines Litchfield's data with Lewis' comments in mind the pat-
tern of calculated ignition energies and measured ignition energies is
anything but clear. If there is a lesson, it is that the process is incredibly
complex and simple answers to combustion questions seem ever to be
elusive.
The critical diameter, Dq, can be inferred from several kinds of measure-
ments. The diameter of the smallest tube through which flame will
propagate (or the largest through which it will not propagate) provides an
estimate of Dq. The tube length is many times the diameter. Similar esti-
mates of Dq can be determined by measuring the distance between flanges
or plates that form a narrow channel through which flame tries to propa-
gate. A third method, described in Chapter 9, measures the distance
between the flanges on electrodes between which a spark passes. The dis-
tance at which there is a large increase in the amount of spark energy
required to cause ignition is taken to be the critical diameter.
The critical diameter is frequently called by the name of the experimental

measurement. It is sometimes called quenching diameter and, more fre-
quently, quenching distance. In all three measurements a surface—either
the tube, the plates, or the flanges—extracts heat from the combustion
wave. This quenches it before it can reach self-sustaining dimension.
Figure 2-3 is a plot of data taken from Lewis and von Elbe, and similar
data from Calcote. It demonstrates the definite correlation in a particular
experimental arrangement between ignition energy and quenching
diameter.
Figure 2-3 Ignition Energy versus Quenching Distance (in Calcote data taken as
electrode spacing at which ignition energy was twice minimum)
The reader may consider that Figure 2-3 does not provide good support
for the previous qualitative treatment of the relationship between ignition
energy and quenching diameter because the three sets of data seem to
cluster about three different lines. The reason for this is that the datapoints
reported by Calcote are derived from different experiments than those
performed by Lewis and von Elbe. The latter data represent a more certain
determination of quenching distance, as explained in Chapter 9. Calcote
reported the dimension at which the ignition energy doubled. This dimen-
sion is related to quenching diameter, but it is not equal to it. Calcote used
1---
8
-inch-diameter spherical electrodes. For materials with quenching diam-
eter much smaller than 1--8- -inch, the amount of protrusion of the electrodes
into the developing flame sphere to increase the ignition energy to twice
the minimum value is a small fraction of the actual quenching diameter.
For materials with quenching diameter larger than the electrode diameter,
the amount of protrusion into the sphere must be a greater fraction of the
quenching diameter. Thus, for materials with low ignition energy, Cal-
cote's estimate of quenching approximates that of Lewis and von Elbe. For
materials of high ignition energy Calcote's measurement underestimates
the quenching diameter. The essential point demonstrated by Figure 2-3 is
that the energy required for ignition is certainly related to a critical flame
diameter. Were pV the correct estimate of the critical ignition energy, the
3
ignition energy should be proportional to D q .
Lewis and von Elbe discuss the relationship between ignition energy and
quenching diameter at some length in the second edition of their book.
They discuss two estimates of ignition energy. The first is of the same form
as that presented by Fenn, but without the arbitrary constant. This equa-
tion gives the sensible heat in a flame sphere of the critical diameter,
assuming that the sphere contains only burned gas. This quantity is
always more than the ignition energy, but the two values are not related
by a constant. Typically, the ratios are in the range 4 to 10 for mixtures of
fuel and air. For the few mixtures with oxygen for which values are given
the ratio is between 1 and 2. They also discuss another equation that is a
2
function of D q to elucidate what is happening in the developing wave, but
reach no conclusions that are of computational value to us. In summary,
they conclude that the ignition source contributes the excess energy neces-
sary to allow the wave to develop.
Although the author and others have often spoken of the relationship
between quenching distance and ignition energy as though it were an
immutable law, it must be recognized that in the experiments of Lewis,
von Elbe, Calcote, et al. there is an implicit condition on the relationship.
The energy is discharged into the cloud almost instantaneously. A little
reflection shows that if this is not so then the rate of injection of energy
into the mixture must also have a strong influence on the total energy
required to initiate an explosion. Most investigations of this phenomenon
have been carried out with low voltage arc ignition. Some of the results
are discussed in Chapter 9.
Ignition Energy and Flame Velocity

Development of a combustion wave depends on thermal energy flowing
from a zone of burning gas to an adjacent zone of unburned gas. One
would expect that altering conditions so that the burning gas produces
more heat per unit volume, or so the heat transfer to the adjacent gas layer
is faster, would speed up the flame development. Not only does the flame
develop faster, the initial energy requirement is reduced. On the other
hand there can be no combustion at all if the gas is 0 or 100% air, and the
ignition energy required will be infinite. Flame velocity will be zero. Fig-
ure 2-4 illustrates that there is indeed an inverse relationship between
ignition energy and the velocity of flame propagation. The correlation is
surprisingly good, despite the fact that ignition energy data and flame
velocity data are from different sources. This relationship provides a use-
ful qualitative means for considering the effect of changing the
combustible mixture. However, the correlation is not sufficiently good to
estimate the magnitude of ignition energy from the value of the flame
velocity. The velocity of flame in a tube or in any other experimental
geometry is highly dependent on the experimental apparatus and tech-
nique as well as on the properties of the combustible mixture.
One of the tantalizing aspects of combustion and explosion studies is that

almost all measures of ease of ignition correlate approximately with all oth-
ers. However, the second-order differences and experimental influences
forestall attempts to state simple theories, although a sense of universal
order insists that such ought to exist. In Chapter 3, in the discussion of
material classification, and in Chapter 5, at the beginning of the discussion
of explosionproof enclosures, additional examples of the close relationship
among some combustion properties of gases and vapors are presented.
Figure 2-4 Relationship Between Ignition Energy and Flame Velocity in 2.5-cm Tube
(ignition energy of propane, methane, ethane, and hydrogen from Lewis and
von Elbe, flame velocity from International Critical Tables)
Lower and Upper Explosive Limits

When one determines the critical energy required to ignite various con-
centrations of a gas- or vapor-air mixture, curves such as those shown in
Figure 2-5 are obtained. At some concentration the amount of energy
required for ignition is minimal. At any other concentration, the energy is
greater, increasing as the concentration approaches the lower explosive
limit (LEL) or the upper explosive limit (UEL). At these concentrations the
energy required for ignition is so high that it is not practically possible to
obtain a self-sustaining flame. If sufficient energy is injected into a mixture
below the LEL or above the UEL some material will burn, but a combus-
tion wave will not propagate from the energy source and continue to
spread. The LEL and UEL are also referred to as the lower and upper flam-
mability limits, LFL and UFL. This book uses the LEL and UEL notation;
however, in many fields, LFL and UFL are gaining preferred status. Near
the limits of flammability the mixture is not “explosive.” Pressure rise is
nearly zero, and flame speed is very low. (In the limit pressure rise is zero
and flame speed is zero at the LFL and UFL.)
Figure 2-5 Effect of Concentration on Ignition Energy (data from Lewis and von Elbe,
Table 10)
Although in practice the limits of flammability are often treated as inher-

ent properties of a combustible material, the values of the LEL and UEL
depend on (1) the amount of energy available to test the mixture, (2) the
size and geometry of the test chamber, and (3) the initial temperature and
pressure of the mixture.
When testing near the LEL or UEL it may be necessary to use an ignition
source capable of delivering large amounts of energy. If the source has
insufficient energy, the determination will only be a limit of ignitability for
that source.
Use of laboratory-determined flammability limits of the more easily

ignited materials, that is, those we usually consider to be flammable, in
practical situations offers little cause for concern. A laboratory test of a
mixture with high ignition energy and a low vapor pressure at ambient
temperature may, however, significantly underestimate the hazard of the

material handled at a somewhat higher temperature in a much larger
chamber. Whether dissimilarity of conditions will invalidate the data at
hand should always be a consideration with any combustion data. ASTM.
E689, for example, recommends that, if a material is to be handled in a
large chamber near the AIT determined by the standard method, several
determinations of AIT should be made in test chambers of different vol-
umes and the results should be extrapolated to assess the possible hazard.
The effect of increasing temperature is to broaden the flammability limits.

The combustion zone doesn't have to contribute as much energy to main-
tain flamespread. For the paraffin hydrocarbons, for example, the LEL
drops by about 8% — and the UEL increases by about 8% — when the
temperature is raised from ambient to 100°C.
The effect of increasing pressure is also to broaden the flammability limits.

Increasing pressure increases the amount of energy released by a unit vol-
ume of gas. The UEL is most sensitive to pressure increase, whereas the
LEL is much less affected. For propane-air, the upper limit rises from 9.5%
to 26% when the pressure rises from 1 atmosphere to 100 psig, but it
increases only to 32% for further increase in pressure to 200 psig. (For a
much more detailed discussion of explosion limits and related flammabil-
ity characteristics of gases and vapors consult the report by Zabetakis.)
For the purposes of instrument safety it is usually sufficient to define the

LEL for a specific flammable gas or vapor, and the lowest ignition energy
for a typical ignition mechanism. If the concentration can be maintained
with certainty, below the LEL or above the UEL, no hazard exists. If the
energy can be kept below the lowest ignition energy, no hazard exists for
any mechanism of ignition comparable to that with which the ignition
data were obtained.
Most Easily Ignited Concentration

The concentration at which ignition energy is minimum is the Most Easily
Ignited Concentration (MEIC). This concentration is seldom the stoichio-
metric concentration, that is, the concentration at which all material reacts
so that there is neither excess fuel nor excess air (or oxygen). Whether the
deviation from stoichiometry is to the lean or rich side depends on the
chemical structure of the gas or vapor. Most hydrocarbons are most easily
ignited on the rich side. The most easily ignited concentrations of hydro-
gen, methane, and acetylene are on the lean side. The value of the most
easily ignited mixture, like the flammability limits, is somewhat affected
by the nature of the experimental procedure, but for practical purposes it
may be considered a constant.
Minimum Ignition Energy

Earlier discussion used the concept of a critical ignition energy. The varia-
tion of critical ignition energy with concentration has been noted. The
smallest amount of energy required to ignite the most easily ignited atmo-
spheric concentration of the gas or vapor is called the MIE (minimum
ignition energy), Wm. Minimum ignition energy Wm must be measured
under carefully controlled conditions. Tests must be conducted with the
most easily ignited mixture. The means for supplying energy to the mix-
ture must not conduct heat away from the incipient flame sphere. This
would increase the required initial energy input. Much of the ignition
energy data in the literature is not minimum ignition energy data, because
it is the ignition energy required for stoichiometric mixtures, not the most
easily ignited mixture. Other data in the literature were obtained with
electrode configurations that extracted heat from the developing flame
sphere, thereby yielding ignition energy values higher than minimum
ignition energy Wm.
The most reliable determination of MIE of a gas or vapor, the method now
used in many laboratories, uses high-voltage capacitive discharges. This
method approaches the ideal. By using voltages above 10,000 volts, elec-
trode separation can be greater than the critical flame ball diameter, Dq,
which is typically less than 3 mm. Careful mixing controls concentration.
Equipment is carefully constructed to minimize unmeasurable energy
losses from corona or unmeasured stored energy contributions from stray
capacitance. To determine MIE, a capacitor of known size is charged to a
known voltage and discharged through an arc into the mixture. Either the
size of the capacitor, C, or the charging voltage, V, is varied to change the
amount of stored energy in the capacitor. Since the capacitor discharges
only through the arc, in fractions of a microsecond, the energy release to
the gas is assumed to be instantaneous. All the energy from the capacitor
is presumed to be dissipated in the arc. Most workers have assumed that
all the energy dissipated in this high voltage arc is transferred to the gas.
Stored energy is calculated from CV2/2 and is taken to be the MIE for the
gas or vapor. A more detailed description of one experimental apparatus
used by Lewis et al. is given in Chapter 9.
Effect of Test Conditions

Temperature of the Gas or Vapor
In the limit, if the temperature of the gas or vapor mixture is raised to its
autoignition temperature, no electrical energy is required to initiate a com-
bustion wave. One intuitively concludes, therefore, that increasing gas
temperature decreases the required electrical ignition energy. Fenn pre-
sented the data in Table 2-2. This author inserted the numbers in
parentheses. They are autoignition temperatures from NMAB 353-5, and
are listed for reference only. They cannot be compared directly with the
tabulated temperature data because the indicated autoignition tempera-
ture may be for a different concentration than that used for electrical
ignition tests. For many materials, the lowest autoignition temperature is
for rich mixtures (concentration above stoichiometric). Fenn's data are for
stoichiometric or lean mixtures. The fact that Fenn had to add electrical
energy to ignite a 100°C mixture of carbon disulfide does not mean that
the experiment is seriously flawed.
The ratio of energy required to ignite a mixture at T degrees K to that

required at 298°K (25°C) can be approximated from the equation:
WT T –4
- = § ---------·
-----------
W 298 © 298¹
At the extremes of the temperature range of Fenn's study this equation

predicts his data only within a factor of 2. The reader should remember
that Fenn's work is based on the apparatus used by Calcote, The ignition
energies listed are not minimum ignition energies because they are not
measured at the most easily ignited concentration. These ignition energies
were determined at sufficiently large electrode spacings that electrode
quenching is not a significant factor. Therefore, the equation given above
is not valid for most practical situations of interest to instrument people,
where a large fraction of the “ignition energy” is lost to the ignition pro-
cess, as discussed in Chapter 9.
Fuel (AIT-C) Temperature (°C) Ignition Energy (mJ)

Carbon disulfide (90) 25 0.076
100 0.05
n-Heptane (204) 25 1.45
100 0.67
171 0.32
iso-Octane (415) 25 2.7
100 1.1
171 0.48
n-Pentane (260) -30 4.5
-20 1.45
25 0.78
100 0.42
171 0.23
175 0.25
Propane (432) -40 1.17
-30 0.97
-20 0.84
25 0.84
57 0.42
82 0.36
100 0.35
204 0.14
Propylene oxide (449) 25 0.24
100 0.15
182 0.09
Table 2-2 Effect of Temperature on Ignition Energy
The data in Table 2-3 are adapted from the paper by Bartels and Howes.
These data, obtained in a 24-V, 1-mH inductive circuit, using the standard
IEC ignition testing apparatus described in Chapter 9, also demonstrate
the reduction in igniting current due to increase in temperature.
Mixture AIT (°C) Igniting Current (mA)

20°C 200°C
22% Hydrogen-air 585 280 200
7.8% Ethylene-air 450 600 500
5.25% Propane-air 432 850 800
Table 2-3 Effect of Temperature on Igniting Current
The ignition data for each fuel clearly shows that addition of thermal
energy to the mixture by raising its temperature reduces the amount of
electrical energy required for ignition. But the reduction in energy is very
low compared to what one would predict from the Fenn data. The reason
is (as explained in Chapter 9) that the electrodes are within the developing
flame sphere and conduct a large percentage of the spark energy from the
ignition zone. The temperature of the mixture does not significantly affect
these energy losses because they are conduction losses proportional to the
difference between the electrode temperature and the temperatures of the
flame and burned gas, a linear function.
Increasing the mixture temperature also slightly broadens the observed

limits of flammability when the limits are determined using electrical igni-
tion sources. Higher mixture temperature reduces the amount of heat that
must be transferred from the reaction zone to the unburned gas to main-
tain the flame. Fenn states that increased initial temperature also increases
the flame temperature. For a fixed amount of energy a plot of energy ver-
sus concentration will, therefore, give an appearance of wider
flammability limits. The source energy, aided by the energy due to tem-
perature, will ignite higher and lower concentration mixtures.
Pressure of Gas or Vapor
The number of molecules of gas or vapor per unit volume is approxi-

mately proportional to pressure. Decreasing pressure, therefore, decreases
the energy per unit volume produced by combustion and slows down
heat transfer to the unburned gas. Conversely, increasing pressure
increases both the energy per unit volume and the rate of heat transfer.
The limits of flammability are also somewhat affected by pressure, being

broadened by a pressure rise. In Figure 2-6, the effect of pressure on
inflammability limits is not large enough to be clearly demonstrated. The
ignition energy, however, is approximately proportional to 1/p2. In his
article, Thomas illustrates the same inverse square dependency for the
ignition energy of methane. The effect of increasing pressure also seems to
follow the inverse square law, but there are indications that the effect lev-
els off after the pressure reaches several atmospheres. Fenn's equation for
ignition energy shows the ignition energy to be inversely proportional to
pressure squared, and his plot of the effect of pressure extends from about
1/20th atmosphere to 5 atmospheres. Calcote also investigated pressure
effects on ignition energy. The inverse square relationship holds well for
vacuum and pressures to 2 atmospheres, but the slope of ignition energy
versus pressure seems to decrease above 2 atmospheres. His data covered
a range of 1/4 to 3 atmospheres.
Figure 2-6 Effect of Pressure on Ignition Energy (data on hydrogen from Table 10, from
Lewis and von Elbe, first edition)
Green and Agnew presented data on the quenching distance of propane-

air mixtures determined by inserting parallel plates in the flame path in a
constant volume bomb. It is not clear whether ignition took place between
the plates or near the slot formed by the plates. This investigation covered
2 to 100 atmospheres. For comparison they included data from Friedman
that were obtained in a constant volume bomb over a pressure range to 2
atmospheres. These latter data indicate an inverse relationship between
quenching distance and pressure, which one would expect if the ignition
energy is assumed to be proportional to the square of quenching distance.
If W = kD2, then
2 2
W1 p D
-------- = ----0- = ------1-
W0 p1
2
D0
2
and
D1 p
------- = ----0-
D0 p1
Above 2 atmospheres the decrease in quenching distance with pressure

increase had a slope lower than one. These data must be taken as indica-
tive only. The values of quenching distance reported are consistent with
those of Lewis et al. for a 5% propane-air mixture, but the values for richer
mixtures are smaller. One would conclude from the Lewis et al. data that
they should be larger. The 5% mixture gave the smallest quenching dis-
tance, not only in electrical ignition experiments, but also in experiments
with flat plates.
Figure 2-7, adapted from Bartels and Howes, illustrates the variation in
igniting current with pressure in a 24-V, 1-mH circuit. If ignition were
occurring under ideal conditions so that the minimum ignition energy is
inversely proportional to the square of absolute pressure, the minimum
igniting current would be inversely proportional to pressure, if it can be
assumed that W = LI2/2. L and I are values of the circuit inductance and
current, respectively. (This assumption is not completely valid because in
the standard IEC apparatus used the ignition energy is proportional to the
square of current only down to about 2 mH.) Although the inverse rela-
tionship between pressure and igniting current is approximately true for
the hydrogen-air data, the data for propane and ethylene mixtures do not
decrease as fast as the inverse law would predict. The ignition energy level
in this apparatus at atmospheric pressure is about twice the minimum
ignition energy, and there is no reason to believe that the energy losses to
electrodes should be inversely proportional to pressure. Deviation of the
results from the ideal inverse relationship is, therefore, not surprising.
Figure 2-7 Effect of Pressure on Igniting Current
Effect of Changing Inert Gases

Most data presented in this chapter have been for atmospheric mixtures in
which 79% of the air consists of nitrogen and other inerts that do not take
part in the combustion reactions. Occasionally it is necessary to predict
what would occur if some gas other than nitrogen were used. Certainly, if
all or part of the nitrogen in a fuel-rich atmospheric mixture is replaced by
additional oxygen, one would expect that a given volume of gas would
produce more thermal energy, thereby decreasing the amount of electrical
energy required for ignition.
As noted by H. Lewis, decreasing the amount of nitrogen also results in

higher flame temperatures because less of the mixture is braking the reac-
tion. If one adjusts the fuel percentage for methane, ethane, and propane
to obtain the most easily ignited mixture and changes the oxygen percent-
age in the oxygen-nitrogen remainder from 0.21 for air to 1.0 for pure
oxygen, the minimum ignition energy decreases somewhat faster than the
cube of 0.21/%O2. (Estimated by the author from the curves in Lewis and
von Elbe, first edition). For ethylene, hydrogen, and acetylene the decrease
in minimum ignition energy seems to be somewhat less steep. If the fuel
percentage in the mixture is held constant, and the oxygen ratio is

increased the ignition energy decreases at a much slower rate, perhaps as
the square of 0.21/%O2. These approximations were developed by the
author to give some idea of the effect of changing oxygen content of a mix-
ture. They are order of magnitude estimates only.
If the atmospheric nitrogen is replaced by another inert gas, however, the

effects are more difficult to predict. Figure 2-8 shows the expected result—
that changing the inert gas from nitrogen to CO2, helium, or argon does in
fact affect the amount of energy required for ignition. Lewis and von Elbe
state that the difference between the argon and helium curves results from
the difference in heat conductivity of the mixtures, but that the effect of
replacing the nitrogen with CO2, though similar to that of using helium, is
caused by the large heat capacity of CO2.
Figure 2-8 Effect of Inert Gas on Ignition Energy of Hydrogen (from Lewis and von
Elbe, first edition)
Electrode Geometry
In considering electrical ignition thus far, we have assumed an energy
source that injected a critical amount of energy instantaneously into a
small volume of a combustible mixture. We have also assumed that the
flame sphere develops unimpeded by electrodes. In the work of Calcote,
Lewis, et al., the ignition source was an arc from a capacitor charged to a
high voltage. Calcote and his associates used --18- -inch spherical electrodes.
Lewis and colleagues used -----1-
16
-inch electrodes with glass flanges. Calcote's
comparison of the ignition energy measurements of several materials by

both teams of investigators shows some differences, but they appear to be
the differences one expects in combustion experiments, which seldom, if
ever, yield precisely reproducible results. The differences are random in
sign, and can not be interpreted as being caused by the difference in elec-
trode geometry. In both investigations the electrodes were outside of the
incipient flame sphere.
In most practical situations the energy is not delivered to the combustible

mixture in this ideal fashion. Rather, the arc that delivers energy to the
combustible gas passes between electrodes that are spaced at a distance
less than the critical quenching distance Dq. Electrodes within this critical
diameter will conduct energy away from the combustion zone, thus
increasing the amount of loss. This deficit must then be made up by the
release of additional energy into the arc. Chapter 9 shows in greater detail
that any such intrusion of the electrodes into the volume defined by the
critical sphere diameter Dq of an unimpeded combustion wave results in a
significant increase in the amount of energy required to cause ignition.
Time Scale of the Ignition Process

Ignition by arcs is ignition by tiny high-temperature sources. Whether a
discharge causes an explosion is determined in a fraction of a millisecond
after the arc is struck. Though the flame may propagate for a somewhat
longer time before dying out, Linton and Wooding showed that after a
few hundred microseconds the rate of growth of the incipient flame
sphere is observably different, depending on whether the flame eventu-
ally dies or becomes self-propagating.
High-voltage capacitance discharge injects energy into the combustible

mixture in less than a microsecond. Slowing the rate of injection so that
discharge lasts for tens of microseconds raises the total energy needed for
ignition somewhat. The discussion in Chapter 9 shows that if the rate of
energy is slowed still further, so that the total time of release is comparable
to the 200-microsecond critical period noted above, the total energy
required for ignition increases substantially.
References
A Review of Electrical Research and Testing with Regard to Flame-Proof Enclo-
sure and Intrinsic Safety of Electrical Apparatus and Circuits, Ministry of Fuel
and Power, London, 1943.
Fenn, J. B., “Lean Flammability Limit and Minimum Spark Ignition

Energy,” Industrial and Engineering Chemistry, 43, 11 (December 1951),
2865–2869.
Calcote, H. F., C. A. Gregory Jr., C. M. Barnett, and R. B. Gilmer, “Spark

Ignition Effect of Molecular Structure,” Industrial and Engineering Chemis-
try, 44 (November 1952), 2656.
Maltby, F. L., “Safety and Electronic Control Systems,” ISA Journal (Sep-
tember 1956), 318–322.
Khitrin, L. N., Physics of Combustion and Explosion, translated from the Rus-
sian, National Science Foundation, Washington, D. C., 1957.
Wolf, I. W., and V. T. Burkett, “A Method for Determining Minimum Igni-

tion Energies: Results for a Neo-pentane-Air Mixture,” Combustion and
Flame, 1 (1957), 330–338.
Linton, D. R., and E. R. Wooding, “Investigation of the Ignition of a Gas by

an Electric Spark,” British Journal of Applied Physics, 10 (April 1959), 159–
166.
Litchfield, E. L., Minimum Ignition-Energy Concept and its Application to

Safety Engineering, RI 5671, U. S. Bureau of Mines, Pittsburgh, 1960.
Lewis, B., and G. von Elbe, Combustion, Flames and Explosions of Gases, 2nd
ed., Academic, New York, 1961 (Chapter V).
Gordon, R. L., L. C. W. West, and D. W. Widginton, “The Ignition of Meth-

ane-Air Mixtures by Arc Discharge of Controlled Duration,” IEE
Conference Report Series No.3, Institution of Electrical Engineers, London,
1962, 15–20.
Thomas, V. M., “Design of Intrinsically Safe Apparatus for Use in Coal

Mines: A Review of Data and Techniques,” The Mining Electrical and
Mechanical Engineer (May 1964), 295–329.
Lewis, H., “Initiation of Explosions,” ISA Monograph No. 111, Instrument

Society of America, Pittsburgh, 1967, 44–51.
Zabetakis, M. G., Flammability Characteristics of Combustible Gases and

Vapors, Bulletin 627, U.S. Bureau of Mines, Pittsburgh, 1967.
Bartels, A. L., and J. A. Howes, “The Variation of Minimum Igniting Cur-

rents with Pressure, Temperature and Oxygen Enrichment,” Conference
on Electrical Safety in Hazardous Environments, Institution of Electrical
Engineers, London, 1971.
Bone, W. A., and D. T. A. Townend, “Explosions and Gaseous Explo-

sives,” International Critical Tables.
ASTM E659-1979, Standard Test Method for Autoignition Temperature of Liq-

uid Chemicals, ASTM, Philadelphia.
Green, K. A., and J. T. Agnew, “Quenching Distances of Propane-Air

Flames in a Constant Volume Bomb,” Combustion and Flame, 15 (1970) 189–
171.
NMAB 353-5, Classification of Gases, Liquids and Volatile Solids Relative to

Explosionproof Electrical Equipment, National Academy Press, Washington,
1982.
Calcote, A. F., et al., “Minimum Spark Ignition Energy, Correlation with

Ramjet and Turbojet Burner Performance,” Final Report No. 1 to Dept. of
Navy, Contract NOA (S) 10115, March 1959.
Kuchta, J. M., Investigation of Fire and Explosion Accidents in the Chemical,

Mining, and Fuel-Related Industries—A Manual, Bulletin 680, U.S. Bureau of
Mines, Pittsburgh, 1985.
CHAPTER 3
CLASSIFICATION OF
HAZARDOUS LOCATIONS
AND COMBUSTIBLE MATERIALS
Why Area Classification?

Every area classification method includes a statement of the following:
• a succinct description of the hazardous material that may be
present, and
• the probability that it is present, in ignitable concentration, so
that appropriate equipment may be selected, and safe, economi-
cal wiring and installation practices may be used.
There are many techniques for reducing the explosion of hazardous elec-
trical installations. There are important differences in the combustion
properties of flammable or combustible materials used in industry. If one
defines the relative explosion hazard of the material that is likely to be
present and the probability that it is present, then one can make a wiser
decision about the kind and degree of protection that must be provided to
ensure a safe installation.
The North American System

In the United States, area classification principles are stated in Articles 500
and 505 of the NEC, ANSI/NFPA 70. Articles 500-505 of the NEC consti-
tute the fundamental reference in the United States for installation of
electrical equipment in hazardous locations. Articles 500 and 505 broadly
define the classification of hazardous areas in accordance with the nature
of the hazard and the degree of hazard. Other articles in NEC Chapter 5
specify the kind of equipment and installation practices to be used in haz-
ardous locations of a particular classification. NEC Article 505, first
introduced in 1996, defines an alternative classification scheme that recog-
nizes the IEC Zone classification and material grouping for flammable
gases and vapors used in most industrial nations. This classification sys-
tem is discussed later in this chapter. Similar, but not identical,
classification rules are given in the Canadian Electrical Code, Part 1.
44 Classification of Hazardous Locations and Combustible Materials
Because the NEC enjoys widespread acceptance and use, the provisions of
Articles 500-503 have been used as the basis for American Petroleum Insti-
tute (API), NFPA, and ISA Recommended Practices and Standards. Article
504, introduced into the Code in 1990, incorporated much of the philoso-
phy of ISA RP12.6, “Installation of Intrinsically Safe Instrument Systems in
Class I Hazardous Locations.” Similarly, many other parts of NEC Chap-
ter 5 are derived from more detailed NFPA standards.
The predominant classification scheme in the United States and Canada is

that presented in Article 500. It defines the nature and degree of the haz-
ard existing in a particular location by specifying that the location is Class
I, II, or III; Group A, B, C, D, E, F, or G; and either Division 1 or 2.
Class defines the generic kind of hazardous material that may be present;
Group comprises materials with similar explosion hazard properties; and
Division is an indication of the probability that an explosive concentration
of material may be present.
Classes and Groups of Materials

In North America the three classes of materials are defined as follows:
Class I: Flammable gases and vapors
Class II: Combustible dusts
Class III: Flyings
Flyings are materials that are not normally suspended in air; that is, they
are larger than dust particles. Flyings include materials, such as cotton
linters, sawdust, textile fibers, and other large particles, that usually are
more a fire hazard than an explosion hazard.
Materials with similar explosion-hazard properties are identified by stat-

ing their group classification.
For Class I gases and vapors there are four groups:

(1) Group A: acetylene
(2) Group B: hydrogen and materials of similar hazard properties
(3) Group C: diethyl ether, ethylene, some aldehydes, etc.
(4) Group D: methane and other alkanes, most common industrial

solvents, most alcohols, gasoline, jet fuel, etc.
Classification of Hazardous Locations and Combustible Materials 45
For Class II hazardous dusts, three groups are defined:

(1) Group E: metallic dusts
(2) Group F: carbonaceous dusts, including carbon black, charcoal

and coal (The 1996 NEC eliminated references to volatile content
above 8% and sensitized materials, but these will almost certainly
be restored in the 1999 NEC.)
(3) Group G: agricultural dusts and plastics dusts
For Class III flyings, no groups are defined.
NFPA 497 and NFPA 499 contain the complete list of materials that have
been grouped by NFPA.
How Groups are Determined
During the early years of area classification in the United States groupings
of gases and vapors were based on tests of the explosionproof properties
of commercial enclosures. Group D materials, such as methane, propane,
ethane, the alcohols, and many common industrial solvents, were found to
require about the same degree of tightness of enclosure joints to prevent
transmission of an explosion. They were, therefore, grouped together.
Similarly, Group C materials such as ethylene, most ethers, and some of
the aldehydes were found to require tighter construction, and sometimes
stronger construction, than that required for Group D materials. Group B
materials such as hydrogen sometimes were found to produce very high
pressure rises and to always require very tight joints to prevent transmis-
sion of an explosion outside the enclosure. Group A, which includes only
acetylene, was established because of the propensity of acetylene to form
copper and silver acetylides, which are easily ignited by friction. Acety-
lene is, therefore, presumed to pose a much greater hazard of explosion
than hydrogen, although other explosion properties of the two materials
are similar.
At present, more definitive methods are used to classify Class I gases and
vapors. Classification may be based on (1) ease of ignition by electric
sparks, (2) the difficulty of containing an explosion in an explosionproof
enclosure, as measured in a standardized apparatus, or (3) on similarity of
the chemical structure to that of a material whose classification is known.
A philosophy for grouping of dusts into Groups E, F, and G was not specif-
ically stated until the 1981 edition of the NEC. However, in the 1960s some
ISA SP12 Committee members noticed that the electrical resistivity of the
dust was, implicitly, a basis for classification, even though there is no nec-
essary correlation between resistivity and flammability properties. For
some years the use of resistivity to classify dusts was controversial, and it
disappeared as a criterion for classification in the 1993 edition of the NEC.
Divisions
A measure of the probability that an explosive mixture is present in a haz-

ardous location is given by stating the Division.
Class I Locations
A Class I, Division 1 location is where there is a high probability (by

underwriting standards) that an explosive concentration of gas or vapor is
present during normal operation of the plant.
A Class I, Division 2 location is where there is a very low probability that

the flammable material is present in explosive concentration during nor-
mal operation of the plant; so an explosive concentration is expected only
in the case of a failure of the plant containment system.
This description of the divisions defined by the NEC is short when com-
pared to the language of the NEC. The intent of this succinct summary is
to emphasize the core concepts of the Code's definitions.
The assignment of a Division 1 or Division 2 classification to a location is

based on the probability that the mixture is within the explosive range.
Were it possible to guarantee a uniformly distributed mixture at a concen-
tration of 50% of the Lower Explosive Limit (LEL), there would be no
hazard because the mixture would not sustain an explosion. Smell is no
criterion. Most flammable materials, with the notable exceptions of meth-
ane, hydrogen and a few others, have strong odors and may be irritating
or toxic at concentrations far below that which will support an explosion.
The NEC does not quantify the definitions of Division 1 and Division 2.
Many experts agree that the dividing line between Division 1 and Division
2 is the presence of an explosive concentration about 100 hours/year. Sim-
ilarly, the dividing line between a Division 2 location and one that is not
classified is about 10 hours/year. It is also assumed, but not always stated
in American references, that in a Division 2 location a release is not only
infrequent, but also of short duration.
Class II Locations
A Class II, Division 1 location is where:

• there is a combustible cloud of dust in normal operation of the
plant; or
• electrically conductive (Group E) dust is present in hazardous
quantities; or
• an equipment failure simultaneously produces an ignitable cloud
and an electrical ignition source.
A Class II, Division 2 location is where:

• a cloud exists only if there is a failure of the processing equip-
ment, or
• a dust layer exists that may interfere with safe dissipation of heat
from electrical equipment, or that may be ignited by abnormal
operation or failure of electrical equipment.
In language more like that of the Code, a Division 1 location is one where:
• hazardous concentrations exist continuously, intermittently, or
periodically under normal operating conditions; or
• hazardous concentrations exist frequently because of repair or
maintenance operations or leakage of equipment; or
• breakdown of equipment or process failure might simulta-
neously release hazardous concentrations of flammable gas,
vapors or dust and cause failure of electrical equipment.
A Division 2 location is one that is presumed to be hazardous only in

abnormal situations; that is, as a result of accident, such as when process
equipment or a container fails. Division 2 locations are as follows:
• locations in which flammable liquids or gases are handled, pro-
cessed, or used but are normally confined in closed containers or
closed systems;
• areas that are rendered nonhazardous by forced ventilation,
which would become hazardous if the ventilating equipment
failed;
• areas adjacent to Division 1 areas in which hazardous concentra-
tion of vapors or gases could be communicated, unless prevented
by positive ventilation with adequate safeguards against ventila-
tion failure;
• areas in which dust layers accumulate and may be ignited by
electrical equipment.
Class III Locations
A Class III, Division 1 location is one in which easily ignitable fibers or

materials producing combustible flyings are handled, manufactured, or
used.
A Class III, Division 2 location is one in which easily ignitable fibers are
stored or handled, except in the process of manufacture.
The notion of normal presence compared to accidental presence of combus-

tible fibers as a distinction between Class III, Division 1 and Class III,
Division 2 locations is not as strong as in the definitions for Class I and
Class II locations. In the 1931 NEC, what are now called Class III, Division 1
and Division 2 locations, were identified as Class III and IV locations,
respectively. Class IV was dropped in 1947 when Divisions 1 and 2 were
introduced, becoming Class III, Division 2.
Article 505—Class I, Zone 0, Zone 1, and

Zone 2 Locations
The introduction of Article 505 into the 1996 NEC was the first step in rec-
ognizing the IEC method of area and material classification in the United
States. This will allow eventual recognition of methods of protection, such
as increased safety and encapsulation, that have been standardized for use
in Zone 1 locations in Europe and in other industrial nations, but have not
been recognized for use in Division 1 in the United States.
Presumably, when IEC agrees on definitions for zones in locations where

combustible dust is the hazard, these will be proposed for addition to Arti-
cle 505.
Article 505 defines zones in terms of the presence of ignitable concentra-

tions of combustible gases and vapors, as follows:
• Zone 0: continuously, or for long periods of time
• Zone 1: in normal operation of the plant, or a location next to
Zone 0
• Zone 2: if plant containment system fails or a location adjacent to
Zone 1
Thus, Zone 1 and Zone 0 together constitute Division 1 in the customary

North American classification, and Zone 2 equals Division 2.
Materials are classified as follows:

• Group IIA equivalent to Group D
• Group IIB equivalent to Group C
• Group IIC equivalent to Groups A and B
Determination of Area Classification Relative to

Class I Hazards
The definitions of the types of hazardous locations in the NEC serve to
establish guidelines. The code-enforcing authority determines classifica-
tion of specific areas. The code-enforcing authority may be a
representative of the insurance underwriters, a municipal inspector, or a
member of the corporate safety organization (if the corporation is
self-insured). In the NEC the “code-enforcing authority” is now referred
to as the “authority having jurisdiction.” Whoever classifies the location
must consider the quantity of hazardous material available at the location,
the topography of the site, the construction of the plant or building, and
the past history of fire and explosion of both the particular plant and the
industry with which it is affiliated.
Installations in Division 1 locations can be substantially more costly than

installations in Division 2 locations. Increased use of electrical apparatus
in hazardous locations in the middle of this century led organizations such
as the American Petroleum Institute to publish guidelines for classifica-
tion of locations in which flammable materials may be present. These
guidelines are intended to promote uniformity of practice and minimize
the common tendency to classify a location as though it were more haz-
ardous than it is.
API RP500A, RP500B, and RP500C are written for petroleum installations.
The principles are applicable to other industries with Class I hazards.
NFPA 497 is a similar guide to area classification, intended primarily for
chemical plants. The principles are similar to those of the API RP500
series, but the recommended dimensions of the hazardous zones differ in
many cases because of differences in assumed size of plant equipment and
the amount of flammable material that may be released at open vents or
other points of emission. Many industrial organizations maintain corpo-
rate guidelines for area classification that are similar to the NFPA and API
recommendations.
All these classification guides, as well as diagrams given in the NEC (in
Article 516, for example), are geometric approaches to area classification.
Guides define the distance a Division 1 or Division 2 location extends from
the source of release. The expected size and duration of the release, the
effects of wind, dispersion, and other factors are implicit in the recommen-
dations for distance, which tend to be representative of typical situations,
weighted conservatively. Specifying different distances for different vessel
sizes sometimes indirectly recognizes size of release.
In recent years considerably more attention has been directed toward

methods of area classification that explicitly recognize the expected type
and quantity of material released, the effects of height of the point of
release, the adverse effects of wind, and the duration of the release.
Two factors have motivated more sophisticated approaches. There has

been increasing recognition that the “typical” diagrams do not always
yield safe classifications if applied to situations in which the total release
may be much greater than assumed by those who drafted the diagrams.
Conversely, use of diagrams causes overclassification of plants where
releases may be much smaller than the “typical” one.
Secondly, since the advent of OSHA in 1971, which caused many users of
electrical apparatus to adhere to the letter of the NEC rules rather than the
intent, many plant operators have recognized the severe economic penal-
ties for overclassification, particularly if installations are strictly in
accordance with the NEC. Formerly, the economic penalty for overclassifi-
cation was often lower because installation practices were used that were
safe for that location but that did not meet all NEC requirements. Article
90-4 of the NEC states that the authority having jurisdiction may waive
requirements or permit alternative methods of installation if safety is
assured. OSHA, as the authority having jurisdiction, has generally not
permitted such deviations from the text of the NEC.
The manual of one large chemical company specifies the following steps
leading to classification of a location.
(1) Determine the flammability and physical properties of materials
being handled, for example, flash point, vapor density, NEC Group,
boiling point, handling temperature, and auto-ignition temperature.
(2) From process data, flow diagrams, and layouts, determine the proba-
ble location, rate, and quantity of release. Typical points of release are
packings, sampling valves, relief devices, and drains. Guides are
given for low, moderate, and high flow ranges and for small medium,
and large quantities.
(3) Determine environmental considerations such as elevation or depres-

sion relative to grade and degree of ventilation. A location blocked on
four of six sides (for example, floor, roof, and two sides) is frequently
considered to have restricted ventilation. Indoor locations where air-
flow is 1 ft3/min per square foot of floor area are considered to be well
ventilated.
(4) Specify NEC group classification of the material at the point of release.
(5) Specify division classification and the distance that classification

extends from the source by (a) reference to geometric diagrams of the
type discussed above, (b) use of curves given in the guide to calculate
the division classification and extent, or (c) direct calculation (for fre-
quent large spills).
Method (a) in step 5 is the conventional method of classification. Method

(b) uses precalculated curves that can be used if one estimates the number
of spills per year, their average duration, and the average spill rate of
release and if the molecular weight and lower explosive limit of the mate-
rial are known. The calculations are based on acceptable risk of 1 ignition
in 1,000 years, spills in the range of 100 to 500 lb./min, and conservative
assumptions about the environment.
Method (c) is rather lengthy and involves the following:

• calculating the extent of the flammable cloud under steady-state
conditions
• estimating the number of pieces of electrical apparatus in the haz-
ardous zone and their failure rates
• calculating the spill-minutes per year that would yield an accept-
able risk of one explosion in 1,000 years
If the expected number of spill-minutes exceeds this value the location is

classified Division 1. Otherwise the location is classified Division 2.
The reason for referencing this approach to area classification is to high-

light the limitations of the commonly used geometric approach to area
classification. The details of curve derivations and the exact methods of
calculation are not given here because they are not pertinent to this over-
view. At least one large company provides alternative methods for unique
situations that justify the added engineering investment to determine a
more realistic area classification.
Considerations in Classification of Class I Locations

The predisposing factors to a fire or explosion include (1) the presence of a
flammable liquid, vapor, or gas in ignitable concentration; (2) a source of
ignition; and (3) contact of this source with the ignitable material. These
conditions are commonly represented graphically by the fire triangle (see
Figure 3-1). In this diagram the presence of the flammable material in
ignitable form is depicted by the presence of the flammable material and
oxidizer legs of the triangle. If any one of the legs of the fire triangle is
removed there can be no ignition. Showing the oxidizer leg as a separate
leg emphasizes its importance and makes more obvious that one approach
to preventing an ignition is to prevent the presence of the oxidizer.
Figure 3-1 The Fire Triangle
When considering the classification of a specific area made hazardous by

the presence of flammable vapors or gases it must be remembered that
gases or vapors lighter than air diffuse rapidly. Therefore, they seldom
produce hazardous mixtures close to grade at great distance from the
source. Methane gas and light hydrocarbon vapors from volatile flamma-
ble liquids, for example, rapidly become diluted with air. Also, liquids of
low vapor pressure seldom render an area of significant size hazardous.
Vapors heavier than air, such as propane, butane, and many other Group
D materials, stay close to grade, so the hazard does not exist long at ele-
vated locations but may exist close to grade for great distances. For
example, NEC paragraph 514-2 states that the Division 1 area around a
gasoline dispenser extends 18 in. (0.45 m) around the dispenser, 4 ft (1.2
m) above ground level. The Division 2 area extends 20 ft (6.1 m) from the
dispenser at a level 18 in. (0.45 m) above the ground. In Canada this level
is 2 in. (50 mm).
Air currents, the degree of ventilation, and the topography of a location

are critically important in determining its area classification. If there are no
walls or obstructions one can safely assume dispersal of the hazardous
material in all directions; modified, of course, by the effects of material
density. Air currents can seriously modify the conclusions based on con-
siderations of diffusion only—a light breeze can extend a hazardous area
for great distances, whereas a heavy breeze will disperse the vapor or gas
and eliminate the hazard.
Compressed liquefied flammable gases released as a liquid pose a special

problem. When released as a liquid they readily pick up heat and vapor-
ize, producing a volume of cold gas much larger than the volume of liquid
spilled. This cold gas may remain at grade for a large distance until the gas
warms. The resulting hazardous area may be of much greater extent than
one would judge solely from consideration of nominal vapor density.
NFPA 30, Standard on Basic Classification of Flammable and Combustible

Liquids, classifies flammable and combustible liquids according to their
closed-cup flash points. The flashpoint is the temperature at which the
vapor pressure is high enough to produce a flammable mixture in air.
Classifications are as follows:
• Class I—Flashpoint below 37.8°C (100°F)
• Class II—Flashpoint from 37.8°C to 60°C (100 to 140°F)
• Class III—Flashpoint above 60°C (140°F)
IIIA—Flashpoint from 60°C to 93.4°C (140-200°F)
IIIB—Flashpoint above 93.4°C (200°F)
When Class I liquids are released in quantity, they begin to evaporate at a

rate that depends on their flashpoint. The lower the flashpoint, the faster
the evaporation. Because the vapors produced by evaporation are not nec-
essarily at ambient temperature, they act as though they are heavier than
air until they warm and disperse. Emission of Class I liquids normally
requires the assignment of area classification to ensure proper electrical
system design.
At normal handling and storage temperatures, Class II liquids emit vapor

at a relatively slow rate. They will, in general, not form flammable mix-
tures in air at ambient temperature unless they are heated above their
flashpoints. Vapors that are emitted tend not to travel great distances
because they condense at ambient temperature. Therefore, Class II liquids
are considered capable of causing an ignitable mixture near a point of
release of liquid heated above its flashpoint.
Because of their high flashpoint and the tendency of emitted vapors to

condense at ambient air temperatures, Class IIIA liquids seldom justify a
hazardous area classification. Any classified area will be very small. Class
IIIB liquids seldom cause an area be classified because their vapor pres-
sures are so low that ignitable mixtures are not likely to form.
This discussion of the applicability of the flashpoint in judging whether a

release requires a location to be classified assumes that the flashpoint does
not change as time passes. One must be very careful when assessing the
hazard of mixtures. If one component of the mixture has a low flashpoint
the flashpoint of the mixture will be lowered. But if vapors escape from a
tank holding the mixture, for example, the concentration of the low flash-
point material decreases, and the mixture flashpoint rises, a change
toward a somewhat lower level of hazard. But what if two materials are
mixed, one of which is nonflammable, but has a vapor pressure lower than
that of the flammable liquid? As time passes the nonflammable liquid
evaporates, leaving behind the flammable liquid. A mixture whose vapors
were at first difficult to ignite becomes more easily ignited.
A similar situation of increasing hazard is met with solutions of low-boil-

ing-point materials in water. At the outset, the vapor may be well above
the UEL, and therefore poses no hazard. However, escape of the vapors of
the flammable material can lower the concentration in the vapor phase
below the UEL, producing a hazard. In general, most experts do not con-
sider a location to be unclassified because the flammable vapor or gas is
above the UEL. It would be classified Division 2, at least, because failure of
the containment system might admit air and produce a flammable
mixture.
Another caution: if a flammable liquid is released as fog or mist, vapor

pressure or flash point may not be the sole determinant of hazard. A mist
or fog of a high-flashpoint liquid can be vaporized and ignited by a hot
surface or flame.
In general, a Division 2 location will always exist between a Division 1

location and an unclassified location. The exception is when an impenetra-
ble barrier, such as a wall, is interposed between the Division 1 location
and an unclassified location.
Unclassified Locations
Experience has shown that it is not necessary to classify areas (i. e., the
areas are considered to be nonhazardous) where:
(1) The systems are closed, including only pipes, valves, fittings, flanges,
and meters, and the area is well ventilated (i. e., substantially open to
the free passage of air). The area may be roofed or closed on one side
or may be enclosed with forced ventilation and appropriate safe-
guards.
(2) The system is closed (i. e., piping without valves, flanges, and so on),
even if the area is not well ventilated.
(3) The area is for storage in containers that meet NFPA and/or DOT reg-
ulations for the material involved.
(4) Permanent ignition sources other than electrical installations are

present.
Some Guides for Class I Area Classification

Division 1 Locations
If flammable gases or vapors are likely to be present during periodic main-

tenance or during normal operation, the location is probably Division 1.
Vents, relief valves that operate frequently, seals or packing glands on
moving or rotating shafts, and sampling valves are typical points of
release. Below-grade depressions may be Division 1 if heavier than air
vapors are likely to collect, as they might if at grade the location is Divi-
sion 2. Similarly, roofs that can trap lighter-than-air vapors above a
Division 2 location may enclose a Division 1 location.
A location is not likely to be Division 1 if workmen or operators are in it

daily for extended periods without breathing apparatus. Except for hydro-
gen and methane, most flammable vapors are too irritating or toxic for
lengthy exposure even to concentrations much below the lower explosive
limit. This generality must be applied with judgment. One can easily visu-
alize situations in which a hazard exists close to grade, but does not exist
at head level. Such a situation may exist where batch retorts are located
indoors. Because they are periodically opened for recharging releases of
solvent, vapors are expected in normal operation. In some industries the
entire room in which the retorts are located is customarily classified Divi-
sion 1. In most cases, however, the Division 1 location does not extend
much above floor level. A more realistic classification would be Division 1
to a defined distance above the floor and Division 2 throughout the rest of
the room.
It is quite easy to acquire an exaggerated concept of what constitutes a

Division 1 location. In actuality, most petroleum and chemical installa-
tions are composed of a multiplicity of Division 1 locations of extremely
limited extent. For example, a packing gland leaking 1 liter/min would
not be considered a common occurrence; yet, if a liter per minute were
emptied out of doors, the area rendered hazardous would be difficult to
locate with a gas detector. Leakage from a heavily frosted light-ends
pump gland is difficult to sense with a detector only 1 meter away in a
freely ventilated area. For this reason less than 5% (some say 2%) of instru-
ments in refineries and chemical plants are likely to be located in Division
1 locations.
Topography and ventilation strongly influence hazard. A gasoline pool in a

sizable open manifold pit was reported to cause no dangerous readings of
a detector beyond 3 or 4 ft (0.9–1.2 m) from the pit in an 8–10 mph (4–5 m/
s) breeze. However, a smaller area of more volatile product blocked on one
side produced a detectable hazard in a gentle breeze 100 ft (30 m) down-
wind at grade, but vapor was not detectable 18 in. (0.45 m) above grade,
30 ft (9.1 m) away.
Division 2 Locations
A location that is neither nonhazardous nor Division 1 is classified Divi-

sion 2. Generally a Division 2 classification relates to the need for
equipment to fail and then release a flammable cloud or cause the concen-
tration of a vapor or gas to pass from above the UEL into the flammable
region. The Division 2 classification may also be assigned to a transition
zone between a Division 1 location and a nonhazardous (unclassified)
location. Pumps and similar devices are generally considered to be sur-
rounded by Division 2 locations if they are located in a well-ventilated
location. This assumes that a failure must occur to produce a significant
release. Were a pump to experience frequent seal failure because of
adverse conditions of process pressure or temperature, or chemical attack,
then a Division 1 classification may be warranted. In theory a Division 1
location exists around the packing gland of a valve or the seal of a pump if
there is any leakage of fluid above its boiling point. The Division 1 location
is small, usually considered to be about 0.3 m diameter. If the liquid is
released below its boiling point the Division 1 location is much smaller
and is often ignored. Electrical apparatus is not customarily installed
within the limited confines of such a Division 1 location. The objective of
the Environmental Protection Agency's (EPA) fugitive emissions reduc-
tion program has been to reduce the size of all such point sources. The
abnormal leakage that might occur should the seal or packing gland fail
completely justifies a Division 2 classification for an area 10 to 20 times the
diameter of the Division 1 location. The size depends on whether the liq-
uid is released at a temperature above or below its boiling point. If the
liquid is released at a temperature below its boiling point it must evapo-
rate to form a flammable mixture.
Figure 3-2 is a diagram of a hypothetical vented tank. The vapor space

above the liquid may be above the UEL much or most of the time; but,
because the tank is intended to be filled and emptied frequently, this space
is classified Zone 0. During the filling and emptying operation, the vapor
space will be within the flammable range. The sides of the tank are classi-
fied Zone 1/Division 1 because vapors escaping from the vent are
presumed to flow down the sides. The area around the tank, out to the
dike or berm, is classified Division 2 because it is not expected to be haz-
ardous except in the case of a large spill or overfilling. In North America
the sides of the tank would usually be classified Division 2. It is assumed
that the quantity of vapor escaping from the vent is not sufficient to cause
a flammable atmosphere to exist on the sides of the tank.
Figure 3-2 Hypothetical Vented Tank
Toxicity or irritancy of a heavier-than-air flammable liquid is not always a

certain criterion for deciding whether a location is Division 2 in indoor
locations. As Freytag has noted, in locations where such liquids are han-
dled, pockets of flammable vapors may collect at grade or floor level as the
result of spillage by operators or equipment failure. Physiological effects
at head level from these spills may not be detected, especially if ventilation
does not cause the vapors to rise.
Class II Locations
More extensive discussion of area classification of Class II locations will be

found in Chapter 13.
Special Cases of Area Classification in Instrument

Systems
The foregoing paragraphs discussed some of the factors involved in classi-
fying process areas with respect to explosion hazard. Some instrument
installations require specific consideration of the proper classification of
the inside of an instrument enclosure, particularly with regard to reduction
of hazard by pressurization. The question of the area classification inside
an instrument enclosure arises when the instrument enclosure is part of a
system that, as installed, provides a means for transporting a flammable
material from one location to another. Under most circumstances the clas-
sification of the inside of an instrument enclosure is the same as the
classification of the area in which it is located. A pressurization system can
be considered a reasonable way to reduce the classification of the instru-
ment enclosure below that of the surrounding atmosphere. However,
reduction of area classification by the use of pressurization implies that the
pressurization supply pressure inside the enclosure is higher than the
pressure tending to drive flammable gas or vapor into the enclosure.
Therefore, whenever the process material can be transmitted into an enclo-
sure at pressure higher than the pressure of the pressurizing medium, the
classification of the inside of the enclosure, even when purging (continu-
ous flow) is used as well as pressurization, must be carefully considered.
Although this situation is frequently treated in terms of consideration of
singly or doubly sealed systems, with or without vents, this author finds it
most convenient to consider the problem from the standpoint of a single
consideration. If the pressurizing medium is always at a pressure higher
than the pressure forcing flammable material into the enclosure, pressur-
ization can be considered to be effective in reducing the classification of the
inside of the enclosure below that of the surrounding area.
In singly sealed systems, where process pressures are appreciably above

atmospheric, this will never be the case. The most common example is a
Bourdon tube or bellows-actuated instrument that is connected by tubing
to a process located some distance away, with the instrument located in a
nonhazardous area. Process fluid brought into the Bourdon or bellows by
the connecting tubing will establish a flammable atmosphere within the
instrument should tubing, bellows, or Bourdon fail. The inside of the
instrument enclosure must, therefore, be considered to be Division 2, even
though the area surrounding the enclosure is classified nonhazardous. No
practical pressurization installation can be arranged to reduce the classifi-
cation to nonhazardous. If the pressurization supply pressure were
always higher than the pressure inside the Bourdon tube, the inside of the
instrument case would not become hazardous when the Bourdon leaks,
but the system could not measure accurately. However, safety can be
assured if the flow of pressurizing gas is high enough to dilute the incom-
ing material and maintain concentration inside the enclosure below the
lower explosive limit even in the event of the largest probable release.
Such systems, introduced primarily for use with analyzers and other
instruments with possible internal sources of release of flammable mate-
rial, are called continuous dilution systems. The protective gas may be
either air or an inert gas. The design requirements of continuous dilution
systems are discussed in Chapter 7.
If an instrument is compartmented so that the Bourdon tube is in one com-

partment, vented to the atmosphere, an adjacent compartment could be
considered to be nonhazardous if pressurized. Although failure of the
Bourdon tube would release flammable vapors to its own compartment,
pressurization would prevent communication of the flammable vapors
under pressure into the adjacent compartment. This construction is fre-
quently called a doubly sealed and vented installation; that is, there is one
seal between the process fluid and the atmosphere and a second seal
between the compartment to be protected and the compartment in which
a flammable atmosphere can exist.
A thermocouple in a well presents a similar situation. Such a thermocou-

ple assembly may be connected to an instrument through conduit. If
conduit or cable extends from the connection head to an instrument with-
out any break, then this, too, is considered by some experts to be a singly
sealed system. Failure of the well could result in process fluid being forced
into the instrument through the interstices of the thermocouple or exten-
sion wire, or through the conduit. However, if the system is vented at the
connection head, failure of the well will cause the process fluid to dis-
charge through the vent, and the instrument system can be considered to
be doubly sealed and vented. (Environmental considerations probably
require that the vent be connected to a closed waste system, rather than
being vented to atmosphere.) The second seal is the conduit seal that will
normally be present, the restriction of the cable, or a seal around the cable.
The vent ensures that there can be no pressure in excess of pressurization
pressure to force flammable material into the instrument enclosure. Pres-
surization can, therefore, reduce the area classification in the enclosure
below that of the surrounding area. Other experts would consider a well
to be part of the containment vessel and no more likely to fail than the ves-
sel itself, so well failure need not be considered. Whether well failure need
be considered may depend on the physical, thermal, and chemical stresses
to which it is subjected in a flowing process stream.
The objective of the foregoing presentation has been to emphasize the lim-
itations of pressurization in reducing the classification of an instrument
enclosure. Though it is an important point to be understood about pres-
surization as a hazard reduction technique, as a practical matter the need
for concern about the issue by the designer of electrical installations has
largely been removed by the requirements of 501-5(f)(3) of the NEC. This
clause of the NEC was greatly clarified after a serious accident resulted
from failure of a compressor seal, which forced natural gas into the con-
duit system of a plant. The gas was subsequently ignited. This accident
demonstrated a widespread misunderstanding about the pressure tight-
ness of standard conduit seals. Many people did not realize that the seals
required by the preceding paragraphs of 501-5 are not pressure tight. As
can be inferred from 501-5(e)(2), the standard conduit seal is intended to
prevent the flow of flammable material from one point to another in the
conduit system only under a low driving pressure. Standard conduit seals
leak enough so that they cannot be depended upon as a safeguard against
flow of flammable material driven by process pressure. Regardless of
whether pressurization is being used the Code requires a backup seal that
will prevent entry of flammable material into the conduit system should
the primary seal fail. This secondary seal must have the same pressure
resistance as the primary seal, or venting must reduce the pressure to a
level that the seal can resist. If the secondary seal is a standard conduit
seal, the venting must reduce the pressure differential across it to less than
6 in. of water. (The Code does not specify this figure. This figure can be
inferred from the context of the entire clause on sealing.) In all instances,
regardless of the type of secondary seal used, a vent must be provided so
that any leakage of the primary seal is made apparent.
Area and Material Classification Outside the

United States
Area Classification—Gases and Vapors
Most countries outside the Americas have adopted the IEC area classifica-
tion scheme defined in IEC Publication 79-10. This scheme defines three
levels of classified areas, rather than two, as shown in Figure 3-3.
Figure 3-3 NEC Area Classification Scheme vs. IEC Area Classification
Scheme
The zone classification definitions of IEC Publication 79-10 grew out of a

long evolutionary process. The concept of Division 2 was initiated in the
United States when it was recognized that it does not make economic
sense to treat all locations in a plant as though they are equally likely to
have an explosive cloud present. The safeguards necessary to prevent an
explosion in a location in which a flammable cloud of gas or vapor is likely
to exist in normal operation of the plant are not necessary if the cloud is
likely to be present only when some part of the plant fails. A much lower
level of protection can be justified in the locations where the presence of a
flammable gas or vapor is less probable. This philosophy strengthened as
more electrical apparatus was installed in plants handling flammable
materials, and it became obvious that requiring all the electrical installa-
tions in a plant to conform to the same standards did not make economic
sense. By defining two levels of hazard, Division 1 and Division 2, differ-
ing rules could be written for selection and installation of electrical
equipment. The requisite high level of safety could still be maintained, but
at much lower cost. This concept of Division 1 and Division 2 first
appeared in the 1947 NEC.
This concept spread worldwide in plants of multinational petroleum and

chemical companies. By the mid-1960s, Division 2 was not only adopted as
a concept in such plants, but was recognized in the Codes of Practice and
Standards in the United Kingdom. At about the same time the concept of
Division 0 arose in the UK. A Division 0 location is a location that is haz-
ardous such a high percentage of the time that extraordinary precautions
must be taken to prevent electrical installations from causing an explosion.
Such locations were recognized in the practices of plants in other countries
that formally defined only Divisions 1 and 2. It was understood that electri-
cal installations should not be placed in such locations unless precautions
were taken in addition to those required by the Division 1 rules. By the
mid-70s most European nations had recognized Divisions 1 and 2. In some,
like Germany, which did not have a formal definition for Division 2, rules
for electrical installations seem nonetheless to have made allowances for
locations of lower hazard than Division 1. During this same period the UK
concept of Division 0 gained acceptance among experts, so that when the
IEC drafted a standard on classification definitions the concept of three lev-
els of hazard was readily adopted. IEC chose to use the more general term
“Zone,” rather than the North American two-dimensional term, “Area,”
but this is a purely semantic difference and does not imply a difference in
basic intent between the American and IEC experts. The IEC classifications
are now recognized in most industrial nations of the world outside of the
Americas. Those whose national standards do not yet recognize the IEC
definitions will almost certainly do so in the coming years.
IEC Publication 79-10 began as a general discussion of area classification,

but it has now been expanded to give geometric examples of classified
locations and calculations to assess the degree of ventilation, both artificial

and natural. It is now an excellent reference for area classification.
Area Classification—Dusts
The first edition of IEC 1241, published in 1993, recognized two zones for
locations made hazardous by the presence of combustible dusts. Zone 21
is defined much like U. S. Division 1, without mention of conductive dusts
(i. e., a combustible cloud is present or may be present, during normal pro-
cess operation or cleaning operations). Zone 22 is defined as an area not
classified Zone 21 in which a combustible dust cloud may occur infre-
quently, for short periods, or in which accumulations of layers of
combustible dust may be present under abnormal conditions. If, after an
abnormal event, removal of accumulations of dust cannot be assured, the
area shall be classified Zone 21.
The approach to classification in the responsible CENELEC committee

defines the following three zones.
• Zone 20—A location where a hazardous atmosphere formed by a
dust cloud in air is present continuously or for a long time, and
where dust layers of excessive thickness may be formed. (As a
rule, Zone 20 occurs only inside of containers, pipes, and appara-
tus.) Zone 20 does not include areas where dust layers are present
but a cloud is not present continuously or for long periods of time.
• Zone 21—An area where a hazardous cloud is likely to be present
during normal operation and in which accumulations or layers of
combustible dust will in general be present (e. g., near filling and
emptying points).
• Zone 22—An area where formation of a hazardous dust cloud is
not likely in normal operation and, should it occur, will exist only
for a short time (e. g., a location where dust can escape from mill-
ing machinery and be deposited).
It is highly probable that the IEC document will be revised to accept these
concepts.
The Practice of Area Classification
Although the definitions of classified locations are becoming very similar

in all nations, the way in which they are applied is not necessarily uni-
form. For example, NFPA 497 recommends that the zone 5 ft (1.5 m)
around a vent on a fixed roof tank be classified Division 1 and that the
area around the tank is Division 2 for a distance of 10 ft (3 m) One British
source suggests that a Division 1 location exists all around and above the
tank for a distance depending on tank size. A Division 2 location extends

for some distance at grade. Although investigation sometimes shows that
there is a good technical rationale for such differences in practice, based on
some real but not obvious difference in situation, this example illustrates
well that mere standardization of language is not standardization of prac-
tice. The author, of course, is not taking a position that one of the practices,
or neither, is correct.
The application of the concept of Zone 0 is also subject to differences. The

author understands that German law mandates a Zone 0 classification for
the vapor space in fixed roof tanks and within gas pipelines. In North
America one also hears references to vapor spaces in fixed roof tanks
being Zone 0. Sometimes this is just a matter of careless expression, but
there are some that agree with the German view. It is important to remem-
ber that the criterion for classifying a location is the probability that an
explosive mixture will be present. In pipelines extraordinary precautions
are taken to ensure that the pipe is not filled with a flammable mixture but
that the contents are above the UEL. Some would classify the interior of a
pipeline as Division 2 because a flammable mixture will not exist except in
the case of an accident to the pipeline. Similarly the vapor space in a fixed
roof tank is above the UEL unless the tank is frequently filled and emp-
tied. In such use the tank vapor space should be classified Zone 0.
In some countries, Japan, for example, whose electrical equipment selec-

tion and installation practices show the strong influence of earlier German
practice, Division 1 and Division 2 are defined, but equipment design
rules are essentially the same for both kinds of location.
In Germany the definition of Zone 1 in the national regulations does not

include the concept of “in normal operation.” Therefore, the author under-
stands that installation practices in Zone 2 have not been relaxed as much
as they could be were there a clear distinction between normal and abnor-
mal circumstances.
Material Classification
The IEC method of classifying flammable gases and vapors is discussed at

length later in this chapter. The list of materials classified by the IEC has
been included in the CENELEC standard giving general requirements for
all explosion-protected apparatus, EN50014, and is therefore accepted in
Europe and Scandinavia.
Table 3-1 relates designations of material groups that have been used in the
past, some of which are still in use or may be referenced in the literature, to
those now current. The equivalency shown is approximate. As when one
compares the materials classified in NFPA 497 to those classified in IEC
Publication 79-12, the agreement between groupings is generally good, but

some materials have been classified as having a higher or lower hazard.
Early British Former British Japan &

North America IEC
Intrinsic Safety Flameproof Germany
A IIC 2e IV 3c
B IIC 2e IV 3a & 3b
C IIB 2d III 2
D IIA 2a,2b,2c II 1
Methane(D) I 1 I 1
Note: Carbon disulfide is classified equal to Group B outside North America. 3n includes
all materials in 3a, 3b and 3c. 3a means water gas, and hydrogen; 3b means carbon dis-
ulfide; 3c means acetylene.
Table 3-1 Approximate Equivalency of Material Groupings
The Future for Classification of Materials
The author foresees no major changes in the method of classifying materi-

als in those countries and regions that now use the IEC standard methods.
In the United States there is only a slowly growing recognition that the
current practices for classifying materials need reconsideration. However,
most major changes in standards and practices usually occur only to cor-
rect or improve a situation that is recognized as being too costly compared
to the proposed alternative, or which imposes costs which those using dif-
ferent standards do not bear. The inconsistencies of the U.S. classification
of gases, vapors, and dusts are probably most disturbing to those who
would like the system to be more rational, and to those who see the adop-
tion of international standards as a beneficial move in the long run, even
though the immediate impact on costs does not seem to justify action.
Proposals to adopt Division 0 or Zone 0 in the United States were intro-

duced to Panel 14 of the National Electrical Code Committee many times
during the past two decades. Until adoption of Article 505 in the 1996
NEC there was little interest because there is not a strong definitive
answer to the question, “What is there to gain?” or a suitable counter to
the plaint, “We have enough trouble training people to understand the
difference between Division 1 and Division 2. Why complicate matters by
introducing a third level of classification?”
In neither case is a compelling response available. In principle, the lower

degree of protection that could be used in Zone 1 locations (which consti-
tute the majority of Division 1 locations) will reduce equipment and
installation costs. It is not easy to demonstrate that this reduction is large
enough to counterbalance the trauma and confusion of changing to an

unfamiliar scheme. One cannot easily argue that adoption of a Zone 0
location will increase safety by focussing attention on these especially haz-
ardous locations. Most users of electrical apparatus already recognize that
“Zone 0" locations exist and do not install equipment in such locations
unless it is intrinsically safe or of an equivalent level of protection. Adop-
tion of the IEC area classification definitions is, today, of most interest to
those in international businesses who find the necessity of working with
two systems of classification a confusing nuisance and to those who accept
as a matter of principle that world-wide standardization must yield bene-
fit, even if we are not now able to quantify it. As more and more
businesses in the United States become multinational or are owned by
companies headquartered in countries who now use the IEC standards,
the pressure to adopt the IEC standards will increase.
The adoption of Article 505 will demand no step changes in practice by

manufacturers or users of electrical apparatus, but will allow those who
favor adoption of the international practice to do so.
Quantitative Area Classification
Numerical measures of zone or division classification have long been of

interest to those in the electrical safety field. In order to make an assess-
ment of the probability of an explosion occurring under a posited set of
circumstances, it is necessary to assign numerical values to the probability
that a flammable cloud of gas, vapor, or dust will be present and the prob-
ability that an event or series of events will provide an ignition source. The
words “normally, frequently, and periodically” are imprecise. Even if the
agreed quantitative definitions of zones were not quite correct, having
uniform criteria would be beneficial.
Two sets of quantitative definitions have been proposed. One, which

seemed to be most popular during the 1970s, is as follows:
Zone Hours/Yeara Probability

0 Above 1000 0.1-1.0
1 10-1000 0.001-0.1
2 0.1-10 0.00001-0.001
a
That the atmosphere is between the LEL and the UEL.
The second, first proposed by Italian experts, and which appears now to
be gaining acceptance, is as follows:
Zone Hours/Yeara Probability

0 Above 100 .01-1.0
1 10-100 0.001-0.01
2 1-10 0.0001-0.001
aThat
the atmosphere is between the LEL and the UEL.
It is not likely that one set of numbers or the other, or some combination of
the two, will ever be adopted in the United States. A fear exists that were
NFPA or NEC to adopt a numerical measure, enforcement authorities
and, during litigation, the courts, would soon demand proof from the
operator of a plant that his classification is correct. Because the probability
of a flammable cloud being present is low, only measurements of concen-
tration of flammable material over a period of years would validate a
Division 2 classification, for example. The fear of being forced to do the
nearly impossible is therefore strong, even to the extent that use of proba-
bilities in an illustrative manner in the NEC has met strong resistance.
A problem with both of the above proposals is the range of values

assigned to Zone 2. The first proposal would classify as Zone 2 a location
that is hazardous as infrequently as 6 min/yr, which many people feel is
much too conservative. The second proposal would seem to declare that a
location is nonhazardous even though there is a flammable atmosphere
present almost 1 hr/yr. An open flame could be brought into the area.
This is considered to be too dangerous by some.
Although the Buschart paper referenced at the end of this chapter is per-
haps controversial to the extent that the numerical conclusions are not
acceptable to all, the paper points the way to a rational resolution, but not
an easy one. The starting point for defining hazardous locations and con-
comitant protective measures must be a definition of the level of
acceptable risk; that is, the probability of an explosion that one never
wishes to exceed. This probability, together with the probability of equip-
ment failure in a way that creates an ignition source, can be used to derive
area classification values.
Classifying Materials That Are Not Already Classified
There are five options for classifying a new material:

(1) Have the material tested for MESG by UL using the Westerberg
Apparatus. Use that value as the basis for an estimate of the
grouping. The boundary values are given on page 80.
(2) Have the material tested for MESG by a lab that has an apparatus
complying with the IEC Standard, and use the IEC rules given
later in this chapter.
(3) Have the material tested for minimum igniting current by FM,
UL, or other lab using the IEC standard intrinsic safety test appa-
ratus to determine MIC ratio and use the IEC classification rules.
(4) Estimate the classification by calculation and comparison with

materials of known grouping.
(5) Estimate the classification from chemical similarity or some other

property of the material. Grove, for example, proposed using the
quenching distance measured in a tube, especially for materials
that are too poisonous to allow safe determination of the MESG
or MIC. He showed that such quenching distances are well corre-
lated with MESG and MIC values.
Unless values of MESG or MIC Ratio are available the assignment of a

Group classification to an unclassified material cannot be done with cer-
tainty. However, it is possible to estimate the classification with a degree
of accuracy that, in the author's opinion, is adequate to ensure safety.
Because there is no standardized rationale for the process, the grouping
will have no consensus standing. It will only be acceptable in an organiza-
tion that is willing to recognize the value of engineering judgment,
understands the nature of the classification process, and appreciates the
safeguards inherent in the use of material classifications in selecting and
installing electrical equipment in classified locations.
Carhart summarized some of the methods for classifying materials that

have been tried. These include:
• LEL/UEL
• LEL/Stoichiometric concentration
• (LEL)(H) H is molar heat of combustion
The author and other investigators have tried other ratios and products.
Bossert, in his book, discusses the correlation of classification and flame
temperature. Those discussed by Carhart and others yield good correla-
tion when applied to compounds that have already been classified, but
there is always scatter in the results. The predicted classification of some
materials does not agree with those that have been assigned. These inexact
correlations can be used only when some estimate is needed, and lack of
time or money necessitates using an approximation.
NMAB 353-1 presented guidelines for classification based on chemical

families. These are presented below, somewhat altered to reflect the classi-
fications in NFPA 497M-1997. With the occasional exception of the lighter
compounds in a homologous series the ease of ignition decreases and the
MESG increases as molecular weight increases. In the aliphatic hydrocar-
bon series propane is the most easily ignited. Chlorination reduces
ignitabilities. Esters are somewhat more difficult to ignite than the corre-
sponding hydrocarbons. But the relationships between chemical structure
and flammability characteristics of compounds are still only poorly
understood.
Group D
• Aliphatic hydrocarbons: methane, ethane, propane, butane,
n-pentane, iso-pentane, iso-hexane, n-heptane, octane, etc.
• Alicyclic hydrocarbons: cyclohexane, cyclopropane
• Higher olefins: propylene, isoprene, diisobutylene (cf. dicyclo-
pentadiene - C)
• Aromatic hydrocarbons: benzene, toluene, xylenes, styrene
• Alcohols: methyl, ethyl, propyl, butyls, amyls, (cf. allyl—C)
• Phenols
• Ketones: acetone, butanone, methyl isobutyl ketone, and mesityl
oxide
• Organic acids and androids: acetic acid
• Esters: ethyl acetate, isobutyl acetate, vinyl acetate, ethyl acrylate,
propiolactone
• Ethers not containing −CH2 or −CH3: isopropyl ether
• Glycols and their esters
• Chlorinated hydrocarbons: ethylene dichloride
• Primary amines: ethylamine, ethylenediamine
• Alkanol amines
• Aromatic hydrazines
• Aromatic nitro compounds
• Cyanides, nitriles, isocyanates, acrylonitrile
• Aromatic heterocyclic nitrogen compounds: pyridine
• Amides
• Organo phosphates
• Organo sulfates and sulfones
Group C
• Higher acetylenes: methyl acetylene (cf. acetylene—A)
• Small unsaturated alcohols: allyl
• Aldehydes: acetaldehyde, n-butyraldehyde, (cf. formaldehyde—
B, and acrolein—B(C))
• Ethers containing −CH2 or −CH3: ethyl ether
• Glycol ethers
• Cyclic ethers larger than 3-membered ring
• Secondary and tertiary amines: diethylamine, triethylamine
• Aliphatic nitro compounds: 2-nitropropane
• Aliphatic heterocyclic nitrogen compounds: ethyleneimine
• Organo lead compounds
• Mercaptans and sulfides: hydrogen sulfide (cf. carbon disulfide—
A, were it classified in the U. S.)
Group B(C)
(Group C apparatus maybe used if all conduits are sealed, but otherwise
Group B apparatus must be used.)
• Ethers, 3-membered ring: ethylene oxide, propylene oxide
The Woinsky and Chamlee work noted in the references proposes a

method by which the classification of a material can be estimated. This
method perhaps deserves further investigation. They used the following
equation.
2
( t F – t ai )
TG = -----------------------------------------------------------
-
3⁄4
( t ai + 460 ) ( F a – 3F R )
where:
TG = the group classification temperature
tF = the pseudo-flame temperature for a stoichiometric mixture
in air
tai = the autoignition temperature
Fa = the number of atoms in a molecule of fuel
FR = the number of rings
Their assignment of group classification was based on the following table.

TG Group
> 6400 Worse than A
5800 < TG < 6400 A
3900 < TG< 5800 B
1700 < TG< 3900 C
1700 < TG D
The rationale for their method is that the difference between autoignition
temperature and the pseudo-flame temperature is related to the degree of
quenching and cooling required in a flange gap. Autoignition temperature
is related to the ease of ignition by hot gases emerging from the flange
gap. The number of atoms in a molecule of fuel is a factor because large
molecules and their decomposition products react with free radicals and
reduce their concentration, permitting quenching by adsorption of the
radicals on the flange and on the wall of the explosion vessel. But ring
compounds are more refractory and scavenge free radicals to a lesser
extent, so the 3FR term must be included.
Their predicted classifications of 32 materials classified in the 1971 NEC

agreed with the NEC classifications, with the exception of acrylonitrile,
which was classified by them as a Group C material, and by the NEC as a
Group D material. In comparing their predictions with the code classifica-
tions they used the classification based on sealing of the explosionproof
enclosure, the less hazardous classification. They presented a method of
predicting the classification of mixtures using the same equation.
Classification of Mixtures
Most materials found in classification lists are pure, though there are a few
exceptions, like gasoline. However, many materials in commerce are not
handled in their pure state, but exist as mixtures. Someone has to estimate
the classification. How can it be done? One method, which is applicable
only to mixtures that are not solutions, is described below. If this method
is applied to solutions—for example, a solution of a liquid such as acetone
in water—the errors are very large.
The classification of a mixture can be estimated as follows:

(1) Determine the group classification of each of the components of
the mixture. If all are Group D, the mixture will be Group D. If all
are Group C, the mixture will be Group C. (This assumes that
there are no reactions between components that will release a
flammable gas or vapor.) If the components of the mixture are
from different groups, continue.
(2) Estimate the concentration of the constituents in the vapor using

Raoult's Law.
(3) From the Le Chatelier equation, calculate the LEL and UEL of the
vapor above the mixture.
(4) Compare the LEL and UEL just calculated with the values for
other materials, or with the values for the components of the mix-
ture, and make a judgment based on the similarity of the spread
between LEL and UEL for the mixture and that of the reference
materials.
Example
A mixture shipped in tankers is defined as follows:

Component % by Weight Group LEL UEL
Acetone 80.2 D 2.15 13
Cyclohexane 16.2 D 1.3 8
Isopropyl alcohol 1.0 D 2 12 (est.)
Isopropyl ether 1.5 D 7.9 7.9
Water 1.0 —
Because all constituents of the mixture are Group D, the vapor can be con-
sidered to be Group D, so it is not necessary to compare the flammable
limits and estimate the classification. In this case, the water vapor in the
vapor above the mixture will be a very small percentage of the total, so it
can be ignored. In other cases, as discussed below, the influence of the
water must be considered.
Now assume that material AA, with the following characteristics, replaces
the cyclohexane.
Vapor pressure 447 mm
Molecular weight 58
LEL 2.8
UEL 37
Raoult's Law states that the partial pressure p1 of a component in the

vapor equals the mole fraction in the liquid, x1, times the vapor pressure
P1 of the pure component at the same temperature. Table 3-2 shows the
fractional volume calculations for the vapor of this mixture.
Molecular Moles per Mole Vapor Partial Volume

Material Weight %
Weight 100 g fraction Pressure Pressure fraction
Acetone 80.2 58 1.38 0.79 186.7 147 0.67
AA 16.2 58 0.28 0.16 447 71.5 0.32
Isopropyl 1.0 60 0.016 0.009 36.8 0.33 0.001
alcohol
Isoprophyl 1.5 102 0.015 0.009 1387 1.25 0.006
ether
Water 1.0 18 0.055 0.03 17.5 0.5 0.002
Mixture 100 — 1.75 1.0 — 220.6 1.0
Table 3-2 Fractional volume calculations for the example
Now calculate the LEL and UEL of the vapor ignoring the water vapor.
The flammability limits can be calculated using the Le Chatelier equation:
100
Flammibility limit = -----------------------------------------------------------------------
a⁄A+b⁄B+c⁄C+ + + +
where
a, b, c = proportions by volume of mixture on an air-free basis
A, B, C = the limits of flammability of the constituents
100
LEL = ----------------------------------------------------
67 32 0.1 0.6
------- + ------- + ------- + ---------
2.2 2.8 2 1.4
100
= -------------------------------------------------------------------
30.34 + 11.42 + 0.05 + 0.42
100
= ------------- = 2.4%
42.23
100
UEL = -------------------------------------------------
67 32 0.1 0.6
------ + ------ + ------- + ---------
13 27 12 7.9
= 100
--------- = 16.4%
6.1
The estimated flammability range is 2.4% to 16.4%. This is a broader range

than those of acetone and isopropyl alcohol, both Group D materials, but
not greatly so. This author would, therefore, classify the mixture as a
Group D material. Others might choose Group C to be especially safe, but
the author feels that in practice Group D classification poses no significant

risk.
Suppose the mixture is one that is being handled or transported. It is rea-

sonable therefore to assume that its temperature is close to ambient
temperature. Of the materials classified in NFPA 497-1997 with flash-
points below 100°F and 140°F, respectively, the statistics are as follows:
Below 100°F Below 140°F
Group A 1 1
Group B 8 10
Group C 39 46
Group D 86 115
The Group A material is acetylene. Of the 10 Group B materials, two are

gases not likely to be encountered in liquid mixtures. Another six materi-
als are classified B(C) or B(D), which means that flammability
characteristics are more like Group C or Group D materials, but the pres-
sure rise or rate of pressure rise was judged by UL to be similar to that of
Group B materials. Thus, most mixtures will be composed of Group D and
Group C materials. Such a mixture can never be more hazardous than
Group C, even were it erroneously classified Group D.
In the author's opinion, the practical degree of risk of using Group D

apparatus because a mixture is erroneously classified Group D is not wor-
risome. Requirements for Group C and Group D explosionproof
enclosures are not greatly different. In fact, because they are so similar,
many enclosures are approved for both groups. Under laboratory test con-
ditions the pressure rise of the worst-behaving Group C materials may be
significantly higher than that of Group D materials. It is unlikely that such
pressure rises will be experienced in a mixture of vapors or that the con-
centration that produces the highest pressure rise will be encountered in
practical circumstances. The fact that design criteria for explosionproof
enclosures contain large safety factors and the conditions of compliance
testing are worst case also contribute to the low real risk from misclassify-
ing a mixture of vapors. These arguments are not intended to justify the use of a
Group D enclosure in a Group C atmosphere. They are only addressed to the
risk resulting from the possible misclassification of a mixture.
If intrinsic safety is to be used, the risk is even lower. Almost all commer-
cial systems are approved for Groups C and D, and most for A, B, C, and
D. In the unlikely case of using a system approved for Group D only, the
safety factors inherent in the design and approval of intrinsically safe sys-
tems makes the risk in practical situations even lower than that for
explosionproof enclosures.
A key factor in the author's low assessment of the risk resulting from esti-
mating the grouping of a material is the fact that classifying materials is
the drawing of an arbitrary dividing line in a continuum of values of some
characteristic of the material. Materials just below the dividing line are not
significantly different from those just above the dividing line. Misclassifi-
cation is most probable in borderline situations. Where there is significant
doubt the prudent person will choose the more hazardous grouping.
The reader should question the classification of a mixture using the simi-
larity of LEL and UEL values of the mixture to those of the component
liquids, which have already been classified. If there is a good correlation
between those values and group classification why isn't it used for group
classification of new materials? As Carhart pointed out there is good cor-
relation between grouping of materials and their flammable limits, but the
correlation is not a strong enough basis for a classification scheme. How
can it be justified here? The answer is—it's the best tool available in the
absence of MESG or MIC ratio data for the mixture, and the consequences
of error are not serious.
If a mixture is mostly Group C materials, the vapor will probably be Group

C. If the mixture is mostly Group D materials, the vapor will probably be
Group D. The calculations presented above are a way to estimate whether
the classification of a mixture will shift from one group to another.
Why use this relationship rather than one of those mentioned by Carhart?
Is this the best method to use? Perhaps not, but it is straightforward, and it
uses data readily available for most flammable materials. There is no evi-
dence that it is the best approach, but there is also no evidence that one of
the other relationships is better.
Another approach to classifying a mixture can be used if the MESG values

are know for all the components. This approach uses a form of the Le
Chatelier equation in which A, B, and C are the MESG values for the con-
stituents of the mixture and the calculated value is the estimated MESG of
the mixture.
A Caveat About Classification of Mixtures
If vapors can escape from the mixture, the composition of the mixture will
change as time passes. The change may be in the direction to make the
vapor above the mixture more hazardous, or it may make the vapors less
hazardous. Suppose a mixture in an open or vented container consists
only of two components, one that is nonflammable and one that is flam-
mable. If the vapor pressure of the nonflammable component is the higher
of the two, the vapors will become more easily ignited as time passes. If
the flammable component has the higher vapor pressure, the vapor above
the mixture will become less easily ignited as time passes. The classifica-
tion of a mixture, unlike that of a pure liquid, is always changing if the
escape of vapors causes the mixture composition to change.
Aqueous Solutions of Flammable Liquids
Raoult's Law cannot be used accurately for solutions, so an attempt to use

it to predict the flammability of the vapors above a solution, such as ace-
tone in water, will yield useless results. Instead one must find data
relating the vapor pressure of the flammable material to its weight per-
centage in the mixture. From this data one can determine whether the
vapor above the aqueous solution is flammable or not.
At 25°C the vapor above acetone has a partial pressure of 227 mm Hg. At a
tank pressure of 1 atmosphere, the concentration is therefore 30%, above
the UEL of 13%. If water is added, the vapor pressure of acetone
decreases. At a weight percentage of about 25% acetone in water the par-
tial pressure of acetone is 100 mm Hg, and the vapor is at the UEL, 13%.
The LEL is reached when the acetone content of the solution is about 4%
acetone.
Material Classification in the United States—Tutorial

Discussion and Opinion*
The classification of hazardous gases, vapors and dusts, as presently prac-
ticed in the United States, has developed over several decades without an
underlying unifying structure. The result, in the case of dusts, is a classifi-
cation in the NEC that has repeatedly been the object of proposals to
amend it since 1978. The problems with classification of gases and vapors
are much less apparent, but they become obvious as soon as one tries to
classify new materials. Resolution of these problems will only result from
recognizing the fundamental nature of any classification scheme and
thinking through the objectives of hazardous material classification in the
NEC. A single amendment of the NEC and straightforward amendments
to NFPA 497 could give the classification system a unified structure with-
out causing changes that would adversely affect manufacturers or users of
electrical apparatus. In the long term they could yield positive benefits to
the electrical industry. As this chapter is being completed, there are indica-
tions that some of the actions needed to revamp the U. S. material
classification system might be taken in the foreseeable future.
* Some of the concerns expressed by the author have been addressed in NFPA 497-
1997, which was published after this manuscript was completed.
What is a Classification System?
Any classification system divides a population into groups according to

the value of some attribute of each individual in the group. In most cases,
the values of the selected attribute for all members of the population form
a continuum. The values picked as the dividing lines between groups are,
therefore, somewhat arbitrary, and the characteristics of members of the
population just above and just below the arbitrary dividing line do not dif-
fer significantly.
If the resulting classification has only one dimension (i. e., a material is
placed in Group A, Group B, Group C, or Group D), there can be only one
criterion for selecting the appropriate grouping. A basket of peas can be
sorted into four groups according to pea diameter. The same basket of
peas can be sorted into four groups according to pea color. The group of
peas with the largest diameter would not have the same members as the
group with the greenest color, unless color and diameter are always corre-
lated. In that case one could sort by either color or diameter. But suppose a
new strain of peas has large peas that are yellower than medium size peas.
Using the old sorting rules, it will be impossible to sort the new peas into
groups which would yield Group A peas of the same characteristics as the
old Group A peas. A one dimensional classification system must sort on only
one characteristic.
Secondly, the method used to obtain the numerical value of the criterion
used for sorting must not affect the value in an unpredictable way; i. e., the
measurement errors must be regular and reproducible. It would be accept-
able if the measurement error for small peas were higher than that for
large peas, but the error must be the same for all small peas. If some peas
were softer than others were so that they deformed during measurement
and appeared to have a smaller diameter the method of measurement
used is not appropriate. A parameter other than the one being measured must
not affect the measurement, except in a regular, predictable way.
A classification system must produce a grouping that is relevant to the way in

which the grouping will be used. A grouping of peas according to color is of
no value to a canner that is trying to ensure that all small peas go into the
same can.
The classification of gases, vapors, and dusts in the U. S. has, at one time
or another, violated one or more of these requirements of a useful classifi-
cation system. To further confuse matters for anyone trying to classify
new materials or ameliorate the problems resulting from the published
groupings of materials, the basis for establishing the groupings has never
been defined clearly.
Classification of Gases and Vapors in the United States
Groupings for Class I materials were first published in the NEC in 1935. It
was recognized that the degree of hazard is not the same for all materials.
It was also recognized that it would be impractical to require testing of
apparatus for every gas or vapor to which it might be exposed, or to
require all apparatus to be suitable for gases and vapors of the highest
level of hazard potential. These early classifications were based on experi-
ence with commercial explosionproof enclosures. Criteria were not
publicly available. Autogenous ignition temperature, however, was one
criterion for assigning a material to a group. In 1962 the NEC listed 17
gases and vapors which had been classified, only a few of the hundreds of
materials in commerce. About 1965 the U.S. Coast Guard asked the
National Academy of Sciences to form a panel to classify 200 materials of
commerce likely to be carried in vessels under Coast Guard jurisdiction,
and to develop a rationale for classifying materials in the future.
In 1968 UL tested 15 materials in a study jointly funded by the API, MCA,

and NEMA. 14 of the chemicals were added to the groups listed in the
NEC. Carbon disulfide was not included because of its 100°C ignition tem-
perature and the fact that a very small gap is required to prevent
transmission of an explosion through the flange gap of explosionproof
apparatus.
The materials were tested in the Westerberg apparatus, which consists of

two cylindrical chambers separated by a wall in which a flange gap is
installed to simulate the flange gap of an explosionproof enclosure. Both
chambers are filled with an atmospheric mixture of the material under
test. Ignition takes place in one chamber, with or without turbulence gen-
erated by a fan, or at the end of a length of conduit. The width of the gap is
adjusted to determine the widest gap that will just prevent transmission of
the explosion from the first chamber into the second. The value of this gap
width is called the Maximum Experimental Safe Gap (MESG). Classifica-
tion of the new materials was based on consideration of the MESG, the
pressure rise in the first chamber, and the autoignition temperature of the
material. Classification of the 14 new materials was based on similarity of
their measured properties to those of previously classified materials—
acetylene, hydrogen, diethyl ether, ethylene, gasoline (Naphtha R),
butane, and propane. The properties of gasoline were taken as the bound-
ary between Groups D and C, and diethyl ether represented the boundary
between Groups C and B.
The Electrical Hazards Panel, which undertook the task requested by the
U. S. Coast Guard, concluded that inclusion of the autoignition tempera-
ture (AIT) in the classification scheme was illogical because AIT is not
correlated with other hazard properties of flammable gases and vapors. It
was well known that there is good correlation among combustion parame-
ters such as MESG, flame speed, ignition energy, and quenching distance,
but none of these parameters are correlated with AIT. The NFPA Sectional
Committee on Electrical Equipment in Hazardous Chemical Atmospheres
(the EECA committee) proposed groupings based on the UL data, without
taking AIT into account. The 1971 NEC contained this grouping, and insti-
tuted a requirement for marking of apparatus with the temperature of the
hottest surface that might ignite a flammable atmosphere or a temperature
code. The user could then ensure that the apparatus was not installed
where a gas or vapor with an AIT lower than the temperature or tempera-
ture code marked on the enclosure might be present. Table 3-3 lists the
temperature codes, commonly called T codes, and the equivalent maxi-
mum surface temperature of the equipment, based on an ambient
temperature of 40°C. The codes without an alphabetical suffix are used in
IEC and CENELEC documents. Those with suffixes exist in the NEC and
CEC to recognize temperature limits which had been standardized much
earlier.
Maximum temperature
°C °F Identification
450 842 T1
300 572 T2
280 536 T2A
260 500 T2B
230 446 T2C
215 419 T2D
200 392 T3
180 356 T3A
165 329 T3B
160 320 T3C
135 275 T4
120 248 T4A
100 212 T5
85 185 T6
Table 3-3 Maximum Surface Temperature Identification
Subsequently, other panels of the National Academy of Sciences, sup-

ported by OSHA and the National Institute for Occupational Safety and
Health (NIOSH) classified additional materials. Most have been incorpo-
rated in NFPA 497, which is now the reference document for hazardous
material classification. The number of materials now classified is too great
to list in the body of the NEC. Some of the additional materials classified
had been tested in the Westerberg apparatus. Some materials bear a dual
classification, based on whether explosionproof seals are installed in all
conduits entering an explosionproof enclosure. For example ethylene
oxide is classified B(C), meaning that if seals are not installed it is to be
treated as a Group B material, but if seals are installed it may be consid-
ered to be a Group C material. This practice was instituted because these
materials had an MESG typical of one group, and a pressure rise during
the test typical of another group. Many materials have been classified on
the basis of chemical and structural similarity to compounds whose prop-
erties are known.
Classification by the International Electrotechnical Commission

(IEC)
In Europe, material classification in regard to flameproof (explosionproof)

enclosures and intrinsic safety progressed independently for many years.
The U. K., for example, published two different lists of materials: one
related to the use of flameproof (explosionproof) enclosures and one
related to the use of intrinsic safety. By 1966 it was recognized that the two
lists were similar, and Slack and Woodhead published their correlation
between MESG, measured in a spherical test bomb, and the current
required to ignite the same materials in the British Break-Spark apparatus
used for early intrinsic safety testing.
During this same period, the IEC had become very active in producing
standards for apparatus for use in hazardous locations. IEC developed a
procedure, given in Publication 79-12, for classifying gases and vapors in
three groups: IIA, IIB, and IIC, which are similar to NEC Groups D, C, and
B, respectively. Acetylene is classified in Group IIC.
This method of classification recognized the correlation between the

MESG and the Minimum Igniting Current (MIC) in the IEC standard
intrinsic safety test apparatus. About 150 materials were classified. The
MESG values are those determined in the 20-ml vessel standardized by
IEC in Publication 79-1A. Some materials were classified based on chemi-
cal similarity. This method has also been adopted as a European standard.
The MIC is not used directly. Instead, the MIC of a material to be classified
is divided by the MIC of methane, and this MIC ratio is used for classifica-
tion. The rules for classification are:
Group MESG MIC Ratio
IIA > 0.9 mm > .08
IIB > 0.5 to > 0.9 mm 0.45 to 0.8
IIC < 0.5 mm < 0.45
Either the MESG or MIC ratio may be the basis for classification if:
• Group IIA: MESG > 0.9 or MIC ratio > 0.9
• Group IIB: 0.55 < MESG < 0.9 or 0.5 < MIC ratio < 0.8
• Group IIC MESG < 0.5 or MIC ratio < 0.45
Both measurements must be available if:

• 0.8 < MIC < 0.9
• 0.45 < MIC < 0.5
• 0.5 < MESG < 0.55
The need for both measurements in some cases is, of course, because the
MESG and MIC measurements are not exact, and the correlation between
them is, therefore, not perfect.
Comparison of U. S. and IEC Classifications—Results
Despite the disparate methods of classification used by the IEC and U.S.
material classification committees, the groupings of IEC Publication 79-12
and NFPA 497 are similar. Of the approximately 140 compounds classified
in both NFPA 497 and IEC Publication 79-12, only 19 differ in classifica-
tion. Of these, both committees had classified 6 on the basis of chemical
similarity. The data for the remaining 12 materials are shown in Table 3-4.
Classification MESG-mm MIC Ratio

Material
U.S. IEC U.S. IEC
Carbon monoxide C IIA .63 .94 1.06
Dipropyl ether C IIA .94
Acetaldehyde C IIA .43 .92
Ethyl mercaptan C IIA .9
Diethylamine C IIA .68
Triethylamine C IIA .53
Butadiene B(D) IIB .79 .79 .76
Acrylonitrile D IIB .81 .87
Ethylene oxide B(C) IIB .33 .59 .46
Propylene oxide B(C) IIB .53 .7 .59
Methyl acrylate D IIB .85
Ethyl acrylate D IIB .81 .86
Table 3-4 MESGs and MIC Ratio Data for 12 Selected Materials
It was noted above that gasoline and diethyl ether are the reference mate-
rials for the D-C and C-B boundaries, respectively, in the U. S. As
measured in the Westerberg apparatus, the MESG for gasoline is 0.74 mm.
The MESG for diethyl ether is 0.30 mm. The NEC classifications of the
materials in Table 3-4 are in accord with these values, taking into account
that the classification in the parenthesis is the one based on MESG. The
other is based on observed pressure rise. The IEC classifications are in
accord with the stated IEC rules. In those cases where MESG data were
not available for the NFPA classification work, the classification is based
on a judgment of chemical similarity to other classified materials. Where
Group D was assigned, it can be seen that, although the materials were
classified Group IIB in the IEC publication, the MESG values are close to
the IIA-IIB dividing line of 0.9 mm. Bearing in mind that materials above
and below any arbitrary dividing line in a continuum of values of MESG
are not significantly different; the two classifications can be accepted as
essentially the same. The most obvious differences are for those materials
that have been given a dual classification by the NFPA and those with
U. S. values of MESG relatively much smaller than that of gasoline, and
which were, therefore, classified Group C.
Comparison of NFPA and IEC Classifications—Methods
IEC
It can be seen from the rules given in IEC Publication 79-12 that the IEC
classification method comes close to meeting the requirements of a valid
classification system. The classification of a material depends, in principle,
on only one parameter, either MESG, or MIC, which is correlated with
MESG. This correlation is illustrated in Figure 3-4, which illustrates that
the correlation between the MESG and MIC ratio is not perfect. Therefore,
the classification rules require that both parameters must be known in the
region near the boundary values. This is a deviation from an ideal classifi-
cation system.
Figures 3-5 and 3-6 illustrate the correlation between the classifications of
materials and the MESG or MIC ratio. In Figure 3-6 there is some overlap
between Groups IIA and IIB that further illustrates the requirement that
both measurements must be made in the boundary regions.
Figure 3-4 Correlation Between MESG and MIC Ratio
Figure 3-5 IEC Classification by MESG

Figure 3-6 IEC Classification by MIC Ratio
The IEC classification method fulfills the requirement that the measured
parameters are not affected by some other characteristic of the gas or
vapor. The IEC 20-ml test apparatus for determining MESG was adopted
only after extensive testing to show that its volume lay in a range where
chamber volume no longer affects the relative order of MESGs of materials
being tested. Details are given in the Redeker report. All materials are
tested with the same ignition source location and strength. Secondary
characteristics, therefore, do not enter into the measurement of MESG. The
measurement of MIC for all materials is carried out in the standard IEC
test apparatus for intrinsically safe circuits defined in IEC Publication
79-3. Although the absolute values of minimum igniting current are
known to vary from laboratory to laboratory, the use of the MIC ratio
somewhat ameliorates these variations in the measurement.
The apparatus used to make the measurements from which IEC material
classifications are derived is directly related to the uses to which the classi-
fication is put. The apparatus for determining the MIC ratio is the same
apparatus used for testing circuits for intrinsic safety. The apparatus for
determining MESG is an analog of an explosionproof enclosure. The clas-
sification of a material is, therefore, useful for both intrinsic safety and
explosionproof types of protection.
NFPA
The rationale used to establish the NFPA groupings of gases and vapors
fails to meet the requirements of a valid classification system because the
materials have been grouped by considering two parameters: MESG and
pressure rise. It is well known that pressure rise in an explosion is not a
unique characteristic of an explosive gas or vapor. In fact, in spherical

bombs without venting of burned or unburned mixture, the maximum
pressure for almost all hydrocarbons is nearly the same value, about seven
atmospheres, even though the MESGs of the gases differ by a 25:1 ratio
and the time to maximum pressure varies by a factor of 10. When differ-
ences in explosion pressure among materials are measured they depend
on the specific geometry of the test vessel. How this relates to the Wester-
berg apparatus, how it is used, and how the data are interpreted is
explained below.
Westerberg Apparatus and MESG Values
Most importantly, the Westerberg apparatus design sometimes causes the

MESG values determined in it to be a function of some parameter of the
material being tested other than the MESG. Also, the Westerberg appara-
tus does not emulate an explosionproof enclosure because the gas passing
through the test gap does not exit into a volume much larger than the
enclosure, but into one of about the same size.
The Westerberg apparatus is described in U. L. Bulletin of Research No.

58. It consists of two cylindrical chambers separated by a wall where the
test gap is located. The test gap is 100 mm long. The width of the flanges
can be set to 9.5 mm, 19 mm, or 31.8 mm. The gap width is adjustable.
MESG is the value of the largest gap that will prevent an explosion in the
primary chamber from propagating into the secondary chamber. The gas
can be ignited near the flange gap, at the wall of the chamber opposite the
gap, or at the end of a 3-meter length of 11/2-inch pipe. The gas in the pri-
mary chamber may be quiescent, or it may be made turbulent by a fan
located near the wall opposite the test gap.
For many materials, the MESG values determined in the Westerberg appa-
ratus agree closely with the values determined in the IEC 20-ml apparatus,
in which the gases exit the ignition chamber into a volume about 10 times
larger. For other materials the MESG values are smaller than the IEC val-
ues. This effect has been studied by several investigators. They agree that
the small size of the receptor chamber in the Westerberg apparatus causes
precompression of the unburned gases in it, which sometimes causes
reported MESGs to differ from the IEC values. The fact that some values of
MESG are different would, in itself, not be crucial. However it is not com-
pletely understood why some values agree and some do not. There is no
certain way to know whether the data for a particular material has been
affected by the test method or not. Figure 3-7 illustrates that these effects
cause significant differences from the IEC values of MESG, and that these
differences are not regular and predictable. As Figure 3-7 shows, there is a
poor correlation.
Figure 3-7 Comparison of Westerberg and IEC MESGs
A supplement to Technical Report 58, dated May 1993, stated that reduc-
tion of the primary volume by 30% increased the diethyl ether MESG from
0.3 mm to 0.64 mm. The IEC value is 0.87 mm. This supplement also
reported that numerous tests since the original data were published estab-
lish a 0.23-mm MESG for Hydrogen, not 0.08 mm as originally reported.
The IEC value is 0.29 mm. Corrected values for other materials were not
published.
MESG values are only moderately affected by whether the ignition is at

the end of conduit, opposite the gap, at the gap, or whether the mixture is
turbulent or quiescent. The changes are in the direction that would be pre-
dicted if there were choked flow in the gap, as discussed by Thibault et al.
The method of ignition does, however, affect maximum pressure. Ignition
in turbulent mixtures produces a somewhat higher pressure rise than igni-
tion in quiescent mixtures because turbulence produces higher flame
speeds and consequent lower heat losses to chamber walls during the
shorter time to maximum pressure. The highest pressures result from igni-
tion at the end of the length of conduit, almost certainly caused by
pressure-piling. (See discussion in Chapter 5.)
Discussion of Rationale for NFPA Classifications
The uncertainty generated when using the method of classification

described above is illustrated by Figure 3-8. It is clear that MESG is not
well correlated with the group, to which a material is assigned, because
MESG is not the only criterion for classification. Were the materials with
dual classification grouped only in accordance with their MESGs, Figure
3-8 would become much more orderly. There would be a demarcation
between Groups B and C at 0.3 mm, but there would not be a clear demar-
cation between Groups C and D. The material with an MESG of 0.55 is
pentane, which has combustion properties similar to the other alkanes,
and, like them, is certainly a Group D material.
Further research might yield an understanding of the reasons why some

materials have unexpected values of MESG; however, it is questionable
that the research effort would be worthwhile since an international stan-
dard has already been defined.
Figure 3-8 NFPA Groups and MESG
Another question that must be addressed is whether the complication of

including maximum explosion pressure in the classification scheme is a
necessary one. When commercial explosionproof enclosures are tested for
conformance to a recognized standard, the ability of the enclosure to con-
tain the explosion is determined by using a test gas representative of the
group of materials with which the apparatus will be used. The explosion
pressure is measured and used to determine the value of the hydrostatic
proof-testing of the enclosure strength. In domestic practice, this is usually
four times the maximum measured explosion pressure.
One concern that has been expressed is that if the enclosure is later used in
the presence of a gas or vapor that develops a much higher explosion pres-
sure than the test gas the enclosure may not withstand the stress. Because
the highest pressures are generated in the Westerberg apparatus when
ignition is at the end of a length of conduit, the dual classification was
introduced. If the conduit is sealed, the classification based on MESG can
be used because ignition in the conduit will be confined to the conduit. If
the conduit is not sealed, the classification based on pressure rise must be
used. It is this author's opinion, for the reasons given below, that this con-
cern is not necessary and that no meaningful increase in hazard would be
experienced if classification in accordance with pressure rise were
discontinued.
The NEC mandates that all conduits be sealed if the enclosure contains
ignition-capable arcs or hot surfaces. The seals may be omitted in enclo-
sures that do not contain ignition sources if the conduit size is 2 inches
trade size or smaller. This seal prevents ignition in a conduit from causing
high pressure in an attached enclosure. More importantly, the seal pre-
vents transmission of an explosion in one enclosure through the conduit to
another enclosure, which might then experience abnormally high pres-
sure. Precompression of a mixture before ignition increases the explosion
pressure in proportion to the increase in initial absolute pressure.
Because conduit is unsealed only if the enclosure contains no normal

sources of ignition, an ignition source becomes available to ignite a gas or
vapor in the enclosure only when the enclosed apparatus fails. The proba-
bility of this happening is very low. In power equipment the duration of
the fault is likely to be short because of the action of overcurrent protective
devices. In communications and metering equipment the probability of
fault is low, and the probability that the fault is ignition-capable is lower
still.
The pressure rise measured in the Westerberg apparatus is obtained by

varying the mixture to find the mixture that gives the highest pressure.
The variation of pressure rise with concentration, like that of MESG, is
roughly parabolic. As the mixture deviates from the optimum concentra-
tion the pressure rise decreases. This is especially true of pressure rise
caused by transmission through conduit because the flame speed
decreases as the concentration departs from the optimum mixture. There-
fore losses to the conduit will increase, and the overpressure in an
attached enclosure will be lower. In industry it is highly unlikely that the
conduit and an enclosure would be filled with the optimum concentration
of gas or vapor. The butadiene concentration for maximum pressure rise
was reported by UL to be 4.2%. As discussed in Chapter 7, a cloud of buta-
diene of 100% concentration would have to surround the enclosure for
many hours for this concentration to be reached inside the enclosure. The
probability of this happening is nil.
The presence of a gas or vapor mixture in any concentration above the

lower explosive limit is of low probability. A Division 1 location is pre-
sumed to have a concentration above the lower explosive limit for 10 hr/yr
or more. Only a few percent of the electrical devices in a typical petro-
chemical plant are located in Division 1. Were spills and releases frequent
enough to greatly exceed the 10 hr/yr figure, a plant would likely be vio-
lating EPA or OSHA regulations.
Therefore, the probability of the concurrent presence of an ignition source,

a flammable cloud around the enclosure, and the optimum concentration
inside the enclosure is vanishingly small. In practice, pressures as high as
those measured in the Westerberg apparatus will not occur.
Author's Conclusions
(1) The historical basis for NFPA classification of gases and vapors in the
U. S. does not satisfy the requirements for a valid classification sys-
tem.
(2) There is no obvious way to use the MESG data from the Westerberg
apparatus as a basis for an unambiguous classification scheme.
(3) A standard method for classifying gases and vapors has been adopted
by the IEC and has been accepted in Europe and in many other indus-
trialized nations.
(4) NFPA should adopt the IEC method of classifying materials. It would
not be necessary to change the designation of the groups or to elimi-
nate Group A.
Benefits
(1) Adopting the standard IEC classification method would clarify the
basis for determining classifications and would provide an easily
understood rationale for classifying new materials. The U. S. Coast
Guard, other regulatory agencies, and industry need a rationale, by
which to make the first judgment about the classification of any new
material.
(2) Classification could utilize MIC data. Many laboratories have stan-
dard IEC intrinsic safety testing apparatus that would be less costly to
use than the Westerberg apparatus because less material must be
used.
(3) The standard 20-ml apparatus is smaller than the Westerberg appara-
tus, so it is less costly to use because the quantity of material that must
be available for testing is much smaller.
Dusts
Classification of Dusts—Historical Review
The 1965 edition of the NEC classified dusts as follows:

• Group E—atmospheres containing metal dust, including alumi-
num, magnesium, and their commercial alloys, as well as other
metals of similarly hazardous characteristics
• Group F—atmospheres containing carbon black, coal, or coke
dusts
• Group G—atmospheres containing flour, starch, or grain dusts
A location where dusts are present was to be classified in accordance with

the following summary.
• Class II, Division 1 locations—Locations in which
(1) combustible dust is or may be in suspension in the air continu-

ously, intermittently, or periodically under normal operating
conditions, in quantities sufficient to produce explosive or ignit-
able mixtures;
(2) mechanical failure or abnormal operation of machinery or equip-

ment might cause such mixtures to be produced, and might also
provide a source of ignition through simultaneous failure of elec-
trical equipment, operation of protection devices, or from other
causes; or
(3) dusts of an electrically conducting nature may be present.
• Class II, Division 2—locations in which combustible dust will not:
(1) normally be in suspension in the air; or
(2) be likely to be thrown into suspension by the normal operation of

equipment or apparatus, in quantities sufficient to produce
explosive or ignitable mixtures, but where:
(a) deposits or accumulations of such dust may be sufficient to

interfere with the safe dissipation of heat from electrical
equipment or apparatus; or
(b) such deposits or accumulations of dust on, in, or in the vicin-

ity of electrical equipment might be ignited by arcs, sparks or
burning material from such equipment.
It was further explained that electrically conducting dusts include dusts

from pulverized coal, coke and charcoal, as well as metallic dusts.
The essence of the definitions of Class II, Division 1 and Division 2 loca-
tions have not changed over the years. This discussion addresses the
method of dust grouping and the influence of a dust's conductivity on
area classification.
To a user of electrical equipment, the significance of the classification of a

location as Class II, Division 1 or Class II, Division 2 is the constraint it
places on equipment selection. Although pressurization or intrinsic safety
are also permitted, the NEC specifies that electrical equipment in Class II,
Division 1 locations be dust-ignition-proof (DIP). DIP construction has
tight fitting joints (to prevent entry of dust) and heavy walls (to prevent an
internal fault from raising the surface temperature to an ignition-capable
level). The code-enforcing authority usually requires DIP enclosures to be
listed or approved by UL, FM, or other approval agency. This subjects
them to rigorous testing and assessment for conformance to standards of
construction. In Class II, Division 2 locations the apparatus must be dust-
tight, which means that general purpose equipment may be used. In the
past this equipment was not necessarily approved by a third party for use
in Class II, Division 2 locations.
In 1971 Moodie, chairman of ISA Committee S12.10, which was preparing

recommended practices for electrical instruments in Class II locations,
observed that Groups E, F, and G are respectively, conducting, semi-con-
ducting, and non-conducting. He proposed that the low conductivity coal
and coke dusts should be classified as Group G dusts, rather than treated
the same as the metallic dusts in Group E. Though these ideas were incor-
porated in ISA S12.10, no changes to the NEC were made. S12.10 proposed
guidelines for dust classification as follows:
Resistivity Ohm-cm. Group
< 100 E
100 to 108 F
8
> 10 G
The Committee on Evaluation of Industrial Hazards of the National

Research Council was formed at the request of OSHA to classify flammable
gases, vapors, and dusts listed in the OSHA regulations. In its publications
it recommended that Moodie's proposals be adopted in the NEC. It further
recommended that Group F be discontinued and that all dusts could be
grouped according to their resistivity. As a result of subsequent proposals
from the EECA committee, the 1981 NEC adopted resistivity guidelines for
Groups E, F, and G. The 1984 NEC eliminated Group F. Conductive dust
was defined as having resistivity less than 105 ohm-cm.
In 1987 Group F was restored because of the concerns of manufacturers of

DIP motors. Motor designs that were formerly suitable for Group F could
not meet the surface temperature limits of Group G or the shaft clearances
of Group E. The maximum permissible surface temperature for Group G
motors is lower than that formerly permitted for Group F motors under
normal and overload conditions. Because UL testing of Group E motors is
with magnesium dust instead of the wheat and/or corn dust used for
Group G motors, shaft seals of Group E motors must be of a more compli-
cated design, adding cost. In the author's opinion, the issue of surface
temperature could have been dealt with by amending the surface temper-
ature limits in the NEC to make them relative to dust layer ignition
temperatures. This was not a practical alternative in the short time avail-
able to NEC Panel 14 for considering the proposal. (Amendments to
500-3(f) in the 1993 and 1996 Codes have taken care of this issue.) The
issue of shaft seals is not an NEC matter and has to be resolved by changes
to UL standards. The EECA committee was not consulted about these
amendments before they were proposed.
The 1990 NEC retained the resistivity guidelines despite proposals to

eliminate them. The impetus for these proposals was the action of regula-
tory authorities at some plants to insist that coal and similar dusts be
considered to be conductive if wet dusts met the Group E resistivity
guideline, even though wet dust is not a hazardous dust.
At present this issue is far from rational resolution. NFPA 497M-1991

eliminated resistivity as an aid to classification and states that Group F
dusts are not considered to be conductive. The NEC no longer gives resis-
tivity values, even to distinguish conductive from nonconductive dusts,
and quotes the classification from NFPA 499, which replaces NFPA 497B
and the Class II parts of NFPA 497M.
The Division 1 Classification for Conductive Dusts
Recent history of NEC treatment of dust classifications demonstrates

widespread lack of understanding of the intent of using resistivity as a
convenient way to classify dusts and ignorance of the fact that conductiv-
ity is relevant only when the dust is in a form that can become a
flammable cloud or hazardous layer. It has been argued correctly that
resistivity is not related to the combustion properties of dusts. Resistivity
entered the picture only because the NEC says that presence of a conduc-
tive dust makes a location Division 1 and because resistivity is a
convenient property to measure. It is time to rethink that position. Is it a
necessary requirement to ensure safety? Some agree with the author that it
is not. A proposal to eliminate the requirement was submitted for consid-
eration during preparation of both the 1990 NEC and the 1993 NEC, but it
was not adopted. The 1996 NEC continues to mandate Division 1 classifi-
cation if conductive dusts are present.
The rationale commonly stated for requiring that a location be defined as

Division 1 if conductive dust is present is that if conductive dust bridges
bare live parts, it can cause an arc which will ignite itself. A nonconductive
dust requires that some other event provide the ignition source. The
alleged dual hazard of conductive dusts, therefore, mandates special pro-
tective measures.
In both Division 1 and Division 2 locations the primary method of pre-

venting ignition of a dust cloud or layer is to prevent contact between the
dust and the live or hot parts of electrical apparatus. One method is to pro-
vide an enclosure that prevents the admission of dust, that is, DIP in
Division 1, and dust-tight in Division 2. Another method is to pressurize
the enclosure in accordance with the requirements of NFPA 496. The use
of intrinsically safe apparatus in Division 1, or nonincendive apparatus in
Division 2, also reduces hazard. These methods limit the energy that may
be released to values that will not ignite the dust, whether it is a conduc-
tive or nonconductive dust.
Consider first the mechanics of dust entering an enclosure. If dust accu-

mulates on the joints of the enclosure it will try to enter because vibration
causes the dust to enter gaps in the joints; or, if there is large enough
cycling of internal temperature, the alternating pressure difference across
the joint will try to move material through the joint. However, fine dust
tends to agglomerate and large particles cannot move far. The joint will
clog with dust and tend to seal itself. Some dust may enter, of course,
before the joint seals itself, but will it be enough dust to pose an explosion
hazard? The first dust entering the enclosure is most likely to come to rest
on the internal surfaces of the enclosure because there is no mechanism to
spread the dust around on the enclosed apparatus. But suppose that the
dust does bridge bare live parts. If the dust is conductive it will likely
cause malfunction of the equipment which will be dealt with by the over-
current protection, or in the case of low power apparatus, by replacement
or repair of the apparatus. This will happen long before the amount of
dust can support combustion, as combustion requires an appreciable layer
thickness.
The difficulty dusts meet in entering an enclosure is evidenced by the UL

requirement that the permitted diametral clearance for motor shafts
ranges from 0.01 in. to 0.022 in. (0.25–0.55 mm) as joint lengths range from
1/2 to 1-1/2 in. (12.5–37 mm). After six heating and cooling cycles of the
motor heated to the equivalent of full load operation, lasting at least 30
hours, and while the motor is exposed to a swirling cloud of dust, which
has been sleved to ensure fineness, there shall be no visible entrance of
dust. Equipment without rotating shafts will have flat or threaded joints,
which are much tighter. Flat joints are intended to have clearance of less
than 0.003 in. (0.075 mm) Magnesium dust for testing Group E devices has
particles as large as 0.01 in. (0.25 mm). Forty-four percent will be between
0.003 and 0.006 in. (0.075 and 0.15 mm); 22% are smaller than 0.003 in.
(0.075 mm). The wheat or corn dust for testing Group G devices has parti-
cles smaller than 0.059 in. (1.5 mm).
Passage of a significant quantity of dust through a flat joint is difficult to

imagine, and passage through the diametral clearance of a rotating motor
shaft is certainly not easy, especially when one considers that the pressure
gradient is outward on the heating cycle when the motor is running. One
might argue that if the equipment is a rotating machine then conditions
are different because of the turbulence inside the machine. The author
agrees to the extent that any dust that enters the enclosure is more likely to
be distributed throughout the enclosure, but the rest of the rationale is still
applicable. Dust will still tend to clog gaps in the enclosure, and any con-
ductive dust entering will likely cause malfunction long before there is
sufficient dust to cause an explosion.
Another argument against the mandated Division 1 for conductive dusts

is that it results in an irrational area classification. It has been argued that
it is nonsensical that the presence of any conductive dust, no matter how
small the amount, represents the danger presumed by the NEC Division 1
classification. Even if there is some amount that justifies a Division 1 clas-
sification, there must be an amount that justifies only a Division 2
classification. Where does one now draw the line between a Division 1
location and a nonhazardous location? The 1996 NEC says, “may be
present in hazardous quantities,” but this statement is not enlightening.
Another way to view the issue of hazard where conductive dusts are
present is to recognize that, if electrical apparatus is being used in a plant
that is handling conductive dust, as a practical matter, the plant will not
operate for very long unless the dust is excluded from the equipment. If
the dust is excluded from the equipment to maintain function, there will
not be enough dust to pose an explosion hazard.
Author's Conclusions
The author concludes that there is not sufficient evidence to justify con-
tinuing the mandate to classify a location Division 1 if conductive dust is
present. He believes that the NEC should discontinue this requirement,
and that all dusty locations should be classified using the same rules.
If conductivity is not an issue in area classification, it is no longer an issue

with regard to grouping of dusts, and the arguments about the relevancy of
resistivity as a classification tool can cease. If there are reasons for motors
exposed to metallic dust to be different from motors exposed to coal dust,
this becomes an issue for the standards for motors, not for the NEC.
References
NFPA 70, National Electrical Code, National Fire Protection Association,
Quincy, MA.
NFPA 497, “Recommended Practice for Classification of Flammable Liq-

uids, Gases or Vapors and of Hazardous (Classified) Locations for
Electrical Installations in Chemical Process Areas,” National Fire Protec-
tion Association, Quincy, MA, 1997.
Bulletin of Research No. 58, “An Investigation of Fifteen Flammable Gases

or Vapors With Respect to Explosionproof Electrical Equipment,” Under-
writers Laboratories, Northbrook, IL, 1969.
“Fire Hazard Classification of Chemical Vapors Relative to Explosion-

proof Electrical Equipment,” Report IV National Academy of Sciences,
Washington, DC, 1975.
NMAB 353-5, “Classification of Gases, Liquids, and Volatile Solids Rela-

tive to Explosionproof Electrical Equipment,” National Academy Press
Washington, DC, 1982.
Bulletin of Research No. 58A, “An Investigation of Additional Flammable

Gases or Vapors with Respect to Explosionproof Electrical Equipment,”
Underwriters Laboratories, Northbrook, IL, 1976.
Bulletin of Research No. 58B, “An Investigation of Additional Flammable

Gases or Vapors with Respect to Explosionproof Electrical Equipment,”
Underwriters Laboratories, Northbrook, IL, 1977.
Slack, C., and D. W. Woodhead, “Correlation of Ignitabilities of Gases and

Vapors by a Break Spark and at a Flange Gap,” Proceedings IEE, 113 (Feb-
ruary 1966), pp. 297–301.
Publication 79-12, “Electrical Apparatus for Explosive Gas Atmospheres,”

Part 12: Classification of Mixtures of Gases or Vapours with Air According
to Their Maximum Experimental Safe Gaps and Minimum Igniting Cur-
rents, International Electrotechnical Commission, Geneva, 1978.
Publication 79-1A, “Electrical Apparatus for Explosive Gas Atmospheres,”

Part 1, Construction and Test of Flameproof Enclosures of Electrical Appa-
ratus, First supplement: Appendix D: Method of Test for Ascertainment of
Maximum Experimental Safe Gap, International Electrotechnical Commis-
sion, Geneva, 1975.
Redeker, T., “Classification of Flammable Gases and Vapours by the

Flameproof Safe Gap and the Incendivity of Electrical Sparks,” Report
W-18, Physikalisch-Technische Bundesanstalt, Braunschweig, 1981.
Strehlow, R. A., Nicholls, J. A., Magison, E. C., and P. J. Schram, “An

Investigation of the Maximum Experimental Safe Gap Anomaly,” Journal
of Hazardous Materials, 3 (1979), pp. 1–15.
Brown, G. K., Dainty, E. D., and A. D'Aoust, “The Variation of Maximum

Experimental Safe Gap with Secondary Explosion Chamber Relief for
Ether-Air Mixtures,” CEAL No. 242, Department of Energy, Mines and
Resources, Ottawa, 1971.
Phillips, H., “Differences Between Determinations of Maximum Experi-

mental Safe Gap in Europe and U. S. A.,” Journal of Hazardous Materials, 4
(1981), pp. 245–256.
Phillips, H., “Maximum Explosion Pressure in Flameproof Enclosures:

The Effects of the Vessel and the Ambient Temperature,” Journal of Hazard-
ous Materials, 8 (1984), pp. 251–259.
Thibault, P., Liu, Y. K., Chan, C., Lee, J. H., Knystautas, R., Guirao, C.,
Hjertager, B., and K. Fuhre, “Transmission of an Explosion through an
Orifice,” Nineteenth Symposium (International) on Combustion, The
Combustion Institute, 1982, pp. 599–606.
Simons, C. F. E., Kogeler, W. J., and P. C. Bijl, “Safety of Electrical Instru-

ments in the Oil Industry,” Paper No. 4001M, IEE Conference Report
Series No. f3, Institution of Electrical Engineers, London, November 1962.
NMAB-448, “The Explosion Hazard Classification of Gases and Dusts Rel-

ative to Use of Electrical Equipment,” National Academy Press,
Washington D. C., 1988.
Moodie, T. W., “Class II Dust Classification,” Electrical Safety Practices,

Monograph No. 113, Instrument Society of America, Pittsburgh, 1972, pp.
87–89.
NMAB 353-2, “Test Equipment For Use in Determining Classifications of

Combustible Dusts,” National Academy of Sciences. Washington DC,
1979.
NMAB 353-3, “Classification of Combustible Dusts in Accordance with

The National Electrical Code,” National Academy of Sciences, Washing-
ton, DC, 1980.
NMAB 353-4, “Classification of Dusts Relative to Electrical Equipment in

Class II Hazardous Locations,” National Academy of Sciences, Washing-
ton DC, 1982.
NMAB 353-6, “Rationale for Classification of Combustible Gases, Vapors,

and Dusts with Reference to the National Electrical Code,” National Acad-
emy of Sciences, Washington, DC, 1982.
McCarron, R., “Considerations in Area Classification,” Electrical Safety

Practices, ISA Monograph No. 110, Instrument Society of America, Pitts-
burgh, 1965, pp. 4–15.
“Recommended Practice for Explosion-Protected Electrical Installations in

General Industries,” Ministry of Labor, Research Institute of Industrial
Safety, Japan.
“Vorschriften fuer schlagwettergeschutzte/explosionsgeschutzte Betrieb-

smittel,” VDE 0170/0170d/2.65, Verband Deutscher Elekrotekniker E. V.,
1965.
“Electrical Apparatus and Associated Equipment for Use in Explosive

Atmospheres of Gas or Vapour Other Than Mining Applications,” British
Standard Code of Practice CP1003, Part 3: 1967, British Standards Institu-
tion, London.
“Guide to the Selection of Electrical Equipment for Use in Division 2

Areas;” BS4137:1967, London.
Pinder, K., “Considerations in Classifying Areas for Electrical Installation

in Chemical Processes,” Electrical Safety Practices, ISA Monograph No.
111, Instrument Society of America, Pittsburgh, 1967, pp. 3–43.
Westerberg, W. C., “Classification of Chemical Atmospheres,” Electrical

Safety Practices, ISA Monograph No. 112, Instrument Society of America,
Pittsburgh, 1967.
Burgoyne, J. H., “Division Two As An Objective,” Electrical Safety Prac-

tices, ISA Monograph No. 113, Instrument Society of America, Pittsburgh,
1972, pp. 90–92.
Carhart, H. W., “Classification of Chemicals Relative to Explosionproof

Electrical Equipment,” Electrical Safety Practices, ISA Monograph No.
113, Instrument Society of America, Pittsburgh, 1972, pp. 60–73.
Hilado, C., and S. W. Clark, “Autoignition Temperatures of Organic

Chemicals,” Chemical Engineering (September 4, 1972), pp. 75–80.
Hunt, G. O., “Outside Division Two,” Electrical Safety Practices, ISA

Monograph No. 113, Instrument Society of America, Pittsburgh, 1972, pp.
60–73.
Schram, P. J., “Ignition Temperature of Gases and Vapors as Applied to

Electrical Equipment for Use in Hazardous Locations,” Electrical Safety
Practices, ISA Monograph No. 113, Instrument Society of America, Pitts-
burgh, 1972, pp. 47–59.
Woinsky, S. G., “Predicting Flammable Material Classifications,” Chemical

Engineering (November 27, 1972), pp. 81–86.
Chamlee, R. D., and S. G. Woinsky, “Predicting Flammable Material Clas-

sifications for the Selection of Electrical Equipment,” IEEE Transactions on
Industry Applications, Vol. 1A-10, No. 2 (March/April 1974), pp. 288–298.
Buschart, R. J., “An Analytical Approach to Electrical-Area Classification,”

IEE Conference Publication No. 134, London, 1975, pp. 195–207.
Harris, N. C., “Risk Assessment,” IEE Conference Publication No. 134,

London, 1975, pp. 132–137.
Freytag, H. H., “Physiological Identification of Explosive Atmospheres,”

Ex Magazine, No. 2, (June 1976), pp. 29–30.
“Classification of Locations for Electrical Installations in Petroleum Refin-

eries,” API RP500A, American Petroleum Institute, Washington.
“Recommended Practice for Classification of Locations for Electrical

Installations at Drilling Rigs and Production Facilities on Land and on
Marine Fixed and Mobile Platforms,” API RP 500B, American Petroleum
Institute, Washington.
IEC Publication 79-10, “Electrical Apparatus for Explosive Gas Atmo-

spheres,” Part 10: Classification of Hazardous Areas.
“Classification of Gas Utility Areas For Electrical Installations,” XF0277,

American Gas Association, 1977.
Grove, J.R., “The Measurement of Quenching Diameters and Their Rela-

tion to the Flameproof Grouping of Gases and Vapours,” I. Chem.E.
Symposium Series No. 25, pp. 51–54, Instn. Chem Engrs, London, 1968.
NFPA 499, “Recommended Practice for Classification of Combustible

Dusts and of Class II Hazardous (Classified) Locations for Electrical Instal-
lations in Chemical Processing Plants,” National Fire Protection
Association, Quincy, 1997.
CHAPTER 4
PRACTICE AND PRINCIPLES OF
HAZARD REDUCTION PRACTICE
United States
In the United States the governing standard, adopted as law in most juris-
dictions, is the National Electrical Code. Table 4-1 summarizes NEC
requirements for installations in classified locations. The reader should,
however, consult the Code itself for the details.
Location Wiring Methods Type of Apparatus

CLASS I Threaded rigid or steel intermediate Explosionproof.
DIVISION 1 metal conduit or mineral insulated (Type Pressurized for Division 1.
MI) cable with approved fittings. All boxes Intrinsically safe apparatus.
shall be explosionproof. Fittings and Surface temperature (T code) must not
joints shall be threaded, five threads mini- exceed ignition temperature of the gas or
mum, for cable and conduit entry. Note: vapor likely to be present.
Wiring of intrinsically safe systems need
only meet requirements for ordinary (non-
hazardous) locations.
All conduit 2 in. and larger and conduit or
cable entering enclosures containing igni-
tion-capable arcs, sparks, or hot surfaces
must be sealed within 18 in. of the enclo-
sure. Cables not capable of transmitting
gases and vapors need not be sealed.
All conduits leaving the Division 1 loca-
tion must be sealed.
Rigid nonmetallic conduit may be used if
it is in a concrete envelope at least 2 in.
thick and buried under at least 2 ft of
earth. Connections to above-ground
raceway must be made with rigid metal
conduit or threaded steel intermediate
metal conduit at least 2 ft from emer-
gence above ground. A grounding con-
ductor must be included to maintain
continuity of raceway grounding and to
provide for grounding non-current carry-
ing metal parts.
100 Practice and Principles of Hazard Reduction Practice

In industrial establishments with
restricted public access, and where only
qualified persons will service the installa-
tion, MC or ITC cable approved for Class
I, Division 1 locations with a gas-tight,
continuous corrugated aluminum sheath
and an overall polymeric jacket; separate
grounding conductors (per Art. 250-95)
and listed termination fittings is permitted.
CLASS I Threaded rigid or steel intermediate con- Any suitable for Division 1.
DIVISION 2 duit, mineral insulated cable, medium Explosionproof if apparatus contains igni-
voltage cable, metallic or nonmetallic tion-capable arcs or hot surfaces. Gen-
sheathed cable (Types MI, MC, MV), or eral purpose enclosure if there are no
enclosed gasketed wireways or busways. ignition-capable arcs or hot surfaces (i.e.,
PLTC cable may be installed per Art. 725 arcs are in nonincendive components or
or ITC may be installed in trays, race- circuits, or in sealed components).
ways, on messenger wire, or buried if Pressurized for Division 2.
listed for burial. Surface temperature not to exceed the
Boxes and fittings explosionproof only if ignition temperature of the gas or vapor
they contain ignition sources. which may be present.
Where flexibility is required, as at motor

terminals, flexible metal fittings, flexible
metal conduit with approved fittings, liq-
uid-tight flexible metal conduit with
approved fittings, or flexible cord
approved for extra-hard usage and pro-
vided with approved bushed fittings shall
be used. An additional conductor for
grounding shall be included in the flexible
cord.
Note: Wiring in circuits which under nor-
mal conditions do not release enough
energy, when opened, shorted, or
grounded, to ignite gases or vapors that
may be present need meet only ordinary
location requirements.
Seals: For explosionproof enclosures,
same as for Division 1, and in all conduits
or cables leaving Division 2 and entering
nonhazardous locations. Seals are not
required if the cable cannot transmit gas
or vapor or if the enclosure will not be
subjected to more than 6-in. water inter-
nal pressure.
Practice and Principles of Hazard Reduction Practice 101

Conduit seals at the exit from division 2
are not required if the conduit run con-
tains no unions, etc. and passes com-
pletely through the Class I, Division 2
location and terminates in a nonhazard-
ous location. No seal is required where
conduit passing from a Class I, Division 2
location into an unclassified location tran-
sitions to cable tray, cablebus, ventilated
busway, type MI cable, or open wiring,
and it does not terminate in an enclosure
containing an ignition source in normal
operation.
CLASS II Threaded rigid or intermediate metal con- Dust-ignition-proof (heavy dust-tight con-
DIVISION 1 duit or mineral insulated cable (type MI) struction), pressurized or intrinsically
with approved fittings. safe.
Fittings and boxes shall have threaded Maximum surface temperature: (Note 1)
bosses for connection of conduit or cable Group E - 200°C
fittings and shall have no openings Group F - 150/200°C
through which dust may enter. If taps, Group G - 120/165°C
joints or terminal connections are made
in a box, it must be approved for Class II
locations. (dust-ignition-proof)
Flexible metallic or nonmetallic conduits
or flexible cord approved for extra-hard
usage may be used, if necessary.
In industrial establishments with limited
public access, Type MC cable may be
used if listed for Class II, Division 1 loca-
tions.
CLASS II Rigid or intermediate metal conduit, elec- Dust-tight enclosure (can be gasketed
DIVISION 2 trical metallic tubing, or dust-tight wire- sheet metal) Type tests are less onerous
ways; mineral insulated cable, Type MI or than for dust-ignition proof.
sheathed cable Type MC with approved General purpose enclosure if apparatus
termination fittings, or Type PLTC or Type is incapable of causing ignition by arcs or
ITC cable in cable trays, or Type MC or hole surfaces.
TC cable installed in ladder, ventilated Surface temperature limit - 120°C max.
trough, or ventilated channel cable trays
in a single layer, with a space between
adjacent cables of not less than the larger
cable diameter.
Wiring which can't ignite dust by opening,
shorting, or grounding in normal opera-
tion may use ordinary location wiring
practice.

CLASS III Rigid metallic or nonmetallic conduit, Tight covers, no holes.
DIVISION 1 Type MI, or MC cable, intermediate metal Rotating machinery must be of totally
conduit, or dust-tight wireways. enclosed type.
Boxes and fittings must be dust-tight. Intrinsically safe apparatus.
Surface temperature limits - 120/165°C
(Note 1).
CLASS III Essentially same as Division 1. Essentially same as Division 1.
DIVISION 2
Note 1. Lower temperature applies to apparatus subject to overload (motors, transformers) in nor-
mal operating conditions. Higher temperature applies under overload conditions or to apparatus
not subject to overload.
Table 4-1 NEC Wiring Methods and Types of Protection
Canada
The requirements of the Canadian Electrical Code, Part 1, are similar in
intent and philosophy to those summarized above, but there are impor-
tant differences. It is to be hoped that many of these differences will be
eliminated as the result of a harmonization movement begun soon after
the Free Trade Pact was signed by the United States and Canada under the
auspices of the NEC/CEC Bi-National Correlating Committee.
In Class I, Division 1 locations, only rigid metal conduit is recognized.

Intermediate metal conduit is not permitted. In addition to MI cable, the
CEC permits cables approved for the particular hazardous location. The
CEC does not include all the relaxations for nonsparking apparatus in
Division 2 locations found in the NEC. Although the safety of nonspark-
ing equipment in Division 2 locations is recognized, squirrel cage motors
may be of a nonexplosionproof type, but process control instrumentation
must be approved for use in Division 2. In the CEC “approved” means
“certified by the laboratories of the Canadian Standards Association.”
Countries Following IEC or CENELEC Standards

Most countries whose national regulations have been derived from IEC or
CENELEC standards permit the use of electrical apparatus in locations
that are hazardous because of possible presence of flammable gases or
vapors in accordance with Figure 4-1. In locations made hazardous by the
presence of combustible dusts the apparatus requirements are as follows:
• Zone 21 and Zone 22 with conductive dust⎯IP6X plus some
additional restrictions on plastics, 5 straight threads or 3 tapered
threads on cable or conduit entry, etc.
• Zone 22⎯IP5X and fewer special requirements than apparatus
for Zone 21.
Figure 4-1 Use of Protection Techniques-IEC Nomenclature
The Types of Protection

Ex d
Type of Protection d is flameproof construction, similar in principle and

practice to explosionproof construction. See Chapter 5 for more details.
Ex e
Type of Protection e is increased safety construction. This construction is

used for motors, generators, transformers, terminal boxes, and wiring. The
principle of this technique is to ensure that mechanical construction, insu-
lation systems, and connection facilities are constrained to particularly
robust standards so that failure in a way that will make the apparatus an
ignition source is extremely unlikely. In rotating machinery particular
attention is paid to coordinating overcurrent and overtemperature protec-
tion systems with the construction of the apparatus to avoid a winding's
reaching the ignition temperature of the gas or vapor to which it is
exposed. See Chapter 8.
Ex i
Type of Protection i is intrinsic safety. Requirements of the IEC and CEN-

ELEC standards for ia intrinsic safety are similar to each other and to the
requirements now in effect in the United States. There are some differ-
ences resulting from the fact that there is yet no equivalent to ib intrinsic
safety for use in Zone 1 in North America. Other differences arise from
differences of opinion between European and American experts on the
safety factors needed to ensure safety.
The principle difference between ia intrinsic safety and ib intrinsic safety is

that safety is judged after two faults in ia intrinsic safety and after one fault
in ib intrinsic safety. EN 50154, “Electrical Installations in potentially
Explosive Gas Atmospheres (other than mines),” is a standard for applica-
tion in the countries recognizing CENELEC standards. It includes
intrinsically safe systems, category ia, in accordance with EN 50039 as
acceptable for installation in Zone 0, and category ib systems for Zone 1.
See Chapter 10.
Ex m
Type of Protection m is encapsulation for Zone 1. The generic technique is

similar to encapsulation used in conjunction with intrinsic safety, but
much more stringent controls are imposed because encapsulated circuits
and devices may be inherently ignition-capable in normal operation. See
Chapter 7.
Ex n
Type of Protection n has only the status of an IEC Report, which does not
have the authority of a standard. At present there is no equivalent CEN-
ELEC document. Apparatus covered by EX n includes nonsparking
industrial apparatus, sealed components, nonincendive components,
enclosed break components, nonincendive circuits, and restricted breath-
ing. Restricted breathing, not recognized in North America, is discussed in
Chapter 7, where the reader will also find more extensive discussion of
sealed components. Enclosed break components are treated in Chapter 6.
Nonincendive circuits and components are discussed in Chapter 10. The
IEC report is patterned after British Standard BS 4683, Part 3:1972, now
revised as BS 6941:1988. Although most requirements are similar to those
of ISA S12.12, there are significant differences. The most significant differ-
ence for instrumentation and other low power equipment is the treatment
in BS 6941 of creepage and clearances and terminations. Apparatus that
conforms to BS 6941 is called Type N apparatus.
Ex p
Type of Protection p is pressurization, similar in principle to pressuriza-

tion as standardized in NFPA 496 (see Chapter 6). In North America
pressurization has been considered suitable for use in a Division 1 loca-
tion, which includes locations that would be classified Zone 0. The
European view has been that only ia intrinsic safety is suitable for Zone 0,
although dual protection, such as nonsparking apparatus in an explosion-
proof enclosure is acceptable to many experts. Type ia intrinsic safety
requires three faults before it can be an ignition source. The equivalent
pressurization system would have nonsparking equipment inside the
enclosure and an interlock between pressurization failure and the electri-
cal supply to the enclosed equipment.
Ex s
Ex s does not actually refer to a type of protection. It is a symbol indicating

that apparatus is constructed in a fashion that is considered safe but does
not conform to a recognized standard. At one time this symbol was used to
indicate use of more than one protection technique, but in recent years the
practice has been to mark such apparatus with the symbols of all the types
of protection which have been incorporated in it (e.g., EEx d ia IIC T5).
Figure 4-1 contains no reference to Type of Protection o, oil immersion.

This protection technique is similar to oil immersion as standardized in
the United States, but it is not included in Figure 4-1 because, in most
countries, as in the United States, oil immersion is no longer frequently
used, though IEC and most countries have standardized requirements.
Likewise, Type of Protection q, sand filling, is not listed because it is not a

common technique. This technique uses sand to enclose equipment so that
any incipient explosion is quenched and prevented from spreading. It is
the author's understanding that sand filling is sometimes used as explosion
protection for line voltage circuitry and components in portable apparatus
where flameproofing would make the apparatus too heavy. ISA is prepar-
ing versions of the CENELEC/IEC standards for these types of protection,
which are modified to be consistent with North American practices.
Wiring Methods
Wiring methods used in countries whose standards and regulations are

based on IEC or CENELEC standards include the use of both cable and
conduit. The wiring method used in any installation probably depends
most on the history of practices and standards in that country. Though
conduit may be a recognized method, for example, it may not be often
used in some countries because use of conduit was formerly forbidden, or

at least deprecated.
IEC Publication 79-14, a Report, “Electrical Installations in Explosive Gas

Atmospheres (Other Than Mines),” includes recommendations for selec-
tion of a wiring method, as discussed below.
For All Hazardous Areas
Nonsheathed insulated single core cables should not be used for live con-
ductors, except in conduit systems or inside enclosures or panel boards.
Connections to apparatus shall be made in accordance with the relevant
type of protection.
Unused cable or conduit openings shall be sealed with closures relevant to

the type of protection.
Cables and conduits shall be sealed if necessary to prevent the passage of

gases or liquids. Wiring systems passing from one area to another shall
prevent the passage of gases, liquids or vapors, and shall prevent their col-
lection in trenches.
Metal, thermoplastic or elastomeric sheathed cables, including mineral

insulated cables, may be used for fixed wiring. Cables with folded metal
sheath or cables with steel wire braiding shall be used only if they have an
impervious nonmetallic sheath.
For portable or transportable apparatus with rated voltage not exceeding

1000 V ac between phases (600 V ac to ground) or 1500 V dc between poles
(900 V dc to ground) the supply cable shall have a heavy tough rubber or
polychloroprene sheath, or an equally robust construction. The grounding
(protective) conductor, where required, shall be insulated like the other
conductors, and shall be inside the supply cable sheath, unless the ground-
ing conductor is a screen. Portable apparatus, rated not more than 250 V to
earth and 6A, may be connected by ordinary tough rubber or polychloro-
prene cables, or equal. They shall have a minimum cross section of 1.5
mm2. These cables are not permitted for hand lamps, foot switches, or bar-
rel pumps, or similar apparatus that is exposed to heavy mechanical
stresses. Cables that are fitted with supports, and portable telecommunica-
tion apparatus may have a cross-section of 0.7 mm2. Flexible cables in
hazardous areas may be ordinary tough or heavy tough rubber or poly-
chloroprene sheathed flexible cables or equivalent plastic insulated cables.
Outer sheaths of cables that are not laid in earth or sand-filled conduits
shall be flame retardant, unless otherwise protected from fire.
Conduit Systems
The rules for conduit systems are similar to those of North American
practice.
Mixed Wiring Systems
Wiring systems consisting of both conduit and cable may be linked if fit-
tings conforming to the requirements of each system are employed.
Zone 0 Locations
No specific guidance is given for wiring in Zone 0. Intrinsically safe wiring

must meet Zone 1 requirements, plus, perhaps, additional precautions
which are unspecified. What is intended by this clause is to give room for
requirements such as galvanic isolation of intrinsically safe circuits in
Zone 0, so that there is no direct conductive path into Zone 0, and the
installation of strong defenses against surges cause by lightning or power
system faults.
Wiring that is not intrinsically safe shall be protected mechanically, by

ground fault interrupters, or other electrical means, or by control of their
environment to ensure that damage to insulation is not a cause of the cable
becoming an ignition source. Conduit systems suitable for Zone 1 are suit-
able for installation in Zone 0.
Zone 1 Locations
Intrinsically safe wiring may be installed either isolated from earth,

grounded at one point (if grounding is required for function or safety) or
connected at one point to the potential equalization conductor. The pur-
pose of the potential equalization conductor, which is widely used in
Germany, is to bond all apparatus and structures together to minimize
potential differences within a plant.
More than one grounding connection is permitted if the system is divided

into galvanically isolated segments such there is not more than one
ground in any segment.
Other requirements for intrinsically safe wiring are as follows:

• observing separation from non-intrinsically safe circuits
• guarding against intrusion of other types of circuits
• observing the maximum voltage limitations in apparatus con-
nected to intrinsic safety barriers
• ensuring that the capacitance and inductance of the system con-

forms to the limits imposed by the barrier ratings
No additional requirements are suggested for wiring in Zone 1 locations

other than those noted for hazardous locations in general.
Zone 2 Locations
No requirements for wiring are specified beyond those for hazardous

locations in general. EN 50154 requirements are similar to those of IEC 79-
14, but given in somewhat greater detail. Conduit systems are not cov-
ered. Use of conduit and mixed systems shall be in accordance with
national regulations.
Principles
This part of the chapter focuses on the philosophy and methods of reduc-
ing the risk of using electrical apparatus in locations where combustible
material may be present. The primary aim of this presentation is to estab-
lish a conceptual foundation for more detailed consideration of specific
means of hazard reduction in following chapters. Secondary goals are to
provide an organized conceptual basis for establishing safe requirements
for a special situation not easily handled by conventional protective tech-
niques and to provide a firm basis in principle for assessing the validity of
proposed safety requirements.
In most situations “reducing explosion hazard” means reducing the prob-

ability of significant property damage or loss of life as the result of an
explosion. This book emphasizes the relationship between electrical appa-
ratus and safety, so the aim of hazard reduction can also be stated: ensure
that adding electrical apparatus to an installation does not significantly
raise the probability of explosion-caused damage or personal injury. If a
combustible material is present, there will eventually be a fire or explo-
sion. Even if no humans are present, even if no electrical equipment is
used, at some time a source of ignition will become available, and there
will be an ignition. The probability of fire or explosion is never absolutely
zero if a combustible material is frequently present. Similarly the explo-
sion probability is never zero if an ignition source is continuously or
frequently present. Eventually combustible material will also be present
and there will be an explosion. Lightning may strike a barn and ignite the
contents. A propane tank may leak and the vapors may run into a base-
ment and be ignited by the pilot flame of a gas appliance.
Hazard reduction is therefore not an exercise in absolutes. It is an exercise

in low probabilities, especially low incremental probabilities. The safe
electrical installation is that one that does not significantly raise the proba-
bility of fire or explosion above that which existed before the equipment
was installed.
This focus on low probabilities is an implicit statement that in most situa-

tions the risk of fire or explosion must be very low to be acceptable.
Risk is usually defined in terms of the probability of an event taken

together with the consequences of the event. In simplest terms “risk” is
defined as the product of probability and cost. Cost is used here in the
general sense, including not only monetary cost but loss of life or other
quantity. But the relationship need not be a simple multiplicative one. The
result of an event may be defined by a curve, for example, expected fre-
quency of injury as a function of age. An event might have different
perceived risks for individuals of different age if injury frequency is a
strong function of age. Similarly, two events with the same probability
might have different perceived risks for individuals of different ages
because the curves of injury frequency versus age for the two events had
different shapes.
Another important aspect of risk is that the acceptable level of risk is not a
constant, but a value determined by culture, personal history, mental
state, and many other factors that are not easily quantified. A race car
driver willingly accepts a risk of injury or death that is much higher than
that which is considered acceptable for the average citizen traveling the
highway. The preparation of standards and regulations that intend to
reduce risk is very difficult if all the parties interested in the document do
not accept the same level of risk as being reasonable.
There is no standard for the level of risk that is acceptable in the industries
where explosion hazards are of concern. But, in our culture the risk of
death in an industrial plant is expected to be very low. The cost of replac-
ing a plant that has been destroyed by an explosion and fire is high. It is
accepted that the total risk must be maintained at a very low level. There-
fore it is reasonable to discuss hazard reduction in terms of probability
alone.
There are many approaches to reducing the risk of using electrical equip-
ment in locations where flammable materials may be present. The
diagram in Figure 4-2 illustrates the logical relationships among the many
methods. Not all the methods are commonly used, especially in instru-
mentation. Some are listed only for completeness.
Figure 4-2 Methods of Reducing Explosion Hazard
Methods Based on Permitting Ignition
The methods based on allowing ignition to occur depend on ignition

being so well controlled that combustion does not cause damage. The use
of an explosionproof housing to contain an explosion and cool escaping
gases so that combustion cannot spread beyond the enclosure is the most
common technique of this sort in electrical installations.
Enclosed break devices are intended for use in Zone 2/Division 2 loca-
tions. The function of an enclosed break device is, like that of an
explosionproof enclosure, to contain any ignition that occurs within it. The
enclosed break device does not rely on heavy construction or flanged
joints, but rather on its small size. If a switch or relay is small the incipient
ignition may be partially quenched by the walls of the component. The
pressure rise will be reduced below that which would be developed in a
larger chamber, and a component of ordinary design can contain the
explosion safely.
Although the technique is not frequently used in industry, a continuous

source of ignition to localize combustion is commonplace in other contexts.
An ignition source near the source of flammable material forces ignition to
take place within predetermined bounds where damage does not occur.
The continuous pilot in gas appliances, boilers, and furnaces performs this
function. It confines combustion to a predefined safe location by igniting

the gas before a potentially hazardous, uncontrolled volume of unburned
gas has been liberated. Methods of reducing hazard that depend on pre-
venting ignition are based on eliminating one of the legs of the fire triangle
(see Figure 3-1). These methods eliminate the source of ignition, control the
composition of the atmosphere at the ignition source, or prevent contact
between the ignition source and the flammable atmosphere.
Methods Based on Eliminating the Source of Ignition

One way to eliminate a hazard by removing the source of ignition is to
apply the principle of intrinsic safety. Equipment and wiring that is intrin-
sically safe is incapable under normal or abnormal conditions of igniting a
specified hazardous atmospheric mixture. Therefore, for practical pur-
poses the source of ignition has been eliminated.
In Division 2/Zone 2, an electrical device or circuit must not be an ignition

source when it is in its normal operating condition. Although ISA SP12
originally coined the term “nonincendive” to mean all such equipment
suitable for Division 2, two terms, “nonincendive” and “nonsparking,” are
now used by some. Nonsparking apparatus includes squirrel cage motors,
electronic devices, and other equipment that in normal operation does not
spark or have high-temperature components or surfaces that are capable
of causing ignition.
Nonincendive apparatus includes contacts that are normally sparking but

which operate in circuits that are not ignition capable, or circuits that, if
opened, shorted, or grounded, would not release enough energy when the
equipment is in its normal operating condition to ignite a specified flam-
mable mixture. Nonincendive components are components that, when
operated in circuits of specified current and voltage do not cause ignition
because the volume is so small or the contacting mechanism is so massive
that incipient explosions are quenched.
In some countries, the term “nonsparking” also connotes construction

requirements more stringent than are mandated for general purpose
apparatus in order to reduce the probability of sparking due to failure of
connections, reduction of spacings between conductors, generation of hot
spots, and the like. This view is contrary to the philosophy in North Amer-
ica (since the 1940s) that equipment for use in Division 2 locations need
only meet ordinary location requirements and be free from ignition-capa-
ble arcs and hot surfaces in normal operation. Many North American
standards developers have forgotten this simple criterion, and now seem
willing to join others in writing unnecessarily detailed requirements into
standards for Division 2 apparatus.
Another way to eliminate the source of ignition is to use type of protection

e, “increased safety.” This approach was pioneered in Germany, but is
now an accepted method in Europe and in many other countries.
Increased safety is considered to be a suitable method of protection for use
in Zone 1. It has been used as an economical alternative to explosionproof
enclosures for devices that produce no sparks in normal operation. In gen-
eral, the approach of “increased safety” is to require the following:
• especially large clearances and creepage distances between live
parts
• unusual attention to excellence of insulation, both in amount and
quality
• protection against development of hot spots
• special closures, such as shrouded bolts, to prevent unauthorized
entry
• large rotor-stator clearance in rotating machinery
All these design considerations are intended to ensure that there is an

extremely low probability of electrical or mechanical breakdown that
could produce a spark. In addition to robust construction, Type e rotating
machinery must be provided with overcurrent protection devices that are
coordinated with the thermal characteristics of the machine to ensure that
the machine will be deenergized before it reaches an ignition capable tem-
perature under overload or locked rotor conditions.
Terminal boxes of flameproof enclosures may be of type e construction.

The concept of especially robust construction is also applied to Type e
cables.
Methods Based on Controlling the Atmosphere at the

Ignition Source
Select Location
All standards and regulations for selecting and installing electrical appa-
ratus in explosion-endangered locations emphasize the most obvious way
to reduce hazard, that is, install electrical equipment outside the hazard-
ous location. Electrical equipment should be installed in a location where
flammable material is likely to be present only when the apparatus is
essential to the operation of the plant and can be installed nowhere else.
Controlling Concentration
Inerting is the introduction of a material that prevents the formation of a

combustible mixture. Often this technique is applied by providing a blan-
ket of nitrogen or carbon dioxide in the vapor space of a tank to lower the
oxygen concentration so that the mixture can never become explosive.
This concept is also used in mines. Rock dust is suspended above the gal-
leries or shafts of a coal mine, or it may be sprayed on accumulated layers
of coal dust so that an explosion will propagate only with difficulty. The
inert rock dust adds thermal capacity without increasing the energy
released by combustion and requires greater amounts of energy to raise
the combustible elements of the atmosphere to their ignition temperature.
A combustion wave therefore fails to propagate.
Pressurization or ventilating an enclosure or room is perhaps the most

common use of concentration control to prevent accumulation of a flam-
mable atmosphere at an ignition source. In enclosures not containing an
internal source of flammable material, pressurization prevents entry of
flammable gases or vapors from the outside. If the enclosure contains an
internal source of release of flammable material, as many analyzers do, it
is necessary, in addition, to ensure that a flow of noncombustible gas
through the enclosure is maintained at a volume which will dilute any
release of gas to a concentration below the LEL. Most standards cover only
the case of an analyzer being protected by this “continuous dilution”
when the analyzer is located in a hazardous location. In this case it is nec-
essary also to maintain internal pressurization to prevent flammable
material from entering from the outside. However, continuous dilution
without pressurization is a useful technique to avoid an internal explosion
when the analyzer is located in a nonhazardous location.
Another way of controlling the concentration to prevent the formation of a

flammable atmosphere is to ensure that the mixture is always above the
UEL. Operating practices in pipelines ensure that this is the case. In
annealing furnaces the concentration of hydrogen is maintained above the
UEL. Hydrogen escaping around the doors and vents will burn if the tem-
perature is above the ignition temperature but the combustion does not
propagate back into the furnace cavity itself. Similarly, the concentration
of hydrogen in hydrogen-cooled generators is always maintained above
the UEL so no self-sustained combustion can occur.
Isolating the Ignition Source from the Flammable Material
Two relatively uncommon techniques for preventing contact between an

ignition source and a flammable material are oil immersion and sand fill-
ing. in both cases immersion in a nonflammable medium effectively
isolates the ignition source from the flammable material. Even though the
sand may be somewhat permeable to the gas, any incipient ignition will be
quenched; so, for practical purposes, isolation is achieved.
Sealing and encapsulation are the two most common ways in instrument
practice to isolate an ignition source from the flammable atmosphere. Her-
metic sealing has long been recognized as effective, but standards now
recognize that a lower grade of seal will prevent effective contact in Zone
2/Division 2 locations. The function of encapsulation is also to isolate an
ignition source from the flammable atmosphere. CENELEC Standard EN
50028 defines a quality of encapsulation that may be used in Zone 1 loca-
tions. A less controlled level of encapsulation has long been recognized as
safe in Zone 2/Division 2 locations.
Restricted breathing is a technique that is recognized safe for Zone 2 loca-

tions in Europe, but it is not yet considered to be an acceptable protection
technique in North America. The philosophy of restricted breathing is like
that of nonhermetic sealing. It is only necessary to ensure that an enclo-
sure is tight enough to ensure that the internal concentration of flammable
material reaches only a fraction of the LEL during the short-term presence
of a flammable cloud of gas or vapor outside the enclosure.
The Degree of Protection Required For Different

Degrees of Hazard
As Schön has pointed out in his article, those responsible for preparing
regulations or standards intended to minimize risk characteristically
adopt a very conservative posture when they have difficulty quantifying
the risk. Anyone who has been active in drafting safety standards has
observed the tendency of his colleagues to decide issues on the basis of
“What if …?” or “It's possible that …!” Such questions avoid quantifica-
tion and probability. It is especially frustrating when clauses of a standard
are drafted on the basis of such questions even when quantitative esti-
mates or experiential evidence is available on which to base a more
rational approach to risk reduction.
The approach to considering hazard reduction taken in this section is a

probabilistic approach. It would be impossible to apply any means of
reducing hazard effectively if one did not try to adopt this approach. The
only alternative to a probabilistic approach is to consider all possible con-
tingencies and eliminate all conceivable sources of hazard. For example, it
is possible for an airplane to crash into the building in which you are now
reading. But it is not an event of even modest probability. Any reader who
interrupts his reading to go outside and search the sky for an airplane
should probably seek counseling to help him find the reason for his com-
pulsive fear! Other readers will continue reading.
The probability PE that an explosion will occur is equal to the product of

the probability PC that combustible material is present and the probability
PI that sufficient ignition energy will be released. For example, if the igni-
tion source is continuous, the probability PI that the energy will be
released is 1. If a combustible material is present 1 hr in 10,000 hr, the
probability PC is 0.0001 that the material is present in any 1 hour. The
probability of an explosion PE in any given hour is therefore PI × PC = 1 ×
0.0001 = 0.0001. If the ignition source were present only 1 hr in every
10,000 hr, the probability PE of an explosion in any given hour would be
0.0001 × 0.0001 = 10-8 or 1 in 100,000,000. For almost any purpose, a proba-
bility of 10-8 would be considered to be so low as to be essentially zero.
Even a probability of 10-6 for a single event would be considered a good
underwriting risk, one of essentially zero probability.
The presence of an atmosphere above the LEL for 1 hr in 10,000 hr (1 hr in

13 months, usually rounded off to 1 hr/yr), as assumed above, is a more
severe hazard than exists in most locations classified as Division 2, though
it is within the numerical range that is usually discussed. Presence of an
ignition source for 1 hr in 10,000 hr is certainly a much more severe hazard
than is presented by a recording potentiometer of conventional design. It
is orders of magnitude higher than one would expect for a modern 4−20
mA differential pressure transmitter or other instrument with nonspark-
ing semiconductor circuitry and no hot surfaces. Even a probability PE of
10-8 is, therefore, orders of magnitude greater than that which exists when
a conventional 4−20 mA instrument that is nonsparking in normal opera-
tion is installed in a Division 2 location.
On the other hand, if one expected that an atmosphere were to be above

the LEL as frequently as 1 hr in 1,000 hr, strict rules to prevent entry of
ignition sources would be enforced. Although the location would be clas-
sified Division 1, it would be far more hazardous than most locations
which are now classified Division 1.
Quasi-Quantitative Approach to Safety

Area classification is not, in general, a quantitative procedure at the
present time. The foregoing examples do not imply general acceptance of
numerical criteria for classification of Division 1 and Division 2 locations.
However, where the probability of the atmosphere being above the LEL is
as high as 0.001, most people would classify the location as Division 1 and
act as though the location were nearly always hazardous (i.e., PC = 1). In
the paragraphs that follow a high probability is, therefore, represented by
a probability of 1 for convenience of discussion. Similarly, a very low
probability, less than 10-6, is considered for convenience to be 0. Moderate
or low probability, therefore, implies an absolute probability on the order
of 10-3 to 10-6. This discussion is tutorial, not practical. Were one to assume
that there is, in practice, a probability of 1 that a combustible atmosphere

exists in a Division 1 location, one would draw completely erroneous con-
clusions about the rules for designing and installing electrical systems to
ensure safety. They would be so conservative that useful systems would
be difficult to design.
A fundamental consideration for establishing the degree of protection

required in a particular hazardous situation is that no single occurrence
must raise the probability of explosion from zero to one, that is, no one
event shall significantly increase the hazard. This requirement is equiva-
lent to stating that the probability of there being an explosion shall be
essentially zero. The probability that two events, each of low probability
(e.g., 10-4) will occur simultaneously is 10-8 and may, therefore, for practi-
cal purposes be considered to be zero. Neither event alone raises the
probability of an explosion to a high value. A single event with a probabil-
ity of 10-8 would, of course, also provide the necessary degree of
safeguard.
If an atmosphere is above the LEL frequently and the equipment in that

location is not normally a source of ignition, the probability of explosion
while the equipment is in normal working condition is effectively zero. If
the equipment fails, releases a spark, and becomes a source of ignition-
⎯even once per year⎯the probability of explosion in a given year would
be undesirably high. A single failure of the equipment raises the probabil-
ity of explosion from a very low value to a very high one. A prudent
individual would desire additional protective measures so that two fail-
ures, each of low probability⎯failure of the equipment once per year and
failure of the protective means⎯must occur simultaneously to have an
explosion. When this additional protective means is supplied, the proba-
bility of an explosion is reduced to effectively zero.
The foregoing is not a new philosophy. This rationale is implicit in the

way in which the NEC and other electrical codes define area classifications
and the rules for selection and installation of electrical apparatus. NEC
definitions are not made in probabilistic terms. No probability numbers
are specified. However, classification of a particular area as Division 1 or
Division 2 is made primarily on the basis of loss experience, a consider-
ation of the definitions in Article 500, and the judgment of the individual
who is classifying the site as follows:
(1) A Division 1 location is one which is always, intermittently, or fre-
quently hazardous, that is, the probability is, in underwriting terms,
“high” (although, as stated earlier, on an absolute basis the probabil-
ity may be a very low number) that a flammable atmosphere will be
present.
(2) A Division 2 location is hazardous if the process equipment fails or if

there is some other unexpected occurrence, such as a flammable vapor
or gas traveling beyond the predicted extent of the Division 1 location.
According to underwriting criteria the probability that the atmo-
sphere is hazardous in a Division 2 location is low. When process
equipment is operating normally, the location is nonhazardous. It
becomes hazardous only if the equipment is in some abnormal condi-
tion.
(3) An unclassified or nonhazardous location has essentially zero proba-

bility that a flammable mixture will be present.
Equipment and wiring can be considered in a similar manner:

(1) Intrinsically safe equipment is, by definition, equipment that is inca-
pable of releasing sufficient thermal or electrical energy under normal
or abnormal conditions to ignite a specific hazardous atmospheric
mixture. The probability of intrinsically safe equipment releasing suf-
ficient energy to cause ignition is, therefore, zero.
(2) Nonincendive equipment is equipment that, by definition, will not

release sufficient energy in its normal working condition to ignite a
specific hazardous atmospheric mixture. An ignition source is, there-
fore, not normally present. The probability is low that the equipment
will fail and become ignition-capable.
(3) Ignition-capable equipment is equipment that has sparking contacts

or hot surfaces that release sufficient energy in normal operation to
ignite the flammable atmosphere. The probability of the ignition
source being present in this case is one.
Figure 4-3 illustrates the possible combinations of the probability of the

location being hazardous and the probability of the equipment being a
source of ignition. The product of the two probabilities determines the
probability of an explosion. The number of additional safeguards needed
to reduce the probability of an explosion to effectively zero is also shown.
As specific means of eliminating explosion hazards are considered in later

chapters, it will be seen that recommendations for equipment design and
installation are based on the principles presented above. In other words,
no single event of low probability will raise the ignition probability from
zero to one. The equivalent of two independent occurrences, each of low
probability, is required.
Figure 4-3 Relationship Between Area Classification, Equipment Classifications, and

Additional Safeguards Required
However, one apparently paradoxical situation exists. The NEC permits

the use of ignition-capable equipment in a Division 1 location if it is
enclosed in an explosionproof enclosure. In the context of the philosophy
expounded in the previous paragraphs, this would seem to be imprudent,
because a single failure — that of the explosionproof housing — could
cause the probability of explosion to rise from zero to one. One might,
therefore, conclude that the use of explosionproof housings in this manner
is unsafe. This is not so. Experience has proven such use to be a safe prac-
tice. The apparent contradiction can be rationalized as follows. The fact
that explosionproof housings are designed with extremely conservative
flame path and pressure test requirements constitutes in itself a safety fac-
tor considerably greater than that which could be considered minimum
safe practice. The statistically safe gap values published in British stan-
dards were based on a one-in-a-million chance of causing ignition under
ideal test conditions. The maximum permitted gaps in enclosures are 40%
narrower than the statistically safe gaps. This remarkable safety factor is
also present in the gap values permitted by IEC and CENELEC standards.
Other factors also contribute to the safety of using ignition-capable equip-

ment in explosionproof enclosures in Division 1 locations.
Under practical conditions, even were combustible gas present in ignitable

concentration, ignition would probably occur near the LEL and flame
would not penetrate a gap that is even much larger than that permitted.
Secondly, although in the foregoing discussion we have considered a Divi-
sion 1 location to have unity probability of being hazardous, there are very
few locations classified Division 1 that are hazardous even a small fraction
of the time. If a significant portion of a processing plant were hazardous

for extended periods of time, even if the electrical equipment were all
deenergized, the probability of explosion from other causes, for example,
static electricity, or sparks caused by tools or shoes, is high. In support of
this viewpoint is the fact that the LEL of most gases and vapors is a much
higher concentration than would be tolerable for prolonged exposure of
human beings.
An additional factor of safety is inherent in explosionproof enclosures

because they possess the properties of restricted breathing enclosures,
which are discussed in Chapter 7. The small gaps of the explosionproof
enclosure restrict the rate of flow of flammable mixture into the enclosure
so that the LEL can be reached inside only if the flammable mixture is
present for a long time. Even in Division 1 this situation is usually
improbable.
Because of the simplifying assumption of unity probability that a flamma-

ble atmosphere exists in a Division 1 location, Figure 4-3 cannot be
expanded to include consideration of Zone 0 locations, where the proba-
bility of a flammable atmosphere may indeed be unity. This is a failing of
this quasi-quantitative tutorial approach to safety. Where Zone 0 locations
are defined, three approaches are recognized. The first is to use only
intrinsically safe apparatus. The second is to use ventilation to reduce the
classification to Zone 1 so that explosionproof or other apparatus suitable
for Zone 1 may be used safely. The third is to use dual protection: pressur-
ization and explosionproofing, pressurization and increased safety, or a
similar combination. The second approach has not been codified as far as
the author is aware. He and many of his colleagues feel that a Type X pres-
surization system, which requires an interlock between the electrical
supply to the apparatus and the pressurization system meets the two fault
criterion and would be safe. This opinion is supported by the Benjaminsen
and van Wiechen calculations discussed later in this chapter. These calcu-
lations show that a pressurized system with interlock has the same order
of safety as intrinsic safety Type ib.
Figure 4-3 must be used with care and understanding of its intent. That
intent is to help develop an understanding of the universal principal of
safety that two faults, each of low probability, should lie between a safe
situation and a catastrophe. If one does not understand that the assign-
ment of unity probability of a flammable mixture being present to
Division 1 is a gross oversimplification of what is true in most Division 1
locations, completely erroneous conclusions will be drawn about the prob-
ability of an explosion when considering practical situations.
Quantitative Approach to Safety
The quasi-quantitative probabilistic approach to safety adopted in the pre-

vious section has been a common way of thinking about the subject for
many years, and it provides a strong conceptual and philosophic basis for
making decisions. However, this approach doesn't satisfy those who feel
that safety decisions can be based on more quantitative analyses, and a
number of papers have presented frameworks in which one might work
with probability theory to quantify decisions on equipment selection and
related matters. Some have dealt primarily with intrinsic safety and are
discussed in that context in Chapter 10. Bijl's paper defined the number of
explosions, Nex as:
N ex = N at P p P a P h or N ex = N at P p P s P h
where
Nex = number of explosions in 30 years
Nat = number of attempts to ignite the local mixture in 30 years
Pp = probability of pressurization failure
Ph = probability that flame is transmitted through a flange gap
of an explosionproof enclosure
Pa = probability that atmosphere surrounding a device is com-
bustible
Ps = probability that process fluid enters the enclosure to pro-
duce a hazardous mixture
Although most authors consider only the probability of a flammable mix-

ture surrounding electrical apparatus, Bijl considered also the catastrophic
explosion resulting from ignition of process fluid inside an enclosure.
Direct entry into an enclosure might result from failure of a seal, a Bour-
don tube, or other element containing process fluid. (See discussion of
singly-sealed systems in Chapter 3.)
When these equations are used, the equation that yields the higher num-
ber of explosions is controlling. In most situations Pa and Ps are of greatly
different magnitude; thus, only the larger is considered. If they are equal
the error caused by using only one equation is a factor of 2, which is not
significant in the context of these calculations.
For devices that are not pressurized, Pp is 1. Likewise, for devices in other
than explosionproof enclosures, Ph is 1. Bijl did not discuss restricted
breathing because it was not a generally recognized protection technique
at the time he wrote his paper.
The values Bijl used in his calculations are given in Table 4-2. Some of the
values are derived from the assumptions that (1) one explosion will occur
in 30 years and (2) present practices have been shown safe by experience.
Bijl's 30-year figure is a period which he states is an order of magnitude
larger than write-off times for typical plants.
The basis of Bijl's calculations may be questioned because the number of

devices does not appear explicitly in the calculations. One can also ques-
tion whether his assumed probabilities are correct. When one examines
the values he assigned to the probabilities in Table 4-2 one finds that all of
his assigned probabilities are high compared to what this author would
consider to be realistic. The probability of ignition capable sparks in non-
sparking apparatus and the assumed frequency of sparking of ignition
capable apparatus are both especially high.
Bijl's important contribution, however, is that he attempted to quantify the

factors that must be considered in setting the rules for safe electrical
installations.
Parameter
Parameter Condition Basis of Estimates
Values
Pat Division 0 1 Definition of Division 0
-1
Division 1 10 2.4 hr/day
Division 2 10-4 1 min/wk
-11 Derived from Nat = 10-11 for ignition-capable
Safe Area 10
equipment and Nex = 1 explosion in 30 years
Ps No seals 1 Device immersed in process fluid
1 seal 10-4 1 hr/yr, T/C well or bourdon tube failure
2 seals 10-12 Electric control valve
Pp No purge 1
Alarm purge 10-4 Manual shutdown if purge fails
Interlock purge 10-12 Automatic shutdown if purge fails
Nat Ignition-capable 1011 Twice per cycle of 60 Hz for 30 years
4
Nonsparking 10 1 dangerous moment per day
Intrinsically safe 1 1 dangerous moment in 30 years
Ph General purpose 1 No protection against passage of flame
Explosionproof 10-10 Derived from Pat = 10-1 in Division 1, Nat = 1011,
Nex = 1
Table 4-2 Bijl’s Proposed Probability Values

A major shortcoming of the approaches discussed above is that the calcu-

lation of explosion probability using a simple product of probabilities
yields unrealistically low values if the apparatus fails and the failure is not
detected and corrected for a long period of time. In such cases the proba-
bility of explosion is nearly the same as the probability of the apparatus
becoming ignition-capable. In a series of papers published in 1967−1969
Benjaminsen and van Wiechen attacked the problems of safety by apply-
ing probability theory to their assessment of the relative safety of the
different types of protection in classified locations.
Although these papers were discussed in the third edition of this book,
this author and his colleagues did not understand the full implications of
their conclusions. This, in all likelihood, may still be the case. There are
few people who are familiar with this work. Even fewer probably appreci-
ate its potential positive impact on the ability to place the classification of
hazardous locations and the selection of electrical apparatus for use in
hazardous locations on a sounder philosophical basis. Their work also
provides a basis for better understanding why present practices are safe,
and how safety can sometimes be improved at little or no additional cost.
Before discussing the work of Benjaminsen and van Wiechen, the philo-
sophical undergirding of their work should be reviewed. The exponential
probability function is defined as
Ps = e–λt
where
Ps = probability that equipment has survived
λ = failure rate of equipment or system being studied in aver-
age failures per unit in unit time
t = time
If one integrates this expression from t = 0 when survival probability is 1

to t = ∞ when survival probability is 0 and divides the result by probabil-
ity 1, the resulting number is the mean time before (or between) failure of
a large number of devices, usually abbreviated MTBF.
1
MTBF = ---
λ
λ can be determined for a component by testing hundreds or thousands of
components for many thousands of hours. At the end of the test period,
the average life before failure is computed. This is the MTBF. In practice,
the application of statistical techniques permits an estimate of MTBF based
on smaller scale testing. If one knows that failures of a particular compo-
nent follows the exponential function, as almost all do throughout most of
their life, and knows from calculation or experimental that the average
time between failures is 106 hr the survival function for this component is
as follows:
6
– t ⁄ 10
Ps = ε
6
– t ⁄ 10
The probability of failure, P F = 1 – ε
A two-component system that malfunctions if either component fails

would have a survival probability Ps equal to the product of the survival
probability of each component.
–λ1 t –λ2 t – ( λ 1 + λ 2 )t
P s = ( P s1 ) ( P s2 ) = ε ε = ε
For such a system,
1 ( MTBF 1 ) ( MTBF 2 )
MTBF = ---------------------- = ----------------------------------------------------
-
( λ1 + λ2 ) ( MTBF 1 ) + ( MTBF 2 )
This kind of calculation would be applicable to the case of two explosion-

proof enclosures, each housing ignition capable apparatus. Failure of
either enclosure would provide a source of ignition.
If a two-component system will continue to function safely if either com-

ponent survives (i.e., it fails only if both components fail) the probability
of failure PF is as follows:
–λ1 t –λ2 t –λ1 t –λ2 t – ( λ 1 + λ 2 )t
PF = ( 1 – ε )(1 – ε ) = 1–ε –ε +ε
–λ1 t –λ2 t – ( λ 1 + λ 2 )t
Ps = 1 – PF = ε +ε –ε
Integrating, one finds the MTBF of such a system to be
1 1 1
MTBF = ----- + ----- – ----------------------
λ1 λ2 ( λ1 + λ2 )
( MTBF 1 ) ( MTBF 2 )
= ( MTBF 1 ) + ( MTBF 2 ) – ----------------------------------------------------
-
( MTBF 1 ) + ( MTBF 2 )
If the MTBF of the two components are equal, the MTBF of the system is
1.5 (MTBF1).
Many methods of reducing hazard utilize two protective devices. The sys-
tem is safe as long as either functions. Redundancy using identical
components increases MTBF 50% compared to that using only one device.
However, the time of safe operation is significantly lengthened, if, as is the
case for most equipment with a recognized type of protection, the MTBF
of each device is a significant fraction of the expected operating life of the

installation.
For a single component with an MTBF of 106 hour, Ps = 0.905 at t = 105 hr.,
and the probability of failure, PF = 0.095. For two identical components
arranged so that both must fail to produce an undesired result, the proba-
bility of failure in 105 hours is 0.009. This result can be calculated from the
equation for PF given above or it can be calculated by multiplying the fail-
ure probability of the two components (0.095 × 0.095) Following the
pattern presented thus far in this chapter, one could multiply this proba-
bility of failure, which results in an ignition source being present, by the
probability of a flammable atmosphere being present to obtain the proba-
bility of an explosion occurring.
Note: The MTBF to be used in the calculations discussed in this section is

the mean time between failures that make the device ignition-capable. For
nonsparking apparatus, this value of MTBF will be many times the MTBF
relative to loss of function. MTBF values quoted by manufacturers are
usually values for failures that shift operation outside of the device specifi-
cations. A nonsparking device with 100,000-hour MTBF relative to
satisfactory operation may have an MTBF relative to becoming ignition-
capable that is hundreds of times that value.
For their calculations Benjaminsen and van Wiechen introduced a compli-

cating but realistic consideration, that the failure of the equipment may be
observed after a time interval and the failed equipment is then removed
from service. Now the probability of a flammable atmosphere being
present is an insufficient description. One must consider both the fre-
quency of occurrence of the flammable atmosphere and its duration,
because one now is concerned with the probability that the ignition source
is present during the period that a flammable atmosphere is present.
If one knows the average failure rate λ, the probability that K failures
occur in a time t is given by the Poisson distribution.
K
( λt ) –λt
P K ( t ) = ------------- ε
K!
This distribution is closely related to the exponential distribution. If K = 0
the Poisson distribution becomes the exponential distribution.
The objectives of the Benjaminsen and van Wiechen work were to:
• Give a quantitative comparison of the protective techniques com-
monly used (e.g., flameproof, increased safety, nonsparking,
restricted breathing, etc.).
• Provide guidelines for selecting a protective technique appropri-

ate to the location.
• Provide a quantitative estimate of the level of safety.
The first assertion in their work is that area classification, that is, the
assignment of a level of hazard, must not only the consider the probability
that the atmosphere surrounding equipment will be flammable but must
also take into account the duration of the presence of a flammable cloud.
In conventional practice a location might be classified Zone 1 if a flamma-
ble atmosphere is expected to be present 100 hours per year, or 1% of the
time. But the degree of hazard if the release occurs 100 times a year and
each time it lasts 1 hour may be different from the situation where the
release occurs 10 times a year and lasts 10 hours each time. Table 4-3 lists
the values of frequency g and duration τg used in the calculations. These
values are in general accord with consensus on the probability of a flam-
mable atmosphere being present in a specified zone.
COLUMN 1 COLUMN 2 COLUMN 3 COLUMN 4
Total hr/year of Frequency and Parameters

Area Classification
Explosive Mixture Duration Per Year g τg
Div. 0 10,000 1 × 10,000 hr .... ....
Div. 0 1,000 100 × 10 hr
or
1,000 × 1 hr* 10-1 1
-2
Div. 1 100 100 × 1 hr 10 1
Div. 1 10 100 × 0.1 hr 10-2 0.1
Div. 2 1 10 × 0.1 hr 10-3 0.1
Div. 2 0.1 1 × 0.1 hr 10-4 0.1
-5
Div. 2 0.01 0.1 × 0.1 hr 10 0.1
Unclassified areas 0.001 0.01 × 0.1 hr 10-6 0.1
0.00001 0.001 × 0.1 hr 10-7 0.1
0.00001 0.001 × 0.1 hr 10-8 0.1
*Calculations are based on 1,000 × 1 hr, which gives less favorable figures.
Table 4-3 Area Classification Compared with Frequency and Duration of Explosive
Mixture
The classification of ignition sources must also include the frequency of

failure and the period of time that elapses before the failure is recognized
and the equipment is shut down. For flameproof apparatus and restricted
breathing apparatus it is assumed that loss of protection will be detected
only during a periodic inspection, which is assumed to occur at intervals
of τe hours (104 hours in Table 4-4). For other apparatus an interval before
the failed equipment is switched off, τm is defined.
Table 4-4 lists the numerical assumptions made for each type of equip-
ment and the predicted mean time between explosions, MTBE, for each
type of equipment in the various zones. (The term “division” is used
because this work was done prior to adoption of “zone” by the IEC.)
Area
Classification Div. 0 Div. 1 Div. 2 Unclassified
Total hr/yr 10,000 1,000 100 10 1 0.1 0.01 0.001 0.0001 0.00001
Values of 10 log MTBE Frequency 1× 1,000× 100× 100× 10× 1× 0.1× 0.01× 0.001× 0.0001×
and Duration 10,000 1 hr 1 hr 0.1 hr 0.1 hr 0.1 hr 0.1 hr 0.1 hr 0.1 hr 0.1 hr.
hr
Protective Time Delay g 10-1 10-2 10-2 10-3 10-4 10-5 10-6 10-7 10-8
Equipment Technique Protecting MTBF
Device tm tg 1 1 0.1 0.1 0.1 0.1 0.1 0.1 0.1
Producing Open spark -∞ 1.0 2.0 2.0 3.0 4.0 5 6 7 8
sparks in Flameproof 106 hr 6.0 6.0 6.0 6.0 6.0 6.4 7.3 8.3 9.3 10.3
normal 107 hr 7.0 7.0 7.0 7.0 7.0 7.4 8.3 9.3 10.3 11.3
operation 108 hr 8.0 8.0 8.0 8.0 8.0 8.4 9.3 10.3 11.3 12.3
Pressurized 6
10 , 10 6
8.3 8.3 8.3 8.3 8.4 8.9 9.8 10.8 11.8 12.8
hr**
Nonincendive 104 hr 6.0 6.0 6.0 6.0 6.0 6.4 7.3 8.3 9.3 10.3
Intrins. safe 106 hr*** 8.0 8.0 8.0 8.0 8.1 8.6 9.5 10.5 115 12.5
Restr. breath 107 hr .... .... .... .... 7.0 7.4 8.3 9.3 10.3 11.3
Not Nonsparking 1,000 hr 106 hr 6.0 6.0 6.0 6.0 6.2 7 8 9 10 11
producing standard 0.1 hr 106 hr 6.0 7.0 8.0 8.7 9.7 10.7 11.7 12.7 13.7 14.7
4
sparks in 0.01 hr 10 hr 6.0 7.0 8.0 9.0 10 11 12 13 14 15
normal
operation
Increased safety 1,000 hr 107 hr 7.0 7.0 7.0 7.0 7.2 8 9 10 11 12
0.1 hr 107 hr 7.0 8.0 9.0 9.7 10.7 11.7 12.7 13.7 14.7 15.7
0.01 hr 7
10 hr 7.0 8.0 9.0 10 11.0 12 13 14 15 16
Equipment Nonsparking 1,000 hr *104 hr 8.3 8.3 8.3 8.3 8.5 9.3 10.3 11.3 12.3 13.3
with two + *107 hr 9.3 9.3 9.3 9.3 9.5 10.3 11.3 12.3 13.3 14.3
different flameproof 0.1 hr *104 hr 8.3 9.3 10.3 11 12 13 14 15 16 17
protective *107 hr 9.3 10.3 11.3 12 13 14 15 16 17 18
techniques 0.01 hr *104 hr 8.3 9.3 10.3 11.3 12.3 13.3 14.3 15.3 16.3 17.3
in combina- 7
*10 hr 9.3 10.3 11.3 12.3 13.3 14.3 15.3 16.3 17.3 18.3
tion
Nonsparking 1,000 hr *107 hr .... .... .... .... 9.5 10.3 11.3 12.3 13.3 14.3
+ 0.1 hr *107 hr .... .... .... .... 13 14 15 16 17 18
restr. breath. 0.01 hr *107 hr .... .... .... .... 13.3 13.3 15.3 16.3 17.3 18.3
*The figure given refers to the MTBF of the housing. For the MTBF of the nonsparking equipment inside the enclosure the figure 106 hr, is
used.
**106 hr for press, medium, 106 hr for trip circuit.
***Two independent current or voltage limiting devices in series each with MTBF = 106 hr.
Table 4-4 MTBE Values for Various Protecting Techniques. Inspection Interval, One
Year. te ≈ 104 Hours.
Some of the MTBF values and values of τm assumed in this table need
explanation.
Flameproof Apparatus:
106 hours—enclosures with flanged joints, windows, or direct entry of
cables.
107 hours—enclosures with threaded joints or indirect cable entry.

108 hours—housings of very small volume which are not repairable.
Nonsparking Apparatus:
τm = 1,000 hours for failures that are difficult to detect, such as loose
rotor bars in squirrel cage motors, loose connections, broken outer
bulb of high pressure mercury lamp.
τm = 0.1 hour for failures that will be removed by overload protection.
τm = 0.01 hour for failures that will be removed by short circuit or earth
fault protection.
Note also that the term “nonsparking apparatus” is applied to heavy

equipment such as rotating machinery and that the term “nonincendive”
is used for other types of apparatus which are not ignition capable in nor-
mal operation.
Increased Safety Apparatus:

The MTBF values are assumed to be 10 times that for ordinary non-
sparking apparatus because of the more onerous construction require-
ments for this kind of equipment
The results of the calculations are presented in Table 4-4 as the logarithm
of the MTBE, so a difference of 1 between table entries represents a factor
of 10 in MTBE. An entry of 6 means that the MTBE is 106 hours, or 100
years.
This author believes that the MTBE values in Table 4-4 were calculated
using the following approximate equations. (See the PCI 68-48 paper.) τf is
the length of the dangerous period, the sum of the period during which
the apparatus is ignition capable, and the time during which a flammable
cloud is present.
• Sparking apparatus:
log (MTBE) = -log g (g is probability that flammable atmosphere
is present. See Table 4-3.)
• Sparking apparatus in explosionproof enclosure:
log ( MTBE ) = log ( MTBF ) enclosure + P g § τ = τ----e + τ ·

© f 2 g¹
• Non-sparking apparatus:
log ( MTBE ) = log ( MTBF ) apparatus + P g ( τf = τm + τg )
• Nonsparking apparatus in explosionproof enclosure:
log ( MTBE ) = log ( MTBF ) enclosure + log ( MTBF ) apparatus

– log τ e + 0.3 + P g ( τf = τm + τg )
• Intrinsically safe apparatus - Type ib (ignition capable after 2

faults):
gτ e » 1: log MTBE = 2 log ( MTBF ) device – log τ e

gτ e « 1: log MTBE = 2 log ( MTBF ) device – 2 log g – log g + 0.5
• Pressurized apparatus:
gτ e » 1: log MTBE = 2 log ( MTBF ) device – log τ e + 0.3

gτ e « 1: log MTBE = 2 log ( MTBF ) device – 2 log τ e – log g + 0.3
These equations were derived from much more complicated solutions

described in the Edinburgh paper. Pg is a valued derived from a nomo-
gram which is not reproduced here, in which the input values are g and τf
as defined above for each case. This nomogram was included in the IEEE
PCI paper, 1968. Typical values can be back calculated from the results in
Table 4-4. If τf is greater than 500 hours and g is greater than 10-3 the value
of Pg is 0.5 or less.
The Edinburgh paper states the basis for the more rigorous formulae. For
explosionproof, nonincendive, nonsparking and restricted breathing
equipment, the probability of failure of the equipment is given by:
– λt
PF = 1 – ε
For pressurized equipment, an interlock is assumed that has the same

MTBF as the pressurization source. A dangerous period exists only after
the interlock system has failed and then the pressurization system fails. If
the failure is in the reverse order no hazard exists because the interlock
shuts down the system, which will be noticed and repaired. The hazard
exists after a time, t = t1 + t2, at which the two failures occur. They derive a
probability distribution function after assuming probability density func-
tions for the failures at times t1 and t2 of the form, λ-λt.
Calculations for intrinsic safety were made for what are now known as
Type ib systems. An ignition source is presumed to be present after failure
of two identical protective elements, each with an MTBF of 106 hours. It
was assumed in the first paper, as judged from the derived equations, that
neither of the failures is obvious, so the failure probability is the product
of the failure probabilities of the two components. The resulting MTBF is

1.5 × 106. The failure probability at 10,000 hours is (0.01)2, and the equiva-
lent MTBF for t = 10,000 hours of a single apparatus would be 108. If one
calculates for Type ia intrinsic safety, the failure of three identical compo-
nents with MTBFs of 106 hours has a probability of 10-6. The calculated
MTBF for a system with three components in series, each with MTBF of
106 hours is 1.9 × 106 hours.
The approximate equations discussed above, from which the MTBE values
were calculated for Table 4-4, have the same form for intrinsically safe
equipment as for pressurized apparatus, rather than the form for nonobvi-
ous or undetectable failures. At the time the Benjaminsen and van Wiechen
papers were being written, there was considerable discussion of obvious
and nonobvious faults in IEC SC31G; and, in the second paper, it may have
been assumed that the first fault would be detectable during an inspection.
The derivation of equations for apparatus with two protective techniques

in series assumes that each has a survival distribution like that of explo-
sionproof enclosures, e–λt, with different values of λ. The dangerous period
is given as above by t = t1 + t2, with t2 beginning only after t1 has occurred.
Calculations for ignitions in Zone 1 and Zone 2 where the flammable gas
is present with a probability g in any period of time, and a duration τg, fol-
low the pattern for pressurized enclosures. t1 is the time of failure of the
equipment or combinations of types of protection, and t2 is the time at
which gas becomes present with a probability distribution function e–gt.
Examination of Table 4-4 shows some interesting results.

• Nonsparking equipment, even with limited protection, 1,000 hr
between failure and shutdown, is at least as safe as flameproof
enclosures with flanged joints containing normally sparking parts.
• All apparatus having nonobvious faults (i.e., flameproof, pres-
surized, nonincendive, and intrinsically safe) have MTBE values
that are independent of the area classification as long as the value
of g, the frequency of presence of the explosive mixture, is larger
than 1/τe, the frequency of inspection. One needs not calculate to
reach this conclusion. If a failure occurs so that an ignition source
is present, there will be an explosion as soon as the flammable gas
is present, regardless of area classification. In a Zone 1 location
the explosion may occur early in the year whereas in a Zone 2
location the explosion may not occur until the end of the year.
• The protection provided by restricted breathing is remarkably
high in Zone 2 locations, equaling or exceeding that of flameproof
enclosures with normally sparking apparatus inside. Almost all
enclosures of process control equipment installed outdoors are
rated NEMA 4, CSA Enclosure 4, or IP65 to ensure continuous

function in adverse environments. These enclosures inherently
possess the property of restricted breathing enclosures to delay
the entry of flammable gases and vapors, though perhaps not to
standardized restricted breathing requirements. This raises the
MTBE for all types of apparatus.
• Two protective measures, such as a flameproof enclosure con-
taining nonsparking apparatus or a restricted breathing enclo-
sure containing nonsparking apparatus, greatly increase the
MTBE, as one would expect.
In some of their publications the application of these calculations to the

process of area classification was discussed. For apparatus that contains
ignition-capable sparking the MTBF of the protective measure has little or
no effect on the MTBE. Therefore one can question whether area classifica-
tion has a useful place in the design of safe installations. The authors
proposed that areas be classified as Division 1 if 1/g < τe and that they be
classified Division 2 if 1/g > τe. However, the primary thrust of the Ben-
jaminsen-van Wiechen work was to argue that making judgments about
selection of apparatus should begin with an agreement about a target
MTBE for a plant and that a set of MTBF figures for apparatus be devel-
oped based on experience. In summary, they propose taking steps toward
a more rational integration of area classification and equipment selection,
which takes into consideration inspection times and other factors not now
considered.
The values of MTBE in Table 4-4 are for a single piece of equipment. If one
considered a plant with 10,000 pieces of equipment, all installed in the
same area classification, one would subtract 4 from the entries in the table
to obtain the MTBE for an explosion in the plant. One would conclude that
10,000 pieces of flameproof apparatus would cause an MTBE of 100 hours
in Division 2. There would be an explosion approximately every month.
The authors point out that this is silly, completely contrary to experience.
They note that the flaw in the calculation is the assumption that all devices
would be in the presence of a flammable cloud at one time. A plant with
10,000 devices would never be completely enveloped by a single cloud.
Consequently, they did alternative calculations assuming that 10 times a
year a cloud enveloped one of the pieces of equipment at random, and cal-
culated an MTBE of 105.3. This corresponds to an explosion probability of
39% in 10 years. They then assumed that 10% of the devices were flame-
proof and the rest were normally nonsparking or of restricted breathing
construction or some other technique which has a higher MTBE than
flameproof enclosures, so that the MTBE is determined by the flameproof
enclosures. For this case they calculated an MTBE of 106.3, which yields a
4.9% probability of an explosion in 10 years. They conclude that the real
level of risk lies somewhere between the two cases and that an MTBE of
106 for a single apparatus offers an acceptable degree of safety—a conclu-

sion validated by experience.
The reader should remember that the calculations presented in Table 4-4
are all based on an inspection interval of 10,000 hours. Shorter inspection
intervals increase the MTBE when 1/g > τe, and a longer inspection inter-
val decreases MTBE. The article in the Electrical Review compared values
calculated with no inspection and inspection intervals of 1,000, 10,000, and
100,000 hours. In all cases, the flameproof enclosure with sparking con-
tents has the lowest MTBE. With no inspection the MTBE of apparatus
with a single level of protection ranged from 106 to 107.3. With inspection
at 1,000-hour intervals the range was from 106 to 1011.3.
Benjaminsen and van Wiechen's calculations are a convincing demonstra-

tion that there is a way to evaluate matters of electrical safety using good
judgment and common mathematical techniques. Despite the many
assumptions that they had to make, the results are consistent with practi-
cal experience. One can quibble about the MTBF values assigned to
various kinds of apparatus, but it is not likely that the MTBFs of real appa-
ratus are lower than those assumed.
The MTBF of flameproof apparatus used in this analysis is, in the author's
opinion, probably very low because the usual field-mounted enclosure
has a degree of protection provided by its tightness. As discussed in Chap-
ters 5 and 7, the usual flameproof enclosure is also a restricted breathing
enclosure because of the length of time it takes a flammable atmosphere
surrounding it to raise the interior to the lower explosive limit.
The typical field-mounted transmitter is housed in a NEMA 4 enclosure to

withstand its outdoor environment. The tightness of such enclosures
exceeds that of a restricted breathing enclosure so the entries in Table 4-4
for nonsparking apparatus in a restricted breathing enclosure give the per-
tinent MTBE values. Though values for this combination when located in
Division 1 are not shown in the table, they are the same as those for non-
sparking apparatus in a flameproof enclosure with threaded joints. Those
transmitters that are NEMA 4 and explosionproof may not be safer than
those that are only NEMA 4 because the same kind of failure that invali-
dates the restricted breathing characteristics of the enclosure would likely
invalidate the explosionproof protection as well.
Estimating the probability that nonincendive apparatus and intrinsically

safe apparatus will fail in a manner that makes them ignition-capable is a
very difficult task. Although an MTBF can be calculated based on accepted
failure rates of the components this figure deals only with the ability of the
equipment to function within its tolerances. Of the many possible failures
in a complicated electronic apparatus, only a few affect the amount of
energy that the device might release. If the device has no normally operat-
ing contacts, there is also the need to experience a failure that provides a
mechanism to release the energy. This is discussed in Chapter 10.
The simple multiplication of probabilities, those of presence of a combusti-

ble atmosphere and presence of an ignition source, relate to only the first
two levels in a fault tree describing events that might occur and lead to an
explosion. A partial fault tree is shown in Figure 4-4. There are many pos-
sible reasons for a flammable material to be present, and many others for
an ignition source to be present. Some of these are indicated. Each of these,
in turn, are usually analyzable as the result of the coincidence of two or
more additional events (another set of events linked by an AND gate), or
another set of events linked in a subsidiary fault tree.
Figure 4-4 Example of Simple Fault Tree

Conn and Metzler wrote a very enlightening introduction to safety proba-

bility that uses such subjects as an explosionproof enclosure becoming an
ignition source to explain the methods of analysis. Their fault tree analysis
of an explosionproof enclosure becoming an ignition source considers fail-
ure due to internal explosion, volatilization of electrical insulation, and
development of a hot surface. Failure due to an internal explosion requires
an explosion and a failure of a flame path or a strength failure. He lists 10
causes of flame-path failures and 14 causes of strength failures to be con-
sidered. The internal explosion itself is the result of a subsidiary branch
with an AND gate to take into account the reasons for the presence of a
flammable gas in the enclosure and the possible igniting sources. Though
the complete fault tree of a system may be very complicated, and it is nec-
essary for completeness to include many possible events, much of the
analysis simplifies quickly after probability values have been assigned.
Usually a few of the events have probabilities much higher than all the
others, so it is not necessary to evaluate all eventualities in great detail.
Conn has used the fault tree analytical method to estimate the degree of
hazard of a number of possible events related to mining operations. In one
case he concluded that two events, which had been proposed to have
additive effects by some laboratory-oriented experts and which therefore
should demand amendment of test methods, were, in the context of min-
ing operations, not probable of being coincident, and that the proposed
additional hazard did not exist.
In the continuous process industries attention to safe automation of pro-

cess systems has increased in recent years, especially after OSHA
published its rule on process safety management. Increased attention to
more formal methods of analysis, such as fault trees, has also resulted
from the complexity of process control systems, and the necessity for
addressing issues of safety that arise because of hardware, software, and
process complexity. Simple rules such as “The safety shutdown system
must always be completely separate from the process control system” may
not always yield correct answers. Examples are when the process is com-
plicated, or unstable; the process equipment is easily damaged and not
easily replaced; and when the control system must necessarily be of excep-
tionally high reliability. While most design and analysis will still be done
using relatively crude tools because fault tree analyses are very demand-
ing, it is certain that the thinking patterns underlying fault trees and
similar tools, such as event trees, will become more common. As those in
the process industries use the tools one can hope that the same tools will
be used more frequently by those concerned with fires and explosions.
References
Hickes, W. F., “Intrinsic Safety,” Proceedings of 1960 Symposium on Safety for
Electrical Instrumentation in Hazardous Areas, Instrument Society of Amer-
ica, Pittsburgh.
Maltby, F. L., “History of ISA Committee on Hazardous Area Instrumen-

tation,” Proceedings of 1960 Symposium on Safety for Electrical
Pittsburgh.
Simons, C. F. E., Kogeler, W. J., and P. C. J. Bijl, “Safety of Electrical Instru-

ments in the Oil Industry,” Paper No. 4001M, IEE Conference Report
Series No. 3, Institution of Electrical Engineers, London, November 1962.
Bijl, P. C. J., “Numbers Yield Instrument Safety Balance,” ISA Journal

(October 1966), pp. 62−64.
“Electrical Apparatus and Associated Equipment for Use in Explosive

Atmospheres of Gas or Vapour Other Than Mining Applications,” British
Standard Code of Practice CP1003, Part 2, 1966, British Standards Institu-
tion, London.
Benjaminsen, J. M., and P. H. van Wiechen, “Probability Factors in Select-

ing Electrical Apparatus or Use in Dangerous Areas,” ERA Distribution
Conference, Group 5, Paper No. 4, Edinburgh, 1967.
Wilkins, J. F., “A User's Viewpoint of Safety in Dangerous Areas,” ERA

Distribution Conference, Group 5, Paper No. 1, Edinburgh, 1967.
Benjaminsen, J. M., and P. H. van Wiechen, “Mean Time to Electrical

Explosions,” Hydrocarbon Processing, 47, 8 (August 1968), pp. 121−126.
Benjaminsen, J, M., and P. H. van Wiechen, “Probability Factors as a Basis

for Area Classification and Selection of Electrical Equipment,” Paper PCI
68-48, IEEE, New York.
Benjaminsen, J. M., and P. H. van Wiechen, “Selecting Electrical Appara-

tus for Use in Hazardous Areas,” Electrical Review, 12 January 1968.
Wilkins, J. F., “A Study of 'Hazardous Location' Practices,” Paper PCI 68-

40, IEEE, New York.
Dreier, H., et al., “Explosion-Protected Machines of the Type of Protection

'Increased Safety' (Ex) e,” PTB Testing Memorandum (English translation
by BASEEFA in cooperation with PTB), Braunschweig, 1969.
Dreier, H., and U. Engel, “Measures for Protection of and Practical Experi-
ences with Type 'e' Explosion-Protected Electrical Machines,” Conference
on Electrical Safety in Hazardous Environments, Institution of Electrical
Engineers, London, 1971.
Engel, U., and H. Wickboldt, “Special Types of Explosion-Protected Elec-

trical Machines of Type of Protection 'Increased Safety',” IEE Conference
Publication No. 134, London, 1975, pp. 118−123.
“Electrical Instruments in Hazardous Atmospheres,” Recommended Prac-

tice RP12.1, Instrument Society of America, Research Triangle Park.
National Electrical Code, NFPA 70, National Fire Protection Association,

Quincy.
Wilkens, J. F., “European Practices in Electrical Design and Construction

for Petroleum Installations,” Paper CP62-1325, IEEE, New York.
EN 50154, “Electrical Installations in Potentially Explosive Gas Atmo-

spheres (other than mines),” CENELEC, Brussels.
IEC Publication 79-14, “Electrical Installations in Explosive Gas Atmo-

spheres (other than mines),” International Electrotechnical Commission,
Geneva.
Conn, J. W., and R. W. Metzler, “An Introduction to Safety Probability,”

Internal report, MSHA, Triadelphia, 1989.
Conn, John W., “The Noncoincidence of a Methane Explosion in an Explo-

sionproof Enclosure with Other Pressure Producing Processes,” MSHA,
Triadelphia, 1989.
Benjaminsen, J. H., and P. H. van Wiechen, “Probability Factors as a Guide

for Area-Classification and Selection of Electrical Equipment?,” Ex Maga-
zine, Oct. 1977, Stahl, Kunzelsau, pp. 4−9.
CHAPTER 5
EXPLOSIONPROOF ENCLOSURES
Why an Explosionproof Enclosure Works

Many types of electrical equipment release enough energy during normal
operation or after probable and unpreventable faults to ignite a combusti-
ble mixture. One way to operate such equipment safely in a hazardous
location is to provide an enclosure so constructed that, if ignition does
occur, the flame cannot propagate outside the enclosure and spread to the
surrounding atmosphere. For an enclosure to perform this function it
must contain the internal explosion without damage. Escaping gases must
not ignite a flammable mixture surrounding the enclosure.
Explosionproof enclosures are not necessarily vapor-tight. Gas or vapor

will enter the enclosure and be ignited if there is an ignition source. To
meet functional requirements, explosionproof housings have usually been
distinguished by heavy-wall cast construction for strength, by threaded
covers, or by bolted, close-fitting flanges to provide only long narrow gaps
through which hot gases can escape.
Chapter 2 discussed the effect of cool surfaces on the development of a

combustion flame. One might expect that if an explosionproof chamber
were constructed so that the gap width (perpendicular distance between
flanges) were equal to the quenching distance, this flanged joint would
cool the flame and prevent the combustion wave from propagating out-
side the enclosure. However, the width of gaps that do not transmit an
explosion are considerably less than the quenching distance. Because of
the hot expanding combustion products inside the enclosure, pressure
inside the enclosure is higher than the pressure outside the enclosure. Hot
gas issues from the flange gap under conditions much different from those
existing in a freely developing combustion wave, and a gap of quenching
dimension will not cool the escaping gas sufficiently. More importantly, it
is not necessary that a combustion wave (flame) be propagated through
the gap to ignite the atmosphere surrounding the enclosure. It is only nec-
essary that burned gases be transmitted at a temperature somewhat
higher than the ignition temperature of the mixture outside. Conversely, it
is not true that passage of flame through the flange gap will necessarily
ignite a surrounding flammable atmosphere.
The mechanism of ignition by an escaping jet of gas is hypothesized to be

as follows. When an explosion is initiated at the center of an explosion-
138 Explosionproof Enclosures
proof chamber the combustion wave propagates outward toward the

walls of the chamber. As the combustible material inside the chamber
burns, the pressure and temperature increase, pushing unburned gas
through the flange gap. The burned gas, heated and under pressure, is
then forced out through the gap.
For an 8.5% methane-air mixture the temperature is on the order of 1,500–

2,200°C. For hydrogen-air mixtures the temperature is much higher. The
jet is cooled by the flange and then issues into the surrounding combusti-
ble atmosphere where it may be cooled somewhat further by adiabatic
expansion. As the jet enters the surrounding atmosphere, it entrains sur-
rounding combustible material. Mixing of this combustible material into
the jet provides fuel for combustion outside the chamber, but also further
cools the jet from the temperature at which it issued from the gap. If both
the rate of issuance of hot gas and the gas temperature are high enough,
combustible material will enter the jet without cooling the jet below the
ignition temperature of the flammable gas, and the atmosphere external to
the explosionproof housing will ignite. If the jet is small and entrains too
much combustible material, the cooling effect will be greater than the heat
supplied by combustion, and no ignition will occur. The jet temperature
may be sufficiently high to produce a luminous flame without igniting
outside gas. IEC Publication 79 states that the maximum gap that will not
allow ignition is 8 to 10 times as large as a gap that quenches visible flame.
Luminous flame temperature is much higher than the spontaneous igni-

tion temperature (SIT) reported in handbooks, but the SIT is a lower-limit
ignition temperature. It is measured with a large area of gas exposed to the
hot walls of a chamber, with no loss of heat from the gas volume. Typi-
cally, several seconds of exposure to the heated surface is needed before
ignition occurs at published SIT values. The escaping gas jet, like a hot
wire, must be at a considerably higher temperature than the SIT to cause
ignition with a short-enough time lag. Delay allows convection currents
and turbulent mixing of unburned gas with the jet to cool the gas jet and
make ignition impossible. Phillips has treated this aspect of ignition theo-
retically and has calculated a critical entrainment rate. Generally, small or
high-velocity jets entrain more external gas mixture than larger or
low-velocity jets. As the flange gap is narrowed, entrainment increases
until a critical entrainment rate is reached, at which cooling of the emerg-
ing jet is so great that ignition cannot occur.
This emphasis on jet entrainment of combustible gas is not to imply that

quenching at the gap is unimportant. It is significant. However, the tutori-
ally useful view of a flange acting as a thermal sink is probably too simple.
It is also probable that thermal processes are only a part of what actually
happens to quench flame in a flange gap. It is well accepted that between
any solid surface and a combustion wave there is a thin layer of unburned
gas. This layer exists partially because of chilling by the solid wall, but its
Explosionproof Enclosures 139
existence is also due to inhibition of chain carriers. The combustion pro-

cess does not proceed in a single chemical-reaction step, but rather in a
series of steps. If the active molecules from an early step — the chain carri-
ers — are inhibited by contact with the wall, succeeding reactions do not
occur, and there is no combustion.
For conceptual purposes and qualitative evaluation of the effect of chang-

ing experimental parameters, the concept of quenching is a useful one.
Changes in geometry that affect heat transfer usually affect chemical
quenching similarly. Therefore, in the following sections the terms
“quenching” and “cooling” are used inclusively to describe what happens
at the flange gap, though the actual processes may not be simple thermal
ones.
The effects of some parametric variations in ignition experiments with

flanged enclosures are more readily explained in terms of Phillips' analy-
sis of the emerging jet than in terms of quenching. (Smith explained many
of his experimental results with similar concepts.) Other effects are more
directly rationalized in terms of quenching. In the following sections the
explanation that seems simplest is used; the objective is understanding,
not theoretical precision.
Though explosionproof enclosures have been studied since about 1906,

much about the mechanism by which they are effective is still not com-
pletely understood. The references listed at the end of the chapters are
informative but represent a limited survey of the literature.
Influence of Enclosure Design and Test Procedure

Previous paragraphs presented a simplified qualitative view of how an
explosionproof housing works. In this section the effect of changing the
test procedure or the enclosure design is considered in terms of that view.
Evaluating the effectiveness of explosionproof enclosures or studying
flange-gap quenching is still highly empirical. As in the study of other
combustion phenomena, results depend on the technique and equipment
used by the investigator. There are few absolute numbers in combustion
and ignition studies. Though minimum ignition energy, Wm, is probably
absolute, SIT is certainly not. With explosionproof housings, as with elec-
trical ignition, the work of different investigators must be compared and
evaluated carefully if one is not to be misled. Seeming differences in
results are often caused by differences in experimental or equipment
design and may not be of fundamental significance.
In the following sections the dependent variable of most interest is the

maximum experimental safe gap, MESG. MESG is measured by igniting a
gas mixture inside the test enclosure and observing whether the atmo-
sphere surrounding the enclosure ignites. If it does, the gap is made

smaller. The MESG is the largest gap size in a series of tests that does not
permit ignition outside the enclosure.
Explosion testing of this kind is even more tedious and time consuming
than electrical ignition testing. Only a limited number of tests can practi-
cally be run with a given set of conditions. When comparing and
interpreting data from different investigators, it is, therefore, desirable to
know how many tests were run at each gap width and the increment by
which gap width was changed. Smith identifies these as n and k. He fre-
quently used values of n = 10, k = 0.002 in. (0.05 mm) and n = 20,
k = 0.001 in. (0.025 mm) in his experiments. Interpretation of data for
experiments where n = 5 and k = 0.005 in. (0.13 mm) might lead to spe-
cious conclusions. On the other hand, values of n higher than 20 are
usually impractical. Some investigators report influence of parametric
variations in procedure or equipment design in terms of experimentally
determined probability of ignition. Unless n is between 10 and 20, the
curves of ignition probability versus MESG often show changes in slope.
These changes are irritating, though they may not prevent one from reach-
ing a valid conclusion from the data.
MESG data are not really complete unless, in addition to the values n and
k, the report states the composition of gases inside and outside the enclo-
sure (they are often different), the initial pressure and temperature of the
mixtures, the location of the ignition source, and a description of the
enclosure geometry. It is necessary to specify the location of the ignition
source because it affects the velocity of the gas through the flange gap. In
large chambers central ignition, especially of mixtures with high flame
speed, causes a high exit velocity, frequently sonic, in the flange gap,
which makes the exit jet highly turbulent. Ignition near the flange gap
results in lower velocity and less turbulent exit jets. Discussions in this
chapter, in the interest of brevity, are not this complete.
Composition of Gas Mixture

For type tests of enclosures and for investigating other aspects of the igni-
tion process the gas mixture inside the enclosure is often specified to be
the most incendive mixture. The mixture outside the enclosure is the most
readily ignitable mixture. It is difficult to define these except circularly.
For a specified ignition source, location, pressure, temperature, enclosure
geometry, and gas mixture outside the enclosure, the most incendive mix-
ture is the composition inside the enclosure that most readily ignites the
outside gas (i.e., the one that requires the smallest MESG). The most easily
ignitable mixture is the concentration of outside gas, other conditions as
before, that most readily ignites when the most incendive mixture is
ignited. Reference is made in some documents to testing with the mixture
that produces maximum explosion pressure. This is not usually the mix-
ture that is either most easily ignitable or most incendive.
All three mixtures can be precisely defined only for a given test apparatus
or enclosure geometry. For type testing purposes, many laboratories
always use the same standard test mixtures. The effects of enclosure
geometry, though measurable, are usually overshadowed by other safety
factors in the test series.
The precision of most experimental work is limited by the number of tests

that can practically be run rather than by tight control of all variables.
Therefore, many investigators use the same mixture inside and outside the
enclosure being tested, especially in studies of the influence of enclosure
geometry or other mechanical parameter.
An example of the kind of testing necessary to verify proper mixtures is

given in the Titman-Torry reference. These workers started with 9.5%
methane air (approximately stoichiometric) inside their chamber and
tested mixtures external to the chamber in increments of 0.5% over the
range 5.0 to 9.5%. At each concentration, ignition attempts were made at
gap widths of 0.050, 0.045, 0.040 and 0.035 in. (1.27, 1.14, 1.02, and
0.89 mm). The number of attempts is not clearly defined, but n appears to
be between 5 and 8. By examining the number of times ignition occurred
for each test condition, they concluded that the most readily ignitable mix-
ture lay between 5.5 and 8.5%. A set of similar tests were run with a 9%
internal mixture. They concluded the most readily ignitable mixture was
about 7.0%.
They repeated the test series with 7% composition outside and mixtures
inside from 8.0 to 10.5%. They concluded that mixtures between 9.0 and
10.0% were equally incendive, and they repeated the test series with 9.5%
methane-air inside, varying gap and outside composition. In this test
series 6.5% outside proved to be the most readily ignitable concentration.
Even though the precision of these determinations was not high (because
n was low), the amount of effort expended in trying to define the optimum
internal and external mixtures was considerable. It is understandable that
many workers assume that previously determined compositions for simi-
lar equipment will apply to the new investigation. Others have used a
single composition inside and outside the chamber for convenience in
investigating the effects of changing a single test parameter.
When type testing commercial enclosures, the procedure at some labora-

tories is to select a mixture near stoichiometric as a starting mixture and
independently vary the mixtures inside and outside the chamber to
bracket the mixture of highest explosion pressure and the most dangerous
combination of mixtures for transmission through the gap. The rationale
for this approach is that the internally generated pressure and likelihood
of ignition through the flange gap is a function of so many geometric vari-
ables that it is safest to sweep through a wide range of test gases. The
optimum will not be the same for all equipment.
Though it is not possible to define precisely the effects of gas composition

on MESG independent of other variables, it is possible to relate the effects
of composition to jet ignition. For the stoichiometric composition, fuel-air
ratio is defined as 1. Data in Lewis and Von Elbe indicate that for many
gases a 0.05 unit change in fuel-air ratio around 1 does not drastically
change flame temperature or speed. For fixed geometry of gap one would
not expect radical changes in the properties of the emerging jet as compo-
sition varies by a small amount around the stoichiometric mixture.
The rationale of the type tester who sweeps a range of compositions is

supported by the fact that changes in geometry or turbulence inside the
enclosure because of the nature of the enclosed equipment can markedly
affect pressure rise and burning speed. These alter the jet characteristic
and, therefore, the MESG to some extent.
Effect of Enclosure Volume
The picture of flange-gap suppression of an explosion presented qualita-

tively by Smith, and quantitatively by Phillips, does not clearly show the
effect of change in enclosure volume. One must estimate the influence of
volume changes by considering secondary effects. Decreasing the volume
of a flanged enclosure leads in the limit to a pair of parallel plates. One
would expect that the quenching distance of the gas mixture would then
represent the upper limit of flange-gap width for enclosures of very small
volume.
For enclosures of significant volume, one can attempt to estimate the effect
of varying volume by considering the effect on jet velocity. Both Phillips
and Smith have adopted the view that other parameters, especially jet size,
remaining unchanged, lower pressure results in lower velocity of emerg-
ing gas. There is less turbulent cooling of the jet, which leads to higher
ignition probability. To obtain the same entrainment rate requires a nar-
rower gap. Very small enclosures have a high surface-to-volume ratio and
high venting-area-to-volume ratio as compared to large enclosures, so that
one would conclude that pressure rise, the speed of the jet, and MESG
would all decrease as volume decreases. But a high surface-area-to-volume
ratio also increases thermal losses to the chamber walls. This decreases gas
temperature, which increases MESG. Lower velocity through the gap also
implies increased thermal loss in the gap and higher MESG.
The pressure effect also depends on the location of the ignition source.
There is relatively little effect of volume if the source is near the flange so
that the driving pressure is from the initial explosion process itself. When
ignition is in the center of a large chamber, the effect of volume is substan-
tial, because unburned gas is forced through the gap while the combustion
wave develops, forming a turbulent jet. Most data seem to indicate that
thermal losses predominate in small enclosures and that MESG increases
as volume decreases.
Smith's investigations of the effect of enclosure volume led him to the con-
clusion that with methane-air mixtures inside and outside the chamber the
effect of increase in volume (except when the ignition occurred near the
flange gap) was to increase safety (i.e., to increase MESG). When he
ignited an arc gas-air mixture inside the chamber (arc gas is 85% hydrogen
and 15% acetylene) with methane outside the chamber, the effect of
increase in volume was to decrease MESG. Work by Riddlestone led to the
conclusion that for hydrogen, acetylene, ethylene, and propane atmo-
spheres ignited near the flange, reduced volume leads to increased MESG.
His data extended over a volume range of 2 to 8,000 cc. Brown and Dainty,
working with hydrogen in volumes from 200 to 14,000 cc observed a
decrease in MESG with increase in volume. MESG dropped from 0.012 in.
(0.31 mm) to 0.0085 in. (0.22 mm) as volume increased from 200 to 1,000 cc
with no further decrease for additional increase in volume. Ignition was
near the flange.
There seems to be no ready explanation for Smith's conflicting results,

except that methane is often atypical of the other alkanes. However, the
consensus appears to be that small-volume enclosures require wider
MESG for ignition to occur. This is recognized in many national require-
ments for explosionproof enclosures in which, for the same flange width,
wider gaps are allowed for small enclosures than for large ones. Smaller
volumes decrease pressure, temperature, and flow rate past the flanges. If
the flanges can, therefore, quench more effectively, the MESG should be
larger. On the other hand, large-volume enclosures are capable of forcing
correspondingly larger volumes of hot gas past the flanges. To quench this
larger volume of gas the flanges must be closer together to compensate for
flange heating. That this effect cannot be the only important one is demon-
strated by the fact that MESG does not continue to decrease as volume is
made increasingly larger. Most probably this is due to the pressure in the
enclosure being high enough that exit velocity is sonic. Further increase in
volume or pressure cannot change velocity, only second-order thermal
losses to the walls.
Effect of Flange Width
When one views the functioning of an explosionproof housing primarily in

terms of flange quenching it is easy to reach the conclusion that decreasing
flange width should decrease MESG. A narrower flange gives less thermal
and chemical quenching, so the gap must be reduced. Phillips rationalizes
that the boundary layer on narrow flanges is not as thick as on wide
flanges. The jet is, therefore, larger, and for equal cooling the gap must be
reduced. He also showed the effect of increased thermal quenching.
All investigators agree that increasing flange width increases MESG.

Dainty and Brown showed that for a 3280-cm3 enclosure the MESG for
29.6% hydrogen increased from 0.15 to 0.29 mm when flange width was
changed from 6.3 to 25 mm.
Riddlestone found that for a 50-cm3 enclosure the ratio of MESG for 25-
mm-wide flanges to MESG for 6.3-mm-wide flanges was 1.4 to 1.5 for pro-
pane, ethylene, hydrogen, and acetylene.
Riddlestone and Nethercot reported a ratio of about 1.3 for MESG of 6.3-
and 3.2-mm flanges, for approximately 4% propane-air mixtures, and also
for a 5.7% propane, 27% oxygen, 67.3% nitrogen mixture. They quote data
from Smith as yielding MESG ratios of 2:1.33:1 for flange widths of 25, 8.3,
and 3.2 mm.
Titman and Torry reported a ratio of 1.15 for MESG in an 8-liter sphere
with 25- and 12.5-mm flanges, using methane-air mixtures.
UL data on diethyl ether showed that as flange width was varied from 9.5
to 19 to 31.8 mm, the MESG increased from 0.25 to 0.30 to 0.40 mm.
These data indicate that, other conditions of test remaining unchanged,

doubling the flange width increases gap by a factor near 1.3, up to about
25-mm flange breadth. Much, but not all, of Smith's data seems to agree
with this generality. Wider flanges are not common and little data are
available. The Titman-Torry data are supported by other references
quoted by Smith. These indicate that as the flange nears 25-mm width the
relative effect of widening decreases. Some early results referenced by
Smith suggested that flanges above 50 mm wide do not offer additional
advantage. This conclusion may depend on enclosure volume. Examina-
tion of Smith's data suggests that MESG is most strongly influenced by
flange width when ignition is near the gap and that for other locations of
the igniting source MESG varies less with flange width. This is consistent
with the view that velocity of gas through the gap is low when ignition is
near the flange because internal pressure is low. Variation in flange width
would then have relatively greater effect, through both heat loss and jet
characteristics, than when exit velocity is high.
Effect of Enclosure Shape
This brief discussion concerns only gross variations in enclosure shape, as

among a sphere, a cube, and parallelepipeds. Effects of arrangement of
internal hardware and the influence of interconnected volumes of similar
or different shape are considered under the discussion of pressure piling.
There is not much data. Dainty et al. investigated ignition by three 246-cc
enclosures of differing proportions. The width of all enclosures, and the
length of the gap was 76 mm, but the distances from the gap to the oppo-
site wall in the three boxes were 42, 84, and 169 mm. To achieve a 246-cc
volume, the heights of the boxes were 76, 38, and 19 mm, respectively. For
all tests, ignition was 12.5 mm from the gap in a stoichiometric hydro-
gen-air mixture. Flange width was 12.5 mm.
As expected, when the flange gaps were sealed the maximum pressure
was approximately the same in all boxes, the highest being in the box with
the shortest distance from the gap to the opposite wall. When vented
through the flange gap, the pressure dropped. The drop was least in the
box with the gap and opposite wall closest together and greatest for the
box with the greatest distance to the opposite wall. Lower pressure corre-
lated with increased MESG.
These results may be rationalized in several ways. The large sur-

face-to-volume ratio of the elongated enclosures increases thermal losses;
therefore, it decreases both the pressure driving the gas jet through the
gap and the jet temperature. From the point of view of quenching, both
imply wider MESG. If one considers critical entrainment rate, lower jet
temperature also implies wider gaps but lower pressure may reduce
MESG. It appears that effects of lower jet temperature override the effect
of pressure decrease on jet characteristics.
Dainty's data also show that the time to reach maximum pressure was
longer in the shapes with lower maximum pressure. The combustion
wave had to travel from the ignition point near the gap to the back corner
of the volume before combustion was complete. Because the volume was
the same for all shapes, lower pressure implies lower gas temperature,
which is in the direction of larger MESG.
Explosion Pressure and Rate of Pressure Rise
Measurement of maximum explosion pressure and rate of pressure rise

are often secondary measurements in studies of MESG. Oftentimes neither
quantity is measured. Few generalizations are possible. Both quantities
depend strongly on enclosure geometry and the mixture being tested.
The most significant fact is that, in general, higher maximum pressures

and high rates of pressure rise are associated with fast-burning mixtures
and small MESG. Table 5-1 lists pressure rise in a sealed chamber, the time
to maximum pressure, and the MESG for some typical materials. These
data are from IEC Publication 79. Pressure rise and time to Pmax are mea-
sured in a 4-liter sealed spherical chamber with 25 mm flanges. MESG was
measured in an 8-liter spherical chamber. The mixture for measurement of
Pmax is optimum for pressure development. MESG is measured with the
most incendive mixture inside the enclosure and the most easily ignited
mixture outside. The pressures listed in Table 5-1 are indicative only. The
pressure actually reached during a test will depend on enclosure geome-
try and size and on the arrangement of the equipment in the enclosure.
Pmax—lb/in2 Time to Pmax

Material MESG—in. (mm)
(KPa) ms
Methane 102 (704) 70 0.046 (1.17)
Propane 122 (842) 46 0.038 (0.96)
Pentane 123 (847) 53 0.040 (1.00)
Ether 131 (804) 49 0.034 (0.86)
Ethylene 126 (870) 25 0.028 (0.71)
Hydrogen 105 (725) 7 0.004 (0.10)
Acetylene 146 (1002) 14 0.001 (0.025)
Table 5-1 Explosion Parameters—Closed Chamber
In an enclosure of simple geometry, venting caused by MESG can be con-

siderable. In the SMRE Review of Progress, 1943, it was shown that, in an
8-liter sphere, the venting effect of the MESG reduced maximum pressures
of slow-burning materials like octane, pentane, and methane to a few
pounds per square inch. The MESG for hydrogen is quite small, and vent-
ing by the MESG did not reduce pressure build-up significantly. Another
way of summarizing the data is that for fixed-gap width, reduction of
pressure was greatest for methane, the slowest-burning material, and least
for hydrogen, the fastest-burning material.
Data in UL Bulletin of Research No. 58 show maximum explosion pres-

sures with MESG venting that are considerably higher than those reported
in IEC Publication 79. The ratio of enclosure volume to gap area was about
25 times that of the standard 8-liter sphere, so venting effects were smaller.
Kisselstein reported maximum pressure measurements in 12 commercial

enclosures tested with gasoline, ether, hydrogen, or acetylene. Each enclo-
sure was tested with at least two of the materials. Acetylene, in any
enclosure, produced the highest pressure. The other materials were, in
descending order: hydrogen, ether and gasoline. However, the values of
Pmax varied considerably among enclosures. The variations were as
follows:
• Acetylene: 228 to 485 psi (1570 to 3350 kPa)
• Hydrogen: 168 to 352 psi (1160 to 2440 kPa)
• Ether: 70 to 230 psi (480 to 1590 kPa)
• Gasoline: 72 to 180 psi (495 to 1240 kPa)
The enclosure that produced the highest pressure with acetylene gener-
ated relatively low pressure when tested with gasoline. The highest
pressure developed with gasoline was in an enclosure that generated the
lowest acetylene pressure. His data illustrated well the influence of enclo-
sure geometry and the nature of the enclosed equipment on pressure rise.
Since some of the pressures are far above those listed in IEC Publication
79, it is possible that pressure piling occurred in some enclosures.
Theory does not relate maximum pressure rise and rate of pressure rise
directly to MESG. However, the materials with low MESG also have small
quenching distances, high flame speeds, and high flame temperatures. All
these phenomena are related to low activation energy. Phillips, in his
treatment of jet entrainment, showed that the critical entrainment rate is
higher for materials with low activation energies. MESG is lower because
the entrainment rate is greater in a small jet.
A secondary effect of maximum developed pressure, which is of consider-

able practical importance, is distortion of the flange gap. Many research
investigations are carried out with flanges constrained so that no likely
explosion pressure will increase gap width. In commercial housings, how-
ever, pressure bows the flange between bolts. The gap during the
explosion is wider than when measured before or after the explosion. As
Dainty and Brown have shown, the apparent safe gap between bolted
flanges of usual commercial construction is considerably smaller than
would be predicted from tests on immovable flanges. “MESG” data from
enclosures that may deform during the explosion must be used with care.
Smith gives a lucid explanation of tests at the Safety in Mines Research

Establishment to determine what safety factor must be applied in hydro-
static testing to properly simulate stress during the explosion. Comparison
of the deflection of an enclosure during an explosion and during hydro-
static testing showed that an enclosure of simple configuration is not
overstressed by the rapid rate of pressure rise during an explosion. There-
fore, a hydrostatic pressure test can be equated directly to the stress of the
maximum pressure observed during an explosion. However, Smith's
work was based on enclosures in which no pressure piling occurred.
Riddlestone showed that if pressure piling or turbulence causes a higher

rate of pressure rise than is typical in enclosures of simple geometry such
as discussed by Smith, the stress in the walls of the enclosure is enhanced.
The enhancement of stress was calculated to be a factor of 1.7 for pen-
tane-air mixtures and as high as 2 for a 85/15 hydrogen/methane-air
mixture. He proposed that when the pressure rise time is 5 ms or less, rou-
tine hydrostatic pressure testing should be based on twice the maximum
smoothed peak pressure measured during explosion testing rather than
1.5 times.
Bartels reported the results of extensive tests of short-rise-time explosions

and compared them with calculated values. Though most of the experi-
mental points were lower than calculated values, the scatter of the data
was such that the calculated values represent a reasonable estimate of the
maximum strain enhancement that might be experienced. For very high
rates of pressure rise in enclosures with a high resonant frequency of the
walls, the strain enhancement approaches 2, as calculated by Riddlestone.
Bartels' data shows maximum actual enhancement ratio of 1.75 for oxy-
gen-enriched hydrogen atmospheres. Because the enhancement ratio
depends on the ratio of pressure rise time to the natural frequency of the
enclosure wall, enhancement is likely to be greatest in large enclosures.
Effect of Ignition Source Location
It is generally accepted that location of the ignition source near the flange
gap, in the plane of the gap, yields the narrowest MESG. Most investigators
test with an igniting spark gap 0.4 to 0.5 in. (10 to 12.5 mm) from the gap.
Phillips asserted that when the ignition source is near the gap the combus-
tion wave passes through it at low pressure into undisturbed gas.
Conditions are most favorable for ignition because entrainment cooling is
minimal. He argued that when the ignition source is far from the gap, pas-
sage of unburned gases through the gap while pressure in the enclosure
rises, generates turbulence that cools the hot gases when they enter the jet.
In later work Phillips revised this position, as it was discovered that hydro-
gen-air mixtures had the same MESG with both central and side ignition in
an 8-liter sphere, but the pressure rises were significantly different. He
went on to show that if pressure in the chamber were high enough, as
might be the case were the hydrogen-air mixture enriched by oxygen (as in
a battery container), the safe gap would fall below the conventionally mea-
sured MESG. In pursuit of amendments to his theory, he also modified the
treatment of the entrainment at the head of the emerging jet. The signifi-
cance of this work is not great for normal applications of flameproof
enclosures, but it is another step towards a quantitative understanding of
the flange-gap explosion quenching process.
Smith experimented with ignition of methane-air at several points in an

upright cylindrical chamber with a circumferential gap at one end. He
found that MESG was smallest, about 0.05 in. (1.25 mm), and little affected
by pressure venting in the opposite end of the cylinder when the source
was located near the gap. The gap was widest, about 0.09 in. (2.25 mm),
when the source was placed in the geometric center of the cylinder with-
out pressure relief and decreased to 0.05 in. (1.25 mm), with substantial
pressure relief. Ignition near the pressure-relief vent produced MESG, 0.08
to 0.09 in. (2 to 2.25 mm.), with little pressure relief, but the MESG
decreased rapidly as explosion pressure was lowered by pressure relief.
Ignition near the center of the gap-end of the cylinder produced results
intermediate between central and gap location of the source.
Smith's explanation is similar to that of Phillips. He argued that central

ignition caused the flame to pass through the gap at the same time that
pressure peaked so that turbulence of the emergent jet was maximum.
When the chamber was vented, pressure was reduced; turbulence was
also reduced, and a smaller MESG was measured. With ignition at the bot-
tom of the cylinder and no venting, cooling behind the flame front
dropped explosion pressure slightly, but turbulence was still high when
the flame reached the gap. Venting dropped the pressure and reduced the
MESG.
Dainty et al. investigated energy loss to enclosure walls from hydrogen

flames. They concluded that the maximum pressure and rate of rise in a
given enclosure depend on the location of the ignition source. If, as in a
long enclosure with ignition at one end, the flame makes substantial con-
tact with the walls while traversing the length of the enclosure, energy will
be lost and Pmax, when the flame has completely traversed the box, will be
lower. The rate of rise will also be lower. They demonstrated this by ignit-
ing in the center of a cylinder and near a wall midway between the
cylinder ends. In these investigations the enclosure was sealed, so MESG
was not measured. These pressure measurements seem, however, to be in
general agreement with Smith's explanation.
Effect of Flange Material
Most explosionproof enclosures are metallic; usually cast iron, aluminum,

or formed steel. But there is no assurance in the literature that metallic
construction is essential for function. Fleisig reviewed the development of
small enclosures, 2,000 to 105 mm3, in steatite ceramic, phenolic resins,
and glass-fiber-reinforced polyester. Such enclosures are typically used for
switches. A tight enclosure of plastic or metal protects it from the elements
and provides termination space. Most certification authorities will now
accept nonmetallic enclosures for examination.
Lewis and von Elbe's observation that material is not significant in deter-
mining quenching distance in spark ignition between flanges would lead
one to suspect that flange material may be of secondary importance in
determining MESG. Smith refers to Staples' work, which showed that
bronze flanges and bakelite flanges gave essentially equal MESG.
There is room for considerable progress in investigation of new materials

and manufacturing techniques for explosionproof enclosures.
A secondary effect of enclosure material of present concern to many

authorities is the possibility of ignition by frictional sparks from
light-alloy enclosures or by ejection of light-alloy particles from an enclo-
sure. Neither concern is of particular importance in instrument
applications.
Ignition by frictional sparks generated by contact of iron or steel tools with

aluminum enclosures, especially if tools are rusty, has been the subject of
much investigation. Most national authorities limit magnesium content of
enclosures to less than a few percent.
Though not all investigators have reached the same conclusions regarding
the degree of hazard posed by frictional sparks, all agree that if ignition is
to occur at all, it is produced by relatively high velocity or high-energy
impacts. This matter is of importance in mines, where such accidental con-
tact from picks and cutters is likely, but is not of significance in
fixed-instrument installations where high-impact contacts are unlikely.
It has been shown that ignition of an external flammable mixture may be

caused by ejection of hot particles or vapor from power arcs inside an
explosionproof enclosure. Aluminum electrodes appear to be somewhat
more likely to cause ignition than copper. It has been proposed also that
collision between an ejected particle and the flange may dislodge fine alu-
minum particles that then oxidize exothermically, igniting the external
flammable vapor. No consensus presently exists about measures that must
be taken to prevent ignition by any of these mechanisms. Many mining
authorities, as well as other U. S. authorities, test explosionproof appara-

tus, especially circuit breakers, under overload conditions. However, this
hazard is not pertinent to instruments.
Silver and Brown have shown that aluminum particles trapped between
flanges are a potential hazard. Should their geometry and mode of entrap-
ment permit them to be ignited by escaping hot gases, then discharged
into the external flammable atmosphere, ignition could occur, even
though the explosionproof housing would normally contain the explo-
sion. The hazard is most severe with an easily ignited gas like hydrogen,
but ignition of methane atmospheres has been demonstrated.
Internal Temperature and Pressure
Initially, a pressure higher than the atmospheric pressure inside a com-

mercial explosionproof enclosure usually must be contrived. Most often
the flange gap leak will prevent pressure inside the case from rising above
atmospheric, except transiently. However, if gasketing closes off the gap
and the enclosed equipment provides a means of raising pressure inside
the enclosure, the resulting explosion pressure will, as a rule of thumb, be
proportionately higher. If Pmax is 100 psia for 1 atm initial pressure, Pmax
will be 200 psia for 2 atm initial pressure.
Phillips reported data from Grobleben which showed that increasing ini-
tial pressure decreased MESG significantly. Increasing the initial pressure
to 2 atm approximately halved the MESG for methane, town gas, or
hydrogen mixed with air.
Precompression as the result of an explosion is considered under pressure

piling, which is discussed later in this chapter.
If thermal quenching is the prime mechanism for safety, raising the initial
temperature of an enclosure would lead to higher probability of explosion
transmission or lower MESG. Smith refers to Staples' work showing that
flange temperature of 110°C did not affect MESG and to James' work
showing that 427°C flange temperature assisted passage of flame. Phillips
theorizes that because raising initial gas temperature raises flame temper-
ature and reaction rate, the MESG should be smaller. He refers to the work
of James and Beardshall, et al. The latter found that increase in tempera-
ture from 293°K to 523°K decreased the size of an orifice through which
pentane-air mixtures would ignite a flammable atmosphere from 0.11 in.
to 0.08 in. He reports also on Nabert's work, which showed that an
increase in temperature from 20 to 250°C reduced the gap for methane-air
from 1.12 to 0.94 mm.
Obstacles in Flange Gap
Woodhead, in commenting on Phillips' work (1962), presented data that

showed that the MESG for pentane was not greatly affected by obstacles
near the gap. Holding a 1-in. (25-mm) wide iron band 0.25 in. (6 mm) from
the gap or closing the gap with adhesive tape or a rubber band decreased
MESG slightly. For ethylene, such measures reduced the MESG to 40% to
50% of the unobstructed value. With hydrogen similar reductions were
observed, but wrapping the gap with tape dropped MESG to 20% of the
unobstructed value. All tests were made with the standard 8-liter spheri-
cal enclosure, with central ignition to achieve high explosion pressures. In
the case of pentane and ethylene, MESG did not fall below the maximum
permissible gap, which is approximately 40% of the unobstructed MESG.
It is not likely that use of gaskets or O-rings to make explosionproof enclo-

sures weather-tight has any significant effect on safety of enclosures
designed to U. S. standards. Permissible gaps are 0.003 in. to 0.0045 in.
(0.076 to 0.114 mm), based on no flame transmission rather than on no
explosion transmission, Therefore, the initial safety factor is quite high. A
1996 draft of a harmonized North American explosionproof enclosure stan-
dard (which will replace the individual standards of CSA, FM, and UL)
mandates that installation instructions for enclosures with gaps larger than
those given in the table and accepted by test shall warn the installer to keep
the flange gap opening a specified minimum distance from any obstacle.
Effect of Humidity
Phillips calculated a linear effect of humidity on MESG. He predicted an

MESG increase from 0.046 in. (1.2 mm) to 0.054 in.(1.4 mm) for methane,
as moisture content changed from zero to 4%. Moisture reduces flame
temperature and reaction rates as well as reducing the critical entrainment
quantity, which increases safe gap.
Torry found that a 1.3% increase in moisture raised MESG from 0.04 in. to
0.047 in. (1.1 mm to 1.2 mm), a rate of change close to Phillips' calculation.
Using Phillips' entrainment formula, Dainty calculated that a zero to 100%

relative humidity change in a stoichiometric hydrogen-air mixture would
increase MESG by 0.0007 in. (0.00018 mm). Nominal MESG is about 0.01 in.
(0.25 mm).
Turbulence Outside an Enclosure
Turbulence of the combustible mixture external to an enclosure will usu-

ally increase heat loss from the hot jet, increasing MESG. Smith showed
that a fan stirring the methane mixture in a 2-ft (0.6-m) cubic chamber sur-
rounding an 8-liter sphere increased MESG from 0.046 in. to 0.048 in.
(1.17 mm to 1.22 mm).
Pressure Piling
In the preceding sections, discussion was limited to enclosures of simple
geometry in which explosion pressures of 100 psi (700 kPa) are typical. In
practice, much higher pressures can be generated in enclosures, either by
“pressure piling” or by detonation.
“Pressure piling” is an increase in explosion pressure caused by precom-

pression of the gas before ignition. Explosion pressures in closed
chambers are typically 5 atm to 8 atm if the gas is initially at 1 atm. Raising
the initial pressure to 2 atm raises explosion pressure from 10 atm to 16
atm. Explosion pressure is approximately proportional to initial pressure.
If two chambers are connected by a pipe, ignition in one chamber at the

wall farthest from the pipe will cause abnormally high pressures in the
second chamber. Expanded combustion products push unburned gas
through the pipe into the second chamber, raising the pressure. When the
combustion wave enters the second chamber, it ignites precompressed
gas. The resulting explosion pressure may be significantly higher than
would be observed in a single, closed volume. A similar effect would be
observed if ignition occurs at the end of a long conduit so that the combus-
tion wave pushes gas from the conduit into the enclosure.
Not only is the absolute explosion pressure higher, but the rate of change
of pressure is higher. As unburned gas ahead of the advancing combus-
tion wave is compressed, its temperature is raised. Much less energy must
be extracted from the approaching combustion wave to raise a new layer
of the preheated gas to the ignition temperature, and the flame, therefore,
propagates at a faster rate.
If, as a flame propagates, interaction between the combustion wave and

conduit walls produces turbulence and enhanced mixing, further
increased flame speed may produce shock waves that will propagate
through the conduit, reflecting off the walls and further increasing turbu-
lence at points of interaction between the shock wave and combustion
wave. This results in transition from a thermal combustion wave to a deto-
nation wave. Velocity increases from tens of feet per second to hundreds
or even thousands of feet per second. The prime requisite for transition
from a combustion wave to a detonation wave in situations of interest to
instrument users is the existence of a tube or conduit of length many times
the diameter. Conduit connecting explosionproof housings fits this
requirement remarkably well, and it is partially out of concern for pres-
sure piling and detonation that the NEC stipulates that conduit entering
or leaving most explosionproof housings be suitably sealed. In rotating
machinery, mixing and compression of gases is aided by mechanical rota-
tion, and higher flame speeds and detonation may occur.
Kisselstein reported that ignition of gasoline at the end of 5-, 10-, and 15-ft
(1-1/2, 3, and 4.5 m) lengths of 1-1/2-in. conduit produced pressures of
117, 125, and 110 psi (810, 860 and 760 kPa) in a 3,500 in.3 (0.05 m3) enclo-
sure. He did not specify the pressure in the enclosure without conduit, but
85 to 100 psi is probably typical. These pressure rises are not startling, but
the conduit volumes are approximately 110, 220 and 330 in.3, thus one
would only expect roughly 3, 6, and 9% increases in pressure due to
precompression.
That the pressure rises were small indicates that no detonation occurred in
these tests. One would expect much higher explosion pressures than
caused by volume ratio alone had detonation occurred.
Silver showed that ignition of pentane-air mixtures at the end of 20 in. of

0.5-in. conduit did not enhance pressure rise in enclosures of 15 to 1728 in.3
volume. In some cases the increased volume lowered the pressure rise.
Since the conduit volume is only 25% of the smallest enclosure volume, no
large increase would be expected.
Examination of data in UL Bulletin of Research No. 58 shows that ignition

at the end of 10 ft of 1-1/2-in. rigid conduit substantially raised pressure in
the test chamber. Because the ratio of conduit volume to chamber volume
was approximately 1:8, it is probable that detonation occurred in these
tests. Conduit raised pressure by factors up to about 2 when relatively
slow-burning materials, such as gasoline, diethyl ether, butane, propane,
and ethylene, ignited. Hydrogen and acetylene, notable for high flame
speed, small MESG, and susceptibility to detonation, when ignited at the
end of the conduit, caused chamber pressure rises up to 10 times normal.
However, consistent with Phillips' and Smith's view of jet ignition, the
MESG was never smaller when these high pressures were developed than
when ignition took place near the flange gap and the conduit was not
connected.
Gleim and Marcy investigated pressure piling of methane-air mixtures in

a box 50 in. long by 1 ft square that could be subdivided by a partition into
volumes in 1:1, 3:1, and 7:1 ratios. A central round or square hole through
the partition connected the two volumes. They did not attempt to draw
general conclusions from the data. In the 1:1 configuration, there was
slight pressure piling for holes of 2 in. to 10 in. diameter, the second com-
partment pressure being about 10% higher than the one in which ignition
took place. Pressure in the ignition compartment was somewhat higher
than when there was no partition and ignition took place in a single long
chamber. This may, of course, be because of the lower surface-to-volume
ratio of the smaller compartment. With the compartments' volumes in the
ratio of 3:1, pressure in the smaller compartment, when ignition in the
large compartment was distant from the partition, was about 40% higher
than in the large compartment. When ignition occurred in the large com-
partment, the volumes being in 7:1 ratio, there was no well-defined piling,
but there were pressure surges. That the pressure rises, even with pressure
piling, are lower than would be predicted from volume ratios may have
been due to the central location of the communicating hole. Advance of
the combustion wave would not push all unburned gas into the second
chamber as through a conduit. The central partition may have set up some
turbulence and back-swirl, which would raise pressure in the ignition
chamber. These tests were carried out with methane, a relatively
slow-burning gas, so that losses from the side walls would be much more
important than with a fast-burning gas like hydrogen.
The highest pressure rises in commercial equipment occur in large,

fan-cooled motors. The end bells provide chambers of large vol-
ume-to-surface ratio. Cooling ducts in stator and rotor provide access
from one end bell to another. Fan generated turbulence aids rapid com-
bustion and transports flame from one end of the machine to another.
Turbulence may or may not cause detonation, but pressure piling is com-
mon, often accompanied by pressure surges and oscillation.
In the British and UL investigations listed in the references, no pressure

piling occurred when the rotor was stationary. Equal pressures developed
in both end bells. In one series of UL tests, where the only communication
between end bells was through the 0.095 in. radial clearance between sta-
tor and rotor, no ignition occurred in the opposite end bell. The 3.5-ft-long
gap effectively quenched the flame when the rotor was stationary.
All investigators found that as rotor speed increased, explosion pressure

in the end bell opposite to where ignition occurred also increased. In the
UL tests it was possible to increase the free volume inside the dummy test
motor to 40 ft3. Pressures in both end bells increased, probably because
greater volume to surface area reduced losses during pressure rise. In the
UL tests, pressure rise reached 500 psi with gasoline-air mixtures. British
experiments with pentane-air mixtures, 85% hydrogen/15% methane-air,
and methane-air mixtures did not produce the high pressures of the UL
tests. The British test apparatus was a commercial 800-hp, core-cooled
motor fitted with a dummy rotor. The free volume at each end bell was
about 12 ft3. Losses to the casing perhaps explain the lower pressure. For
most but not all rotor speeds, the pressures developed in both end bells
were in the same order as flame speed—methane, pentane, hydrogen/
methane in increasing order. High flame speed leads to higher explosion
pressure, even in a simple chamber without pressure piling, because the
amount of heat loss to surroundings is less in the shorter time it takes to
reach maximum pressure.
Conn’s report gives some helpful insights into the mechanism of pressure
piling. The combustion wave in the primary chamber is distorted by losses
to the walls, the geometry of the chamber, etc., so that at the instant the
combustion wave reaches the orifice entering into the second chamber all
the material in the primary chamber has not yet been burned. The flame
front is not planar and perpendicular to the axis of the primary chamber
and therefore does not act like a cylindrical piston to push unburned gas-
ses into the receiving chamber. Rather, the piston action is inefficient
because the flame front may be ellipsoidal or otherwise deformed from a
plane. Thus the pressure rise is not always a simple function of the ratio of
volumes of the primary and secondary chambers.
To account for this fact in his mathematical model, Conn assumes that a
fraction of the primary volume is burned and, at the instant that combus-
tion of that volume is complete, the unburned gases in the remainder of the
primary volume flow into the second chamber, pressurizing it. At the same
instant the hot burning gases ignite the prepressurized gases in the recep-
tor chamber. Conn developed this assumption by dividing the primary
volume Vp into two volumes, V1 and V2 by a magic wall that disappears as
soon as all the gas in V1 has burned. V1 is the more distant from the
receiver volume. He postulates that the maximum pressure piling occurs
when ignition of the gases in the receiver vessel occurs at the instant the
pressure in the primary and receiver vessels is the same, that is, the pres-
sure rise in the total primary chamber volume caused by burning the gases
in V1 equals the pressure rise in the receiver volume caused by forcing the
volume of unburned gases, V2, into it. An iterative computation is possible
because for any given primary to receiver volume ratio Vp/VR there can be
only one combustion expansion ratio that satisfies the postulate. There is
only one possible location for the magic wall that will equilibrate the pres-
sures at the same time that all the gas in V1 has burned.
Conn states that his method of calculation yields higher maximum pres-
sures than are usually experienced in practice. Losses of energy in pushing
gas through the connecting orifice, frictional and thermal losses, and back-
flow from the secondary chamber to the primary chamber as combustion
takes place in the receptor chamber are not taken into account.
If the correct mixture is used, and ignition takes place in the correct spot to
provide the optimum expansion ratio, Vp/V1, the explosion pressure in the
receptor chamber increases as the ratio Vp/VR increases, as simplistic think-
ing argues. Conn notes that it is difficult to locate an ignition source to

obtain an expansion ratio of less than about 1.5, the optimum for a Vp/VR
ratio above 6. In practice, therefore, high pressures are not to be expected
with methane-air mixtures. For Vp/VR of 6 his calculation yields a peak
pressure of about 350 psi, which, as noted above, is not to be expected in
practice.
Polka investigated pressure piling in two 5-liter chambers connected by a

35-cm length of 1-1/2 in. pipe. A reducer constricted the entry into the sec-
ond chamber to 1 in. in diameter. A 100-mm-diameter barrier plate
perpendicular to the axis of the connecting pipe could be arranged to be at
different distances from the entry to the second chamber. Both chambers
had an internal height of 185 mm.
Without the barrier plate turbulence of the entering gases prepressurized

the second chamber before they were ignited, and the final pressure was
above that without pressure piling. In the case of a 6% propane-air mix-
ture the prepressurization, was to 1.8 bar abs. and the maximum pressure
in the second chamber was about 14 bar abs.; the maximum pressure in
the first chamber being 8 bar abs. Insertion of the plate to within 10 mm of
the entry did not change the prepressurization, but it dampened the tur-
bulence at the entry and reduced the final pressure to 9.3 bar abs. Tests
with acetylene-air and hydrogen-air mixtures confirmed the role of turbu-
lence in causing enhanced pressure in the second chamber and the
turbulence-dampening effect of the plate at the entry.
Correlation Between MESG and Spark Ignition

Energy
Although MESG determinations and spark ignition energy determina-
tions are measurements under quite different conditions of combustion
and heat transfer, it is not unreasonable to expect that the two phenomena
are closely related. The characteristics of high flame temperature and low
activation energy lead to low MESG also lead to low igniting current in
inductive break spark ignition, low minimum ignition energy, and small
quenching distances. The relationship between MESG and quenching dis-
tance was noted earlier in this chapter. In Chapter 2 the correlation
between minimum ignition energy and quenching distance was dis-
cussed. Correlation between MESG and spark ignition measurements is,
therefore, to be expected.
Slack reported on work at the Electrical Research Association to correlate

MESG and MIC, minimum igniting current, as determined by break-spark
apparatus #2 (similar to #3 at the Safety in Mines Research Establishment,
which is discussed in Chapter 9). For convenience, the MIC used for corre-
lation was that measured in a 24 Vdc, 0.095 H circuit. He concluded that

correlation between MESG and MIC is linear for MESG in the range 0.008
to 0.046 in. (0.2 to 1.17 mm) and MIC in the range 0.06 to 2 A.
The motivation for this effort was that many more materials had been clas-
sified in groups determined by MIC than have been classified by MESG. A
good correlation would permit classification of many additional materials
by MESG grouping, without testing, because MIC data are already
available.
Slack's discussion of the goodness of the correlation is not detailed here. In

summary, the correlation is good, but not sufficiently good that one can
calculate MIC from known MESG, or vice versa, to high accuracy. How-
ever, if one classifies materials in the standard MESG groupings based on
known MIC data, the grouping is correct with very high probability.
Similar work by Helwig and Nabert developed correlations among MESG,

quenching distance, MIC, and minimum ignition energy. MESG measure-
ments were made in a 20-cm3 chamber and are said to agree closely with
values determined in an 8-liter chamber. Quenching distance was deter-
mined in a 0.3-cm3 chamber. Values are claimed to agree closely with
those of Lewis and von Elbe. MIC measurements were made with the
standard IEC test apparatus. Minimum ignition energies were those of
Lewis and von Elbe. The ratio of quenching distance to MESG for 20 mate-
rials was found to average about 1.98. The most discrepant value was 3.06
for carbon disulfide. All others lay in the range 1.76 to 2.1.
Correlations were also found for MESG and quenching distance against
MIC, and for minimum ignition energy versus inductively stored energy
in MIC determinations.
The investigators concluded that one should not use any of the correla-
tions for the purpose of calculating one parameter from another known
parameter. However, a preliminary calculation is a good way to get an
approximate starting point to shorten an empirical determination.
Classification of materials into groups implies establishing arbitrary divid-

ing lines. Table 5-2 shows MESG and MIC (British) data for some typical
materials in NEC Groups D, C, B, and A. Note particularly the data for
propane and diethyl ether. These materials are also classified in adjacent
groups by the British with respect to both MIC and MESG. Since grouping
implies drawing arbitrary dividing lines in a continuum of characteristic
values, there must always be a case where some materials in adjacent
groups are not substantially different. The penalty of small error in calcu-
lating MESG from MIC would at worst be misclassification by one group.
For borderline materials, even empirical testing leaves room for the same
error.
Material NEC Group MESG—in. (mm) MIC (A)

Methyl Alcohol D 0.036 (0.91) 0.167
Pentane D 0.039 (0.99) 0.164
Butane D 0.042 (1.13) 0.162
Propane D 0.038 (0.97) 0.146
Diethyl Ether C 0.033 (0.87) 0.145
Ethylene C 0.027 (0.69) 0.108
Hydrogen B 0.011 (0.28) 0.075
Acetylene A 0.010 (0.25) 0.060
Table 5-2 Correlation Among Grouping, MESG, and MIC
Design Criteria—Introduction
The design criteria for explosionproof enclosures in most nations have
been derived either from North American practice, from IEC standards, or
from CENELEC standards. These standards are continually changing so
the designer should consult the latest edition of the relevant standards
before making design decisions. Some North American standards are
slowly being harmonized with IEC standards. CENELEC and IEC stan-
dards continue to become more alike. In those nations that are not forced
by policy or fiat to adopt CENELEC standards as the basis for national
standards the pace of change is necessarily slow. There is always great
reluctance to change practices that have been in use for decades, especially
if they developed to meet perceived national needs or because of unique
national history. Most nations outside of Europe and North America now
pattern their standards after IEC standards. Because IEC and CENELEC
standards are fast being harmonized, it is often accurate to state that a
requirement has been based on European practice or on North American
practice. In the sections that follow emphasis is on enclosure design for
industrial use. CENELEC standards specify rules for enclosures for mining
applications (Group I apparatus) which are usually more onerous than
those for industrial apparatus (Group II apparatus). In most countries the
standards for mining apparatus are distinct from those for industrial appa-
ratus and the approval of such apparatus may be the responsibility of a
government agency. In Canada mining apparatus comes under the pur-
view of the Department of Energy, Mines, and Resources. In the U. S., the
mining requirements are published in the Code of Federal Regulations,
Title 30.
It is still reasonably accurate to state that many issues can be generalized

on the basis of European practice and North American practice, although
Canadian requirements have moved closer to IEC requirements. One
major difference in the practices is in attitude toward testing. North Amer-

ican practice has been to type test with a substantial safety factor.
European practice is to reduce the safety factor in prototype tests and
insist on routine pressure testing of each enclosure. North American prac-
tice does not require routine production pressure testing but CSA
standards recognize routine testing as an option.
An obvious difference in approach to requirements, the requirements for

gap width, is the result of the differing view toward testing and inspec-
tion, and also a different view of the function of an explosionproof
housing. Permissible gaps in North American requirements are, in gen-
eral, substantially smaller than those permitted in European requirements.
The latter are traceable directly to MESG measurements, being on the
order of 40 to 50% of MESG. In British practice a statistical safe gap was
calculated from the relationship of ignition probability to experimental
gap. Permissible gap was 60% of the statistical safe gap.
Phillips has suggested that the extrapolation of MESG data to a statistical

safe gap is very conservative, that is, the extrapolated values of statistical
safe gap are too small. He argues that the extrapolation is based on an
assumption of a normal probability distribution. This leads to the errone-
ous conclusion that there is a finite probability of explosion transmission
for zero gap width. He suggests that a Weibull distribution should yield
more realistic values of statistically safe gap.
Permissible gaps specified in UL standards are in part traceable to UL Bul-

letin of Research No. 46. These are specified as limiting passage of flame.
This difference in viewpoint does not lead to a really different standard of

manufacture, however, except in the case of shafts for power transmission.
To the author's knowledge all certifying authorities expect that a flange
gap will be as tight fitting as good machining practice permits, that is,
North American practice, and would not approve a flanged enclosure
with gaps deliberately designed to the maximum values without substan-
tial reason. In the case of shafts for transmission of power, though,
Europeans recognize design with large gap to be a practical necessity.
Requirements for radial clearance are a compromise between a safe gap
for explosionproofing and sufficient clearance so that shaft-bearing eccen-
tricity does not cause binding and overheating.
Design Criteria
The following sections summarize the major requirements for explosion-

proof enclosures, using UL1203, FM Approval Standards Class 3600 and
3615, and CSA standard C22.2 No. 30 as references for North American
practice. CENELEC Standards EN 50014 and EN 50018 are used for Euro-
pean practice. Standards of most other countries are similar to the ones
referenced in this chapter. A harmonized North American standard is
being prepared under the auspices of NEMA. This document is intended
to replace the present CSA, FM, and UL standards. It is not expected that
the requirements will differ greatly from those presented here. The docu-
ment will also provide a mechanism for testing and accepting enclosures
designed to IEC and CENELEC standards.
The designer should always consult the standards to which the product
must conform. The presentation in this book aims to give a designer an
overview of the details of importance, but standards change. If there are
areas of a design that are close to the limits of a standard, it is wise to con-
sult the evaluating laboratory. There are some serious differences in
application of the same standards language among the various laborato-
ries. For example, a wall dividing a large enclosure into two smaller
volumes will be treated as an explosionproof boundary by some laborato-
ries, but others will demand only that the wall does not allow an explosion
to pass, and has strength to withstand the static pressure testing of the two
volumes.
Materials
Enclosures
Enclosures must be of substantial construction, must completely enclose

the electrical apparatus and all connections, and must be designed to meet
the normal conditions of the intended use and environment.
Although explosionproof enclosures are usually manufactured of metal,

all testing authorities will accept enclosures that are made of nonmetallic
materials for evaluation. Some agencies also place limits on the kind of
metal that may be used.
CSA
Materials other than iron, copper, aluminum or their alloys are investi-
gated relative to strength, impact resistance, physical and chemical
stability, and resistance to flame. No specific test protocols for chemical
stability and resistance to flame are specified in the standard. The author's
understanding is that chemical stability is assessed after exposure to a 50%
concentration of vapors of the following materials for 10 days: ethanol,
ethyl acetate, ethyl ether, gasoline, methyl ethyl ketone, toluene, and vinyl
acetate. The material shall show no significant degradation in properties.
A similar test, specified by FM, is discussed below.
CENELEC
Industrial (Group II) apparatus shall not be made of alloys containing

more than 6% magnesium.
Nonmetallic materials must be completely specified by type, manufac-

turer's name, color, and the kind and percentage of additives and fillers.
The temperature index, TI, of the material shall be at least 20K higher than
the maximum surface temperature on the enclosure, taking into account
the maximum service temperature. TI is based on the 20,000-hour point on
the endurance curve without loss of more than 50% flexing strength per
IEC 216-1 and 216-2 and based on the flexing property according to IEC
178. If it doesn't break in this testing before exposure to heat, TI shall be
based on tensile strength according to ISO R527 with Type 1 test bars.
Hot and cold endurance shall be satisfactory. Hot endurance is tested at

90±5% R.H. at 80°C, or if higher, 20±2 K above the maximum service tem-
perature, for 4 weeks. If the maximum service temperature exceeds 75°C,
testing shall be for 2 weeks at 95±2 °C and 90±5% R.H. followed by 2
weeks at a temperature 20±2 K higher than the maximum service temper-
ature. Cold endurance testing is conducted for 24 hours at the lowest
specified service temperature. Hot and cold endurance tests are per-
formed before impact testing. The enclosure is then subjected to the tests
pertinent to the type of protection, such as overpressure testing.
A test for resistance to light, 1,000 hours exposure to simulated sunlight, is

also specified. Reduction of impact bending strength shall not exceed 50%
when tested per ISO 179.
EN50014 contains a long section addressed to the alleged hazard of elec-

trostatic charges accumulating on the surface of plastic enclosures and
being discharged in an ignition-capable spark. The requirements are
addressed to parts of portable apparatus or parts of fixed apparatus likely
to be rubbed or cleaned on site. The author's only experience with this
issue has been with a piece of portable apparatus. In that case, the test lab
insisted on applying the surface resistance criteria described below to the
apparatus, even though test data were presented showing that, although a
large charge could accumulate on the part, it was not possible to remove
enough charge from a localized area to cause ignition. The high surface
resistivity of the material that allows charge to accumulate without leak-
ing off also prevents instantaneous redistribution of the charge. An
electrode applied to the charged surface can draw off only those charges
in the local area surrounding the electrode. The test data were not dis-
cussed. They were set aside because the standard does not specifically
permit assessment of hazard based on such testing. In fairness to the test-
ing agency it might have been possible for the agency to consider the test
results in connection with a certificate of conformity to a national stan-
dard, but in this case a certificate of conformity to the CENELEC standard

was at issue.
To eliminate possible ignition by electrostatic sparks, the enclosure mate-

rial must have a surface resistivity less than 1 GΩ at 23°C, 50 ± 5% R.H.,
measured between two 100-mm-long electrodes spaced 10 mm apart. This
is not a requirement for Groups IIA and IIB if the area of the plastic is less
than 100 mm2, or if it is less than 400 mm2 and it is surrounded by an
earthed, conductive frame. For Group IIC the permitted areas are 20 mm2
and 100 mm2, respectively.
The surface resistivity requirement need not be met if the size, shape and
layout, or other protective methods is such that dangerous electrostatic
charges are not likely to occur. No elucidation of this exception is given.
If none of the above are possible, a warning label shall indicate the safety
measures to be taken when the equipment is in service.
UL
Metallic enclosures shall be of iron, steel, bronze, copper or aluminum

alloy containing at least 80% aluminum. Zinc and magnesium alloys are
not accepted for evaluation. Copper is not acceptable for Group A unless
the alloy contains less than 30% copper or the copper is protected by a
coating of tin, nickel, or other suitable material.
Nonmetallic enclosures are evaluated for mechanical strength, impact

resistance, moisture absorption, fungus resistance, combustibility, resis-
tance to repeated explosions, resistance to solvents, resistance to
accumulation of electrostatic charge, and resistance to distortion and creep
at temperatures to be encountered in service. No specific protocols for
testing are specified in the standard, but UL has a chemical vapor expo-
sure test similar to that used by FM.
FM
Metallic alloys must contain less than 6% magnesium. For Group A

approval the alloy shall not contain more than 30% copper.
For nonmetallic materials, a chemical compatibility test is prescribed in

Approval Standard Class No. 3600.
No permanent change in properties shall occur (e.g., less than 10% change
in hardness) after exposure to the vapor of each of the following materials.
Material Representative of
Acetone Ketones
Gasoline Alipathic hydrocarbons
Hexane Alipathic hydrocarbons
Methanol Alcohols
Ethyl Acetate Esters
Acetic Acid Acids
A sample of the material to be tested is suspended for 24 hours in the

vapor formed by 4 fluid ounces of liquid in a container of 1 quart volume.
(120 cm3/l). It is reexamined for changes in critical properties within
1 hour of its removal from the vapor.
Temperature stability of the material is assessed after 14 days aging at

121°C (250°F). There shall be no change in properties that affect the type of
protection.
Author's Opinion
The vapor exposure tests now used by the approval laboratories are defi-
cient in several ways:
(1) They are internally inconsistent. If one looks at differences in severity
of exposure to the chemicals listed, that is, vapor concentration as par-
tial pressure, or as a percent of LEL, it is fairly obvious that there is no
clear rationale supporting the test method or the selection of test
chemicals.
(2) The test is not applied to power cables to be used in exactly the same
environment, yet the possible consequences of chemical attack on
power wiring are at least as serious.
(3) The test is not applied to painted metallic enclosures where paint is
relied upon for corrosion protection, and where attack on the paint
would expose the underlying metal to corrosion.
(4) Including the tests as a criterion for judging apparatus gives a mis-
leading signal to the user. It suggests that chemical compatibility is not
an issue because it has already been subject to assessment. But the fact
that a material has been found compatible with the chemicals used for
testing is no guarantee that it is compatible with all chemicals. Toluene
and xylene, for example, are active solvents. They are not among the
test materials of some laboratories. Chemical compatibility is not an
issue in the standards for other types of protection because it is an

engineering issue, not one to be solved in a standard by simple pre-
scriptions. Would one ban aluminum enclosures because they disap-
pear when sprayed with HCl?
This issue is discussed in greater detail in Chapter 7. The same require-

ments were originally imposed on sealed and encapsulated devices, but
were removed in the second edition of ISA S12.12. The higher average fre-
quency of presence of flammable vapors in Division 1, compared to
Division 2, does not, in the author's opinion, justify continued use of these
tests.
Windows
Windows may be of glass or other material. Materials other than glass are
subject to investigation to determine chemical and physical stability, resis-
tance to flame, and so on.
Strength
In North America the traditional practice has been to test a sample of a

cast enclosure to determine the maximum explosion pressure, then hydro-
statically test a sample of the enclosure (type-test) to 400% of the
maximum explosion pressure. In most enclosures the explosion pressure
is typically 75 to 150 psi for hydrogen and acetylene mixtures, depending
on the enclosure volume, shape, and disposition of contents. If the enclo-
sure is tested with ignition at the end of a long length of conduit, the
pressures may be considerably higher, especially for the faster burning
Group A and B test mixtures. Fabricated steel enclosures shall not perma-
nently deform at a test pressure of 200% of the maximum explosion
pressure and must not rupture at 300% of the reference pressure (225% at
CSA for both rupture and distortion).
CENELEC standards permit testing at 400% of reference pressure to

obtain exemption from routine testing. All welded enclosures must be
routinely tested. UL exempts an enclosure from type test if calculations
show adequate strength at 500% of reference pressure for castings; and for
fabricated steel enclosures, 400% and 300% of reference pressure for dis-
tortion and rupture, respectively.
It has been common for European manufacturers to routinely test every

enclosure at 150% of measured explosion pressure. This test may be car-
ried out hydrostatically or dynamically. In a dynamic test one explodes a
precompressed mixture or a specially formulated mixture to obtain the
required pressure. For Group IIC, routine testing is at 200% of maximum
explosion pressure if the time to maximum pressure is greater than 5 mS,

and 300% if the rise time is less than 5 mS.
FM and CSA will now type test at lower values if the manufacturer rou-
tinely tests 100% of enclosure production. The values of FM are 200% of
reference pressure if the rise time is greater than 5 mS, and 300% of refer-
ence pressure if the rise time is less than 5 mS. CSA uses 150% for
enclosures used in gaseous mines and 225% and 300%, respectively, for
pressure rise times greater or less than 5 mS in other enclosures.
The reference pressure is determined by explosion testing at room ambient

temperature unless the enclosure is rated for use at temperature below −
20°C, in which case the pressure rise is based on the actual rated temperature.
This practice developed after it was found that explosion pressures in enclo-
sures used at −70°C greatly exceeded the pressures measured at normal room
ambient temperature. Bossert's book contains plots of Lobay's data on explo-
sion pressure versus temperature for explosions of methane-air, propane-air,
and ethylene-air, with and without pressure piling. All are linear, pressure
increasing as ambient temperature is lowered. At a lower temperature, a
given enclosure volume contains more fuel and air than the same enclosure
at a higher temperature, in the inverse ratio of the absolute temperatures.
Combustion of the larger volume of fuel produces increased heat and higher
pressure.
The function of the seals that are to be located within 18 in. of the enclo-
sure is to complete the explosionproof enclosure. Otherwise the conduit
runs would be part of the enclosure volume. This seal is required by the
NEC only if the conduit is large (2 in. or larger) or if there is an ignition
source within the enclosure. One can infer that the drafters of this NEC
requirement intended to minimize pressure piling and the development of
high explosion pressure in those situations where it is most likely to occur.
The certifying authorities in North America now require that the user of
any explosionproof enclosure must be instructed to install a seal, unless
the enclosure has been tested with a length of conduit attached and
marked to tell the user that a seal is not required. The lengths of conduit
used differ among laboratories, as follows:
• FM and UL: Groups A, B, C — 5, 10, and 15 ft (1.5, 3.0 and 4.6 m);
Group D is 2 ft (0.6 m)
• CSA — 2.5 and 3.0 m
No distinction is made between enclosures containing sparking apparatus

and normally nonsparking apparatus.
This requirement was implemented by one laboratory more than two

decades ago and was copied by others as time passed. The extremely low
probability of nonsparking apparatus becoming a source of ignition con-

current with the presence of a flammable cloud of a concentration that
would likely support transmission of an explosion into the conduit was
never considered to be worthy of discussion. Neither was the fact that
there was no history to suggest that the provisions of the CEC and the
NEC requiring a seal only for sparking apparatus constitutes a dangerous
practice.
The effect of connecting conduit is to raise the maximum measured refer-

ence explosion pressure, especially for Groups A and B.
All standards except the UL standard specify impact testing of enclosures

and windows, as follows:
• EN50014 specifies 7 J or 4 J impact for an industrial enclosure
depending on the severity of expected exposure in use. Windows
are tested at 4 or 2 J.
• CSA has elected to use only the higher CENELEC values.
• FM specifies 2.7 J for both windows and enclosures.
• UL does not define an impact test.
Thermal shock testing of windows is specified by the FM, UL and CEN-

ELEC standards.
Both the FM and UL test expose the window to a cloth saturated with 10°C
water when the apparatus is operating in an ambient temperature of 40°C.
CENELEC specifies a 1-mm diameter jet of 10°C water impinging on the
window when the apparatus is operating at 40°C.
Joints
The permitted gaps between flanges are tabulated in Tables 5-3, 5-4, and
5-5. Diametral clearance between shafts and bearings are of similar dimen-
sions, as are the gaps in spigot joints, although the permitted gaps in
spigot joints may be somewhat larger because the flame path incorporates
a right angle turn.
Permissible Gap-mm
Flange V < 100 cm3 100 < V < 500 cm3 500 < V < 2000 cm3 2000 < V < 6000 cm3
width
mm UL FM CSA EN UL FM CSA EN UL FM CSA EN UL FM CSA EN
6 — — — — — — — — — — — — — — — —
9.5 04 .05 .1 .1 .04 .05 .04 .1 — — — — — — — —
12.5 04 .05 .1 .1 .04 .05 .04 .1 — — — — — — — —
16 04 .05 .1 .1 .04 .05 .04 .1 .04 — .04 — — — — —
19 04 .05 .1 .1 .04 .05 .04 .1 .04 — .04 — — — — —
25 04 .05 .1 .1 .04 .05 .04 .1 .04 .025 .04 — .04 .025 .04 —
FM Group A
9.5 .02 — — —
12.5 .02 — — —
16 .02 .02 — —
25 .02 .02 .02 .02
Table 5-3 Flange Gaps–Group IIC/Group B
Permissible Gap-mm
Flange V ≤ 100 cm3

100 < V < 500 cm3 500 < V < 2000 cm3 2000 < V < 6000 cm3
width
6 .1 .1 .1 .2 — — — — — — — — — — — —
9.5 1 .1 .1 .2 — — — — — — — — — — — —
12.5 .04 .1 .1 .2 — — .04 .2 — — .04 — — — — —
16 .04 .1 .1 .2 04 .1 .04 .2 — .1 .1 .2 — .08 .08 .15
19 .05 .1 .1 .2 .05 .1 .04 .2 .05 .1 .1 .2 .05 .08 .08 .15
25 .08 .1 .1 .2 .08 .1 .04 .2 .08 .1 .1 .2 08 .1 .1 .2
Table 5-4 Flange Gaps–Group IIB/Group C

Permissible Gap - mm
Flange V ≤ 100 cm3 100 < V < 500 cm3 500 < V < 2000 cm3 2000 < V < 6000 cm3
width
6 .15 .15 .15 .3 — — — — — — — — — — — —

8 .15 .15 .15 .3 — — — — — — — — — — — —
9.5 .04 .15 .15 .3 — — .04 .3 — — .04 — — — .04 —
12.5 .04 .15 .15 .3 .04 .15 .15 .3 — .15 .15 .3 — .1 .1 .2
19.5 .05 .15 .15 .3 .05 .15 .15 .3 .05 .15 .15 .3 .05 .1 .1 .2
25 .08 .2 .15 .4 .08 .2 .15 .4 .08 .2 .15 .4 08 .2 .2 .4
Table 5-5 Flange Gaps–Group IIA/Group D
UL standards include some enclosures with parameters different from

those in the tables. FM accepts enclosures for test that do not conform to
the tabulated values. Assessment is with flange widths reduced to 75% of
the design value, or gaps increased to 150% of the design value, or with
gas mixtures that will penetrate smaller gaps (lower MESG value).
Only FM will permit enclosures for Group A to have flanged joints. They
are subject to testing with artificially enlarged gaps or reduced flange
widths as described earlier. All others require threaded joints only.
Requirements for threaded joints are summarized in Table 5-6.
All standards demand surface roughness not greater than 6.3 μm

(250 μinch).
Groups C and D
UL
Thread diameter Max. Threads/inch Min. Threads Engaged
No limit 20 5
3/8 (9.5mm) 24 5
over 3/8 24 6
over 3/8 28 7
over 3/8 32 8
FM and CSA
≤ 32 5 (CSA, 8mm min.)
Group A
UL, FM, CSA
Class 1 fit ≤ 20 8
Group B
UL and CSA* FM
Class 1 fit 8 7
Class 2 fit 7 6
Class 3 fit 6 5
9.5 mm min. length of engagement, V ≤ 500cm3 (30 in3)
12.5 mm min. length of engagement, 500 ≤ V ≤ 6000cm3
EN50018—all groups
Minimum pitch -0.7mm (≤ 36 threads per in.), medium fit per ISO 965/I & II
5 threads, or 5mm engaged V ≤ 100cm3
8mm engaged, V > 100cm3
Table 5-6 Straight Threads
Bolt Spacing and Location
There are no specifications for the number of bolts required in a flanged

cover. The number needed is determined by the ability of the enclosure to
withstand the explosion test without transmitting an explosion from
within the enclosure to the test mixture surrounding it.
If clearance holes for bolts are located in the flange, minimum spacing to
the interior of the enclosure is specified by all the standards except that of
FM.
CENELEC and CSA — The distance w from the edge of the bolt hole to the
inside edge of the flange is shown in Table 5-7.
Required Flange Width - mm Minimum w - mm
CSA CENELEC
6 < 12.5 6
9.5 8
12.5 12.5 - 25 8
> 16 9.5
> 25 9
Table 5-7 Minimum Distance from Edge of Bolt Hole to Inside Edge of Flange
- CSA and CENELEC
FM — No relationship is specified, but if the bolt spacings are less than the
tabulated values for flange width in the standard the design can be tested
with reduced distance.
UL — A bolt in the flange is permitted in Group B enclosures if the vol-

ume is between 100 and 350 in3 (1.6–5.7 dm3). If the joint is 1 in. or wider,
the distance from the clearance hole to the inside edge of the flange shall
be at least 55/64 in. (21.8 mm). For Groups C and D enclosures the dis-
tance shall be at least 0.5 in. (12.7 mm).
Holes in Enclosure Walls
Holes in the enclosure for securing the nameplate must be bottomed or

closed by welding. Holes for securing mechanism must be bottomed or
have not less than five threads, with the entering threaded part secured
against removal by staking, welding, riveting, or other means of locking.
Removable bolts or screws may not extend through the enclosure wall.
The thickness of metal at the bottom of the hole shall be at least 1/3 the
hole diameter but not less than 1/8 in.
Connections
In North America connections shall be designed to permit five full threads

of rigid conduit to be engaged. Tapered pipe threads shall be used.
Unused openings must be sealed with pipe plugs having at least five full
thread engagement. In Europe connections are customarily straight metric
threads. The installer must provide certified hardware for connecting to
cable or conduit and for closing unused openings. The latter have to be
removable only from the inside of the enclosure, or be removable only
with a special tool. Some authorities, but not all, permit the use of pipe
plugs with an internal hexagonal socket, per EN50014.
Special Fasteners
CENELEC standards have required that fasteners for covers be of a type

that can only be removed by using a special tool so that unauthorized
entry is prevented. One way to meet this requirement is to cast a raised
shroud in the part that surrounds or partially surrounds the head of the
bolt. The shroud ensures that only a specially machined tool can be used
to loosen the bolt. Hexagonal socket screws whose heads are below the
surface of the part are considered to require a special tool for removal.
This requirement was removed from Group II apparatus in the second

edition of EN50018. It is not now enforced by all authorities, although
blanket application of the second edition requires changes in national
legislation.
Marking
Marking of explosionproof apparatus must include, in addition to the

marking required on general purpose apparatus, such as manufacturer's
name, model number, ratings, the following:
• Class and group for which approved
• Temperature code or maximum surface temperature
• Symbol of approving laboratory
• Warning such as “Do not open enclosure while energized”
In Europe the marking may also include the European Community sym-
bol for explosion-protected equipment fully complying with a CENELEC
standard, a script Ex in a hexagon.
• The European marking EEx d IIC T4 X abbreviates much of the
information.
• EEx indicates assessment against the applicable CENELEC stan-
dards.
• d indicates a flameproof enclosure.
• IIC indicates that the device is suitable for above-ground use
where materials of Group IIC may be present.
• T4 is the temperature code.
• X indicates that special conditions of use apply—the user should
consult the accompanying documentation.
Testing
All standards define joint requirements based on net free volume of the
enclosure, but whether the enclosure is explosion tested with the contents
in place is often decided by the test engineer. In principle, all enclosures
should be tested with contents in place because the arrangement of con-
tents may cause pressure piling, which would vitiate an approval based
on lower pressures generated in an empty enclosure. In practice some
engineers test small enclosures such as are used for process control instru-
ments without contents so that changes in the design of contents does not
require complete retesting. Because the volumes are small, and the con-
tents are unlikely to enhance pressure development significantly, no
hazard is incurred by this practice.
Installation of Explosionproof Enclosures
The material in this section is based on the requirements of the NEC and
the CEC.
In explosionproof installations, seals are specified to perform two

functions:
(1) Complete the explosionproof enclosure to prevent propagation of an
explosion down the conduit and prevent pressure piling. These seals
are required for every explosionproof enclosure containing an igni-
tion source in normal operation, or if the conduit size is 2 in. trade size
or larger. One can infer that the drafters of this NEC/CEC require-
ment intended to minimize pressure piling and the development of
high explosion pressure in those situations where it is most likely to
occur. The certifying authorities in North America now require that
the user of any explosionproof enclosure must be instructed to install
a seal unless the enclosure has been tested with a length of conduit
attached, and marked to tell the user that a seal is not required. The
lengths of conduit used differ among laboratories.
If the manufacturer's instructions state that a seal must be installed,

install one.
(2) The second function of a seal is to prevent flammable atmosphere

from being transported through the conduit or cable from a Division 1
location to a Division 2 location, or from a Division 2 location to an
unclassified location. There shall be no union, coupling, box, or fitting
in the conduit between the sealing fitting and the point at which the
conduit leaves an area of one classification and enters another.
Cables must also be sealed to prevent passage of vapors unless the cable is
as tight as a conduit seal. Test methods are specified in ANSI C33.27. They
require that leakage does not exceed 0.007 ft3/hr when pressure of 6 in.
H2O is applied.
Conduit must always be installed so that water drains away from the
enclosure. An explosionproof enclosure must be grounded, as would any
other enclosure. All bolts and threaded joints must be pulled up tight and
the manufacturer's torque specifications, if any, shall be observed.
Whether the installation is covered by the NEC or CEC or some other local
or national code, all hardware related to cable entries and conduit entry
must be approved. In Europe this usually means that certified compo-
nents are needed.
Maintenance of Explosionproof Enclosures
Unless the location is known to be nonhazardous, all servicing and main-

tenance of explosionproof enclosures must be done only after the
equipment is deenergized. If the cover of an explosionproof enclosure is
removed to provide access to the equipment for servicing, protection is
immediately lost. Most plants require that a “hot permit” be obtained
before work is permitted. A combustible gas analyzer is used to determine
that the location is nonhazardous before any work is permitted. After
work is completed, the enclosure shall be tightly closed with all bolts in
place and tightened and all threaded joints completely engaged.
The protection afforded by an explosionproof enclosure depends on

quenching emerging hot gas by close fitting flanged or threaded joints.
The integrity of the joints must, therefore, be maintained throughout the
life of the enclosure. Any foreign material or flange damage can alter the
effective gap dimension, so take the following precautions:
• Treat flange surfaces with respect during disassembly and
assembly. Don't allow tools to contact the joint surfaces. Don't
handle the parts roughly or store them on rough surfaces. Store
covers and enclosures with the joints mated to prevent accidental
damage to the joint surfaces.
• Handle enclosures in clean conditions so that foreign materials
will not be trapped between flanges, enlarging the gap beyond a
safe dimension. Foreign matter is unlikely to enter a joint while
an enclosure is in service, but the joints are much more vulnera-
ble after an enclosure has been disassembled. Clean joints care-
fully with a solvent such as kerosene to remove old grease dirt,
paint, or other material.
• Before reassembly, lubricate the joints if necessary.
Corrosion of threaded and flanged joints is a serious problem in many

plants. While the enclosure is in service corrosion may not enlarge the gap,
but it may weaken a wall. Metals compatible with the specific corroding
atmosphere should be specified, if they are available. Corrosion inhibitors
or lubricants may slow the corrosion process. In locations of severe corro-
sion inspect enclosures periodically to determine whether this may have
occurred.
If corrosion occurs, and the corrosion products cannot be removed with

solvent, or if corrosion has marred the surface of the joints, the enclosure
should be considered to have lost its protective function.
Few, if any, explosionproof enclosures are vapor-tight. When the humid-

ity is high, water vapor will enter the enclosure when the temperature
inside the enclosure falls below the ambient temperature. Further decrease
in temperature will allow the water vapor to condense. Even if the temper-
ature rises, the water may not evaporate and leave the enclosure before
additional water vapor enters. Another cooling cycle will condense more
water. Eventually a considerable amount of water will accumulate in the
bottom of the enclosure. This water must be drained periodically to pre-
vent damage to the contents of the enclosure. Explosionproof drain
fittings are available.
Enclosed Break Devices
Closely related in concept to explosionproof apparatus are “enclosed

break” devices, which are intended for application in Division 2 locations.
The utility of these devices is based on the following:
• Small enclosures are inherently strong enough to contain an
internal explosion. In 1987 PTB investigators tested encapsulated
devices of 2 cm3 internal volume and found that the maximum
pressure rise of an acetylene-air explosion was 5 bar. All other
mixtures of slower burning materials can be expected to be more
efficiently quenched by the chamber walls and yield lower pres-
sures.
• In Division 2 the probability of a flammable vapor or gas entering
the enclosure is very small, both because a flammable cloud is
present only infrequently and for a short time and because if the
enclosure can contain an explosion without transmitting it to the
surrounding atmosphere, the time required for the internal vapor
concentration to reach the LEL will greatly exceed the period
during which the external cloud is present (see Chapter 7).
BS 6941 limits enclosed break devices to 20-cm3 volume and maximum

ratings of 660 V and 15 A. Poured seals or encapsulating compounds shall
have a softening or melting point 20K higher than the maximum operating
temperature of the component.
Elastomeric seals of covers intended for removal in service or that are

unprotected from mechanical or environmental damage shall be removed
before ignition testing. Other seals, if elastomeric, shall be subjected to
oxygen aging (pure oxygen at 72°C for 96 hours at a pressure of approxi-
mately 18.5 atmospheres) or, if thermoplastic, to air aging (7 days at 80°C)
and replaced. Five prepared samples are filled with and surrounded by
the appropriate mixture:
• Group IIA 6.5% ethylene-air
• Group IIB 28% hydrogen-air
• Group IIC 34% hydrogen, 17% oxygen, remainder nitrogen
The devices shall be ignited by the enclosed contacts operating at maxi-

mum ratings of source power and maximum load in terms of voltage,
current, frequency, and power factor. There shall be no external ignition,
no visible signs of damage to the component, and no failure to clear the
arc when the switch is opened.
IEC 79-15 contains essentially the same test protocol. Enclosed break
devices are not at present recognized in ISA S12.12. They were removed
when the latest version was drafted because some argued that there was
not sufficient guarantee of safety, and because there were no vested inter-
ests in support of maintaining them in the document.
References
“A Review of Electrical Research and Testing with Regard to Flame Proof
Enclosure and Intrinsic Safety of Electrical Apparatus and Circuits,” Min-
istry of Fuel and Power, London, 1943.
Titman, H., and J. Haig, “The Pressure Generated During Explosion of

Pentane-Air Mixtures Inside a Core-Cooled Motor,” Research Report No.
11, Safety in Mines Research Establishment, July 1950.
Matson, E. F., DuFour, R. E., and W. C. Westerberg, “An Investigation of

Large Electric Motors and Generators of the Explosion Proof Type for
Hazardous Locations, Class I, Group D,” Bulletin of Research No. 46,
Underwriters' Laboratories, Inc., September 1951.
Gleim, E. J., and J. F. Marcy, “A Study to Determine Factors Causing Pres-

sure Piling in Testing Explosionproof Enclosures,” Paper No. 5, 7th
International Conference of Directors of Safety in Mines Research, Safety
in Mines Research Establishment, July 1952.
Gleim, E. J., and J. F. Marcy, “A Study To Determine Factors Causing Pres-

sure Piling in Testing Explosionproof Enclosures,” Report of
Investigations 4904, U.S. Bureau of Mines, 1952.
Staples, H., “The Pressure Generated During Explosion Inside a

Core-Cooled Motor; Mixtures of Methane In Air and of Hydrogen/Meth-
ane 85/15 in Air,” Research Report No. 38, Safety in Mines Research
Establishment, May 1952.
Smith, P. B., “The Role of Flanges in Conferring Protection on Flameproof

Electrical Enclosures,” Research Report No. 77, Safety in Mines Research
Establishment, August 1953.
Maas, W., “Temporary Distortions of Flameproof Casings for Electrical

Equipment,” Paper No. 22, 8th International Conference of Directors of
Safety in Mines Research, Safety in Mines Research Establishment, 1954.
Maskow, H., “Ignition by Ejection of Particles,” Paper No. 28, 8th Interna-
tional Conference of Directors of Safety in Mines Research, Safety in Mines
Research Establishment, 1954.
Smith, P. B., “The Behavior of Certain Enclosures Under Rapidly Applied

Internal Pressures,” Research Report No. 91, Safety in Mines Research
Establishment, July 1954.
Titman, H., and R. Torry, “Flameproof Enclosures for Mining Electrical

Equipment: The Protection Afforded by Flanges of One Half-Inch Radial
Breadth for Mixtures of Methane and Air,” Research Report No. 123,
Safety in Mines Research Establishment, October, 1955.
Wolfhard, H. G., and A. E. Bruszak, “Suppression and Transmission of

Explosions by Cylindrical Channels,” Report of Investigations 5495, U.S.
Bureau of Mines, 1959.
Vanpee, M. and H. G. Wolfhard, “Ignition by Hot Gases,” Report of Inves-

tigation 5627, U.S. Bureau of Mines, 1960.
Lewis, B., and G. von Elbe, Combustion Flames and Explosions of Gases, 2nd
Edition, Academic Press, New York, 1961.
Phillips, H., “A Reaction Rate Theory for Flame Proof Enclosures,” IEC
Paper No. 3902 M, IEC Conference Report Series No. 3, 1962.
Riddlestone, H. G., and W. Nethercot, “Explosionproof Electrical Equip-

ment for Aircraft,” IEC Conference Report Series No. 3, pp. 26–28, 1962.
Torry, R., “Flameproof Enclosures for Mining Electrical Equipment: Influ-

ence of Atmospheric Moisture on Maximum Safe Gaps,” Research Report
No. 202, Safety in Mines Research Establishment, April, 1962.
British Symposium on Flameproofing and Intrinsic Safety, 1962, IEE Con-

ference Report Series No. 3, Discussion, pp. 34; 59.
Phillips, H., “On The Transmission Of An Explosion Through A Gap

Smaller Than The Quenching Distance,” Combustion and Flame, June 1963,
pp. 129–135.
Riddlestone, H. G., “The Effect of Transient Pressures On Flameproof

Enclosures,” Report D/T 138, Electrical Research Association, 1963.
Kisselstein, C. F., “Explosionproof Enclosures, Design and Tests,” ISA

Monograph 110, pp. 61–74, ISA, Pittsburgh, 1965.
Riddlestone, H. G., “Flameproof Enclosures: Safe Gaps for Enclosures of

Small Volume,” Report No. 5092, Electrical Research Association, 1965.
Slack, C., and D. W. Woodhead, “Correlation of Ignitabilities of Gases and

Vapors by a Break Spark and At a Flange Gap,” Proceedings IEE, Vol. 113,
No. 2, Feb. 1966, pp. 297–301.
Brown, G. K., Dainty, E. D., and S. Silver, “Laboratory Investigations of

Hydrogen Explosion Phenomena Relating to Electrical Apparatus,”
Research Report R182, Department of Mines and Surveys, Ottawa, April,
1966.
Dainty, E. D., Brown, G. K., and A. D'Aoust, “Hydrogen-Air Explosion

Pressure Reduction Due to Heat Absorption Into the Walls of Electrical
Enclosures,” Canadian Explosive Atmospheres Laboratory (CEAL) No.
163, January, 1968.
Dainty, E. D., Brown, G. K., and A. D'Aoust, “The Variations of Stoichio-

metric Hydrogen-Air Safe Gaps Due to Variations in the Shape of Three
Small Rectangular Enclosures of Constant Volume,” Canadian Explosive
Atmospheres Laboratory, No. 168, Feb. 1968.
Helwig, N., and K. Nabert, “Zusammenhänge zwischen Kenngrössen für

explosionsgeschützte Betriebsmittel,” PTB Mitteilungen, April, 1968.
Silver, S., “The Effect of Conduit Attached to Explosionproof Enclosures

On Internal Pentane-Air Explosion Pressures,” CEAL No. 178, June, 1968.
“Recommended Practice for Electrical Equipment for Use in Explosive

Gas Atmospheres for General Industries—1965,” Ministry of Labour,
Japan, 1968.
Brown, G. K., and S. Silver, “An Explosion Hazard Due To Inflammation

of Aluminum Particles in Flat Flanges of Flameproof Enclosures,” CEAL
No. 197, 1969.
Dainty, E. D., and G. K. Brown, “Explosion Gas Temperatures At The

Entrance To Flanges of Enclosures Containing 8.5% Methane-Air Mix-
tures,” CEAL No. 196, 1969.
Stewart, J. A., “Comparison of World Standards for Electrical Enclosures

For Hazardous Areas: Part 1,” IEEE Transactions, Vol. IGA-6, No. 4., July/
August 1970, pp. 313–314.
Short, W. A., “Comparison of World Standards for Electrical Enclosures

For Hazardous Areas: Part II,” IEEE Transactions, Vol. IGA-6, No. 4, July/
August 1970, pp. 315– 325.
“An Investigation of Fifteen Flammable Gases Or Vapors With Respect to

Explosionproof Electrical Equipment,” Underwriters' Laboratories Bulle-
tin of Research No. 58, Chicago, 1970.
Bartels, A. L., “Low Power Arcing in Flameproof Enclosures,” Conference

on Electrical Safety in Hazardous Environments, London, 1971.
Davis, E. H., “A Review of Work on the On-Load Method of Testing

Flameproof Electrical Equipment,” Conference on Electrical Safety in Haz-
ardous Environments, London, 1971.
Phillips, H., “The Basic Theory of the Flameproof Enclosure,” Conference

on Electrical Safety in Hazardous Environments, London, 1971.
Phillips, H., “The Mechanism of Flameproof Protection,” SMRE Research

Report 275, 1971.
Phillips, H., and S. K. Shaw, “A Statistical Evaluation of the Safe Gap,”

14th International Conference of Mine-Safety Research Establishments,
Donetsk, 1971.
Phillips, H., “A Non-Dimensional Parameter Characterizing Mixing Pro-

cesses in A Model of Thermal Gas Ignition,” Combustion and Flame, Vol. 19,
1972, p. 181.
Bartels, A. L., “The Relationship Between Stress and Strain in Flameproof

Enclosures,” IEE Conference Publication No. 134, London, 1975, pp. 153–
157.
Fleisig, D. J., “The German Explosionproof Concept for Hazardous Atmo-

spheres,” IEE Conference Publication No. 134, pp. 174–179, London, 1975.
Canadian Standards Association, C22.2 No. 30, “Explosionproof Enclo-

sures for Use in Class I Hazardous Locations.”
Dainty, E. D., “The Determination By Calculation of the Variation of Safe

Gap With Humidity Changes for Stoichiometric Hydrogen-Air Mixtures,”
Private Communication.
“Industrial Control Equipment for Use in Hazardous Locations,” Under-

writers' Laboratories, Inc., Standards for Safety No. 698.
“Outlet Boxes and Fittings For Use in Hazardous Locations, Class I,

Groups A, B, C and D, and Class II, Groups E, F and G,” Underwriters'
Laboratories, Inc., Standards for Safety No. 886.
“Explosionproof and Dustignition-proof Electrical Equipment for use in

Hazardous (Classified) Locations,” Underwriters Laboratories, Inc., Stan-
dards for Safety No. 1203.
Factory Mutual Research, Class Number 3615, Approval Standard,

“Explosionproof Electrical Equipment, General Requirements.”
EN50014, “Electrical Apparatus for Potentially Explosive Atmo-

spheres-General Requirements,” CENELEC, Brussels.
EN50018, “Electrical Apparatus For Potentially Explosive Atmo-

spheres-Flameproof enclosure `d',” CENELEC, Brussels.
IEC Publication 79–15, Electrical Apparatus For Explosive Gas Atmo-

spheres, Part 15: Electrical Apparatus for Type of Protection ‘n’,” Geneva
1987.
BS 6941: 1988, “Electrical Apparatus for Explosive Atmospheres with Type

of Protection ‘n’,” British Standards Institution, London.”
Conn, J. W., “Pressure Piling,” Engineering Report 87062501, MSHA, Tria-

delphia, 1987.
Wehinger, D., and D. Wassmann, “Flameproof Encapsulation for Small

Volumes,” Ex Magazine, p. 41, December 1988, Stahl, Kunzelsau.
Polka, U., “Pressure Arising from Gas Explosions in Subdivided Enclo-

sures,” Ex Magazine, p. 39, August 1980, Stahl, Kunzelsau.
Lobay, G., “An Investigation of the Effect of Low Ambient Temperatures

upon Explosive Atmospheres,” Report ERP/MRL 77/123 (TR), Canadian
Explosive Atmospheres Laboratory, Ottawa, 1977.
Thibault, P., Liu, Y. K., Chan, C., Lee, J. H., Knystautas, R., Guirao, C.,
Hjertager, B., and K. Fuhre, “Transmission of an Explosion Through an
Orifice,” The Nineteenth Symposium on Combustion, The Combustion
Institute, 1982, pp. 599–606.
Phillips, H., “Maximum Explosion Pressure in Flameproof Enclosures:

The Effects of the Vessel and the Ambient Temperature,” Journal of Hazard-
ous Materials, 8, pp. 251–259, 1984.
Phillips, H., “Ignition in a Transient Turbulent Jet of Hot Inert Gas,” Com-
bustion and Flame, 19, 1972, pp. 187–195.
Phillips, H., “Safe Gap Revisited,” Dynamics of Explosions, ed. by Kuhl, A.

et al., Vol. 114 of Progress in Astronautics and Aeronautics, AIAA, Washing-
ton, D. C., 1988.
Phillips, H., “The Safe Gap: Effect of Explosion Pressure,” IEE Conference
Publication 296, London, 1988.
Phillips, H., “The Physics of the Maximum Experimental Safe Gap,” Pro-
ceedings of the International Symposium on the Explosion Hazard Classification
of Vapors, Gases, and Dusts, NMAB 447, Washington, D. C., National Acad-
emy Press, 1987, pp. 65–82.
Phillips, H., “A Comparison of the 'Standard' Methods for the Determina-

tion of Maximum Experimental Safe Gap (MESG),” Proceedings of the
International Symposium on the Explosion Hazard Classification of Vapors,
Gases, and Dusts, NMAB 447, Washington, D. C., National Academy Press,
1987, pp. 83–108.
CHAPTER 6
REDUCTION OF HAZARD BY
PRESSURIZATION
Recognition in the National Electrical Code

The use of pressurization, often called “purging” in North America, was
recognized for many years in a fine print note in Paragraph 500-1 of the
National Electrical Code (NEC). The note stated: “In some cases, hazards
may be reduced or hazardous locations limited or eliminated by adequate
positive-pressure ventilation from a source of clean air in conjunction with
effective safeguards against ventilation failure.” This note is now in Para-
graph 500-3 of the 1996 NEC.
Paragraph 501-8 recognizes purging more specifically: “501-8(a) Class I,

Division 1. In Class 1, Division 1 locations, motors, generators, and other
rotating electric machinery shall be (1) approved for Class I Division 1
locations, or (2) of the totally enclosed type supplied with positive-pres-
sure ventilation from a source of clean air with discharge to a safe area, so
arranged to prevent energizing of the machine until ventilation has been
established and the enclosure has been purged with at least 10 volumes of
air, and also arranged to automatically deenergize the equipment when
the air supply fails, or (3) of the totally enclosed inert-gas-filled type sup-
plied with a suitable reliable source of inert gas for pressuring the
enclosure, and arranged to automatically deenergize the equipment when
the gas supply fails.” “Totally enclosed motors of Types (2) or (3) shall
have no external surface with an operating temperature in °C in excess of
80% of the ignition temperature of the gas or vapor involved. Appropriate
devices shall be provided to detect and automatically deenergize the
motor or provide an adequate alarm if there is any increase in temperature
of the motor beyond designed limits. Auxiliary equipment shall be of the
type approved for the location in which it is installed.”
In a few places in this book, the term “purging” includes the concepts of
both purging and pressurization. In the United States the terms have been
used interchangeably until recently. In Europe “purging” formerly con-
noted flow, whereas “pressurization” did not. The technique discussed in
this chapter is generally called pressurization. Purging is reserved for the
function of passing a volume of air or other protective gas through an
enclosure to sweep out flammable mixture before the enclosure is
pressurized.
184 Reduction of Hazard by Pressurization
The technique called “pressurization” may be used without continuous

flow of protective gas only if there is no release of flammable material
within the enclosure under normal or fault conditions. If such an internal
release can occur, “continuous dilution,” described later in this chapter,
must be used. The hardware for continuous dilution is the same as the
hardware for pressurization; however, the rationale for selecting the hard-
ware required for a given situation depends, not only on the area
classification outside the enclosure, but also on the type and frequency of
release that may be expected inside the enclosure. “Continuous dilution”
is now considered by many people to be a subset of pressurization in
which there is a continuous flow of protective gas.
The Standardization of Pressurization

ISA Recommended Practice RP12.4, first issued in 1960, was reissued as
ISA S12.4 in 1970. This standard, concerning instrument purging for
reduction of hazardous area classification, was prepared in order to rem-
edy the lack, either in the NEC or elsewhere, of a specific recommendation
for instrument installations. It filled a need for a commonly recognized
safe practice for the instrument industry. However, because ISA Stan-
dards and Recommended Practices were not considered authoritative, if
they were known at all among electrical inspectors, contractors, and regu-
latory authorities, ISA requested NFPA to develop a standard based on
RP12.4 that would bring greater credibility in the electrical community to
the results of ISA's standardization efforts. The NFPA document was
issued as a tentative standard in 1966 and as an official standard in 1967.
The NFPA document, in addition to small equipment enclosures, included

requirements for control rooms and power equipment enclosures. No sig-
nificant changes in philosophy were introduced in NFPA 496, but an ISA
nomograph relating enclosure thickness to the size of the fuse protecting
the circuitry inside the enclosure was not included. The nomograph in the
1960 ISA RP12.4 was based on calculations and demanded wall thickness
much greater than those in common use and considered to be safe. NFPA
argued that the wall thickness requirements in general purpose electrical
safety standards provided an adequate level of safety. The nomograph
was revised in the 1970 edition of S12.4, but the measurements on which it
was based assumed a specific type of fuse. NFPA continued to exclude the
nomograph from NFPA 496 for this reason and continued to maintain that
enclosure thickness requirements in other standards for electrical equip-
ment provided a satisfactory level of safety. The ISA nomographs are still
referenced in some documents, but they are of historical interest only.
In 1970 NFPA 496, Part B, introduced requirements for pressurized enclo-

sures in Class II locations. The 1986 edition added chapters to cover
analyzer enclosures and analyzer houses. This edition also included the
Reduction of Hazard by Pressurization 185
concept of continuous dilution that had been introduced in the 1983 edi-
tion of IEC Publication 79-2. The 1989 edition made minor changes to the
text and illustrations. The 1993 edition made no changes in the philosophy
of pressurization but completely reorganized the document. It introduced
the term “protective gas,” used in IEC Publication 79-2, and replacing the
term “purging” with “pressurization,” except where flow of protective gas
is intended. This edition also replaced earlier installation diagrams with
more current practice. In the early days of pressurization, low-range pres-
sure indicators and alarms were not readily available so alternative
arrangements of pressure gauges and restrictions were permitted. These
alternatives are no longer necessary because low pressure alarms suitable
for use in Class I, Division 1 locations are now available at affordable
prices. NFPA 496 is the American National Standard for pressurization
systems. ISA S12.4 is being recast as a document providing supplementary
guidance for the use of NFPA 496.
The use of pressurization to reduce hazard, that is, to reduce the area clas-
sification inside an enclosure, is not limited to Class I gas or vapor
hazards. Pressurization is also used in Class II locations, to prevent dust
accumulation from interfering with proper function of equipment, as well
as to increase safety. If gases or vapors enter an instrument enclosure
while the pressurization system is shut down, reactivation of the pressur-
ization system will surely reduce hazard by sweeping flammable material
out of the instrument enclosure. Some argued in early discussions that if
the hazard is a Class II hazard caused by dust, a pressurization system
might increase hazard rather than reduce it. They argued that when a
pressurization system is reactivated after a period of shutdown, it might
stir up combustible material that had settled in the enclosure, suspending
it once more in the atmosphere to cause a hazard that had not existed
before pressurization. Others felt that no practical pressurization system is
likely to cause air to flow with sufficient velocity to stir up a dust cloud
and maintain the dust in suspension. A cursory survey of the relevant
technical literature easily resolved the matter.
In theory, an air velocity of 0.75 m/s (150 ft/min) is needed to pick up

spherical particles of 10-μm diameter. Wheat and flour dusts are typically
much larger. Settling velocity of 10 μm particles is about 0.03 m/s (6.0 ft/
min). For 100 μm particles it is about 0.3 m/s. The velocity provided to
maintain dust in suspension in pneumatic conveyors and dust collection
systems is seldom lower than 5 m/s, and is usually greater than 15 m/s.
These figures indicate that any dust that enters a purged enclosure will
soon settle and will not be redispersed by flow of pressurizing or purging
gas supplied at a rate conforming to current pressurization practice.
Pressurization techniques can be adopted to reduce the degree of any

Class I hazard. Pressurization may be the only practical method for install-
ing some equipment in Group A or B hazardous locations. It was noted in
Chapter 5 that the safe gap for Group A and B materials, such as acetylene
and hydrogen, is so small that it is difficult to construct large explosion-
proof enclosures. Most equipment that has been certified for Group A or
Group B has relatively small volume. If equipment must operate at high
energy levels and yet must be easily accessible for adjustments or servic-
ing, explosionproofing, oil immersion, sealing, or other techniques are not
convenient. Pressurization, however, can be adapted to any situation.
Classification of Pressurization Systems

ISA RP12.4 defined three types of pressurization systems. This nomencla-
ture is still used in North America. They are as follows:
• Type Z—pressurization to reduce the classification within an
enclosure from Division 2 to nonhazardous
• Type Y—pressurization to reduce the classification within an
enclosure from Division 1 to Division 2
• Type X—pressurization to reduce the classification within an
enclosure from Division 1 to nonhazardous
The requirements for each type of pressurization can be rationalized in

terms of the philosophy of hazard reduction presented in Chapter 4. Type
Z pressurization, which reduces the area classification within the enclo-
sure from Division 2 to nonhazardous, permits installation of
ignition-capable equipment in the pressurized enclosure. For there to be
an explosion, the pressurization system must fail at the same time that a
process equipment failure releases a flammable cloud of gas or vapor.
Because two independent failures are required, it is not necessary to pro-
vide additional safeguards such as automatic disconnecting of the
electrical system in the event of pressurization failure. To do so would be
to wear a belt and suspenders—safe, but unnecessary.
Type Y pressurization reduces the classification of the enclosure from

Division 1 to Division 2 and permits the use of nonincendive or nonspark-
ing equipment inside the enclosure. Two independent failures, one in the
equipment and one in the pressurization system, must occur before there
can be an explosion. No electrical shutdown safeguard in the pressuriza-
tion system is necessary.
Type X pressurization reduces the classification within the enclosure from

Division 1 to nonhazardous and, therefore, permits the use of igni-
tion-capable equipment inside the enclosure. Because the environment
outside the enclosure is Division 1 and the equipment within the enclo-
sure may be ignition-capable in normal operation, failure of the
pressurization system is a single occurrence that changes the probability
of explosion from essentially zero to essentially one. In this situation it is,

therefore, necessary to provide an additional safeguard in the pressuriza-
tion system. An interlock and other safeguards must be provided to
ensure that the equipment within the enclosure will operate only when
the pressurization system is functioning. The effects of accidental opening
of the enclosure while a hazardous atmosphere exists around it must also
be prevented.
NFPA 496 Requirements for Pressurization Systems

Enclosure and Pressurizing System
A pressurization system prevents flammable material from entering the

enclosure by maintaining a small positive pressure inside the enclosure.
The enclosure to be pressurized must be robust so that accidents of fore-
seeable nature and normal conditions of use will not damage the
enclosure such that the pressurization system cannot maintain the neces-
sary positive pressure. Enclosures shall conform to the accepted levels of
protection against entry of water and the effects of corrosion for the
intended application. Windows should be shatterproof. An enclosure for
power equipment (equipment which switches or uses more than 2500 VA)
shall be substantially noncombustible, that is, equivalent to UL 94 V0 or 94
5V rating per ANSI/UL94, “Tests for Flammability of Plastic Materials.”
When the system is installed, the provisions for sealing conduit or cable in
the applicable installation code must be followed unless the conduit or
raceway is part of an approved pressurization system. In this case conduit
can be both a raceway for wiring and a pipe for bringing protective gas to
an enclosure.
A pressurized enclosure must be protected from damage by overpressure

such as might be caused by failure of the regulator on the protective gas
supply. Any pressure relief device must not provide a path for escape of
sparks or burning material to a Division 1 location.
With the exception of some systems for use in Class II locations, the pres-
surizing system must maintain an internal pressure inside the protected
enclosure of at least 0.1 in. H2O above the surrounding atmosphere. This
pressure is the static pressure developed by a 15-mph (24 km/hr) wind.
The pressurization system can prevent entry of flammable gas or vapor in
a wind up to this velocity. It was the reasoning of the ISA committee that a
spill large enough to maintain a flammable cloud for an appreciable time
in a wind of this velocity is extremely improbable. Should a spill be large
enough to cause a vapor cloud to persist in a wind of this velocity, the
cloud will almost certainly escape from the confines of the plant and be
ignited by a source outside it.
Should 0.1 in H2O pressure not be maintained the user must be warned to
take corrective action by an alarm or an indicator, or in some types of pres-
surization systems the power to equipment inside the enclosure must be
removed. If the supply of protective gas fails an alarm must be provided.
In the vocabulary of NFPA 496 an indicator is a piece of flow or pressure
monitoring equipment which is observed periodically, consistent with the
demands of the application. An alarm is a piece of equipment that gener-
ates a visual or audible signal that attracts attention. Both are intended to
be located at a manned location.
Any components that are not protected by the protective gas or which
may be energized in the absence of protective gas must be suitable for the
area classification of the place of installation when the protective gas is not
present.
Protective Gas System

The protective gas shall be free of contaminants or foreign matter that
would compromise function of the protected equipment and shall have
only trace amounts of flammable gas or vapor. Instrument air is usually an
acceptable protective gas. Plant air is usually not clean enough. The intake
of the protective gas supply system must be in a nonhazardous location. If
any part of the suction line must be run in a classified location, as when a
vertical section passes through a hazardous location to reach a nonhazard-
ous area high above the ground, the line must be constructed of
noncombustible material and must be protected against mechanical dam-
age and corrosion. Electrical power for a blower or compressor in the
protective gas supply system shall be supplied from a power source sepa-
rate from that of the protected equipment or be fed from the protected
equipment supply before any power disconnect to the protected equip-
ment. If ignition-capable equipment is in an enclosure in which
pressurization reduces the area classification from Division 2 to nonhaz-
ardous (Type Z), and the Division 2 classification of the location outside
the enclosure is obtained by pressurizing a room or larger enclosure in a
Division 1 location, the two protective gas supplies shall be independent.
Failure of a common supply would be a single event that could permit an
explosion to occur. If the protective gas is used for cooling as well as haz-
ard reduction, as in ventilated equipment, the discharge of protective gas
must be to a nonhazardous location, unless the equipment does not create
ignition-capable particles during normal operation, in which case the dis-
charge may be to a Division 2 location. The flow of protective gas in
ventilated equipment shall be sufficient to cool it per the manufacturer's
instructions.
Determination of Temperature Marking
The temperature marking, T code or temperature, marked on the enclo-

sure shall be the highest of:
• the highest temperature on the external surface of the enclosure;
• the temperature of the hottest internal component; or
• the temperature of the protective gas leaving the enclosure.
However, internal components may exceed the marked T Code or

temperature
• if the enclosure is marked with a warning not to open it until
after a stated delay sufficient to permit the component to cool
after power has been removed; or
• if the component is housed separately in a pressurized or sealed
enclosure whose surface temperature is less than that of the
marked T Code.
The T Code may be exceeded in the case of small components that have
been shown to be incapable of igniting a test gas with an ignition tempera-
ture lower than or equal to the T Code. See Chapters 9 and 10.
Type Z Pressurization
An indicator will display failure to maintain 0.1 in. water pressure inside
the protected enclosure, but it is not necessary to deenergize the equip-
ment inside the enclosure. If an enclosure can be isolated from the
protective gas supply by a valve, an alarm must be provided, unless an
indicator is located immediately adjacent to the enclosure and the isolat-
ing valve is intended for use only during servicing of the protected
enclosure. The valve must be marked “Warning—Protective Gas Supply
Valve—This valve must remain open unless the area is known to be non-
hazardous or unless the equipment in the protected enclosure has been
deenergized.”
If an alarm is used, it must be located in a constantly attended location.

The device actuating the alarm must take its input from the enclosure and
not from a point between the enclosure and the protective gas supply. It
may be electrical, mechanical, or pneumatic; but if it is electrical, it must be
suitable for the location in which it is installed. No valve shall be installed
between the alarm and the protective gas supply. This alarm may serve
also as the alarm required on the protective gas supply.
If an indicator is used it must not be installed between the enclosure and

the protective gas supply. There shall be no valve between the indicator
and the protective gas supply. The indicator may indicate either pressure
or flow. It must be located for convenient viewing. An alarm must be pro-
vided to alarm failure of the protective gas supply. Figure 6-1 illustrates
arrangements of alarms and indicators that conform to NFPA 493-1993.
Type Y Pressurization
All requirements for Type Z pressurization shall be met. Equipment inside

the enclosure shall be suitable for use in Division 2 locations. Ventilated
equipment shall be automatically deenergized if failure of the ventilation
can cause surface temperatures higher than the marked T Code.
Type X Pressurization
Failure of the protective gas supply must actuate an interlock switch that
will deenergize any circuit or equipment not suitable for Class I, Division
1 locations. If immediate disconnection produces a more severe hazard
than delay of shut down, it is permissible to delay shutdown if both an
audible and a visible alarm are provided in a constantly attended location.
The switch may be either flow or pressure actuated, but it must take its
input signal from the enclosure. It shall not be located between the protec-
tive gas supply and the enclosure. No valve may be installed between the
switch and the enclosure.
Devices shall be provided on transformers, motors, and other devices that

may be overloaded to automatically disconnect them from the power
source should the design temperature limit be exceeded. This requirement
does not apply to motors and transformers in equipment where the load is
imposed internally to the equipment itself. Delay of the shutdown is per-
mitted if immediate shutdown is a greater hazard. Both audible and
visible alarms must be supplied. These devices are not required for venti-
lated equipment if the flow of protective gas is sufficient to cool the
equipment even under overload conditions.
Markings
A pressurized enclosure must be marked with a label that is visible before

the enclosure is opened. Its purpose is to warn the operator not to open
the enclosure unless the location is known to be free of flammable vapors
and gases or all equipment inside the enclosure has been deenergized. The
enclosure must also be marked with the area classification of the protected
enclosure, the type of pressurization (X, Y, or Z), and the T Code. If neces-
sary to protect against ignition by hot surfaces inside the enclosure, a label
shall warn that the enclosure shall not be opened unless the location is
known to be nonhazardous or the internal equipment has been deener-
gized for a stated time that is long enough for the parts to have cooled.
Figure 6-1 Typical Alarm and Indicator Configurations for Types Y and Z Pressurization
(from NFPA 493-1993)
Additional Requirements for Pressurized Enclosures

in Class I Locations
If the enclosure has been opened or if the protective gas supply has failed
to maintain the required positive pressure, the enclosure shall be purged
by flow of protective gas for the time and flow rate marked on the outside
of the enclosure. This time, specified by the designer of the system, per-
mits 4 volumes of protective gas to flow through the enclosure (10
volumes for motors). After purging only positive pressure need be applied
to the enclosure. No minimum flow rate is required.
Compartments within or adjacent to the main enclosure shall either be

pressurized separately (e.g., in series with the main compartment) or shall
have top and bottom vents to the main enclosure. The vent area shall be at
least 6.5 cm2 (1 in.2) per 6560 cm3 (400 in.3) of volume, with a minimum
vent size of 6.3 mm (1/4 in.). These precautions are unnecessary if the
compartment contains only equipment suitable for its location when pres-
surization is not active. Compartments less than 20 cm3 (1 in.3) are not
considered to be internal compartments requiring protection unless the
total volume of all such components is a large fraction of the total pro-
tected volume. Transistors, capacitors, microcircuits, and other such
sealed components need not be counted as part of the protected volume.
Cathode ray tubes are considered to be hermetically sealed components.
Markings
The time and flow rate required for purging the enclosure shall be marked
permanently on a label on the outside of the enclosure. Pressure may be
marked instead of flow rate if pressure level positively determines the
flow rate. In Type Y and Type Z systems the purging may be accom-
plished manually. Purging may be omitted if the location is known to be
nonhazardous while the enclosure was open or the pressurization was
ineffective.
In Type X systems the purging must be governed automatically by a

timer. Also, in Type X systems a door that can be readily opened without
the use of a key or a tool must be interlocked by a switch that disconnects
power from the equipment inside the enclosure when the door is opened.
The timer or switch must be suitable for use in Class I, Division 1 loca-
tions. A Type X pressurized enclosure containing hot parts that require a
cooldown period shall not be readily opened without the use of a tool or
key.
Requirements for Pressurized Enclosures in Class II

Locations
Pressurization systems for enclosures in Class II locations must meet the
general requirements. Because hazardous quantities of dust are not likely
to accumulate in an enclosure while it has not been pressurized, it is
unnecessary to purge the enclosure before energizing the equipment. It is
only necessary to remove any dust accumulations and pressurize before
turning on the power.
The internal pressure required in Class II locations depends on the density

of the hazardous dust. Below a specific particle density of 130 lb/ft3 (2083
kg/m3) the internal pressure is the same as for Class I locations, 0.1 in.
H2O (25 Pa). If the dust has a greater specific particle density the internal
pressure must be maintained at 0.5 in. H2O (125 Pa).
If the ignition temperature of the dust likely to be present is not known,

the surface temperature of the enclosure shall not exceed the limits stated
in the NEC, Table 500-3(f). See Table 4-1.
Type X pressurization of a tight enclosure in a Class II location does not

require an interlock between the protective gas system and the power
energizing the protected equipment. It is likely that only minor amounts
of dust will enter an enclosure after the pressure fails. If the failure is made
known by an alarm at a constantly attended location, administrative pro-
cedures require that the system be repaired or shut down soon after the
alarm is given.
Requirements for Pressurized Control Rooms

Control rooms are manned locations, so the protective gas must be air. It
must be clean and only trace amounts of flammable gas shall be present in
it. The source of air shall not be located in a classified location. The suction
ducting shall be noncombustible, free of leaks, and protected against
mechanical damage and corrosion so that holes in it will not allow flam-
mable gases to be drawn into the control room. When all openings are
closed, the pressurization system must maintain an internal pressure of
0.1 in. H2O (25 Pa). When all openings capable of being opened are open
simultaneously the system must maintain a minimum outward flow
velocity of 60 fpm (0.3 m/s) at each opening. The pressure in the control
room may drop below 0.1 in. H2O (25 Pa).
The positive pressure system may include heating, ventilating, and air
conditioning equipment as well as the auxiliary devices needed for the
protective system. The system must be sized to provide for the needs of air
consuming devices such as compressors and laboratory hoods as well as

the requirements of the positive pressure safety system.
In principle, Type X systems must be provided with an interlock switch to

deenergize all apparatus that is not suitable for operation in a Class I,
Division 1 location if pressurization is lost. However, it is common to per-
mit some time to elapse before equipment is manually deenergized in
order to facilitate graceful shutdown of electrical equipment and avoid
such hazards as that introduced by instantaneously shutting down all the
elements of a process control system. Type Y and Z pressurization systems
need not interlock the power to the protected apparatus with the pressur-
ization system. Failure of the pressurization system must always be
alarmed in a constantly attended location. Failure of pressurization shall
be detected at the discharge of the fan by measuring velocity pressure or
static pressure or by measuring flow. Measurement of power to the drive
motor is not reliable and shall not be used because slippage of the belt
linking the motor to the fan could cause pressurization to be lost without
signaling.
Power to the pressurization system shall be taken off the main power line
ahead of any disconnects for the control room, or it should be derived
from a separate source.
Pressurization of Enclosures with Internal Source of

Release—Continuous Dilution
The term “continuous dilution,” originally used in IEC Pub. 79-2, 3rd edi-
tion, from which the NFPA requirements are derived, is not used in NFPA
496, which refers to pressurization of enclosures having an internal release
of flammable gas or vapor. The objectives of any protective system for an
enclosure with an internal source of flammable materials are as follows:
(1) The system must prevent ignition of the internally released material
and must take into account the characteristics of the material and the
magnitude of the release.
(2) The system must prevent ingress of flammable material from outside
the enclosure if the enclosure is in a classified location.
A pressurization system of the types described earlier in this chapter satis-

fies the second requirement. A continuous flow of diluent gas or a blanket
of inert gas can satisfy the first objective if the system guarantees sufficient
flow of diluent under normal and abnormal conditions.
The type and magnitude of internal release is defined under normal and
abnormal conditions. “Normal” and “abnormal” conditions refer to the
condition of the containment system of the flammable gas or vapor. In

normal condition the release is characterized as “none” or “limited.” In
abnormal condition of the containment system a release is characterized as
“limited” or “unlimited.” A limited release is one that will be diluted by
the pressurization system, that is, by flow of protective gas, to less than
25% of the LEL. An unlimited release is one that cannot be diluted to a safe
level (less than 25% of the LEL) by the pressurization system.
A release inside an enclosure causes a hazard that is more severe and of

longer duration than a similar release in the open. Therefore, the terms
“normal” and “abnormal,” with respect to the containment system, are
defined on the basis of more stringent criteria than are customarily
applied to equipment in the open. In the absence of the protective system,
a leak that would be undetected in the open will slowly raise the concen-
tration inside an enclosure to the flammable range. In the open, the rise in
concentration of flammable vapor would be transient. Within an unpro-
tected enclosure, the flammable material remains for an extended period.
Because of the greater danger from releases inside an enclosure, the term
“normal” includes consideration of expected degradation of seals in the
containment system during the service life of the apparatus, probable
breakage, and other factors not usually considered part of normal opera-
tion. Pipe threads, welds, and similar joints may usually be considered not
to leak in normal service, especially if prototype systems can withstand a
type test of 1.5 times rated pressure without leaking. Plastic or elastomeric
seals would usually be presumed to degrade and leak in normal service.
Such a seal may be considered to have a limited normal release, as would
a flanged joint, a sliding seal, or a similar construction. The gas released if
a flame is extinguished would be considered to be a normal release. Flexi-
ble metallic elements such as bellows, Bourdon, and helixes would usually
be presumed to allow no release in normal condition of the system. Under
abnormal conditions, however, they may be presumed to fatigue and
release a relatively large quantity of material. Such a release must be lim-
ited by some means, such as a restriction in the input line. Similarly, a
release from a seal equivalent to that which would occur were the seal not
present would be considered as an abnormal condition. It might be neces-
sary to limit this kind of release also by fixed restrictions in the input lines,
if the basic construction does not limit flow to the quantity that can be
diluted by the protective gas supply to less than 25% of the LEL.
Table 6-1 summarizes the requirements for pressurizing enclosures in

which there may be an internal release of flammable material.
Additional
Requirements for
Internal No Release Limited Release Unlimited Release
External Area Equipment Under Normal Under Normal Under Abnormal
Classification Suitable for Conditions Conditions Conditions
Class I, Div. 1 None None None
Class I, Class I, Div. 2 Y Y None
Division I Nonclassified X X Inert
Class I, Class I, Div. 2 None Z None
Division 2 Nonclassified Z X Inert
Class II Class I, Div. 2 None Z None
Nonclassified Z X Inert
Class I, Div. 1 None None None*
None Class I, Div. 2 None Z None*
Nonclassified Z X Inert
*Precautions must be taken if unlimited release is large enough to alter the external area
classification.
Table 6-1 Pressurization Requirements for Enclosures Subject to Internal Release
(From NFPA 483-1993)
If the apparatus inside an enclosure is suitable for installation in Division

1, no further protection is required regardless of any release that may
occur in normal or abnormal conditions. If ignition-capable apparatus is
present in an enclosure with no normal source of release, the first failure in
the flammable gas containment system will allow an explosion to occur.
Therefore, a protective system is required. If the abnormal release is lim-
ited, a Type Z system will provide the requisite second failure. However,
if the enclosure is located in Division 1, the pressurization system must be
Type X to prevent an explosion from occurring after a pressurization fail-
ure allows flammable gas or vapor to enter the enclosure from the outside.
If the containment system is judged to permit a limited release under nor-
mal conditions, and the electrical apparatus is ignition-capable, all
protection comes from the pressurization system. Therefore, the pressur-
ization system must be Type X. If the release in abnormal conditions is
unlimited, no flow of air will protect against an explosion, so the protec-
tive gas must be inert.
If the electrical apparatus inside the enclosure is suitable for use in Divi-
sion 2 locations, and there is no release in normal operation, no protective
system is required, except in Division 1 locations. A failure of the appara-
tus and another of the containment system are necessary for an explosion
to occur. In Division 1 locations a Type Y protective system is needed to
protect against entry of flammable gas or vapor from the outside. If the
release in both normal conditions and abnormal conditions is limited, a
Type Y system is needed in Division 1 locations and a Type Z system is
needed in other locations. The requisite two failures are failure of the
apparatus and failure of the protective system.
The decision trees presented as part of the discussion of IEC Pub. 79-2 later
in this chapter may be useful in understanding Table 6-1 if the differences
in terminology are taken into account.
If the enclosure is located in a classified location the protective system

must maintain a pressure inside the enclosure of 0.1-in. H2O. Inert gas
must reduce the oxygen content to 5% by volume or to 50% of the mini-
mum oxygen concentration required for a flammable mixture, whichever
is lower.
If the enclosure contains apparatus suitable for Division 2 locations, that

is, it is not an ignition source in normal operation, and if failure of the con-
tainment system is not obvious, inert gas, not air, shall be used as a
protective gas. An undetected failure of the containment system followed
by failure of the enclosed apparatus might result in an explosion.
Among other design considerations for these protective systems is the

need for inlets and outlets for the protective gas to be placed to ensure
thorough mixing of the protective gas with any release and removal of the
mixture from the enclosure. If inert gas is used, it is permissible to close
outlets after purging to conserve inert gas, provided that this does not
introduce a hazard of too little inert gas flow. If the mixture in the contain-
ment system is within its flammability range, flame arresters or other
means shall prevent an explosion from being transmitted back into the
inlet piping. If air is used as a protective gas the pressure in the enclosure
must not exceed the pressure of the flammable material supply system.
Introduction of air into the process may cause a flammable mixture to
develop, or may cause corrosion or oxidation of process lines.
Analyzer Rooms
Analyzer rooms are designed to permit, or require, personnel to enter.

Analyzer rooms also often contain sampling systems containing flamma-
ble materials and analyzer enclosures that have internal sources of release.
The requirements for safe protective systems, therefore, are a meld of
those for control rooms and those for continuous dilution.
Analyzer rooms shall be separated from control rooms by at least a wall

impermeable to vapors. An analyzer room located in a classified area shall
be designed to exclude entry of flammable gases, vapors, and dusts. Venti-

lation system design shall ensure the following:
• Flow of air is distributed throughout the analyzer house.
• Air flow captures flammable vapors as close to the source as pos-
sible, taking into account that vapors such as heavier-than-air
hydrocarbon vapors should be removed at floor level and
lighter-than-air gases such as hydrogen and methane should be
removed at the ceiling.
• The flow of the ventilating system is sufficient to dilute leakage
after the largest single failure to less than 25% of the LEL.
The pressure and flow criteria for control rooms apply also to analyzer
houses. Inert gas shall not be used as protective gas for an entire analyzer
house. If inert gas is used for protecting apparatus within the analyzer
house, there shall be administrative procedures, warning signs, and train-
ing dedicated to reducing the risk that a person might enter an analyzer
house that is oxygen-deficient because of leakage of inert gas.
Restrictions or orifices to limit flow shall be installed at inlets for process

streams and also at outlets if they could be a source of uncontrolled leak-
age from the process. All flow-limiting devices shall be installed outside
the analyzer house, and, to minimize the amount of piping in the house,
all sample conditioning equipment shall be located outside the house. If
sample lines contain flammable mixtures, flame arresters or other means
shall be provided to prevent an explosion from propagating back into the
sample line.
Discharges from analyzers shall create no hazard inside or outside the

house. Discharge vents should be 5 ft from building openings, and at least
12 ft above grade. False ceilings and floors shall not be used because they
could interfere with ventilation and provide blind volumes for collection
of flammable vapors and gases.
Ventilation fans shall be constructed to minimize sparking. Some authori-

ties may mandate either aluminum blades only or conductive plastic
blades. Motors and control equipment shall be mounted outside the
house. Electrical supply to the house shall be disconnected if ventilation
fails and an audible and visible alarm shall be actuated in a constantly
attended location. Automatic shutdown is not required if the analyzer
house is classified Division 1 and contains no open flames. Interlock of
power with the ventilation system is also not necessary if the house is clas-
sified Division 2, contains no open flames, and has permanently installed
piping or tubing in the analyzer equipment and if flammable vapors are
piped to the outside of the building.
IEC Publication 79-2

The third edition of IEC Publication 79-2 is significant primarily because it
introduced the concept of continuous dilution for protection of apparatus
with internal sources of release of flammable materials. When one takes
into account the differences in the terminology “Zone” instead of “Divi-
sion,” and “first case” and “second case” systems corresponding in
essentials to Type X and Type Y or Z, respectively, the content of Publica-
tion 79-2 does not differ significantly in principle from that of NFPA 496.
However, because of a history of different practices regarding the use of
alarm-only and alarm/shutdown systems the document allows consider-
able leeway for local decision.
The odd terminology, first and second case, derives from the fact that area
classification was not part of the scope of IEC standardization at the time
that Publication 79-2 was written. The committee was, therefore, con-
strained to present pressurization requirements in terms of two
arrangements of apparatus, first case and second case.
First case systems interlock the pressurization system with power to the
enclosure so that failure of the pressurization system disconnects the elec-
trical supply and sounds an audible or visible alarm. If shutdown
jeopardizes the safety of the installation, it may be delayed, or it may be
replaced by a continuous alarm that operates until pressurization is
restored.
Second case systems provide only an alarm on failure of pressurization.

The pressurization shall then be restored promptly or the supply shall be
disconnected manually.
With respect to purging an enclosure before switching on the power,

marking and labeling, use of tools for entry, and so on, the document is
similar to U. S. practice. Like many IEC documents, however, it is deliber-
ately vague in many areas in order to achieve broad consensus. The
mandated minimum internal pressure in an enclosure is 0.2 in. H2O,
rather than 0.1 in.
Table 6-2 summarizes the IEC requirements for continuous dilution.

Though the table differs in appearance from Table 6-1 there are no impor-
tant differences in the result. The concentration of flammable materials
released in the enclosure is to be reduced to less than 25% of the LEL, and
inert gas shall reduce the oxygen content to 5% by volume or to 50% of the
minimum concentration of oxygen required to form a flammable mixture,
whichever is lower.
Figures 6-2, 6-3, and 6-4 show the decision trees from which Table 6-2 was
derived.
Internal release Ignition Apparatus with no

Area
Combination capable sources of ignition in
Normal Abnormal classification
apparatus normal operation
Zone 1 First case Second case
requirements requirements apply
apply
1 None Limited Zone 2 or Second case No protective
non-hazardous requirements measures required
apply
Zone 1 First case Second case
requirements requirements apply
apply1
2 None Unlimited
Zone 2 or non- Second case No protective
hazardous3 requirements measures required
apply1
Zone 1 or First case Second case
3 Limited Limited Zone 2 requirements requirements apply
apply
Zone 1 or First case Second case
4 Limited Unlimited Zone 2 requirements requirements apply2
apply1
1)Dilution gas is inert; air is not permitted. See Appendix B.
2)
Dilution gas is inert if the abnormal release is not self-revealing.
3)The unlimited release in this case is not large enough to alter the classification of the non-
hazardous area around the enclosure; if this is not so, precautions should be taken as a
consequence.
Table 6-2 IEC Publication 78-2 Requirements for Continuous Dilution (adapted)
Figure 6-2 Installation in Zone 2 or Non-Hazardous Areas of Normal Industrial Ignition-

Capable Apparatus (from IEC Publication 79-2)
Figure 6-3 Installation in Zone 1 of Normal Industrial Ignition-Capable Electrical

Apparatus (from IEC Publication 79-2)
Figure 6-4 Installation in Zone 1, Zone 2, or Non-Hazardous Areas of Electrical

Apparatus with No Sources of Ignition in Normal Operation (from IEC
Publication 79-2)
CENELEC EN50016
The first edition of EN50016 was essentially the same as IEC Publication
79-2, 1962 edition, discussing only pressurization of enclosures without
internal source of release. Only one type of protective system was
described. The second edition of EN50016, the final draft of which was
published in 1994, has been expanded to include consideration of enclo-
sure with internal source of release, that is, having a containment system
within the enclosure.
The focus in this edition of EN50016, as in other CENELEC standards for

explosion-protected apparatus, is to provide a document to assist testing
agencies in assessing the suitability of a product. It is not a principal objec-
tive of this document to give guidance to the user of the pressurization
technique. In this regard the document is quite different from other stan-
dards for pressurization.
The document discusses the following three approaches to pressurization:

(1) Static pressurization—filling an enclosure with inert gas and provid-
ing no continuous supply of gas in the classified location.
(2) Pressurization with leakage compensation—pressurization of the sort

discussed in NFPA 496 where protective gas flow is sufficient to
maintain the mandated internal pressure.
(3) Pressurization with continuous flow of protective gas—equivalent to

continuous dilution.
Static Pressurization
Enclosures protected by static pressurization shall be filled with inert gas

to ensure that initial oxygen concentration is less than 1% by volume (v/v).
Internal sources of release of flammable substances are not permitted. Two
automatic devices shall operate when the overpressure falls below the
minimum specified by the manufacturer. Correct operation of these
devices shall be verifiable in service. The minimum overpressure shall be
no less than the pressure loss in 1 hour, or during 100 times the period of
cooling of enclosed components to a temperature consistent with the T
code. However, the minimum specified overpressure shall be at least 50
Pa. The user shall decide whether the automatic devices are used as inter-
locks or alarms. It must be necessary to use a tool to open any doors or
covers of industrial apparatus, which must be marked with a warning
against opening in the hazardous area. Group I (for gassy mines) appara-
tus shall have special fasteners.
Pressurization with Leakage Compensation

The general requirements for constructing and marking pressurized
enclosures with leakage compensation and the provision of protective gas
are similar to those standards already discussed. They also apply to enclo-
sures with containment systems. The enclosure and ducting shall
withstand 1.5 times the minimum overpressure specified by the manufac-
turer, but at least 200 Pa. If, in service, the ducting or enclosure can be
subjected to a pressure high enough to cause deformation, the manufac-
turer shall supply a safety device to prevent any adverse effect on the
protection type.
Such safety devices shall be incapable of causing an explosion or mounted

outside the hazardous area. They form safety-related parts of a control
system and shall conform to Category 3 as defined in CEN EN954-1. The
intent of this requirement, which need not be verified by the testing
agency, is that no single fault shall compromise the safety provided by the
pressurization system. Most systems designed in the past probably met
this requirement. The manufacturer or the user may provide these
devices. In the latter case, the manufacturer must provide the information
about the system necessary to ensure conformity to the standard (Figure
6-5). The manufacturer shall also supply information about safety devices’
action levels and tolerances.
Figure 6-5 Draft Recommendations (from EN50016, 2nd Ed., 1994)

Every system shall be provided with time delay relays and devices for
monitoring the flow of protective gas to ensure that the electrical appara-
tus cannot be energized until a purging cycle has been completed. After
the prescribed flow and overpressure have been achieved, a purge timer
runs for the specified purge time and then permits the apparatus to be
energized. The circuit shall reset to the beginning if any step in the
sequence fails. The objective of purging is to reduce the concentration of
the worst case flammable gas for which the system is designed to 25% of
the LEL if the protective gas is air; or to reduce the oxygen concentration
to below 2% volume/volume if the protective gas is inert. The values of
minimum overpressure and, if needed, the minimum flow rate shall be
stated by the manufacturer and verified by type testing. A minimum over-
pressure of 50 Pa must exist at every point where leakage can occur.
Automatic means shall monitor overpressure and flow rate, if a minimum
is specified, and shall actuate if either falls below the stated value. It is the
responsibility of the user to determine whether the action will be used to
alarm or to shut down electrical apparatus. It is also the user's responsibil-
ity to account for the volume of protective gas ducts and adjust the
purging time to ensure that the purging time allows a purge of five times
the volume of the ducting.
Safety devices for a protective gas supply that is common to many enclo-
sures may also be common to all the enclosures. The system design must
account for the most unfavorable conditions in any enclosure in the group.
Opening a cover or door in one enclosure need not interrupt the power to
all enclosures or actuate an alarm if the following three conditions are met:
(1) Opening a door is preceded by switching off the power to all appara-
tus in that enclosure which is not protected by a recognized type of
protection.
(2) The common safety device continues to monitor the overpressure

and, if necessary, the flow through all the other enclosures in the
group.
(3) Purging precedes the subsequent connecting of the supply to the elec-
trical apparatus in the opened enclosure.
Pressurization of an Enclosure Housing a Containment

System
This standard defines three levels of release.
(1) No release
There is no release when the containment system is infallible, that is, it

has no intentional leakage into the enclosure. This can be assumed if
the containment system consists of glass, ceramic, or metal tubes or

vessels having no moving joints and joined by welding, brazing,
glass-to-metal seals, or eutectic seals and if the system passes the test
for infallibility discussed below. Low temperature solder joints (such
as tin/lead alloys) are not acceptable. Infallibility is determined by the
following tests.
There is no release if the gas mixture within the containment system is

always below the LEL.
There is no release if the minimum pressure to be maintained in the

enclosure is at least 50 Pa higher than the maximum pressure specified
for the containment system and if an automatic device is designed to
operate if the pressure difference falls below 50 Pa.
(2) Limited release
The rate of release into the enclosure from the containment system,
including any normal release, and the flow from the process through
the containment system, shall always be predictable under all condi-
tions of containment failure. Flow limiting devices shall be located
external to the enclosure, unless the containment system is infallible as
described above, between the point of entry into the enclosure and the
flow-limiting device. Systems containing a flame shall be judged with
the flame extinguished. Elastomeric seals, windows, and other nonme-
tallic parts may be used in the containment system. Flanged joints and
pipe threads are permitted.
(3) Unlimited release
The release of a flammable liquid from a containment system that

would be considered a source of a limited release, if filled with vapor
or gas, is an unlimited release.
The rules for applying pressurization are summarized in Table 6-3.

Included in these rules is a limitation imposed by an assumption that is
repeated in several forms in the standard. The assumption is that it is diffi-
cult or impossible to achieve secure protection using inert gas in apparatus
that may release gases or vapors which are capable of reacting exothermi-
cally with little or no oxygen present (e.g., UEL greater than 80%). This
statement is never explained in the document. In the Austin, Bartels,
Owler paper listed in the references, the authors suggest that a high con-
centration of a flammable material with inert protective gas in the interior
of an enclosure would not be hazardous, but opening the enclosure might
cause an explosion. The paper then asserts, without further explanation,
that inert gas should not be used if the UEL of the gas or vapor is above
80%. One justification for this statement is that inert gas would most fre-
quently be used when there is ignition-capable equipment in the

enclosure, and a small leak might bring the mixture below the UEL. How-
ever, there are instances where inert gas might be used for reasons of
convenience, for example, when no suitable supply of air is available. In
this instance, were there no ignition-capable devices in the enclosure, the
hazard would be low.
NO RELEASE
LIMITED RELEASE UNLIMITED RELEASE
(or no containment)
STATIC INERT GAS (no Not permitted. Not permitted.
PRESSURIZATION containment system
permitted)
PRESSURIZATION AIR OR INERT GAS
INERT GAS if no INERT GAS if no normal
WITH LEAKAGE normal release and release and material in
COMPENSATION material in containment containment has < 2%
has LEL < 80% and < entrained oxygen.
2% entrained oxygen.
PRESSURIZATION AIR or INERT GAS at AIR to reduce If no normal release,
WITH CONTINUOUS flow rate needed to concentration to 25% INERT GAS to dilute to
FLOW maintain 50 Pa LEL or INERT GAS to 2% v/v. Flow may be
overpressure reduce oxygen reduced after purging to
concentration to 2% v/v. amount to maintain
If UEL > 80% only AIR. overpressure if contained
material has entrained
oxygen < 2%.
Table 6-3 Summary of permitted applications of pressurization in EN50016, 2nd

edition
Assessment of Pressurized Apparatus
Because EN50016 is intended to be suitable for certification of apparatus, a

complete set of verification tests is prescribed as well as a set of routine
tests on all production. While these give good guidance to a person imple-
menting a pressurization system for use in the plant, it is unlikely that a
user would perform all the testing covered by this standard. Most likely a
person would assess conformance to some of the requirements on the
basis of judgment.
Type tests to be performed are as follows:

• All relevant tests of EN50014 (General Requirements for all explo-
sion-protected apparatus)—These tests would include impact
tests of windows and the enclosure, temperature tests of plastic
enclosures and similar tests.
• Overpressure test—A test pressure of 1.5 times the maximum

overpressure specified by the manufacturer, but not less than 200
Pa shall not deform the enclosure to invalidate the protection
provided by pressurization.
• Leakage test—The leakage rate when the maximum overpressure
specified by the manufacturer is applied to the enclosure shall
not exceed the maximum value stated by the manufacturer. Stati-
cally pressurized enclosures shall be evaluated against the maxi-
mum leakage and resulting drop in pressure as discussed above.
• Purging test or filling procedure for static pressurization—The enclo-
sure is fitted with a number of sampling tubes and then filled
with a test gas; a nonflammable gas if the protective gas is air or
air if the protective gas is inert. If the enclosure has a limited
release in normal operation, the maximum release is simulated.
The time to purge or dilute the enclosure at the purging rate spec-
ified by the manufacturer shall not exceed the time specified by
the manufacturer. Purging is complete when an inert test gas is
diluted to 25% of the LEL of the gases for which protection is
designed. If both helium and argon or carbon dioxide are used to
simulate light and heavy gases and vapors, the end of purging is
1% helium and 0.25% argon or carbon dioxide. If the protective
gas is inert, and air is used for testing the end of purge time is
when the oxygen concentration is less than 2% volume/volume.
• Dilution test—The dilution test is performed at the dilution rate
specified, with the maximum internal release simulated. Criteria
for passing are the same as above. In addition, a volume of flam-
mable gas equivalent to the volume of the containment system is
released while the specified minimum rate of diluent flows. The
concentration shall not rise above twice the above limits and shall
be below them after 30 min has elapsed.
• Verification of minimum overpressure—Sampling of pressure at var-
ious points within the enclosure when the minimum overpres-
sure specified by the manufacturer is applied to determine that 50
Pa minimum is maintained throughout the pressurized enclosure.
• Test of infallible containment systems—An infallible containment
system shall withstand 5 times the maximum overpressure speci-
fied for normal service for 2 minutes without any permanent
deformation. The containment system shall maintain an internal
vacuum of 103 Torr when surrounded by helium at the maximum
pressure specified for the containment system. Alternatively, the
pressure may be applied inside the containment system and the
vacuum maintained outside it.
• Overpressure test for containment systems with limited release—No

permanent deformation when 1.5 times maximum rated internal
pressure is applied for 2 minutes.
The following tests are also to be performed as routine tests on all

production:
• any required by EN50014
• overpressure test
• leakage test
• tests for infallible containment systems with limited release
Marking
In addition to the markings required by EN50014 the following shall be

marked on enclosures other than those protected by static pressurization:
• internal free volume, including ductwork that is integral to the
enclosure
• type of protective gas, if not air
• minimum quantity of gas required to purge the enclosure, as
• minimum purging flow rate
• minimum purging duration
• minimum additional purging time to purge volume of ducts
not supplied with the enclosure (responsibility of the user)
• minimum and maximum overpressure
• maximum leakage rate from the pressurized enclosure
• temperature of protective gas at inlet if specified by the manufac-
turer
• the point or points at which pressure is to be monitored
If the enclosure has a containment system, the following information must

also be given:
• category of internal release
• minimum flow rate of protective gas, if any
• the maximum inlet pressure to the containment system
• the maximum flow rate into the containment system
• the maximum oxygen concentration in the containment system
Enclosures protected by static pressurization shall be marked with two

warnings:
(1) THIS ENCLOSURE IS PROTECTED BY STATIC PRESSURIZATION.
(2) THIS ENCLOSURE SHALL BE FILLED IN A NONHAZARDOUS
LOCATION IN ACCORDANCE WITH THE MANUFACTURER'S
INSTRUCTIONS.
The text of the CENELEC draft has been submitted to IEC TC31 as a basis
for revision of Publication 79-2. The CENELEC view of pressurizing sys-
tems as certified apparatus and the North American view of pressurized
systems being installed under the approval of the authority having juris-
diction will certainly engender serious discussion.
This author's opinion is that pressurization is not likely to be used in a

location that has a hazard greater than Zone 1, and that it is usually used
because no other type of protection is practical or available. Therefore, cer-
tification of hardware and enclosures should not be encouraged, although
certification of the pressure control and alarm packages may be helpful.
The installer should accept responsibility for safe installation of the system.
Installation of Pressurization Systems

NFPA 496 and the early versions of ISA S12.4 present only the important
functional requirements for the system supplying protective air. EN50016,
intended to be a basis for certification of apparatus rather than as a guide
for the user, leaves to the user the decisions about choice of inert gas or air
as a protective gas and the use of the contacts to alarm or deenergize the
apparatus. IEC 79-2 and EN50016 provide some guidance about expected
overpressure as a function of ducting layout and location of inlet and
outlet.
Ecker et al. offer more detailed guidance to the facility designer. Their arti-
cle considers a design approach for a pressurized equipment panel. For a
large enclosure of this sort, NFPA 496 requires a 10-enclosure-volume
prepurge before energizing the electrical apparatus. The volume of the
exemplary enclosure is 50 ft3 (1.4 m3). Low enclosure pressure is detected
by a differential pressure switch. A pitot tube detects flow in the supply
duct. Ecker et al., assuming a prepurge time of 5 min, compute purge rate
to be
3
( 50 ft ) × 10 × 1.1 × 1.25- 3 3
------------------------------------------------------------ = 137 ft /min (0.065 m /s)
5
The factors of 1.1 and 1.25 are introduced to cover possibility of the pitot
tube reading 10% too high and to cover factors such as piping roughness,
panel components that interrupt smooth flow, panel leakage, and exhaust
port location.
Figure 6-6 (a) Distribution of Overpressure. Examples of the Static Overpressure Along
the Ducts and Through a Pressurized Enclosure (from IEC Publication 79-2,
2nd Ed.)
Figure 6-6 (b) Distribution of Overpressure. Example of the Static Overpressure in a

Pressurized Electric Rotating Machine with Fan (from IEC Publication 79-2,
2nd Ed.)
They calculate the exhaust port area from the orifice equation:
Q
A = ------------------
K 2gh
where
A = effective port area (ft2)
Q = flow rate (ft3/s)

K = orifice coefficient (assumed to be 0.61)
g = gravitational constant (32.2 ft/s2)
h = pressure drop across exhaust port (ft of air)
Q = 137/60 = 2.78 ft3/s
h = 0.7 in. H2O = 45.1 ft of air
Therefore, A = 2.78/0.61(2 × 32.2 × 45.1)1/2 = 0.07 ft2.
A damper, solenoid valve, or butterfly valve is provided behind the

exhaust port to conserve air after the initial purge is completed.
The use of 0.7 in. of water as the operating internal pressure was dictated
by the characteristic of the pressure switch used rather than by safety
requirements.
When considering supply ducting, the authors note that although conduit
or polyvinyl chloride pipe are more costly than sheet metal, ease of fabri-
cation, smoothness, and ease of making large bends may make them
preferable in medium capacity systems.
The authors recommend air velocity of 500 to 5,000 ft/min at the point of
entry into the panel. They establish the lower limit because velometer
accuracy falls off at lower velocity, making it difficult to check proper
operation of the supply system. Above the suggested upper limit the flow
tends to be noisy and turbulent, and power requirements are higher.
For the volume required in this example, 137 ft3/min, a trial calculation
with a 2-in. pipe yields a velocity of 5,870 ft/min. Friction loss is stated to
be 13.5 in. H2O/100 ft, which is an excessive drop. A 3-in. pipe would give
more reasonable values of 2,690 ft/min and 4.5 in. H2O/100 ft.
The authors recommend that the blower be designed to supply the mini-
mum panel pressure plus piping losses, with a safety factor of 1.5 to cover
piping contingencies. They recommend that the blower be mounted above
the panel if possible, well above floor level, to avoid clogging of the intake
filter by dust and dirt.
References
Beutler, J. A., and J. A. Clark, “Heat Transfer and Computational Consid-
erations in the Design of Instrument Cases for Operation in Hazardous
Atmospheres,” Proceedings of 1960 Symposium on Safety for Electrical Instru-
mentation in Hazardous Areas, Instrument Society of America, Pittsburgh.
McCarron, R., “Considerations in Instrument Case Purging,” Proceedings

of 1960 Symposium on Safety for Electrical Instrumentation in Hazardous Area,
Instrument Society of America, Pittsburgh.
Electrical Apparatus for Explosive Gas Atmospheres, Part 2: Electrical Appara-

tus—Type of Protection `p', IEC Publication 79-2, International
Electrotechnical Commission, Geneva, 1983.
McCarron, R., “Considerations in Instrument Case Purging,” Electrical

Safety Practices, ISA Monograph No. 110, Instrument Society of America,
Pittsburgh, 1965, pp. 16–26.
Allman, W., McKinney, A. H., and J. A. Sheldon, “Temperature Rise in

Metal Housings Due to Electrical Short Circuits,” Electrical Safety Prac-
tices, ISA Monograph No. 111, Instrument Society of America, Pittsburgh,
1967, pp. 82–87.
“Purged and Pressurized Enclosures for Electrical Equipment,” NFPA No.

496, National Fire Protection Association, Quincy, 1993.
Lee, R. H., “Fuses,” Electrical Safety Practices, ISA Monograph No. 112,
Instrument Society of America, Pittsburgh, 1969, pp. 58–60.
McCarron, R., “Report of An Investigation of Metal Instrument Cases,”

Electrical Safety Practices, ISA Monograph No. 112, Instrument Society of
America, Pittsburgh, 1969, pp. 53–57.
“Instrument Purging for Reduction of Hazardous Area Classifications,”

S12.4, Instrument Society of America, Pittsburgh, 1970.
Ecker, H. W., James, B. A., and R. H. Toensing, “Electrical Safety: Design-

ing Purged Enclosures,” Chemical Engineering (May 13, 1974), 93–97.
Austin, A. J., Bartels, A. L., and A. M. Owler, “Safety Requirements for

Analytical Instruments,” IEE Conference Publication No. 296, London,
1988.
EN50016—Pressurized Apparatus `p'. Part 3 of BS 5501, “Electrical Appa-
ratus for Potentially Explosive Atmospheres,” British Standards
Institution, London.
CHAPTER 7
ENCAPSULATION,
SEALING, AND IMMERSION
Although the techniques of encapsulation, sealing, and immersion in oil or

sand are not likely to be applied to a complete instrument, the techniques
interest the instrument designer and user because they can be applied to
components or subassemblies to reduce the hazard classification of the
instrument.
All three methods of hazard reduction impede contact between the source
of ignition and the flammable atmosphere and/or quench any incipient
flame.
Oil and Sand Immersion

Oil-immersed power equipment has been widely used worldwide. Sand-
filled apparatus has been used in Europe, especially in France. Sand-filled
apparatus is nonhazardous because the sand restricts the flow of flamma-
ble atmosphere to the equipment (this is discussed under “Sealing”) and
quenches the incipient flame ball. This technique of immersion has little
potential application to instruments and is not considered further here.
For additional information, see IEC Publication 79-5 and CENELEC Stan-
dard EN50017.
Apparatus that is oil-immersed to reduce hazard is different from equip-

ment that is filled with oil primarily to quench the arc and reduce
temperatures in arcing parts. General-purpose oil-filled starters and push
buttons may sometimes sustain arcs that are not contained within the case.
In Division 2 locations, apparatus such as circuit breakers, motor control-

lers, and switches intended to interrupt current in normal operation are
acceptable in general-purpose enclosures if the arcing parts are immersed
in oil. Power contacts must be immersed at least 2 in. (50 mm). Control cir-
cuits may be immersed only 1 in. (25 mm).
Underwriters’ Laboratories Standard No. 698 outlines requirements for

oil-immersed apparatus. In summary, they are:
(1) The enclosure must be of substantial metallic construction designed
to preclude possibility of sparks being produced above the oil level.
218 Encapsulation, Sealing, and Immersion
(2) Normal oil level must extend 6 in. (150 mm) above any electrical joints
or arcing parts. In general, there may be no mechanical or soldered
joints above the oil level. Auxiliary devices used only in control circuit
of the equipment and located 2 in. (50 mm) below the oil level will be
accepted for test.
(3) Oil level must be indicated by a suitable, visible, level indicator with
graduations to show minimum, normal, and maximum oil levels. A
warning label must call attention to the necessity of keeping the oil
level above the minimum level at all times.
(4) The drain hole, if furnished, must be provided with a pipe plug with
five full-thread engagement.
(5) Ordinary fuses are not approved for use within the enclosure.
(6) Marking of enclosure surface temperature or T code is required as for

explosionproof enclosures and all other explosion-protected equip-
ment.
(7) Marking requirements are similar to those for explosionproof and

dust-ignition-proof enclosures:
(a) Identification of manufacturer and model number.
(b) Electrical ratings.
(c) Class and group of location of permissible use.
(d) Maximum operating temperature.
(e) Warning to disconnect device from supply before opening

enclosure and to keep tightly closed while in operation.
Testing of control circuits is conducted with oil level 1 in. (25 mm) below
the marked minimum level. For tests of the primary device the oil level is
reduced 2 in. (50 mm) below the normal level. Tests are also performed
with the equipment in operation, and flammable gas or vapor mixtures
are introduced above the oil level to determine whether ignition will
occur. It is also required that, with the equipment at the full-rated load
and rated voltage, the oil shall produce no dangerous amount of flamma-
ble gases or vapors.
Oil immersion in instrument systems is not attractive. Accessibility for cal-

ibration and maintenance is limited and inconvenient. When an
intrinsically safe system design is not feasible, an explosionproof design or
pressurization can usually provide an acceptable solution for Class I, Divi-
Encapsulation, Sealing, and Immersion 219
sion 1 applications. For Class I, Division 2 locations, encapsulation,

sealing, explosionproofing, pressurization, or nonincendive circuit design
are usually more attractive alternative design approaches.
Sealing
The principle of sealing, commonly, but unnecessarily, called “hermetic”
sealing, has been recognized in the NEC, paragraph 501-3(b)(1) for four
decades. This paragraph states that general-purpose enclosures may be
used in Class I, Division 2 locations if make-break contacts are hermeti-
cally sealed against the entrance of gases or vapors. However, there was
no definition in the NEC and, until the 1980s, no other recognized author-
ity stated any requirements that a seal must meet to be considered suitable
for the use intended in 501(3)(b) of the NEC.
Rules for assessment and test of sealed devices are now published in ISA
S12.12, IEC Publication 79-15, CSA C22.2 No. 213, and BS 6941, as well as
in the internal standards of test houses. A CENELEC standard has been
drafted. The details of present standards are discussed later in this
chapter.
If one were to apply the dictionary definition of a hermetic seal, very few
devices could meet the requirements. The dictionary definition of a her-
metic seal is “made perfectly closed or air-tight so that no gas or spirit can
enter or escape.” In practice, any seal passes gas or vapor to some extent. In
the limit even a metallic or gas container is somewhat permeable to vapors
and gases. In practice, therefore, a hermetic seal is a controlled leak.
Military specifications have defined the quality of seals in several ways.

MIL S-8484, now obsolete, defined quality in terms of the volume of leak-
age per year, per inch of seal, at a pressure differential of one atmosphere
across the seal (see Table 7-1).
MIL R-6106J defined three types of sealed relays:

Unsealed Totally enclosed with respect to mechanical protection and
entry of dust.
Shall be explosionproof, that is, when an internal test
mixture is ignited, the explosion shall not propagate to a
surrounding mixture, nor shall it rupture the enclosure. The
test mixtures are a series of butane-air mixtures at 1/2%
intervals from 3-6%.
Hermetically sealed Sealed by fusion of metal to metal or ceramic.
Evacuated and backfilled with inert gas to 15 psia to
facilitate leak checking.
Maximum permissible leak rates stated as cm3/s/in.3 of net
contained volume:
10-6 for volumes 1 in.3 or less;
10-8 for volumes greater than 2 in.3
Environmentally sealed Not sealed by metal-to-metal or metal-to-ceramic fusion.
Evacuated and backfilled like hermetically sealed units.
Leakage limits defined in individual specification sheets.
Of the approximately 150 specifications listed at the end of the generic

specification, only two were listed as environmentally sealed. These were
60 A relays of 150 cm3 volume. Maximum leakage is specified to be 6 × 10-4
atm-cm3/s. In terms of the definitions in S-8484, this leak rate is that of a
Grade C seal.
Bedwell and Meyer reported the results of their testing of a variety of seal-
ing methods:
Seal Type Results
Aluminum or silver soldering Grade A after 25C, 100% RH for 30 days;
270 min vibration at 1.5 mm amplitude,
10-55 Hz, and three temperature cycles
from –65°C to 200°C
Welded Variable quality - high degree of skill
needed to obtain good seals
Adhesive Variable quality; many were Grade C
O-rings or natural rubber gaskets in Large percentage was Grade A
rectangular grooves
Composition with cork or asbestos High percentage of Grade D seals
Copper or lead gaskets in metal-to-metal 96% of copper were Grade A. Lead -
joints with 90° ridges results were more variable
Lapped joints Variable results
It seems that these results lead to a reasonable conclusion that most seals
likely to be found in relays and similar devices are at least of Grade C qual-
ity. Industrial conditions are less onerous than the test conditions of these
investigations so, in use, the seals will not be stressed to the same level.
To demonstrate what these grades imply in mechanical construction,

Table 7-1 lists the hole size which, for 1 atm pressure differential across a
0.030-in. wall, would give a leak rate corresponding to a 1-in. long seal of
the several grades of MIL S-8484. These hole diameters are calculated from
the following equation:
4
17 r 2 2
Q = 5.98 × 10 ---- ( P 2 – P 1 )
l
where
Q = leak rate in standard cubic centimeters per year per inch
r = radius of hole (in.)
l = hole length (in.)
P1, P2 = pressures on either side of the leak (atm)
Grade Leak rate Hole diameter (in.)

A <1 cm3/yr/in./atm 30 × 10-6
B < 99 cm3/yr/in./atm 95 × 10-6
C < 9999 cm3/yr/in./atm 300 × 10-6
D > 9999 cm3/yr/in./atm > 300 × 10-6
Table 7-1 Example of Leak Rates
In connection with the use of hermetic seals, it is sometimes suggested that

the sealed device be pressurized. However, in the industrial application of
hermetically sealed devices, pressurization does not significantly increase
safety.
Most commercial hermetically sealed devices, if pressurized at all, do not

have initial internal pressure greater than 1.2 atm absolute. This internal
pressure in a typical relay would drop to 1.1 atm in a year or so if the seal
were Grade A. A Grade C seal would allow the pressure to drop from 1.2
to 1.1 atm in about an hour. Even if initial pressurization were at 2 atm, a
Grade C seal would allow the pressure to drop to 1.1 atm in 2-1/2 hr. In
the preceding examples, an end point of 1.1 atm was chosen because it is a
sufficient internal pressure to guarantee that barometric pressure changes
and pressure changes caused by diurnal temperature cycling would not
cause internal pressure to become less than the external pressure. Reversal
of pressure differential would cause pumping through the seal. Since the
useful life of industrial equipment is far longer than that of the safe period
using even a Grade A seal, pressurization alone is not sufficient. Even if
the seal in a metallic container were initially perfect, at almost any time
within the lifetime of the equipment corrosion could cause small unno-
ticed pinholes, and protection would be quickly lost. The basis of these
statements is further discussed in Appendix B.
The foregoing considerations show that hermetically sealed devices are

not safe because they are pressurized or because they do not leak. A her-
metically sealed device is safe in a Class I, Division 2 location because the seal
prevents the atmosphere inside the sealed enclosure from reaching ignitable con-
centration during intermittent exposure of the enclosure to a combustible cloud of
gas or vapor. It is the objective of this chapter to develop a rationale for
defining the essential requirements of a sealed device, to discuss other
applications of the same rationale, and to describe the assessment criteria
of important current standards for sealed devices.
In Class I, Division 2 locations the atmosphere is above the lower explo-

sive limit only infrequently. The duration of the event is short compared
with the period between occurrences. The average concentration sur-
rounding the sealed device is below the lower explosive limit. The average
concentration inside the sealed enclosure is the same as the average con-
centration outside the enclosure. During the infrequent excursions of the
outside concentration above the lower explosive limit, the seal prevents
the rising concentration in the atmosphere inside the enclosure from
reaching the lower explosive limit. The seal acts like a low pass filter to the
occasional short pulses of flammable concentration surrounding it.
The utility of sealed enclosures in Division 2 locations derives from this

ability to prevent transient excursions of the atmosphere outside the
enclosure to combustible concentration from raising the concentration
inside the enclosure to combustible concentration. Note that there is an
essential condition implicit in the foregoing statements that is of para-
mount importance to the safe application of seals. The average
concentration in the atmosphere around sealed devices must always be
nonhazardous because the concentration inside will eventually reach the
average concentration outside the sealed enclosure.
Flammable material will be transported across the seal of an enclosure by

flow caused by a difference between internal and external pressure, by dif-
fusion of material through the seal, or by wind pressure driving material
across the seal. In this discussion, the effect of wind pressure is not consid-
ered further. The effects are too small (see Chapter 6). Most sealed devices
are installed in a general-purpose enclosure according to the NEC. When
sealing is applied to a device not so protected, additional consideration
must be given to protection from the environment. Differential pressure
across the seal is developed by changes in ambient temperature or baro-

metric pressure and by temperature changes inside the enclosure as
equipment is energized and deenergized. Because both causes of pressure
difference may be cyclic, the flow caused by these pressure differences is
often referred to as “breathing.”
In the following paragraphs the effect of breathing and diffusion are con-
sidered. For most applications, any seal is safe with respect to breathing.
Therefore, the quality of seal needed for safety is determined by diffusion
transport.
For consideration of the effects of breathing, assume the limiting case of a

seal that offers no impediment to flow of gas. Any pressure differential
across the seal will immediately drive material into or out of the enclosure.
Assume that the pressure drop across the seal cycles 0.1 atm daily. If this is
caused by internal heating, it implies a 30°C rise of the entire volume
within the enclosure. If caused by ambient temperature and barometric
swings, it implies a daily temperature swing of about 15°C in phase with a
barometric change of 0.05 atm.
If the atmosphere outside the sealed enclosure is 10% hydrogen, and a 0.1-
atm differential pressure drives outside atmosphere into the enclosure, the
mass of gas inside the enclosure will increase 10%. The internal concentra-
tion, if it were initially 0, increases to 1% hydrogen. Return of pressure
outside the enclosure to 1 atm will allow the pressure inside the enclosure
also to return to 1 atm, but the concentration will remain at 1% hydrogen.
Over the next pressure cycle, the concentration inside the enclosure will
rise to about 2% hydrogen, and so on. It will take four pressure reversals
to raise the internal concentration to 4%, the lower explosive limit of
hydrogen. In the unlikely event that the 0.1-atm pressure cycle is caused
by in-phase ambient temperature and barometric pressure changes, a 10%
hydrogen atmosphere would have to exist for four days for the concentra-
tion inside the enclosure to reach the lower explosive limit. If the pressure
differential is caused by an average 30°C rise inside the enclosure, an
unlikely figure, one would have to postulate presence of the 10% hydro-
gen atmosphere outside the enclosure for many hours while the enclosure
heats and cools through four cycles.
To achieve a flammable mixture inside the enclosure, we must assume an

unlikely combination of circumstances. We must assume a more severe
daily cycling pressure differential across the seal than would ever exist in
practice; we must assume 10% hydrogen existing outside the instrument
for many hours or days. If we assume 5% hydrogen, the atmosphere
would have to be present even longer before the 4% LEL could be reached
inside the enclosure. The efficacy of the seal, therefore, lies in its long-time
constant compared to any probable duration of the hazard external to it.

Breathing of a real seal is treated in more detail in Appendix B.
From these considerations, it can be concluded that breathing does not

determine the quality required of the seal. This conclusion is based on the
assumption that the enclosed equipment does not raise the temperature
inside the enclosure significantly. This is a reasonable assumption for
most relays and similar sparking contacts used in instrumentation.
In his analytical and experimental investigations Conn addressed the

entry of methane into explosionproof enclosures under conditions that
might exist in a mine. He assumed a major release of methane, such as
might be caused by breaking through into an abandoned chamber, and a
sharp temperature drop inside an explosionproof enclosure to infuse the
methane into the enclosure. His equation, derived assuming no impedi-
ment to gas flow, and based on the equation of state of the gas inside the
enclosure, has a form for zero initial concentration similar to the equation
derived in Appendix B-2. Conn's relevant ratio is the ratio of the final
internal temperature to the maximum (peak) internal temperature. This is
the temperature after heating by operation of the internal electrical appa-
ratus. It is assumed that combustible gas sensors would cause the power
to be removed from the explosionproof enclosure at a small fraction of the
LEL of methane, thus allowing the enclosure to breathe in the surrounding
methane as the internal air cooled. The hazard of concern is a possible
explosion after the air in the mine had been rendered safe by forced venti-
lation and switches in the enclosure were again operated.
Conn found that his calculations had to assume a 25% concentration sur-
rounding the enclosure to yield a 5% concentration inside the enclosure
after one time constant of cooling from a 150°C initial internal tempera-
ture. This implies 50% methane to obtain the optimum concentration of
10% inside the enclosure. An assumption of 20% methane in the surround
and cooling to mine ambient in 5 hours yielded a 5.5% internal
concentration.
Conn also investigated the results of a release that is exponential in nature

and which has a time constant such that the external concentration drops
from 50% to 30% in about an hour. He found that the concentration in the
box increases to about 8% in the first hour. In his experiments, he found
the calculated results to be supported well by the experimental data. The
exceptions were in those calculations that assume exponential cooling of
the air inside the enclosure. The actual cooling was greater than calculated
because inflow of cool ambient air aided the process. For this reason the
infusion of methane into the enclosure in the early stages is higher than
calculated.
In Technical Note #8, Conn developed equations for estimating internal

air temperature of an explosionproof enclosure based on exterior surface
temperature and interior power dissipation. This equation is for use when
measurements of interior air or wall temperature are impractical. He
states that the interior wall temperature can be 25°F lower than the interior
air temperature and still give calculated values of infused methane within
0.5% for most practical problems.
If the temperature rise is high and/or equipment is energized and deener-

gized frequently, so that the resulting cyclic pressure change pumps flow
through the seal, the safety of the system must be calculated using the
equations developed later in this chapter. Even in such cases, the flow
impediment presented by the leak in a practically sealed enclosure will
likely make the device safe. How leaky a seal may be permitted is deter-
mined by diffusion effects, which are considered in the following
paragraphs.
If two gases are mixed and the concentration of each gas is not uniform
throughout the mixture, both gases will flow to make the mixture homo-
geneous. This flow, called diffusion flow, is related to thermal agitation of
the molecules. It depends on the presence of a concentration gradient. In
gases and vapors, the flow may be described as a transport of material to
make the partial pressure of a component the same at every point in the
mixture. The rate of diffusion is related to molecular weight, with lighter
molecules diffusing more rapidly. The rate of diffusion in gases is not
highly dependent on concentration level.
The fundamental equation for diffusion flow is

dc
dQ = – Δ ------ dydz
dx
where
dQ = the amount of material passing through an area dy dz in the
direction of x in a time dt
dc/dx = the rate of increase of concentration in the direction of x
Δ = the coefficient of diffusion of the material
For considering diffusion flow through a circular hole in an enclosure, the

equation can be written in the form
dC i C o – C i πd 2
-------- = Δ ----------------- ---------
dt V0 l 4
where
V0 = volume of enclosure (cm3)
Ci = concentration inside enclosure
Co = concentration outside enclosure
l = enclosure wall thickness (cm)
d = diameter of hole in enclosure (cm)
Δ = diffusion coefficient (cm2/s)
Solving this differential equation for t, the time to reach concentration Ci,
we obtain
4V 0 l C o – C i0
t = ------------2- ln -------------------
-
Δπd Co – CI
where
Ci0 = the initial concentration inside the enclosure. In the usual
case Ci0 is near zero.
It can be seen that, for a given hole diameter, the most severe case is when
the enclosure volume is small.
Assume V0 = 8 cm3, typical of a small relay; l = 0.075 cm; d = 2 ⋅ 10−3 cm

(Grade D seal); Ci0 = 0; Ct = 4%, LEL of hydrogen; Co = 10%; Δ = 0.65, typi-
cal of hydrogen;
( 4 ) ( 8 ) ( 0.075 ) 10 5
t = ---------------------------------------2- ln --------------- = 1.7 ⋅ 10 s = 48 hr
0.65π ( 2 ⋅ 10 )
–3 10 – 4
If the outside concentration were 5%, the time for diffusion to raise the
inside of the enclosure to the LEL would be 153 hr. These calculations
show that even if a small volume and a highly mobile gas are assumed,
diffusion effects are small enough that most sealing means would give
effective protection as long as the average concentration outside the enclo-
sure is below the LEL. Brown, Dainty, and Silver have verified that the
first-order diffusion equation is valid for diffusion of hydrogen into explo-
sion-proof enclosures. With 44±6% hydrogen-air outside the enclosure,
the concentration reached 4% after the times listed in Table 7-2. Since the
enclosures tested had standard flange gaps, which were not treated to
impede diffusion, these data are especially exemplary of how slow the dif-
fusion process is. Note that the external atmosphere was 11 times the
lower explosive limit.
Enclosure Volume* Flange Width Flange Estimated Gap Time to Calculated

(cm3) (cm) Periphery (cm) (μm) 4% (min) Time to 4%**
516 1.59 43.2 21.3 30 22
606 1.59 43.2 16.5 33 33
1840 1.25 57.5 15.2 81 64
5950 4.1 138.5 24.8 171 174
*Includes gas sampling tubing.
**By this author
Table 7-2 Diffusion into Explosionproof Enclosures
Dainty and Brown subsequently performed further diffusion tests with cyl-
inders with a circumferential gap midway between the ends. The width of
the gap was fixed by shims at spacings of 0.0018–0.020 in. (0.045–0.5 mm).
Flange widths of 0.5 in. and 1.0 in. (12.5 mm and 25 mm) were tested. The
cylinders were 6 in. (150 mm) I.D. and 18 in. (450 mm) internal height. By
installing filler blocks at the top and bottom the volume was changed from
508 in.3 (8300 cm3) initially to 339 or 170 in.3 (5560 or 2800 cm3). In tests of
all combinations of volume, flange width, and gap size, the average inte-
rior concentration of methane agreed with the theoretical value calculated
from first-order diffusion theory.
In other experiments Dainty and Brown showed that the concentration at

the top or bottom of the cylinder was less than that adjacent to the gap, as
one would predict from diffusion theory, until the gas had time to diffuse
to the ends of the cylinder. For a cylinder volume of 508 in.3 (8300 cm3), 0.5
in. (12.5 mm) flange, and 0.020-in. (0.5-mm) gap, the diffusion half-time
was 4 hr. The maximum difference between the concentration at the gap
and at either the top or bottom of the cylinder occurred at approximately
0.5 hr and amounted to about 3% methane. When the gap concentration
was 5.5% the concentration was 2% at the top of the cylinder, 2.7% at the
bottom, and 3.5% at the center of the axis through the cylinder. Dainty and
Brown concluded that these differences were much larger than could be
expected in commercial enclosures, whose gaps would be much tighter. If
the average gap were only 0.001 in. (0.025 mm), for example, the diffusion
half-time would be 20 times longer so that internal diffusion would sub-
stantially reduce the concentration gradient.
It can be concluded from these considerations of breathing and diffusion

that to meet the intent and spirit of paragraph 501-3(b) (1) of the NEC,
“hermetic” seals are not required. All that is required is an effective seal,
which need not be as good as Grade C.
Most people readily accept the fact that the average concentration of flam-
mable material in a Division 2 location, where the NEC permits hermetic
seals to be used, is far below the lower explosive limit. Many now recog-
nize that, in most Division 1 locations also, the long-time average
concentration of flammable material is below the lower explosive limit.
No statistics are available to support this premise. However, the following
arguments support the assumption:
(1) If the atmospheric concentration of the flammable material were
above the LEL for extended periods of time, ignition by electrical
equipment would be an academic matter. Ignition by sparks from
tools, shoe nails, lightning, or electrostatic discharges would almost
certainly destroy the plant even if there were no electrical equipment
present.
(2) Most flammable materials are either toxic or discomforting at concen-

trations of less than 0.1%, much lower than the LEL. Unless the haz-
ard is due to methane, hydrogen, or acetylene, personnel will be
unable to work in an area without breathing apparatus unless the
concentration is far below the LEL. Health and environmental regula-
tions also motivate low emission rates.
(3) Except in closed rooms or closed-process vessels, gas or vapor haz-

ards are self-dispersing. The product loss required to maintain a con-
tinuous supply of flammable material would not go unnoticed.
The above rationale explains why the use of “hermetic” seals was permit-
ted in Division 2 areas for decades before there were standards and
specifications for “hermetically sealed” equipment. Because present area
classification procedures in North America may assign a Division 1 classi-
fication to an area whether it is above the LEL 100% of the time, as in a
process vessel or a very small percentage of the time, as in the usual
instrument installation, it is not possible to recommend the use of “her-
metic” seals unrestrictedly in Division 1 locations. A seal cannot be used in
a location where the average concentration is above the LEL. The concen-
tration inside the sealed device attempts to reach the average level of
concentration outside of the enclosure. If this average level is below the
LEL, an installation with sealed ignition sources is safe whether the loca-
tion is classified Division 1 or Division 2, if the time for diffusion of
flammable material through the seal is long compared to the expected
duration of an excursion above the LEL in the external atmosphere. If the
average concentration is above the LEL, the installation is unsafe. A dis-
tinction between Division 1 areas where the average concentration is
above the LEL and those where the average concentration is below the
LEL is not now recognized in the United States.
Defining Safety Requirements for Sealed Devices

A seal that effectively reduces hazard must, under expected conditions of
use, prevent transient excursions of the ambient atmosphere above the
LEL from raising the concentration of flammable material inside the enclo-
sure to the LEL. It must have structural integrity and corrosion-resistance
suitable to the intended environment. Its surface temperature must be
below the ignition temperature of the material concerned.
The most important requirement is a sufficiently good seal, both initially

and after years of service, to prevent sufficient diffusion of flammable
material across the seal to cause a hazard. How good must the seal be? The
answer to this question depends on what assumptions are made about
how long a flammable atmosphere will be present in a Division 2 location
and on how high the concentration will be. In a survey conducted by the
author in 1968, 16 of the 20 industrial safety people who responded felt
that twice LEL for 2 hours represented a maximally severe definition of
probable hazard in a Division 2 location. One individual suggested that 10
times LEL for 15 min is also possible. Another suggested that the time
should be 8 hr, because an unmanned area might be checked only once a
shift. Many respondents felt that 2 hr at twice LEL would be an extraordi-
nary event; a hazard not removed for 2 hr would surely result in
catastrophe, because were no electrical equipment present, a furnace or
other source would ignite the vapor.
The only fair conclusion from this survey is that there is yet no universal
view of the degree of hazard that is to be expected in a Division 2 location.
However, twice LEL for 16 hr would seem to be an extremely conservative
assumption from which to derive seal requirements, offering a substantial
safety factor for possible degradation during use. As noted in Chapter 2,
many authorities feel that if the atmosphere is within the flammable limits
more than 10 hr per year the location should be classified Division 1.
Diffusion Tests
For the purpose of this discussion, it is assumed that twice LEL for 16 hr is
the criterion for design. Diffusion, therefore, should not cause the concen-
tration to rise higher than the LEL after 16 hr of exposure to twice LEL;
hydrogen being the most hazardous material from a diffusion standpoint.
A direct diffusion test is practical only if the enclosure is large enough to
permit either installation of a concentration sensor or withdrawal of a
sample for measurement outside the enclosure without changing the pres-
sure inside the enclosure.
One form of diffusion test is to fill the enclosure to a convenient concentra-

tion of the test gas, say, 50%, and determine the elapsed time until the
concentration drops to one-half the initial value. This is the diffusion half-
time.
Since diffusion is the controlling transport mechanism, the test gas should
be representative of the hazardous material. CO2 is readily available, and
its diffusion characteristics are typical of common hazardous materials,
except hydrogen, which diffuses more rapidly by a factor of 4.5. Methane
diffuses about 25% faster than CO2. All other hazardous vapors and gases
listed in the referenced handbooks diffuse more slowly, many of them
40% to 60% slower; thus CO2 is a safe test gas for apparatus to be used in
such atmospheres. If hydrogen is the hazardous gas, either it should be
used for testing, or the half-time requirement must be adjusted.
For diffusion-testing, the criterion for approval is easy to define. For what-
ever assumption of maximum concentration-duration has been made, the
enclosure must not let the interior concentration exceed the LEL. If the
assumption is 16 hr at twice LEL, the diffusion half-time must be greater
than 16 hr for a test gas typical of the hazardous material. If the test gas is
not typical of the hazardous material, the required half-time should be 16
ΔH/ΔT, where ΔT is the diffusion coefficient of the test material and ΔH is
the diffusion coefficient of the hazardous material.
The reader is reminded that the 16 hr-twice LEL criterion used in this dis-
cussion is not only arbitrary for the purpose of the presentation, but also
includes, presumably, a substantial margin of safety. This choice of value
was made so that test criteria could be derived simply, without obscuring
the derivation with arbitrary factors on the test parameters. The safety fac-
tor is in the basic assumption.
Even though a diffusion half-time of 16 hr is required, the quality of the

seal is not impractically high. From Appendix B-3,
lV
t DH = 0.693 --------
ΔA
To determine the grade of seal required, assume that the area A is made
up of one hole per inch of seal, each hole being An units in area.
lV
t DH = 0.603 --------------
ΔLA n
where L = length of seal.

Assume
4 2
t DH = 5.78 × 10 s, Δ = 0.65 cm ⁄ s
0.693lV – 5 lV
A n = --------------------------------------------- = 1.85 × 10 ------
0.65 × 5.78 × 10 L
4 L
The diameter of the leak dn is

1⁄2
– 3 lV
d n = 4.85 × 10 §------ · cm
©L ¹
1⁄2
– 3 lV
= 1.9 × 10 §------ · in.
©L ¹
For several typical devices the required grade of seal is shown in Table 7-3
assuming 16-hr diffusion half-time.
Device l (cm) V (cm3) L (cm) lV/L (cm3) dL (in.) Seal Grade

-2
Snap switch 0.075 0.77 5 1.15 × 10 200 × 10-6 C
Small relay 0.05 8.2 7.5 5.5 × 10-2 440 × 10-2 D
-6
Large relay 0.075 984.0 41.0 1.80 2550 × 10 D
Table 7-3 Typical Seal Requirements
Diffusion tests are impractical except in large enclosures and require con-
centration-measuring devices not found in most laboratories. It is,
therefore, desirable to have equivalent means of determining whether an
enclosure is adequately sealed. Three basic methods, from which many
alternative procedures can be derived, are pressure testing, flow testing,
and ignition testing.
Pressure Testing
In a pressure test one raises the internal pressure and measures the pres-
sure-half-time, that is, the time required for the pressure to drop to one-
half the initial value. The test is applicable to most enclosures, although it
may be inconvenient for very small devices such as snap switches.
The equation for the pressure half-time, as a function of the established

diffusion half-time is:
–9
t pH = 4.43 × 10 Δt DH
-----------
-
An
This equation, derived in Appendix B-3, illustrates the greatest difficulty

with pressure testing. The pressure half-time is determined not only by
the diffusion test limits but also by the area An assumed for the holes mak-
ing up the total leak area. But the diffusion characteristics of an enclosure
are determined by total leak area, irrespective of the number or size distri-
bution of leaks. The diffusion half-time is the same whether the leak is
concentrated in a single hole or distributed over many smaller holes.
Because the flow through long, narrow channels is proportional to the

square of the channel area, the flow through a single channel of area A is
larger than the flow through n channels of area A/n.
If the pressure half-time for a single leak channel is tpH, the pressure half-
time for n identical channels is tpH/n. A specification based on too small
an assumption for n (A/n too large) will permit an enclosure with a large
number of small diameter leaks to pass the pressure half-time test,
although it would fail the diffusion half-time test.
Table 7-4 gives the minimum pressure half-time test requirements,

depending on the average leak area An assumed; tpH is based on a 16-hr
diffusion half-time and is tabulated both for a hydrogen diffusion test and
a CO2 diffusion test. Because the enclosure must be tighter to yield a speci-
fied diffusion half-time for hydrogen than for CO2 the pressure half-time
tabulated for hydrogen is longer than that for CO2. Δ is taken as 0.65 cm2/s
for hydrogen and 0.16 cm2/s for CO2, dn is the diameter of a hole of area
An.
tpH(s)
An(cm2) dn(cm)
H2 CO2
−9 −6
4.42 × 10 75 × 10 3770 930
−8
4.91 × 10 250 × 10−6 338 83
4.42 × 10−7 750 × 10−6 37.7 9.3
Table 7-4 Dependence of Pressure Half-time on An
If, for simplicity, the pressure half-time in seconds is required to be equal

to or greater than the diffusion half-time in hours, the average equivalent
hole size would be on the order of 10−3 cm. An enclosure having equiva-
lent total area with larger but fewer holes would not pass the pressure test,
which is, therefore, conservative.
Flow Testing
A pressure test is inconvenient for small enclosures because the volume of

the pressure-measuring device and connections may be large compared to
the volume of the enclosure. The half-time of the enclosure alone can be
computed by applying a factor (V/(V + Vm) to the measured half-time. The
V is enclosure volume and Vm is the volume of the measuring equipment.
However, the corrected value is inaccurate if Vm is much larger than V.
For such enclosures a flow test can be performed. The maximum flow per
second is given by:
0.7 ( P 2 – P 1 )V
Q = ----------------------------------- , ( P 2 – P 1 ) in atmospheres
t pH
This equation is derived in Appendix B-4. (P2 − P1) must be small com-
pared to 1 atm. Because this test is derived from the pressure test, the
maximum flow limit depends on the minimum tpH, and the same diffi-
culty in establishing the limit is present.
Ignition Testing
The most direct test of a sealed enclosure would seem to be an ignition

test. It seems obvious to subject the enclosure to a simulated hazardous
condition and to operate the sparking devices inside. If no explosion
occurs inside the enclosure, it is certainly safe. If an explosion inside does
not cause an explosion outside, it is still safe, because of the quenching
properties of the case, but not because of its sealing properties.
Unfortunately, designing a convenient ignition test is not easy. Under test

conditions the arcing contacts inside the enclosure should operate at rated
dissipation plus an allowance for overload. This poses no problem. The
impediment to convenient testing lies in the length of time the device
must be soaked in the hazardous mixture.
A direct test is to soak the device in the most easily ignited concentration
(MEIC) of a representative hazardous material as given in Table 7-5, then
operate internal contacts. A soak of six minimum diffusion half-times (4.2
time constants) would raise the concentration inside the enclosure to 98%
of the external concentration if the enclosure meets the minimum diffu-
sion-test requirement. This period of soak is inconveniently long.
Group B − 21 ± 1% hydrogen-air (LEL-4%)
Group C − 7.8 ± .3% ethylene-air (LEL-3.1%)
Group D − 3.9 ± .25% pentane-air (LEL-1.5%) or 5.25 ± .25% propane-air (LEL-2.2%)
Table 7-5 MEIC of Class I Test Gases
An alternative test is to let the enclosure soak for a time sufficient to allow
the atmosphere inside a minimally tight enclosure to reach the LEL, then
run ignition tests. For Groups C and D test gases this would be slightly
less than one diffusion half-time, which is still an inconveniently long
time. For hydrogen, the required soak time would be about 5 hr, about
one-third the diffusion half-time.
Raising the pressure outside the enclosure to drive material into the enclo-
sure speeds material transfer. Assuming the test chamber is large relative
to the enclosure volume and using the same reasoning used in the discus-
sion of breathing, the concentration Ci inside the enclosure will, at
equilibrium, be:
Co ( P0 – 1 )
C i = -------------------------- , P 0 in atm
P0
The pressure must be applied for six pressure half-times. At the end of this
time, the minimally sealed case will be near equilibrium. The concentra-
tion Co and the pressure Po must be chosen so that Ci in the equation is the
desired test gas concentration, preferably the LEL. If the pressure is 2 atm,
the concentration outside must be twice the desired inside value. At the
end of the soak period, the external mixture must be changed to the test
mixture at atmospheric pressure and the equipment operated. The soak
time must be controlled. If it is longer than six pressure half-times, the test
will reject good enclosures by allowing the concentration inside to build
up to a higher level than it would reach in a diffusion test. Another diffi-
culty is that the concentration inside an especially leaky device might
reach a value above the UEL, preventing ignition and yielding a false no-
ignition result. This cannot happen if the outside concentration is not
above the UEL.
Another approach to ignition testing is to attach an inlet tube to the enclo-

sure being tested and apply pressure (as in the flow test) for enough time
for an enclosure, which would just pass the flow test, to pass one enclo-
sure volume of the test gas.
One could derive many other test procedures by combining or altering the
methods noted above. All ignition tests performed at the most easily
ignited concentration have a substantial built-in safety factor in that the

enclosure that barely meets the diffusion half-time, pressure half-time, or
flow tests is ignition tested with the most easily ignited mixture inside. In
practice, should the assumed maximum concentration exist outside for the
maximum time (twice LEL for 16 hr, for example), the mixture inside
would be at the LEL. Even for ethylene, propane, and pentane, this repre-
sents a safety factor of about 2.5 on enclosure time constant. It is
considerably higher for hydrogen. Because of the favorable mixture
inside, the minimally adequate enclosure, as judged by the other tests, will
probably have an internal ignition. It may be safe despite internal ignition
because the enclosure contains the explosion and none occurs outside, or
it may be safe because the arc inside is incapable of igniting the test gas. In
the latter case, no seal is required anyway. The device is nonincendive and
meets Division 2 requirements without sealing. An enclosure in which no
internal ignition occurs, because of seal integrity only, has a diffusion or
pressure time constant at least 250% of the minimum required value.
Because of the difficulty in defining a convenient, noncontroversial test

suitable for both small and large enclosures, it is likely that standards for
sealed devices will, in general, rely on pressure, flow, or bubble testing.
Bubble and Leak Tests
Many types of tests have been devised for detecting and/or measuring
leaks in hermetically sealed devices. Many depend on detection of bubbles
escaping from the device when it is submerged in hot oil or water or in a
silicone oil at ambient temperature under vacuum. Others are based on
mass spectrometry or radiation detection using a tracer gas.
Bubble tests have a sensitivity of about 10−5 std cm3/s. Tracer methods are
sensitive to leaks of 10−6 std cm3/s or slower. An enclosure seal passing a
bubble test is at least grade B, much better than is necessary for safety pur-
poses. A seal that passes a bubble test is certainly safe. A seal that does not
pass the bubble test is not necessarily unsafe.
Standards for Sealed Devices

Since the third edition of this book was published several standards that
state criteria for sealed devices have been released. All limit sealed enclo-
sures to an internal volume of 100 cm3. The reason is practical, not
technical. Were the standard to allow large sealed enclosures there must
be greater concern for robustness of the seal, which would have to be dealt
with by appropriate tests. By restricting the volume (albeit arbitrarily) to
100 cm3, most applications of the principle would be permitted, and the
standard would not have to deal in detail with less common cases. All fol-
low the same pattern of prestressing elastomeric or thermoplastic gaskets,

if any, then performing a leak test. These criteria are summarized below:
• Prestressing gaskets and similar items
Elastomeric
Pure oxygen at 68–72°C and pressure of 2100 kPa (300 psi) for 96 hours—
BS6941: 1988, CSA C22.2 No. 213, and IEC Publication 79-15.
Thermoplastic
(1) 7 days in air oven at 78–82°C - BS6941: 1988, CSA C22.2 No. 213, and
IEC Publication 79-15.
– 0.0693 ( T – T 1 )
(2) t = 300e , where:
t = time of exposure in days

T = oven temperature
T1 = max. rated operating temperature of sealed device - FM,
ISA S12.12
Leakage Tests
(1) Immerse sample in 75–85°C water to depth of at least 50 mm for 2

min — no bubbles.
(2) Immerse sample in water of same temperature as sample (15–

35°C) with pressure over water of 5 kPa (0.72-psi) absolute — no
bubbles.
(3) Immerse 23–27°C test sample in 48–50°C water to depth of 25 mm

for 1 min — no bubbles.
(4) Immerse to a depth of 75 mm with pressure over water of 16 kPa

(2.3-psi) absolute — no bubbles.
(5) Gas flow type leak detection at a rate not greater than 10−5 ml/s
at a pressure differential of 101.3 kPa (14.7 psi).
These leakage tests are specified in the following several standards:

• BS6941: 1988 — #1 or 2
• CSA C22.2 No. 213 — #3
• FM (ISA RP12.12) — #3, 4, or 5

• IEC Publication 79-13 — #3, 4, or 5
CSA C22.2 #213 also permits use of one of the bubble test methods in
ANSI/ASTM F98. This standard describes an immersion test, a back-fill
and immersion test, a reduced pressure test, and a pressurization and
immersion test. The tests are said to be suitable for detecting “gross” leaks
only, but the standard suggests that sensitivity can be improved by
increasing the pressure across the leak, using a test liquid with lower sur-
face tension than water or a tracer gas with lower molecular weight or
lower viscosity. “Gross” is not defined, but, as noted above, bubble tests
are usually considered for detection of leaks of 10−4 atm cm3/s or greater.
There is an implicit assumption that the sealed devices are protected from
mechanical damage and the environment by an enclosure, so no shock or
impact tests are prescribed. During the development of ISA Standard
S12.12, a major issue was whether sealed devices for use in Division 2
must be subjected to the chemical compatibility testing many laboratories
impose on plastics enclosures for use in Division 1 locations. In these tests,
samples of the plastic being investigated are suspended in the vapor space
of closed containers, each containing a liquid specified in the standard.
After a specified time of exposure to the saturated vapors (usually 24 hr),
the plastic is examined for crazing, softening, and so on.
Although S12.12-1994 does not contain this testing, it is instructive to

review some of the arguments that led to deleting the clause that had been
in earlier editions. Though the Europeans have agreed thus far that chem-
ical resistance must be a matter for the buyer and seller to negotiate for a
particular application, the issue continues to surface in North America.
One major fact, offered by a plastics expert, is that exposure to chemical

vapors may result in some adsorption of corrosive material, but the con-
tent of the layer of material near the surface will seek the average
concentration surrounding the device. When the concentration is low, any
adsorbed material will be desorbed. The effects, if any, would solely be
surface effects. Long term exposure would not likely be at concentrations
above the toxic limit, which, as previously discussed, is much lower than
the LEL of most flammable gases and vapors.
Other reasons for deleting the test were that different laboratories used
different materials and there was no convincing rationale for selecting the
materials and the test method.
Arguments for Test Deletion
Some of the arguments for deleting the test are summarized below. (See
Table 7-6 for the supporting data.)
Vapor V.P./760
Vapor Conc. Toxic Suitability
Material Pressure Vapor LEL
LEL Limit of PVC*
at 22°C Conc.
Methanol 100 .13 .06 2.2 .002 R
Furfural 1 .001 .021 .05 >>.00015 U
Gasoline 50–200 0.1–0.25 .014 7–18 .001 U
Ammonium 760(NH3) 1 .16 6.3 <<.0005 R
hydroxide
Benzene 63 .08 .013 6.2 .0001 U
Diethyl Ether 488 .64 .019 33.7 >>.1 U
Ethyl Acetate 79 .1 .02 5 .0004 U
Ethylene 74 .1 .062 1.6 .00008 U
Dichloride
Acetone 195 .26 .022 11.8 .0005 U
Methyl Ethyl 89 .12 .017 < >7 .00015 U
Ketone
Acetic Acid 14 .018 .04 .45 .00001 R
2-Nitropropane 15 .02 .026 .8 .00005 ?
R - Resistant
U - Unsatisfactory
*Corrosion Resistance Tables, 2nd Ed., Philip Schroetzer, Marcel Dekker, New York, 1986
Table 7-6 Properties of Chemical Compatibility Test Liquids
First, the test is internally inconsistent. The severity of exposure differs

from chemical to chemical. The vapor concentration ranges from less than
1 to greater than 30 relative to the LEL.
The test is not linked to the nature of the hazard to be encountered. There
is no relationship between the test exposure and the level of exposure that
can be expected in practice. In Division 2, a 0.1 hr exposure, 10 times a
year, would be a severe exposure. Twenty-four-hour test exposure at con-
centrations hundreds of times that which would be permitted for human
exposure for long duration does not seem realistic.
The test to be applied to sealed devices is not performed on power cables

that will be used in the same environment, yet the results of chemical
attack on power wiring are at least as serious. PVC, the most common
insulating material for PLTC and power cables, has a long history of use in
Class I, Division 2 locations. The final column in Table 7-6 shows the rec-
ommendations of one authority regarding the use of PVC in the presence
of these materials. Surely, the fact that most of the materials attack PVC
should cause second thoughts about the nature and severity of the alleged
chemical attack on plastics in Class I, Division 2 locations.
When it was found that UL, FM, and CSA had different lists, the proposed
solution was to incorporate all materials on the three lists into a longer list
of 19 materials. Later some materials were removed from the list: ethanol,
ASTM reference fuel C, hexane, N-hexane, toluene, and vinyl acetate.
Although one can guess that hexane and N-hexane are chemically similar
so both are not needed, the rationale for deleting both was missing, as was
a rationale for deleting toluene and vinyl acetate, but leaving the 12
selected materials in the list.
Including the proposed test as a criterion for judging apparatus gives a

misleading signal to the user. It suggests that chemical compatibility is not
an issue because it has already been subject to assessment. But the fact that
a material has been found to be compatible with the chemicals prescribed
in the standard is no guarantee that it is compatible with all chemicals.
Toluene and xylene, for example, are active solvents. They are not among
the test materials. Chemical compatibility is not an issue in the standards
for other types of protection because it is an engineering issue, not one to
be solved in a standard by simple prescriptions. One would not ban alu-
minum enclosures because they corrode, sometimes to the point of
providing no protection, when sprayed with HCl.
Even if the test protocol could be justified for plastic enclosures, the con-
centrations outside the enclosure are not the concentrations surrounding
components inside the enclosure. The transport of the outside atmosphere
into an enclosure installed in a Division 2 location outdoors will be
extremely slow because this enclosure must be tight in order to be suitable
for field installations. If the enclosure is not tight enough for field installa-
tion, it must be installed in a protected location, which also would impede
free flow of outside atmosphere to the device. As discussed above, an
external atmosphere takes appreciable time to enter even poorly sealed
buildings or enclosures, and these times are long compared to a realistic
estimate of the duration of a vapor cloud outside in Division 2.
This test is not applied to painted metallic enclosures intended for use in
Class I, Division 2 locations. If the hazard is real, then paint peeling off an
enclosure, allowing it to corrode and invalidate its NEMA 4 or CSA Enclo-
sure 4 rating, for example, should be of equal concern.
Swiss Standards
In Switzerland, investigations into requirements for sealed devices began

somewhat earlier than in the United States. This work has resulted in a
standard published by the Swiss Electrotechnical Association (Schweiz-
erischer Electrotechnischer Verein). In 1957 the standard required that
vapor-tight enclosures prevent internal concentration of flammable vapor
from rising above the LEL when high concentrations of explosive gases
are present in the vicinity for 30 min. This standard was based on a great
deal of analytic work similar to that presented earlier in this chapter, sup-
ported by testing of enclosures.
The objective of the Swiss consideration of restricted breathing differed

from that of U. S. studies on sealing requirements. The Swiss focused on
the use of large enclosures to house switchgear, rheostats, and similar nor-
mally sparking devices. The principle assumptions were the following:
(1) There is no appreciable temperature cycling due to operation of the
equipment.
(2) Flammable atmosphere is present for 30 min at 100% concentration,

or the maximum possible concentration at 40°C if the material has a
boiling point above 40°C.
(3) Assumption 2 allows breathing due to external fluctuations of tem-

perature and pressure to be neglected because their cycle is longer
than the period of hazard.
(4) There is diffusion to form a homogeneous internal mixture.
Because of the assumption that the maximum concentration of flammable

material would be present for 30 minutes, the Swiss derivation of require-
ments had to take into account the LEL of the material as well as its
diffusion constant. The Swiss derivation of requirements took the follow-
ing form. (Notation is the same as used previously.)
dC i AΔ
-------- = -------- ( C i – C o )
dt Vl
The critical time, tcrit, for Ci to reach CL, the LEL concentration, is given by
– ln ( 1 – x u ) A C
t crit = ---------------------------- , where φ = ------ and x u = -----L-
φΔ Vl Co
The equation for tcrit shows that, as expected, the time to reach the LEL is
shorter when the value of φ is high (the enclosure is small and leaky), the
LEL is low, and the diffusion constant is high.
The product Δ/[−ln(1 − xu)] is a property of each material. This can be

written as Δ/[ln(1/1 − xu)].
If CL << Co then ln(1 − xu) can be replaced by Co/CL so that the material
property is given by ΔCo/CL.
For materials with a boiling point less than 40°C, Co = 1.0 and the material
property is given by Δ/CL.
Because the Swiss regulations assume testing with CO2, it is convenient to

relate the material constant of the material being considered to the diffu-
sion constant of CO2. A dimensionless constant S is defined. If ΔC is the
diffusion constant of CO2, S is
1Δ ln 2 1 Δ log 2
S = --------- ---------------------------- = --- ----- -------------------------------
2Δ c – ln ( 1 – x u ) 2 Δ c – log ( 1 – x u )
Δc = 0.015 cm2/s, and log 2 = 0.30103 ≈ 2Δc.
Therefore, relating the diffusion properties of the specific material to the

test gas CO2, S is
Δ
S = ------------------------------
-
– log ( 1 – x u )
An approximate value of S is S′ = 2.3ΔCo/CL.
This expression gives conservative values of S′ for Co/CL > 2.5. The error is
reported to be +28% at 2.5. The exact equations should be used for smaller
values of the concentration ratio.
An approximate value of Δ can be calculated from Δ = 1/ m , where m is

the gram-molecular weight of the gas of vapor.
A conservative estimate of Co for materials with boiling points greater

than 40°C can be obtained using the empirical equation:
log C o ≅ 0.015 ( T S – 40°C ), T S > 40°C
Ts is the boiling temperature of the material in °C.
The diffusion characteristic of an enclosure of construction characteristic,

φ = A/Vl is derived by considering the diffusion half-time, tCH, when CO2
is the test gas.
ln 2
t CH = ----------
ΔC φ
If φ has units of cm/liter, and Δc = 0.15 cm2/s,

1.28
φ = ---------- hr
t CH
Now, because of the definition of the characteristic material constant S

t CH
- = 2S
-------
t crit
This relationship derives from the following:

ln 2
t CH = ---------- ;
ΔC φ
– ln ( 1 – x u )
t crit = ---------------------------- ;
φΔ
1Δ ln 2
S = ------------------------------------------
2Δ c [ – ln ( 1 – x u ) ]
Therefore,
Δ 2SΔ
- = -------------C
---------------------------
– ln ( 1 – x u ) ln 2
and
t CH
- = 2S
-------
t crit
Because of the assumption that the flammable mixture is present for 30

min, tcrit = 0.5 hr. Therefore,
t CH = 2St crit = S
This states that the fundamental requirement for a safe enclosure is that
the diffusion half-time of the enclosure when tested with CO2 must be
equal in hours to the value of S (or S′).
The 1965 Swiss standard did not recognize the definition of S used in the
foregoing derivation and specified that an enclosure must have a diffusion
half-time of 20 hr or a pressure half-time of 20 s. An equivalent flow test
was also defined which permitted a flow rate no greater than 0.5 l/hr per
liter enclosure volume. This flow rate was derived assuming a 20-s pres-
sure half-time and a 40-mm water column differential pressure. Most
common materials have values of S less than 20. Some notable exceptions
are acetylene, hydrogen, and isoprene. Therefore, the 1965 Swiss standard
was in accordance with the derivations given previously for materials
with S ≤ 20. It was too conservative for many common materials, which
have values of S less than 10.
It is the author's understanding that some time after 1965 the Swiss
stopped permitting restricted breathing apparatus where hydrogen, acety-
lene, or isoprene are likely to be present.
Included in the Swiss investigations were tests of enclosures ranging in

volume from 4.5 l to 210 l. The test gases were acetylene, propane, and car-
bon monoxide. In only two cases did the concentration rise above the LEL
during the test, in which the enclosures were surrounded with 100% con-
centration of the test gas. Any rise above the LEL was detected by
periodically attempting to ignite the internal mixture. Table 7-7 summa-
rizes important data. Although not all the enclosures were of restricted
breathing quality, the data show that real enclosures tend to be safer than
predicted by the theory.
Enclosure 1 2 3 4 5 6 7
Volume (liter) 200 200 210 150 41 4.5 21.6
tCO(hr) 5 — — 11 19 7 2.5
tPH(s) 1 21 43 22 1.5 4 0.5
Test gas Acet. Acet. Prop. CO Prop. Prop. CO
LEL/UEL 1.5/82 1.5/82 2/9.5 12.5/74 2/9.5 2/9.5 12.5/74
S 30 30 17 3.5 17 17 3.5
Calc. time to LEL 5s — — 1 hr 34 min 34 min 12 s 21 s
No explosion after 13 s 13 s 20 hr 29 hr 15 hr 20 s 20 s
Explosion after 18 s — — — — — 30 s
Table 7-7 Summary of Swiss Experimental Data
The Swiss recognized that diffusion testing is not convenient and estab-
lished a pressure half-time test with an initial 40-mm water gauge
pressure inside the enclosure. The pressure half-time limit was set at the
same value in seconds as the diffusion half-time in hours. To avoid com-
plicated corrections, pressure testing was limited to enclosures with
volume less than 10 l, and rigid enough not to deform under pressure.
IEC Publication 79-15 (a report, not a standard) used the Swiss pattern for
establishing criteria for restricted breathing enclosures. However, as is
often the case when a second committee uses the work of others, addi-
tional factors of “safety” were introduced. With the exception of

luminaires, restricted breathing is not to be employed if the internal air
temperature falls by more than 10 K when the electrical apparatus is deen-
ergized. The minimum diffusion half time is specified to be 80 hr, using
CO2 as the test gas. The pressure half time is 80 s, and the maximum flow
rate is 0.125 (enclosure volume in liters) per hour when 400 Pa is applied
to the enclosure. These values are a factor of 4 more conservative than the
Swiss practice. They must be adjusted if the value of S is greater than 20.
In BS6941 (1988), the approach to restricted breathing was greatly simpli-

fied. The enclosure shall withstand an impact test of 3.5 J (2J for windows).
After prestressing gaskets as for sealed devices the pressure half time shall
be measured with air, and shall be no less than 3 min. The prescribed pres-
sure change is from 300 mm water gauge below atmospheric to 150 mm
water gauge below atmospheric unless testing with positive pressure will
not alter the sealing characteristics of the enclosure. This is an extremely
conservative criterion. A pressure half-time of 3 min corresponds to a dif-
fusion half-time of 180 hr. The Swiss recommended diffusion half time for
hydrogen was 40 hr, equal to the value of S for hydrogen. Most materials
fall below 20. Even though the test requirement is conservative it does not
make the technique unusable. Enclosures containing switchgear, power
relays, and similar items will be much larger than tabulated in Table 7-3
and will typically open on one surface. This opening will have a gasketed
joint, and all the other joints will be welded. For any shape of enclosure, as
all the dimensions increase by the same factor the volume/seal length
ratio increases as the square of the factor. The same leak area per linear
dimension will produce a longer pressure half time in a given test proce-
dure, showing that the diffusion half time is also longer. Many NEMA 4
enclosures have pressure half times much longer than 3 min.
Restricted breathing was not written into ISA S12.12 or CSA C22.2 #213,
most probably because it represented too radical a departure from con-
ventional thinking about ignition-capable power apparatus in Division 2
locations. There was no champion with a strong interest in the benefits the
technique might provide, and strong opposition from those to whom
departure from current explosionproof practice in Division 2 would have
economic significance.
Encapsulation or Potting
Encapsulation or potting is considered to be any embedment of a compo-
nent or assembly in a solid or semisolid medium such as plastic, ceramic,
tar, and grease. There may or may not be an additional enclosure sur-
rounding the solid medium. Tars and waxes for potting or encapsulating
transformers have customarily been used inside a supporting can. How-
ever, many devices are now potted or encapsulated using epoxies, PVC, or
other plastics which do not require additional support.
The application of potting or encapsulation techniques to reduction of

hazard can be justified on the basis of either of two methods of protection
that have already been considered in detail. Because the potting material
effectively seals the ignition source from the surrounding atmosphere, the
potted device can be considered safe because the potting acts as an effec-
tive seal. If the voids within the encapsulated assembly are small, the
assembly may be safe because it is explosion proof.
One important use of encapsulation is in intrinsically safe circuits to

ensure that a particular junction point in a circuit cannot be grounded or
connected to some other point in the circuit. Encapsulation of such a junc-
tion with appropriate concern for mechanical strength and compatibility
with the surrounding environment can reduce to zero the probability that
the encapsulated joint is electrically accessible.
The application of encapsulated assemblies will undoubtedly grow.

Because encapsulation offers protection equivalent to that of a hermetic
seal, such an approach has already been accepted and will continue to be
accepted without serious question in Division 2 areas. However, at the
present time there is no explicit basis in the NEC Code for using such
construction.
Type of Protection m
Type of Protection m is intended for use in Zone 1. Its safety derives from
the enclosing of ignition-capable parts in a compound so that an explosive
atmosphere can not be ignited. The encapsulation technique may be
applied either as embedding, in which a compound is poured over an
electrical part in a mold that is removed after the compound has solidified;
or as potting, in which the mold remains attached to the encased electrical
devices.
Encapsulation shall be without voids, but apparatus such as relays with

internal voids up to 100 cm3 [There's that number again! Repetition is
changing its status from an arbitrary value proposed by an individual
(this author, I think) in the mid-70s, to some kind of a fundamental limit!]
may be encapsulated. The wall thickness shall be 3 mm minimum, except
that it may be 1 mm if the void is less than 1 cm3. Switching contacts shall
be enclosed in another housing before encapsulation; and, if the contact
rating is greater than 6 A, that housing shall be inorganic. Devices con-
nected to the mains power shall be suitable for 4,000 A prospective
current, unless marked otherwise.
The Type of Protection, encapsulation m, shall be maintained after recog-

nized overloads and after a single fault. The faults to be considered are
short-circuiting of any component, open circuit of a component, faults in
printed circuits, and others. Subsequent faults, such as operation of a com-
ponent above its rating, caused by the primary fault are considered to be a
single fault. However, some components and spacings are considered not
subject to fault for the purposes of encapsulation. These are, if used at no
more than two-thirds voltage and power rating:
• film type resistors
• wire resistors with single layer in helical form
• coils with single layer in helical form
• plastic foil capacitors
• ceramic capacitors
• optocouplers whose summed input and output voltages do not
exceed 1000 V and which pass dielectric strength tests specified
in the standard.
Coils and transformers complying with increased safety requirements (EN

50019) also are presumed not to fail, including those with wires less than
0.25 mm and protected against internal overheating. Protective transform-
ers per the intrinsic safety standard EN 50020 are not presumed to fail,
except those of Type 2a (protected by a thermal fuse against damage when
shorted or overloaded, with rated input voltage applied, similar to U. S.
Class 2 transformer construction).
Conductors separated by a minimum spacing of 1 mm for circuits of 380 V

rms potential difference, increasing to 20 mm for a 10,000 volt potential
difference, are presumed not to fail.
Fuses in the encapsulant must be of the sealed type, and batteries shall not
produce gas, release electrolytes, or reach excessive temperature in normal
use.
A compound used for encapsulation is subjected to type tests of dielectric

strength and water absorption (if it is to be used in a moist environment,
but no definition is given).
Encapsulated devices are tested to ensure that both marked surface tem-
perature and the temperature rating of the encapsulant are not exceeded.
Embedded thermal fuses may be used. Internal temperature rise is mea-
sured over three cycles from maximum to minimum rated ambient
conditions. After this test, there must be no visible sign of damage to the
encapsulant.
If the encapsulated component is not to be used inside an enclosure that

meets the impact and similar requirements for all explosion-protected
apparatus, these tests apply. The component is also dielectric-strength
tested in a manner similar to other apparatus with circuits leaving the
enclosure.
ISA SP12 has prepared a draft standard for encapsulation that is mostly
copied from the CENELEC standard, but it is too early to judge its recep-
tion. The intended application of the technique is in Zone 1. The extensive
requirements of EN 50028 are unnecessary for safe application in Division
2 locations, and the simple requirements for sealed components need not
be supplanted by the encapsulation requirements.
Encapsulated relays can be evaluated as sealed relays, as discussed above.

Some relay manufacturers claim that their products are sealed by encap-
sulation to IP67 (see Chapter 15). The type test for IP67 is 30-minute
submersion in water to a depth of 1 meter. No water shall enter.
References
Fowle, F. E., Smithsonian Physical Tables, Smithsonian Institution, Washing-
ton, D.C., 1933.
Perry, J, H., Chemical Engineer's Handbook, 3rd ed., McGraw-Hill, New

York, 1950.
Seals and Seal Testing Procedure, Mil-S-8484 (USAF), June 25, 1954
(obsolete).
Bedwell, D. C., and E. A. Meyer, “Leakage Testing of Sealed Electronic

Enclosures,” Electrical Manufacturing (December 1955), pp. 127–133.
Environmental Testing, Aeronautical and Associated Equipment, Mil-S-

5272C (ASG), 13 April 1959.
Magison, E. C., and L. E. Cuckler, “Reducing the Hazard from Electrical

Components and Assemblies by Hermetic Sealing, Encapsulation or Liq-
uid Immersion,” 1960 Proceedings on the Symposium on Safety for Electrical
Pittsburgh.
Powell, C. A., “Precautionary Measures for Electrical Apparatus for Use in

Hospital Operating Theaters,” IEE Conference Report Series No. 3, Institu-
tion of Electrical Engineers, London, 1962, pp. 66–67.
Test Methods for Electronic and Electrical Parts, Method 112C, Seal, MIL-
STD-202 F.
Magison, E. C., “Reduction of Hazard by Sealing and Encapsulation,”

Electrical Safety Practices, ISA Monograph No. 110, Instrument Society of
America, Pittsburgh, 1965, pp. 109–112.
“Regeln für Schwadensichere elektrische Installationsmaterialien und

Apparate (Regeln für Schwadensicheres Material),” SEV 3050, Schweiz-
erischer Elektrotechnische Verein, Zurich, 1965.
Brown, G. K., Dainty, E. D., and S. Silver, “Laboratory Investigations of

Hydrogen Explosion Phenomena Relating to Electrical Apparatus,”
Research Report R182, Department of Mines and Technical Surveys,
Ottawa, 1966. pp. 47–59.
Dainty, E. D., and G. K. Brown, “Diffusion of External Methane Atmo-

spheres Through Gaps of Various Sizes and Widths Into Enclosures of
Different Volumes and the Effect of Greased Joints,” CEAL Report 134,
Department of Energy, Mines and Resources, Ottawa, 1967.
Dainty, E. D., and G. K. Brown, “Internal Concentration Gradients Due to

Diffusion of an External Methane-Air Mixture Through a Gap Into a
Cylindrical Volume,” CEAL Report 143, Department of Energy, Mines
and Resources, Ottawa, 1967.
Magison, E. C., “Sealing and Encapsulation,” Electrical Safety Practices,

ISA Monograph No. 112, Instrument Society of America, Pittsburgh, 1969,
pp. 67–69.
Fleisig, D. J., “The German Explosionproof Concept for Hazardous Atmo-

spheres,” IEE Conference Publication No. 134, London, 1975, pp. 174–182.
Kirk, R. E., and D. F. Othmer, Encyclopedia of Chemical Technology, Vol. 5,

Interscience Encyclopedia, Inc., New York.
Relays, Electromagnetic (including established reliability types), Mil-R-

6106J (ASG).
IEC Publication 79-15, “Electrical Apparatus for Explosive Gas Atmo-

spheres, Part 15: Electrical Apparatus for Type of Protection ‘n’,” Geneva
1987.
BS 6941: 1988, “Electrical Apparatus for Explosive Atmospheres with Type

of Protection N,” British Standards Institution, London.
EN 50028, “Electrical Apparatus for Potentially Explosive Atmospheres,

Encapsulation ‘m’,” CENELEC, Brussels.
von Angern, K., “Theoretical Principles on the Restricted Breathing Prop-

erties or Materials in Potentially Explosive Environments and their
Utilization,” IEEE Transactions on Industry Applications, Vol. 1A- 4, No. 3,
May/June 1978.
Conn, J. W., “Methane Infusion into Explosionproof Enclosures,” Techni-

cal Note #2, MSHA, Triadelphia, 1987.
Conn, J. W., “Methane Infusion into Explosionproof Enclosures—A Sum-

mary,” Technical Note #6, MSHA, Triadelphia, 1989.
Conn, J. W., “Estimation of Average Internal Air Temperature in an Explo-

sionproof Enclosure,” Technical Note #8, MSHA, Triadelphia, 1989.
Conn, J. W., “Potential Explosion Hazards Related to Explosionproof

Enclosures,” Technical Note #7, MSHA, Triadelphia, 1989.
CHAPTER 8
INCREASED SAFETY,
TYPE OF PROTECTION e
Increased safety, Type of Protection e, is intended for application in Zone 1

hazardous locations. Increased safety apparatus is not designed to contain
an internal explosion, should one occur. Rather, increased safety apparatus
is apparatus that in normal operation is nonsparking and has been
designed conservatively. When used within its ratings, any failure, even
taking into account short circuit and overload conditions, has an accept-
ably low probability of producing an ignition capable spark or hot surface.
One can consider the use of increased safety apparatus as a logical exten-
sion of the principle that transformers, squirrel-cage motors, and similar
nonsparking apparatus of normal industrial quality are suitable for use in
Zone 2/Division 2 locations. It is increased attention to robustness of con-
struction compared to normal industrial equipment that makes Type e
apparatus suitable for use in Zone 1, and therefore, in Zone 2 as well. It is
the author's opinion that the most likely applications of Zone 1 area classi-
fication will be in batch processing plants, which now tend to be
overclassified Division 1. Ex e motors could replace explosionproof motors
in such locations. Motor control and protective gear, with arcing parts in
explosionproof enclosures but with terminals housed in Type e enclosures,
could replace fully explosionproof devices. Other applications for Type e
construction are the exciting windings of electromagnetic flowmeters, sole-
noid valves, and other normally nonsparking devices with windings
operating at power levels too high for intrinsically safe design. It would
seem that Type e construction would naturally be chosen for transformers,
but none of the earlier standards or discussions of increased safety men-
tion them. This is probably because distribution transformers have
undefined loads and can almost always be located outside the hazardous
area. The latter is usually true also of transformers for dedicated supply to
a furnace or other process equipment. Current standards mention trans-
formers other than measuring transformers and state simply that the
temperature rises must be measured with the dedicated load or with the
transformer loaded to its most adverse conditions, including short circuit.
Construction requirements for windings, terminals, temperature rise, etc.
are the same as for other Type e apparatus.
Type e construction was introduced in Germany about 100 years ago. It

was standardized by IEC in Publication 79-7 in 1969, and in CENELEC
EN50019 in 1977. Both documents have been revised significantly since
their original publication. Many of the revisions were additions of detail to
252 Increased Safety, Type of Protection e
ensure similarity of practice and interpretation as the documents became

more widely used for certification outside Germany. Other changes, such
as reductions in permitted creepage and clearance distances, were intro-
duced to take into account improved insulations and better understanding
of the nature of electrical discharges between live parts in air and across
insulators. In many countries, acceptance of increased safety as a suitable
technique for apparatus in Zone 1 was slow at first. Many experts held the
view that it was suitable for Zone 2 only. The latter view did not prevail.
Because North American Division 1 encompasses both Zone 0 and Zone 1,

there was, until the recognition of Zones 0, 1, and 2 in the 1996 NEC, no
conceptually clean way to authorize the use of increased safety techniques
in the NEC. ISA is now preparing S12.16, “Electrical Equipment for Use in
Zone 1 Hazardous (classified) Locations,” Type of Protection - Increased
Safety e. This standard is based on IEC 79-7, adjusted to meet North Amer-
ican needs.
The most common applications of increased safety until the 1970s were to
squirrel cage induction motors driving relatively constant loads, terminal
boxes, lighting fittings, current transformers and lamps, both handlamps
and caplamps. However, the principles may be applied to any similar
nonsparking apparatus. Normal practice in Germany has been to provide
Type e terminal boxes on flameproof apparatus. Electrical connections
between the terminal box and the flameproof apparatus are by flameproof
feed-throughs or conductors in tight-fitting flameproof bushings. Type e
terminal boxes must, in most cases, be at least IP54, which is a much less
onerous standard of construction than flameproof rules demand. The
claimed advantages of the Type e construction are lower cost and easy
accessibility of the terminals and field-wiring conductors for inspection.
Plastic enclosures of Type e design have been approved in Germany for
mounting a variety of safety devices, such as ground-fault detectors and
overcurrent relays, whose current-breaking contacts are flameproof but
whose terminals are Type e. The claimed advantage of these boxes is that
the trip status is easily checked without opening the enclosure, eliminat-
ing the need for electrical status circuits and signal lamps that might be
used in a metallic enclosure.
In recent years Type e construction has also been applied to the nonspark-
ing parts of synchronous machines and to variable speed motors driven
by electronic variable voltage or frequency supplies. The machine, its pro-
tective controls, and its supply must be tested and approved as a system.
Conn has proposed considering the use of enclosures for nonsparking low
voltage apparatus in mines that have many of the features of Type e
enclosures.
Increased Safety, Type of Protection e 253
Construction Requirements
Essentially all the requirements in the General Requirements documents
apply to Type e apparatus. The documents specific to Type e devices
emphasize the characteristics in the paragraphs below, which are intended
to ensure that even under foreseeable abnormal conditions of operation,
Type e apparatus shall not become a source of ignition through arcing or
development of hot Surfaces.
Terminals For External Connections
Terminals that twist, deform or turn during normal tightening or termi-

nals with sharp edges that might damage conductors shall not be used. All
terminals shall be generously dimensioned to permit connection of con-
ductors of cross-section at least suitable for the rating of the apparatus.
These terminals shall not loosen in their mountings in service and shall
not allow conductors to slip from their intended location. Terminals shall
effectively maintain proper contact without deterioration of the conduc-
tors, even if contact is made by direct compression of multi-stranded cores
in terminals intended for direct compression of the conductor. Terminals
intended to clamp stranded wire shall include a resilient intermediate
part. Effectiveness of contact shall not be adversely affected by tempera-
ture changes in normal service. Insulating material shall not be used to
transmit pressure. Terminals shall not be made of aluminum.
Tests on terminals for Type e apparatus include voltage drop, secureness,

temperature rise, dielectric strength, stress corrosion cracking if the termi-
nal is made of copper alloy containing less than 80% copper, thermoplastic
insulator stress relief, and vibration.
Internal (Factory Installed) Connections
Connections inside apparatus shall not be subjected to mechanical stress.

The specifically permitted types of connections are as follows:
• screwed or bolted connections that are locked against loosening
• crimped connection
• soldering, if the conductors are additionally mechanically sup-
ported brazing or welding
• terminals that comply with the requirements for external connec-
tion terminals
If aluminum terminals are used, precautions against electrolytic corrosion

are required. Terminals of copper alloys with less than 80% copper shall
be subjected the stress corrosion-cracking test.
Creepage and Clearances

Increased safety apparatus achieves much of its electrical robustness by
incorporating creepage and clearance distances much larger than those
required for ordinary location equipment. The dimensions in early stan-
dards for increased safety were, like the dimensions for apparatus used in
ordinary locations, based on custom and experience. In the latest stan-
dards the influence of the work done by IEC TC28A, as published in IEC
Publications 664 and 664A, has been taken into account. Type e creepage
and clearances were proposed as the basis for creepage and clearance
tables in the first drafts of the IEC intrinsic safety standard. They were not
accepted because at that time the minimum creepage and clearance recog-
nized was 6 mm at 60 V RMS, a value which was, even then, inappropriate
for low power electronic devices to which intrinsic safety might be
applied. Application of increased safety has been mostly to large machines
and lighting fittings. The technique is not appropriate to small rotating
machines whose no-load and rated load currents differ by only 20% or so.
Use of 6 mm to 12 mm spacings in line voltage motors with ratings of tens
of horsepower are not inappropriate relative to the size of conductors, ter-
minals, and the apparatus, and represents only a doubling or quadrupling
of the kind of spacings found in machines for normal industrial use. By the
time the first edition of CENELEC Standard EN50019 appeared, the creep-
age and clearances had been reduced for voltages below 250 V. Now that
the Type e creepage and clearance distances are based on IEC Publications
664 and 664A, the Type e values are somewhat lower than those specified
for intrinsic safety. At the IEC SC31C and SC31G meetings in Dubrovnik
(1986), both committees considered proposals to alter creepage and clear-
ances based on the work published in IEC Publication 664. SC31C agreed
that further consideration should be given to the proposal, and the latest
Type e rules reflect that consideration. SC31G rejected the proposal of its
Working Group and took no further action. The author's opinion is that
the distances that are suitable for high power Type e apparatus in Zone 1
are also suitable for all intrinsically safe apparatus operating at minute
power levels by comparison and with a restrictive set of construction
rules. The rules for assessing the magnitude of creepage and clearance dis-
tance where there are ribs and depressions in insulating parts are taken
from IEC Publication 664A. These differ from earlier values in that the old
German “3 mm rule” has been replaced by a 2.5-mm rule. The rules in the
intrinsic safety standard follow the older Type e rules.
Solid insulating materials shall be mechanically suitable at temperatures

of 20K above the temperature attained in continuous rated service, but at
least 80°C. Parts made of laminated material or plastic parts whose surface
is damaged or removed in service, shall be coated with a varnish having
the same grade of tracking resistance as the undamaged material. This is
not required if a sufficient length of undamaged material meets the creep-
age requirements or damaging the surface does not impair its insulation
properties. (Not in S12.16) Insulating materials in contact with current-car-
rying parts shall have a flammability classification of at least 94V-2 in
accordance with ANSI/UL 94. The relevant rating is that for 1.6-mm mate-
rial or the minimum thickness in contact with the current-carrying part,
whichever is greater.
Windings
Conductors shall be covered with two layers of insulation, for example,
Class MW35 conductors per ANSI/NEMA MW1000. Enameled, round
winding wires shall conform either to Grade 2 of IEC 317-3, 317-7, or 317-8,
or to Grade 1 of these standards while meeting the Clause 13 dielectric
strength requirements of Grade 2, and shall have no more than six faults in
30 m when tested per Clause 14. The latter is a test for discontinuities in the
enamel coating. Windings shall not be wound with conductors smaller
than 0.25 mm (30 AWG) unless another of the standard types of protection
protects them. Resistance thermometer windings (resistance temperature
detectors, RTDs) embedded in machine windings and impregnated or
sealed with the windings are exempt from this requirement.
Coils shall be impregnated so that the spaces between conductors are

filled as much as possible. If solvent-containing impregnation materials
are used, the process shall be done twice. Motor coils that are impregnated
or otherwise insulated equivalently before they are installed are excepted,
if, after installation they are not accessible for impregnation.
Temperature Rise Limitation
No part of a device, including bare conductors, such as cage rotors of

motors, may exceed the temperature corresponding to the marked T code
during starting, normal operation, or recognized overload. In luminaires,
the surface temperature of the lamp inside the luminaire may exceed the
temperature marked by T code by 50K for those explosive atmospheres
listed on the certificate (not in ISA S12.16). Temperature at the rim and sol-
dering point of the lamp cap shall not exceed the temperature marking, or
195°C, whichever is the lower.
Insulated windings shall not exceed a value which is about 10 K to 15 K

lower than the nominal rating of the insulation system in rated service.
Because a difference in insulation temperature of 8°C 10°C approximately
doubles its service temperature, the reliability of the motor is greatly

enhanced. The standard practice of allowing a lower temperature rise
when measured by thermocouple than by the resistance method is
followed.
The limiting temperature is defined by the temperature code or the short-

term rating of the insulation system, whichever is lower. The short-term
rating of the insulation is about 70°C to 80°C higher than the permissible
temperature in rated service, depending on the class of insulation. A time
tE is defined as the time it takes the machine or winding to climb from the
normal operating temperature to the limiting temperature when it is ener-
gized with the starting current, IA. IA is measured under locked rotor or,
for ac solenoids, under locked armature conditions. The time tE is not per-
mitted to be less than 5 s if a current-dependent protective device is used.
The ratio of IA to IN, the normal rated current, is not permitted to exceed
10. Marking of the device includes the ratio IA/IN and the time tE in order
to allow selection of an appropriate protective overload device. The stan-
dard places a limit on the relationship of tE and the ratio IA/IN, but the
limit can be violated if a special overload device has been tested and if it is
named on the marking. A device, internal or external, is always required
to prevent windings from exceeding the limiting temperature, unless that
temperature will not be reached under continuous overload, such as
locked rotor conditions for a motor; or, if the device (such as a ballast for
fluorescent lamps) cannot be overloaded. The limit control may act indi-
rectly, as on the current drawn by the machine or directly by sensing the
temperature of windings or parts of the machine. S12.16 states that the
temperature control device (except sensor) must be external to the motor.
Enclosures must conform to the requirements of the general requirements

document, and, in addition, must be at least of the degree of Protection
IP54 if bare conductors are contained or IP44 if only insulated conductors
are enclosed. (S12.16 says NEMA Types 3, 3S, 4, 4X, 6, or 6P). If drainage
or ventilating fittings are used that reduce the enclosure protection below
the IP values stated above, the actual IP values shall be marked on the
enclosure. Special fasteners are required on enclosures for installation in
mines to prevent unauthorized entry. In Type e enclosures for industrial
applications, no special fasteners are mandated.
Rotating Electrical Machines
Exceptions to the general requirement of IP54 enclosures are made for

pipe-ventilated machines and those limited to application in clean, super-
vised locations. S12.16 does not include these exceptions.
Internal fans shall comply with the clearance requirements and material
restrictions for external fans, as stated in the general requirements stan-
dard. In summary, this means that the clearance between a fan and the
housing shall be not less than 1 mm, but at least equal to 1/100 of the fan
diameter. This clearance need not exceed 5 mm, however. Fans, fan hoods
and ventilating screens shall have a surface resistivity not exceeding 1
gigaohm. This restriction does not apply to fans in industrial apparatus
that have a peripheral speed less than 50 m/s. Metallic parts of fans and
accessories shall not contain more than 6% magnesium if intended for
industrial service. For mining service there shall be no more than 6% total
of magnesium and titanium, and no more than 15% total of aluminum,
magnesium, and titanium.
Radial air gap between stator and rotor, measured at standstill, shall be at
least the value determined from the formula:
D – 50 0.75nl ½
Min. radial gap (mm) = ® 0.15 + ---------------- § 0.25 + ----------------· ¾rb
780 © 1000 ¹ ¿
¯
where
D = rotor diameter, mm. The minimum and maximum values
to be used in the formula are 75 mm and 750 mm.
n = maximum rated speed in rev/min, but a minimum value of
1,000 shall be used in the formula.
r = determined from the following formula but a minimum
value of 1.0 is to be used in the formula.
core length (mm)
r = -----------------------------------------
1.75D (mm)
b = 1.0 for machines with rolling bearings and 1.5 for machines
with plain bearings.
Cage rotors, whether used for starting synchronous machines or in induc-

tion machines, must have bars brazed or welded to the shorting rings,
unless the rings and bars form a single manufactured unit. Bars must fit
tightly in the slots so there can be no sparking between the bar and the
core during starting. This can be achieved by casting the bars, keying
them, welding them, or providing a supplemental metallic liner. It is being
proposed that nonsparking be assessed by testing a motor filled with
hydrogen-air to see whether ignition occurs. It is also being proposed that
machines with ratings greater than 200 kW/pole shall not have skewed
slot construction in the stator or the rotor. Unless the temperature marking
demands a lower limit the temperature of rotor bars shall not exceed
300°C during starting. This may require that parts in the path of leakage
flux be nonmagnetic or insulated to prevent them from exceeding this
limit during stalled conditions.
Figure 8-1 illustrates the use of temperature rise data to determine the
marked values to tE and the T-code. In a typical machine, the rotor bars
and short-circuiting rings will be uninsulated. The stator winding, of
course, consists of insulated coils. In Figure 8-1 it is assumed that the sta-
tor coils have Class F insulation which can be heated to a maximum
temperature rise above a 40°C ambient temperature of 170°C under short
circuit conditions during the time, tE. It is also assumed that the stator
temperature rise in normal running at full load is 90°C, and the rotor bar
rise in normal full load running conditions is 100°C.
These temperature rises make the motor unsuitable for T4, T5, and T6. The
stator winding reaches its maximum permitted temperature rise of
170°C − 90°C = 80°C after tE3 = 12.5 s of locked rotor operation after the
temperature has first stabilized at the continuous rated load value of 90°C.
For T2 marking, the permitted rise above 40°C ambient is 260°C. If the nor-
mal full-load running temperature of the rotor bars is 100°C, the
additional permitted rise under locked rotor conditions is 160°C, which is
reached in tE2 = 10 s. T3 marking allows locked rotor conditions to cause
an additional rise of 60°C above full-load running conditions. This rise
would occur after tE1 = 2 s. It is not permitted to mark this machine T3,
because the time tE must be at least 5 s. This machine may, therefore, be
marked T2 with a time tE of 10 s. Olenik et al. show a number of examples
of temperature rise determinations, and note that motors may be either
stator-limited or rotor-limited. (This terminology is common in the Ger-
man literature.)
Figure 8-1 Temperature Rises During a Short Circuit Test - Class F Insulation
and Uninsulated Rotor Bars
Luminaires Designed for Mains Supply
The principle limitation on the selection of light sources is that no part of

the source shall reach a temperature exceeding the limiting temperature
should the bulb break. Specifically mentioned as permissible sources are
fluorescent lamps of the cold-starting types, with single pin caps per IEC
61-1; filament lamps in accord with IEC 64, 64A, and 432; mixed light
(MBT) lamps; and others which will not exceed limiting temperatures for
more than 10 s after the bulb is broken. Lamps containing free metallic
sodium are not permitted. S12.16 does not refer to MBT lamps and refers
only to bi-pin fluorescent lamps per ANSI C78.1 and tungsten filament
lamps per ANSI C78.20 and C78.22.
The distance between a fluorescent tube and its outer cover shall be at least
5 mm, unless the outer cover is a tube, in which case the separation may be
2 mm. For other lamps the distance from the lamp to the protective cover
depends on the wattage of the lamp, ranging from 3 mm for P < 60 W to
30 mm for P > 500 W.
Lampholders may be of metallic or nonmetallic construction but, with the

lamp cap, must meet the dimensional and flame transmission tests for
flameproof enclosures for either Group I or IIC or be provided with a
quick-acting switch with an enclosure that will not transmit an internal
ignition per IEC 79-1. S12.16 recognizes only construction with a switch.
Lamps with screw caps shall be designed to prevent self-loosening. After
tightening to a prescribed torque (1.0–3.0 Nm, depending on size) the
torque required to loosen the lamp shall be greater than 1/3 the tightening
torque after the lamp has been rotated 15° towards withdrawal.
The surface temperature of the marked temperature code may be

exceeded if the highest temperature of the lamp is at least 50 K below the
temperature shown by test to be required to cause ignition of the gases
shown on the certificate. (This relaxation is not recognized in S12.16.)
S12.16 proposes sulfur dioxide testing of the connection of bi-pin lamps to
lampholders and vibration testing of luminaires with bi-pin lamps.
The temperature at the rim of the lamp cap and at the soldering point of
the lamp cap shall not exceed 195°C or the limiting temperature, which-
ever is the lower. The temperature of ballasts shall not exceed the limiting
temperature even in the cased of aging lamps. Inserting a diode in series
with the lamp and energizing at 110% rated voltage assesses this. Several
additional requirements for bi-pin lamps are being proposed by CEN-
ELEC and appear in the S12.16 draft. Included is a requirement for an
isolation switch that must isolate the luminaire when the cover is
removed. There is no requirement that the switch have any explosion-pro-
tected properties, but it must be fail-safe to the open position and must not
be capable of being manually defeated.
Hand lamps and cap lamps with their own source of supply, except cap
lamps for mining applications, shall be protected from damage by a pro-
tective cover that is 1 mm from the source of light when it is fully inserted.
If the lamp is inserted in a spring-loaded lamp holder and is in contact
with the cover, the travel shall be at least 3 mm. A guard must protect the
cover unless the area of the cover is less than 50 mm2 and a protruding rim
with minimum height of 10-mm supplies protection. The guard is not
required for covers of larger area if they can withstand the impact test
specified for the protective guard in the general requirements standard.
Measuring Instruments and Measuring Transformers
The limiting temperature of insulation or that determined by T code shall

not be exceeded when the device is energized at 1.2 times rated current or
voltage. Current transformers and current-carrying parts of measuring
instruments shall not reach the lower of the limiting temperature or 200°C
when subjected to Ith = 1.1 times short circuit current for 1 s. No mechani-
cal damage shall be sustained that would decrease safety when a current,
Idyn = 1.25 × 2.5 × short circuit current, is applied for 0.01 s. The 1.1 and
1.25 factors are safety factors so the RMS short circuit current in service
may not exceed Ith/1.1, or the peak value may not exceed Idyn/1.25,
respectively. Moving coil instruments are not permitted. Olenik, et al.
state that moving coil instruments cannot satisfy the overload testing
requirements, which were originally specified as 50 to 100 times rated
current.
Batteries
Batteries shall conform to the requirements of the standard, but additional

requirements apply to those batteries with capacity greater than 25 Ah at
the 5-h discharge rate. These concern primarily robust and reliable con-
struction of the containers and covers, maintenance of creepage distances,
connections between cells, and so on.
General Purpose Connection and Junction Boxes
Tests shall be run with a maximum number of terminals, with the worst-
case temperature rise (according to manufacturer's literature) connected in
series to determine that no terminal exceeds the limiting temperature at its
rated current. A maximum rated power rating for the box is calculated
using the resistance of the test circuit and the terminal rating.
Resistance Heaters
Resistance elements must have a positive temperature coefficient. The

coldstart current shall not exceed 110% of the manufacturer's specification
after 10 s of operation. An electrical protective device, in addition to over-
current protection, shall be supplied to minimize heating or arcing
hazards due to leakage and currents flowing to ground unless the manner
in which the heater is incorporated in other apparatus gives equivalent
protection. If a conductive covering is associated with this protection, it
shall extend over 70% of the area of the insulating sheath and shall have a
resistance lower than the heater, unless overtemperature under fault con-
ditions is prevented by other means or if the heating resistor has self-
limiting properties. Heating resistors may not be in contact with the
potentially explosive atmosphere. Thermal insulation shall not normally
be assumed to exclude contact between the heater surface and the explo-
sive atmosphere.
Conductors for connection to heaters shall be 1-mm2 minimum for

mechanical reasons.
Resistive heating devices shall not exceed the limiting temperature when
energized. Means for achieving this are as follows:
(1) A stabilized design, where the temperature does not exceed the rated
value under stated conditions of use, without the use of auxiliary pro-
tective devices.
(2) A self-limiting property of the resistance heating device, whereby the

power output of the device decreases as the temperature around it
increases, so that at some ambient temperature the heating element
temperature and the ambient temperature are the same.
(3) A protective system, separate from any control system, which discon-
nects the heater from the supply if the temperature exceeds a preset
value.
The manufacturer's instructions shall contain information on the interrela-

tionship of the temperature of the resistance heating device, its heat
output, the temperature of the surrounding gas or liquid or solid, and the
heat transfer characteristics at the boundary of the heating device and its
surroundings.
A protective system shall be achieved by sensing the temperature of the

heating device, or its immediate surroundings; by sensing the surround-
ing temperature and one or more other parameters, such as level, current,
flow, leakage current, or power consumption; or by sensing two or more
parameters other than temperature. The settings and adjustments of pro-
tective devices shall be locked and sealed and shall require manual
replacement or manual resetting of the protective device after it operates,
unless the operation of the system is continually monitored. Sensor failure
shall be fail-safe, ensuring that the device is disconnected before the limit-
ing temperature is reached. The protective system shall be separate from
any device used for control in normal operation. In ISA 12.16 this section
has a parenthetical restriction that it its not applicable to heat-tracing sys-
tems, which are covered by an additional clause. This clause follows the
pattern just described but refers to ANSI standards and test methods, and
specifically mandates grounded systems and the use of ground-fault cur-
rent interruption, for example.
References
Dreier, H., “Uber Erwarmungsmessungen an explosionsgeschützten Dre-
hstrom-Kurzsschussläufer-Moteren der Schutzart ‘erhöhte Sicherheit’ (Ex
e),” PTB Amtsblatt, February 1958.
Dreier, H., “Types of Protection for Explosion-Protected Electrical Appa-

ratus and the Safe Use in Explosion-Endangered Areas,” International
Aspects, Conference on Explosion-Protection, Fagernes, 1973.
Hulsberg, Fr., “The Type of Protection Called Increased Safety,” Paper

#45, The Eighth International Conference of Directors of Safety in Mines
Research, Dortmund-Derne, 1954.
“Electrical Machines in Hazardous Areas,” Publication 601, Laurence,

Scott & Electromotors, Ltd., 1975.
Engel, U., and H. Wickboldt, “Special Types of Explosion Protected Elec-

trical Machines of Type of Protection ‘Increased safety’,” Conference
Publication 134, IEE, P118-123, London, 1975.
Sinclair, R. S., “Aspects of Temperature Rise Measurement and Tempera-

ture Protection of Rotating Electrical Machines, with Particular Reference
to Increased Safety Motors and Operation at Variable Speed,” Conference
Publication 296, IEE, London, 1988.
Kienzle, K., “Sicherungsverteiler in der Technik ‘Erhöhte Sicherheit’ mit

Druckfeste gekapselten Einbauelementen,” Ex Zeitschrift, Stahl, Kunzel-
sau, October 1978, pp. 32–36.
Olenik, H., Rentzch, H., and W. Wettstein, Explosion Protection Manual, 2nd
edition, Brown, Boveri & Cie, Mannheim, 1983.
BS5501, “Electrical Apparatus for Potentially Explosive Atmospheres, Part

6. Increased Safety `e',” EN50019, British Standards Institution, London.
Conn, J. W., “A Safety Requirement Classification System for Housing

Electrical Equipment in Hazardous Locations,” Technical Note #10,
MSHA, Triadelphia, 1992.
CHAPTER 9
IGNITION OF GASES AND VAPORS BY
ELECTRICAL MEANS
Chapter 2 considered the development of an incipient flame ball into a

self-propagating combustion wave, assuming that an ideal source injects
sufficient energy at the center of the flame sphere.
The development of an ideal spherical combustion wave is a useful con-

cept for study of the effects of changing parameters. Seldom, however,
except in artificial laboratory situations, does ignition actually occur in this
ideal way. At voltage levels of greatest practical interest (cathode-ray
tubes excepted), that is, below 8,000 V, the electrode spacing is less than
the quenching distance. The electrodes, therefore, conduct heat from the
incipient flame sphere. The amount of energy provided initially must be
increased to compensate for the additional losses through the electrodes.
The greater the portion of the flame sphere of critical diameter Dq, which
is intercepted by the electrodes, the higher will be the required ignition
energy. For this reason, decreasing the spacing between a pair of elec-
trodes will increase the required ignition energy. Likewise, for otherwise
identical conditions, electrodes of greater area within the critical flame
sphere will require more energy. Lastly, the rate at which ignition energy
increases as electrode spacing decreases is a strong function of electrode
area within the critical diameter Dq.
The effects of electrode geometry are shown in Figure 9-1. Although the
ignition energies plotted are taken from the work of several investigators
who used different vapors and gases of similar minimum ignition energy,
the trend of the data is as expected. The smallest electrodes show the low-
est rate of increase of energy as voltage (and, therefore, spacing) is
reduced.
At the present time there is no accurate analytical way to relate the geome-
try of an electrode system to the amount of energy required to ignite a
specific flammable mixture. It is necessary to use empirically determined
ignition energies. Because there is no exact quantitative way to extrapolate
data from one set of experimental conditions to another, the user of empir-
ical data must carefully consider whether reported ignition data are
applicable to a different situation and whether and how the data must be
adjusted.
266 Ignition of Gases and Vapors by Electrical Means
Figure 9-1 Effect of Electrode Geometry on Ignition Energy
It is convenient to categorize the modes of ignition as follows:

(1) Arcing at closing contacts or between fixed electrodes in a capacitive
circuit. Energy stored in a capacitor is released in the arc as contacts
close or as the gap between fixed electrodes breaks down.
(2) Opening contacts in inductive circuits. Energy stored in the inductor is

released in an arc as opening contacts interrupt current in the circuit.
Ignition of Gases and Vapors by Electrical Means 267
(3) Opening or closing contacts in a resistive circuit. Although this mode

of ignition might be considered to be the limiting case of either mode
(1) or (2) as capacitance and inductance approach zero, the most effi-
cient arcs in resistive circuits are those caused by opening contacts.
The arc is maintained longer between opening contacts, and quench-
ing decreases as the contacts separate.
(4) Ignition by hot wire or surfaces. This mode of ignition depends on the
heating of fine wires or surfaces to temperatures sufficiently above
the ignition temperature of the gas to cause ignition.
This chapter emphasizes the mechanisms of low voltage electrical igni-

tion. The goal is to provide a basis for developing a qualitative
understanding of the processes involved in simple circuits. The ways in
which data for simple circuits are used to assess the safety of complex cir-
cuits will be discussed in Chapters 10 and 11.
Characteristics of Electric Arcs

The first three methods of electrical ignition listed above require transfer
of energy from an electric arc to the gaseous mixture. This is also true
when very small wires fuse and strike an arc. The material presented
below is intended to provide a basis for a qualitative understanding of the
processes of ignition by electric arcs so that ignition data in the literature
can be better interpreted and the effects of changing test conditions can be
appreciated.
The minimum breakdown voltage of a gap between electrodes in air

depends on gas density, electrode spacing, and electrode geometry. At
normal temperature and pressure the minimum breakdown potential
between electrodes in air is approximately 300 V at a spacing of 3.10−4 in.
(7.5 μm). These numbers are not absolute; they depend on the shape and
surface condition of the electrodes. Breakdown between spheres is at
higher voltage than between points. For the type of arc that occurs as a
result of breakdown in air between widely spaced electrodes, the break-
down process is as follows: if a potential difference is applied between
fixed electrodes separated more than 3.10−4 in. (7.5 μm), electrons in the
gap are propelled to the anode, the positive electrode; if the potential
between the two electrodes is gradually increased, the current also
increases and approaches a saturation current level shown as point A in
Figure 9-2. At some critical value of applied potential the current will
increase rapidly. This is the characteristic breakdown of the gap, section B
of the curve in Figure 9-2, drawn for near minimum breakdown potential.
Figure 9-2 Gap Voltage-Current Curve Before and After Breakdown
During the initial application to the gap of voltages below the breakdown
voltage, electrons drawn to the anode are those that were initially present
in the gap. Some electrons are always present, usually from collision of
gamma rays with gas molecules or by photon action on the electrode
material. Many electrons cannot complete the journey from cathode to
anode because they collide with gas molecules along the way and drift
back to the cathode or are forced back by space charge. As the potential
between electrodes is increased, more electrons have sufficient velocity to
survive successive collisions with molecules of gas and reach the anode.
At the saturation current level, the number of electrons reaching the anode
is primarily dependent on the number of electrons initially in the gap. As
the potential between the electrodes is raised further, some electrons gain
sufficient energy so that they displace an additional electron from a gas
molecule by collision, increasing the total number of electrons available to
cross the gap. At the critical potential, called the breakdown voltage, the
field strength is high enough that collision-produced electrons outnumber
those lost by collision and an “avalanche” occurs; that is, on the average,
each electron produces additional electrons by collision. The number of
electrons in the gap increases exponentially, and breakdown is said to
have occurred. After breakdown, if the current is limited to values on the
order of 0.1 A, the voltage across the electrodes generally remains at about
the level of minimum breakdown voltage in air, about 300 V, and the dis-
charge is called a glow discharge. The exact voltage is a function of the
materials of the anode and cathode, the spacing of the electrodes, and the
current. Increasing the current causes a transition from the glow discharge
to a low-voltage arc discharge as shown in region C of Figure 9-2. The
voltage between electrode drops to a lower value. The characteristics of
the low-voltage arc discharge are further described below.
Because the breakdown of the gap between electrodes depends on enough

electrons acquiring sufficient energy to displace additional electrons from
molecules of gas, breakdown does not occur immediately on application
of the minimum breakdown voltage. Although free electrons are always
present, they may be so few in number that significant time can elapse
before a naturally produced electron is so favorably situated in the gap
that application of the minimum breakdown voltage will result in ava-
lanche breakdown. Therefore, it is often necessary to apply a voltage
greater than the normal minimum breakdown voltage to induce break-
down. In ignition research, the possibility of abnormally high breakdown
voltage is often eliminated by irradiating the gap with an ultraviolet
source, a radioactive source, or a small auxiliary arc that produces ultravi-
olet radiation to ensure an ample initial supply of electrons.
Because much electrical equipment contains no voltages, even as high as

the 300-V minimum sparking potential in air, the characteristics of the low-
voltage arc are of greater pertinence to the study of electrical ignition than
those of the glow discharge. Because 300 V is the minimum sparking
potential in air at normal temperature and pressure, one might conclude
that discharge arcs are not possible below this potential. However, anyone
who has accidentally touched a screwdriver to a 12-V automobile battery
knows that sizable (not necessarily ignition-capable) arcs can be generated
at these low voltages. It is certainly not necessary to obtain a glow dis-
charge first.
The initiating mechanism of low-voltage arcs between electrodes that are

slowly brought together may be looked at in two ways. One point of view
holds that electrostatic fields of the order of 106 V/mm are sufficient to
extract electrons even from cold metals and that microscopic surface irreg-
ularities reduce the required applied potential gradient to a value
equivalent to 105 V/mm in terms of macroscopic electrode spacing. A sec-
ond point of view is that when two contacts are brought together current
first flows when ohmic contact is made by a whisker of metal or carbon.
Current vaporizes the whisker and provides conducting vapor to start an
arc. It is not of particular concern in ignition work which theory more
closely describes the actual physical processes. Breakdown data obtained
by investigators of both persuasions gives essentially the same informa-
tion, that is, the breakdown voltage gradient below the minimum
sparking potential in air is approximately 105 V/mm. Figure 9-3 illustrates

the relationship between contact separation and breakdown voltage.
Figure 9-3 Breakdown Voltage Between Electrodes in Air
A short arc may be formed by opening a current-carrying circuit, by clos-

ing a circuit with voltage source lower than the sparking potential and
causing an arc, or by breakdown between two electrodes at voltage above
the minimum sparking potential in air and then drawing enough current
to cause transition from a glow discharge to an arc discharge. The steady-
state characteristics of the final arc are the same in all three cases.
The reader should be aware that, although they are a useful tool for con-
templating the ignition process, the steady state arc characteristics only
approximate the arcs formed during the development of an ignition event.
The transient characteristics of these arcs are considerably more complex.
The characteristics of a stable arc can most effectively be described by a

family of voltage-current curves with electrode spacing as a parameter.
Sets of curves for tungsten and carbon electrodes are shown in Figure 9-4a
and Figure 9-4b. Curves for other materials are similar, but the minimum
current Im and the minimum voltage Vm, to which the limiting (0 mm)
curves are asymptotic, depend on the material. Table 9-1 gives values for
some metals for arcs in air, in nitrogen, and in hydrogen; the values are
taken from Holm. The Vm and Im values reported by other investigators
may differ by 20 to 30%. Where three values are listed, the second was
reported by Widginton, and the third was reported by Zborovszky and
Cortugno.
Figure 9-4a Arc Characteristics-Tungsten (from Holm, Figure 53.05)
Figure 9-4b Arc Characteristics-Carbon (from Holm, Figure 53.04)

Metals Vm, V Im, A

Carbon 20 0.03
Aluminum 14 –
Iron 13–15 0.35–1.55
Nickel 14 0.5
Copper in nitrogen 14 0.6
Copper in air 13,11,13 0.43
Copper in hydrogen 18 1.3
Silver in air 12 0.4
Silver in nitrogen 13 0.8
Silver in hydrogen 21 –
Cadmium in nitrogen 11, 8.5, 8.5 0.1
Tin 13.5, 10.5, 11.2 –
Gold in air 15 0.38
Gold in hydrogen 20 –
Tungsten 15 1
Platinum 17.5 0.9
Stainless steel 15 0.5
Zinc 10.5, 9.5, ∼9 0.1
Table 9-1 Minimum Voltage and Current Values for Stable Arcs for Various
Metals
Widginton reported that oxide films reduce both Vm and Im. He also
reported that the Vm for copper electrodes was not strongly dependent on
separation speed, but the arc voltage at wider spacings increased with
electrode separation speed. At 2 A arc current, for example, the voltage at
1-mm arc length was 27 V, when electrode speed was 1 cm/s. When speed
was 2,000 cm/s, the voltage drop increased to 47 V. Under the same test
conditions, Vm was measured to be 15 and 17 V, respectively.
It is of particular interest that carbon arcs can be stable at lower current

levels than any other material listed. It is likely that in many practical situ-
ations arcing of metallic contacts is at voltage and current levels near those
typical of carbon electrodes, either because of carbonaceous dirt or
because of contamination by surrounding hydrocarbon atmospheres. The
latter effect is notable in affecting the breakdown voltage, as shown in Fig-
ure 9-3. The normal palladium electrode (one that is clean) requires higher
voltage at given separation than a carbon electrode. A palladium electrode
“activated” by arcing in certain hydrocarbon vapors requires still lower
voltage to cause breakdown. These effects provide an explanation,
although not the only one, for ignition in resistive circuits at circuit voltage
and initial current that are incompatible with the arc characteristics of the
electrode material.
Ignition of methane-air has been reported when breaking a resistive cir-

cuit, with an open-circuit voltage of 125 V and a 0.9 short circuit current
with platinum electrodes. The values for platinum in Table 9-1 show 17.5 V
and 0.9 A as asymptotes. Reference to Figure 9-4a shows that a load line for
this circuit would not intercept the steady-state arc characteristic of plati-
num. (Figure 9-4a shows the characteristics of tungsten, which are similar
to those of platinum.) A load line for the circuit does intercept the arc char-
acteristic of carbon, as shown in Figure 9-4b. Therefore, the concept of
sensitization of metallic electrodes by carbonaceous deposits after the con-
tacts have been in use is supported by what is known of the characteristics
of short arcs.
In the early 80s a number of laboratories verified that the IEC standard
apparatus is more sensitive after the cadmium disc has been used. This
sensitivity change was not noted when the apparatus was used with
hydrogen-air mixtures.
Zborovszky and Cotugno report minimum arc currents for cadmium and
zinc of approximately 0.03 A. They, like many other investigators before
them, reported significant changes in ignition characteristics of arcs, usu-
ally in favor of ignition, after electrodes have been used for a time and
buildup of deposits was observed around the electrodes. An extensive lit-
erature exists on such deposits relative to reliability of contacts carrying
very low currents, and it is generally felt that they result from contamina-
tion by hydrocarbons.
Another factor, relevant to the sensitivity of the standard IEC apparatus,

suggested by Widginton, was that heat transfer from the arc to the gas is
aided by the vapors of cadmium and, perhaps, by exothermic oxidation of
the vapor.
Zborovszky and Cotugno also investigated arc duration as a function of

supply voltage in resistive circuits with cadmium as the cathode material.
They found, as one would predict from load-line analysis, that when the
supply voltage is only slightly greater than the minimum arc voltage, Vm,
the arc duration is very short. They analyzed transfer of energy from the
circuit to the arc based on maximum power transfer and concluded that
ignition is most likely to occur in resistive circuits when the circuit is bro-
ken and the arc voltage passes through the maximum power transfer
point, that is, when the arc voltage is half the supply voltage (in circuits
where the supply voltage is at least 2 Vm). They conclude that, except for
quenching differences caused by differences in thermal conductivity of the
cathode material, in circuits with supply voltages greater than approxi-
mately 40 V the effects of electrode material should be small. However, it
is not readily demonstrated that this is true, because when the power sup-
ply voltage is higher than 40 V, the ignition currents are in the same range
as the values of minimum arc current, Im.
The minimum arc current, which depends on the cathode material, deter-
mines the duration of the arc, so Widginton's data do not show the
expected independence of material characteristics.
Capp and Widginton report that earlier work of Capp showed that in
short arcs, where the plasma voltage drop is small so that the anode and
cathode falls account for most of the arc voltage, most of the arc energy in
long-duration arcs is conducted to the electrodes, not to the gas. In tran-
sient arcs, where energy transfer to the gas is significant, they state that the
energy comes from the plasma column voltage drop, Vp. For arcs in nitro-
gen with a cadmium cathode they give the equations:
V p = 0.5 ( V arc – 9 ), V arc < 17
V p = V arc – 13, V arc > 17
These equations imply that the cathode fall, Vm, remains at 9 V, while the
anode fall increases linearly from 0 at Vm to a maximum of 4 V.
They support Zborovszky's view by stating that in studies of arc ignition

no ignition has been seen to occur when the arc voltage is near Vm. Igni-
tions have always been related to arc discharges where a considerable
portion has been at a level substantially above Vm.
Arcing at Closing Contacts and in Capacitive Circuits

Capacitive discharge arcs have been studied by investigators with two
quite different interests. The method of igniting gases with discharge arcs
from small capacitors charged to high voltage was brought to a high state
of sophistication at the U. S. Bureau of Mines in Pittsburgh, Pennsylvania,
where it was used to measure the minimum ignition energy of gases and
vapors.
In these investigations the potentials were sufficiently high that electrode

spacing was greater than the quenching distance of the gas or vapor being
tested. Therefore, thermal losses from the spark, except to the surrounding
gas, were very small. Because electrode spacing was greater than the
quenching distance, when the expanding gas sphere reached the elec-
trodes it was already larger than the critical diameter Dq and was self-
sustaining.
The electrodes did not remove energy during the initial growth of the
flame sphere. Because there are only small thermal losses from the spark
(except to the surrounding gas), it is assumed in such high-voltage, mini-

mum-ignition energy determinations that all of the energy stored in the
capacitor is discharged in the arc and absorbed by the gas. The total
energy delivered to the gas is the energy stored on the capacitor CV2/2,
where C is the value of the capacitor in farads and V is the voltage to
which the capacitor is charged.
The high-voltage discharge test equipment used by Blanc, Guest, von

Elbe, and Lewis was of two types (see Figure 9-5). For capacitors larger
than 100 pF, a dc voltage supply was used. The test capacitor, connected to
electrodes mounted within a test bomb, was slowly charged by a rotating
charging rod that transferred charge from the dc power supply to the test
capacitor. A test capacitor size was selected, and the spacing between elec-
trodes was adjusted for a given test series. The voltage on the test
capacitor was then slowly increased by rotating the charging rod until
breakdown occurred. To remove uncertainties in data due to the statistical
time lag of breakdown, the gap was irradiated by a radium source.
When testing capacitors smaller than 100 pF, the rotary charger was
replaced by a Bakelite rod connected continuously to the test capacitor.
The grounded electrode was then moved to reduce the gap distance and
cause breakdown. In both cases the apparatus was carefully constructed to
minimize stray capacitance in leads, to eliminate any contribution of
energy by continuous steady-state current from the source, and to mini-
mize any bias on data caused by inductance and resistance in the electrode
circuits. The electrodes used in obtaining data were pointed or flanged.
Typical plots of ignition energy versus electrode separation are shown in
Figure 9-6.
Figure 9-5 Bureau of Mines High-Voltage Test Apparatus; (a) Apparatus for
Capacitance > 100 pF; (b) Apparatus for Capacitance < 100 pF
Note that when flanged electrodes are used there is an appreciable range
of spacings just greater than the quenching distance over which the igni-
tion energy is relatively constant. However, the amount of energy
required to cause ignition at the quenching distance increases at an almost
infinite rate. If the flanged electrodes are replaced by pointed electrodes,
the amount of energy required for ignition begins to increase near the
quenching distance, but at a rate that is dependent on electrode geometry,
as shown in Figure 9-6. It was on the basis of such experimental data that
the concept of a minimum flame sphere diameter was developed. If
flanged electrodes are used at spacings below the quenching distance, and
the arc is confined to the center of the flange, even though a large amount
of energy is dissipated between the flanges, the flame sphere cannot grow
to the critical diameter where the rate of heat generated by combustion
equals or exceeds the rate of heat lost through the surroundings. The
flanged electrodes conduct a large amount of heat away from the flame
zone. Pointed electrodes conduct less heat away from the combustion
zone, so that less energy is required to offset electrode conduction at close
spacing. The Bureau of Mines investigators found it necessary to use glass
flanges on the electrodes to confine the arc always to the center of the
flange. They found similar quenching effects with both glass and metal
flanges. The ratio of the thermal conductivity of a solid to that of a gas is so
large that differences between solids are immaterial. Glass flanges were
used because metal flanges, being electrically conductive, occasionally
supported an arc at the edge of the flange, causing ignition even when the
flanges were within the quenching distance.
Figure 9-6 Ignition Energy versus Distance (adapted from Lewis and von Elbe)
Figure 9-6 shows also that at wide electrode separation the ignition energy
increases. This can be explained qualitatively on the basis that as arc
length increases the arc voltage and arc energy increase. Photographs of
ignition by long arcs have shown that instead of a single incipient flame
sphere developing from the arc, several incipient flame balls may develop,
requiring the expenditure of more than the minimum ignition energy.
The results of the Bureau of Mines investigations have contributed substan-

tially to understanding of the ignition mechanism. The method is extremely
important for obtaining a near-absolute measure of minimum ignition
energy. The energies measured in such high-voltage ignition tests are not,
however, directly pertinent to problems of electrical safety, especially those
that are common in the instrument industry. The voltages used in these
tests are impractically high. In addition, all test conditions are carefully con-
trolled to eliminate energy loss. Such conditions are not attained in practical
instrument systems. High-voltage ignition data are, of course, pertinent to
ignition by static discharges, and to possible ignition by cathode-ray tubes.
Most instrument systems operate below 1,000 V. At even 1,000 V break-

down occurs at approximately 0.125 mm electrode separation in air. The
quenching distance for most materials is the range 0.5 to 3 mm. Quenching
by the electrodes is, therefore, significant in practical situations. Quench-
ing is even more important in low voltage systems where arcs can be
initiated only at spacings on the order of microns. Because electrodes are
inside the flame sphere, it is very difficult to obtain reproducible ignition
energy data for low-voltage conditions. Evaluation of data from different
investigators is also difficult. The specific geometry and materials of the
electrodes determine test results. Because the electrode spacing at low
voltages is of the order of a micron, changes in surface geometry invisible
to the naked eye may have large effects on the amount of energy required
to cause ignition.
Typical Test Equipment for Capacitive Circuits

In the United Kingdom the ignition capability of capacitive discharge
sparks was studied using the “intermittent break apparatus.” In this appa-
ratus the circuit is broken between two pieces of platinum alloy strip 0.009
in. (0.23 mm) in thickness. The edge of one of these strips, which is 0.2 in.
(5 mm) wide, is cut to form a sawtooth with 11 teeth. This is drawn across
the straight edge of the other strip to produce a succession of 10 sparks at
each operation. These electrodes are operated in the “break-flash” appara-
tus (to be described later under “Inductive Break-Spark Testing”), such
that the 10 sparks take place in about 0.21 s. The interval between succes-
sive sparks is, therefore, about 0.02 s. Figure 9-7 shows the main features
of this method. Curve F in Figure 9-10a represents data from similar appa-
ratus in France.
Figure 9-7 “Intermittent Break” Apparatus Electrodes
Allsop and Guénault reported results obtained by manually closing elec-

trodes in Safety in Mines Research Paper No. 107. Data for a platinum
point and rod are shown as Curve E1 in Figure 9-10a.
Curve E in Figure 9-10a represents ignition caused by discharge of capaci-

tors through 0.015-in. lengths of 0.0005-in. nichrome wire. Fusing of the
wire causes an arc. This apparatus is not used for routine testing, but it
was constructed at the Safety in Mines Research Establishment because it
has the advantage of using a fresh electrode pair for each test.
In Germany, at the Physikalisch-Technische Bundesanstalt (PTB), a

counter-rotating plate and a slotted disc, as shown in Figure 9-9, were
developed for testing the ignition capability of all kinds of sparks. Sus-
pended from the rotating plate are four tungsten wires of 0.008 in. (0.2
mm) in diameter. The wires are 0.43 in. (11 mm) long. A cadmium disc is
mounted so that its top surface is 0.4 in. below the rotating tungsten wire
holder. The cadmium disc has two chordal grooves 0.080 in. (2 mm) wide
by 0.080 in. (2 mm) deep machined in it. The tungsten wire contact holder
revolves at 80 rpm; the cadmium disc rotates in the opposite direction at
approximately 20 rpm. As can be seen from the figure, the counter-rotat-
ing discs cause the tungsten wire to approach the edge of the disc, scrape
across the surface of the cadmium disc, and bend the wire; thereby provid-
ing a relatively fast break as the wire springs on the edge of the slot. A
slow make will occur as the wire comes against the side of the slot; a slow
break, as the wire and slot move relative to one another; and a relatively
fast break, as the wire and disc break contact for the last time in the cycle.
Some investigators have used steel wires instead of the standard tungsten
wire for long duration test series. The sensitivity is essentially the same as
with tungsten wire. Steel breaks less frequently than tungsten, which
work-hardens and becomes brittle more quickly than steel wire.
Because this apparatus is easier than most to construct and maintain, and
because it tests a circuit with both fast and slow, make and break, sparks,
it has been adopted by the IEC as the international standard test appara-
tus. IEC Publication 79-3 describes its construction. All standards of
intrinsic safety now specify this apparatus with tungsten wires as the ref-
erence for ignition capability of a circuit.
In his own investigations the author used 0.6-mm-diameter tungsten wire

sharpened at the end. One wire was held stationary while the second wire
was rotated at right angles to the first, so that contact was made at a speed
of approximately 2.5 mm/s (see Figure 9-8).
Figure 9-8 Magison’s Test Apparatus

Figure 9-9 PTB Test Apparatus
Precautions in Testing
Although the testing apparatus used in different laboratories differs in
detail, all investigators observe the same precautions.
Ignition capability of a circuit must be tested in the most easily ignited

concentration of the gas or vapor. Once it has been established that the test
apparatus yields the most easily ignited mixture, this fact need not be fre-
quently reverified. However, when setting up new test apparatus, it is

essential to verify by test that small changes in composition increase the
required ignition energy. This is the only way to ensure that the means of
introducing the mixture do not alter the gas composition.
The author, for example, for reasons of convenience, used methanol vapor
for testing the ignition capability of various sparks. Although the volume
of the test chamber was known and an appropriate amount of methanol
liquid to give the desired composition within the chamber after vaporiza-
tion had been computed, the amount of liquid injected into the chamber
was established empirically under the conditions of use. After initial
adjustment of the feed apparatus, there was little difficulty in maintaining
the proper mixture, but this was occasionally checked by making slight
perturbations in the amount of feed to determine that the ignition energy
did in fact increase from a minimum value.
The composition of gas-air mixtures is usually controlled by ratioing the

flow of the gas and air into a mixing chamber. Alternatively a storage res-
ervoir may be filled with the desired gas-air mixture by filling to the
appropriate partial pressures of gas and air. In most laboratories storage
of premixed gas is being abandoned in favor of flow ratioing because of
the greater risk of accident with a stored volume of easily ignitable
mixture.
It is essential that all combustion products from a previous explosion or

series of ignition attempts be completely removed by flushing the cham-
ber with air or a clean gas mixture. If air is used, it, too, must be flushed
from the chamber by an adequate flow of gas mixture.
Design and maintenance of a satisfactory contact mechanism is undoubt-

edly the most difficult problem facing anyone making ignition energy
determinations. All experimenters have reported difficulty in obtaining
reproducible results from electrode systems operated as voltage levels
such that the electrodes are within the quenching distance. Several years
ago the author used a rotating slotted steel plate and a copper wire riding
on the plate to obtain reproducible ignition data at levels comparable to
those determined with the break-spark apparatus used in the United
Kingdom. However, the wire needed frequent replacement. Most experi-
menters have found that high-melting-point wires are required to
maintain stable geometry. Copper electrodes need frequent replacement;
the heat of arc formation melts the copper electrode and changes its shape.
Platinum and tungsten electrodes are not as subject to this difficulty.
There is no general rule except that, unless apparatus is arranged so that a
new contact surface is presented for each ignition trial, the calibration of
the equipment must be checked frequently by determining the amount of
energy that must be dissipated to ignite a known gas with known circuit
conditions. Only in this way can one be assured that changing contact
geometry or contamination of the contact surface has not altered the igni-
tion capability of the test apparatus.
Even though gas composition is carefully controlled, and the equipment

successfully provides grossly stable electrode geometry, it is essential that
changes in ignition energy be related to changes in test conditions only
after numerous trials at each condition. British practice is to make 100 tri-
als at the desired test condition. Others have specified 1,000 trials.
In the years since the introduction of the standard IEC apparatus

described in IEC Publication 79-3 two common methods of establishing
ignition parameters have developed.
(1) One method, similar to the early British practice, is to set a voltage or
current for N revolutions of the wire support, and if an ignition
occurs, drop to a lower level, until no ignition occurs. The MIV or
MIC is reported as the lowest value giving an ignition.
(2) The other method is to set several values of voltage or current to

obtain, for example, 15 ignitions for each value of the variable at vary-
ing numbers of revolutions of the wheel. The data pairs are then plot-
ted on log probability paper. The plot is close to a straight line. A
value for the variable to have a desired ignition probability can then
be read from the plot. Some German investigators have chosen a
probability of ignition of 10−4, but in recent years a value of 10−3 has
been used at PTB. One must be cautious to compare ignition data only
at the same probability of ignition.
Widginton, in his 1988 paper, compared the two methods by using a com-
puter simulation. As a basis for the simulation he used the probability data
for ignition by a 50-nF capacitor obtained using the standard IEC test appa-
ratus. He found that the stepwise determinations yielded a value of
igniting voltage at a probability of about 0.002 per revolution of the wheel
in the standard apparatus and required about a third of the testing time of
the probability method. He found the values measured in successive test
runs to be somewhat less variable than those determined by the probability
method.
It is important to recognize that the probabilities reported in Widginton's

paper are the probabilities for ignition in one revolution of the wheel,
which holds four wires. Presumably, when investigators are quoting prob-
ability of ignition relative to their experiments, they have taken any
multiple sparks per revolution into account.
These are not the only methods that have been used to define ignition val-
ues in recent decades. One method of determining points for the plot on
probability paper is to determine the average number of sparks before the
first ignition for each value of the variable. At least 16 runs must be made
for each value of the variable.
When measuring the amount of energy required to ignite flammable mix-

tures by closing contacts to discharge capacitors, it is necessary that the
contribution of energy from the power supply be nil. The need for concern
in this regard would seem to obvious. However, if data are being sought
over several orders of magnitude of capacitance and, therefore, over
orders of magnitude of voltage, a resistor connected between the dc power
supply and the test capacitor that satisfactorily isolates the low-voltage
power supply when a large capacitor is short-circuited may be inappropri-
ately small when small capacitors are charged to much higher voltages.
The power supply may then cause current to flow in the arc, lowering the
amount of stored energy required for ignition.
If the test apparatus operates on a fixed time cycle, too large an isolating
resistor may increase the time required to charge the capacitor to power-
supply voltage so that the capacitor is discharged at less than the desired
voltage. Continuous monitoring of capacitor voltage could signal this
problem, but it is best practice not to parallel the test capacitor with other
circuit elements that might alter energy storage or release characteristics.
If one is interested in relating energy stored in a capacitor to ignition, all
the energy stored in the capacitor must be discharged into the arc. Any
series inductance and resistance between the storage capacitor and the
point of arcing will absorb energy or lower the rate of energy discharge in
the arc. The measured stored energy required for ignition will be too high.
Typical Ignition Test Results

Figure 9-10a shows the results of typical tests of capacitive discharge igni-
tion at low voltages. The capacitance-voltage curves are not straight lines
as they would be on a log-log plot if the ignition energy, CV2/2, is a con-
stant. At lower test voltages C (and, therefore, ignition energy) is larger
than would be predicted assuming constant ignition energy.
Because the electrodes are always within the quenching distance, energy
must be increased as the voltage is decreased to compensate for increasing
quenching by the electrodes. Unfortunately, there is no analytical expres-
sion known to the author that adequately predicts the shape of the
capacitance versus voltage curve. That for a given electrode geometry the
required ignition energy increases as voltage is decreased is, however, in
accord with our notions of quenching.
Some of the ignition characteristics plotted appear to be asymptotic to a

voltage of 10 V to 20 V in the range of reported minimum short-arc volt-
age. Circuit voltages below the minimum arc voltage could not establish
an arc to cause ignition. Only a small amount of lead inductance is

required, however, to raise the voltage above the minimum arc voltage,
should there be a discontinuity in arc current. Ignition at circuit voltages
below the minimum arc voltage, therefore, cannot be assumed to be
impossible, even with closing contacts.
In circuits with resistance, capacitance, and inductance, ignition may be

initiated by discharging a capacitor on a closing contact. If the circuit is
then broken by the contacting mechanism, the energy stored in the induc-
tor by the current from the capacitive discharge is released in a break
spark. This type of ignition is discussed further in Chapter 11.
Figure 9-10a Low-Voltage Capacitor Discharge Ignition

Effect of Resistance in Discharge Path

Table 9-2 gives the data obtained by Allsop and Guénault. These data
show that moderate resistance in the capacitor's discharge path greatly
increases the amount of energy that must be stored to cause ignition. The
data in this table for paper and electrolytic capacitors of similar size illus-
trate that the internal resistance of the electrolytic capacitor increases the
ignition voltage above the nominal value for a paper capacitor.
Electrolytic CIRCUIT RESISTANCE (Ω)

Capacitance-μF 0 1 2 4 10 15
*605 16 27 30 33 40 44
605 20 31 33 35 44 53
298 21 33 34 36 46 55
170 23 34 35 37 48 56
116 25 34 37 38 48 57
48 28 35 39 39 48 56
Paper
*100 16 28 31 36 44 50
97 21 32 36 38 46 56
50 23 32 37 39 48 58
19 29 36 40 41 51 60
10 42 51 54 55 63 74
*2 60 72 76 84 94 104
*1 88 106 126 146
*Pointed platinum electrodes. Others tested with point and a rod.
Table 9-2 Minimum Voltage (V) to Cause Ignition—8.3% CH4
The effects of series resistance and of series inductance for hydrogen-air

ignition (as reported by Müller) are shown respectively, in Tables 9-3 and
9-4. The method of ignition consisted of a pair of platinum wires, one sta-
tionary and one rotating, so fixed that closure occurred at 100 cm/s.
Capacitance RESISTANCE (Ω)

(μF) 0 100
1.0 25 220
8.0 17 160
66.0 12 170
Table 9-3 Ignition Voltage (V), H2-Air
Capacitance INDUCTANCE (μΗ)

(μF) 2 85
1.0 150 250
8.0 45 75
Table 9-4 Ignition Voltage (V) H2-Air
Widginton's discussion of intrinsically safe circuit design includes curves

of capacitance versus voltage for several values of series resistance. His
ignition data for hydrogen and methane, taken using the IEC apparatus,
are reproduced in Figure 9-10b. These curves have been included in most
standards for intrinsic safety for the evaluation of intrinsically safe appara-
tus. Figure 9-10b is only an approximate rendering of the official curves. It
should not be used for analyzing designs for submittal for approval. The
2nd edition of EN50020 added zero-resistance curves for propane (Group
IIA) and ethylene (Group IIB) and redefined the hydrogen curve. The
curves for ignition of Group II materials by discharge through series resis-
tance were removed. These curves are included for methane (Group I).
The Widginton curves, like the data by Allsop and Guénault, show that at
voltages well above the minimum arc voltage, Vm ignition is substantially
independent of capacitance value when the series R is high enough to
make the time constant of the RC discharge as large as a few hundred
microseconds. The capacitor then acts as a current source during the critical
initial ignition period. Hopefully, one could relate this constant current to
the values determined for resistive circuits, but the correlation is not exact.
The values of current one obtains by dividing the ignition voltage by the
series resistor have no obvious simple relationship to the currents
required for ignition in resistive circuits. The latter relate to opening con-
tacts and the capacitive data relate to closing contacts where the
quenching is much more severe.
Series resistance is probably effective in raising the required ignition volt-

age on a capacitor because it prevents the capacitor from releasing all its
energy into an arc before the circuit is short-circuited by contact of the
electrodes. For capacitors charged to tens or a few hundred volts, an arc
forms at an electrode spacing on the order of 10−2 to 10−1 mm. Typical clo-
sure speeds are 100-1000 mm/s, so the time from arc initiation to contact
closure is on the order of tens of microseconds or less. At some value of
resistance, the time constant of the circuit probably allows ignition when
contact bounce interrupts the current from the R-C circuit initiated by the
closing contacts. The lower level of quenching by the opening contacts
would probably change the ignition scenario. Therefore, one would not
expect a simple relationship between current discharged by the capacitor
and the current required in a resistive circuit.
Figure 9-10b Relation Between Minimum Igniting Voltage and Capacitance (Widginton)
Johannsmeyer reported that for hydrogen-air mixtures the relationship

between the ratio of the ignition voltage required with a series resistor to
that without a series resistor is linear. His data show the ratio to be 1 +
0.0125R, for the range 0-100 ohms, and capacitors from 30 to 1000 nF. His
results are shown in Figure 9-10c. An analysis of the Widginton data, and
the data in Tables 9-2 and 9-3 yield average values of 0.05 to 0.15 for the
coefficient of R in the above equation. Widginton's curves yield a value of
0.15 for the range up to 5 ohms. Johannsmeyer stated that for values of
resistance above 100 ohms the capacitor should be treated like a resistive
circuit.
Figure 9-10c Dependence on Ignition Voltage on Capacitance for Different Series

Resistors: Hydrogen-Air (from Johannsmeyer)
In his monograph on intrinsically safe systems Minard presented a plot of

ignition voltage for hydrogen-air as a function of resistance for values of
1 μF, 10 μF, and 100 μF. For resistance from 1 to 100 ohms the curves of
voltage vs. resistance are linear on a plot of log V vs. log R. The ignition
values for 1 μF are lower than those from the Widginton curves, which
were published in earlier editions of IS standards. Those for 10 μF and 100
μF are in closer agreement to the earlier values, but tend to be somewhat
lower. Neither the basis for the curves nor the raw data points is stated. In
the range from 1 to 100 ohms the curves shown are parallel and can be
represented by an equation of the form V = V1(R)a. The values of V1 and a
derived by this author are as follows:
Capacitance − μF V1 a
1 20.5 0.15
10 16 0.14
100 12 0.16
There are critical values of resistance in Minard's plots below which series
resistance has very small effect on ignition voltage. These values are 1
ohm, 0.4 ohm, and 0.16 ohms, respectively, for 1, 10 and 100 μF. The corre-
sponding time constants are 1, 4, and 16 microseconds, which are
consistent with the comment above about closure time of the electrodes.
One approval lab has published the data in Table 9-5, which states the
value of unprotected capacitance with ignition capability equal to the pro-
tected capacitor. These approximate equivalents are based on the UL913
(Widginton) curves for hydrogen. Though the equivalence is stated in
terms of capacitance, it is really equivalence of ignition capability that is
being stated. These same equivalencies do not apply to other gases. All
capacitance values are in μF.
C+0Ω C + 5.6 Ω C + 15 Ω C + 40 Ω
100 3.2 1.2 0.5
22 2.4 0.85 0.4
10 1.2 0.6 0.36
3 0.46 0.27 0.2
Table 9-5 Capacitance Values of Ignition Capability Equal to Specified R-C
Circuits
Another laboratory uses a proposal to IEC submitted by PTB for analytical

purposes. The ignition voltages for protective resistor values of 5.6, 15,
and 40 Ω are those reported in the Ex Zeitschrift, October 1989. Those for
hydrogen-air are reproduced below. This author has rounded off the val-
ues of capacitance in Table 9-6. The referenced article also included values
for ethylene-air mixtures. The curve for R = 0 is not that of EN50020, 2nd
ed., but a slightly higher curve, that is, 13 V instead of 11.5 V at 10 μF.
Below 0.5 μF the curves agrees with EN50020, 2nd ed.
C – μF R = 5.7 Ω R = 15.4 Ω R = 40.2 Ω

0.03 79 104 146
0.05 63 77 104
0.1 47 54 77
0.3 32 37 47
0.6 27 29 39
1.0 23 25 33
10.0 15 17 23
Table 9-6 Effect of Series Resistance on Capacitive Ignition of Hydrogen
(PTB)
This author has tabulated ignition voltages from all three sources for com-
binations such as those listed by the first laboratory and can find no
simple correlation. In general, the data presented by Schebsdat show igni-
tion at much lower voltages than the Widginton data and somewhat lower
than the Minard curves. There seems to be no pattern of voltage ratios
detectable to this author that unify the several sets of curves. If designers
use the values in the table above to estimate equivalent capacitance, they
should plan to test early in the design cycle to verify safety if 1.5 times the
fault voltage applied is more than half of the voltage read from the Widg-
inton curves. For capacitance of 10 μF and less, the ignition voltage values
from the Schebsdat table may be used to guide design.
In many instances a designer cannot afford to insert a resistor in the dis-

charge path of a capacitor. A small inductor can be substituted.
Schebsdat's equation and the limits of application are given in Ex Maga-
zine, June 1976. Johannsmeyer presented some data for capacitance values
of 300 nF and less discharging through inductance in his dissertation.
Effect of Source Current, and Resistance and Inductance

in Discharge Path
Standard reference curves for capacitive ignition assume that a large resis-
tor isolates the source of charging voltage from the capacitor. In most
work the resistor is 100 KΩ. In practice this cannot be the case, and in bar-
rier-protected systems the source resistance is usually a few hundred
ohms or less. The effect of the current from the source adding, perhaps, to
the igniting capability of the capacitor, as far as this author is aware, has
not been much studied. Because the capacitive-discharge ignition process
is with closing contacts, and the resistive-ignition process is with opening
contacts, it has been assumed in the application of intrinsic safety barriers
that any additive effect is small enough to be covered easily by the safety
factors on the connected capacitance value.
Zborovszky has addressed this issue. She has shown that if one adds
capacitance to the resistive circuit the igniting current increases, peaks,
and decreases to the initial, C = 0, value, at a critical size of capacitance. At
this value of capacitance the arc discharge changes from a breaking arc to
a making arc. In her data for a 30-V circuit with a methane-air mixture, the
critical capacitance is 3.7 μF, and the value for ignition, if the source
impedance is 3,000 Ω, is 4.0 μF. A barrier with a 30-V open circuit voltage
would have a connected capacitance limit of about 3.0 μF, so the circuit is
safer with regard to resistive ignition when operated within this limit than
when operated with no connected capacitance.
In PTB Report W-11, Schebsdat reported his investigations of the effect of

varying source resistance, using 100 KΩ, 1,000 Ω, 500 Ω, 200 Ω and 100 Ω
source resistance; capacitor discharge through resistors of 10 or 25 ohms;
and inductance from 0.5 to 5 mH. He investigated capacitance values of
0.05 to 1 μF. The test mixture was hydrogen-air. The effect he investigated
was the transition from a capacitive ignition on the closing of the contacts
to an inductive discharge as the contacts opened the current established
by the initial capacitive discharge. It is statistically possible, but very
unlikely, that this can happen in a purely resistive discharge circuit where
the only inductance is that of the test apparatus itself. A succinct summary
of his results may be misleading, so the reader interested in using his data
to guide design should consult the original report. Having warned the
reader, however, this author makes the following statements.
For values of source resistance of 1 KΩ and lower all discharges were

inductive, regardless of the combination of R, L, and C tested within the
stated range of values.
For 100 KΩ source resistance the transition occurred for inductance of

about 3 mH in discharge circuits with 10 or 25 ohm resistance. Ignition
voltage is always higher than the value for R = L = 0 values. With few
exceptions, the ratio exceeds 1.5.
In circuits of 1KΩ source resistance, or less, the ratio of ignition voltage to

that of the R = L = 0 value is typically 0.7 to 1.0, depending on component
values. Such mixed circuits require special attention to determine
whether, with rated values of inductance and capacitance of a barrier con-
nected, the assumed safety factor still exists. (As discussed in Chapter 10,
the erosion of the safety factor is, in most cases, primarily a certification
issue, not a safety issue in practice. The real safety factor lies in the fact
that the IEC standard test apparatus is more sensitive than any real
switching mechanism likely to be encountered in practice, and it is used in
the certification laboratory with all other factors adjusted to maximize the
probability of an explosion.)
The Riddlestone, Riddleston-Nethercot, and Bartels papers describe

attempts to elucidate the effects of series resistance in capacitive discharge
circuits. The early Riddlestone work was done by discharging capacitors
charged to 5.9 kV through a 1.25-mm gap between hemispherical elec-
trodes of 10 mm diameter. Series resistance between the capacitor and the
gap was varied. Because of the spacing and size of the electrodes, there
was substantial quenching. The stored energy required for ignition was
found to increase as the series resistance was increased to a value of 1000
ohms. Further increase in resistance caused a decrease in stored energy.
The energy discharged in the gap remained essentially constant at approx-
imately 2 mJ for resistance values from 0 to 200 ohms, then increased
rapidly to about 30 mJ at 400 ohms. Further increase in resistance caused
the discharge energy in the gap to drop to about 15 mJ. As explained in the
Riddlestone-Nethercot paper, the sudden transition in energy required
was seen to be correlated with a change in discharge characteristic from a
2-μs, 45-A peak current discharge to a 77-μs, 17-A peak current discharge.
At currents higher than about 40 A steady-state arcs in air are known to
develop smaller-diameter, higher-temperature arc channels. Further
experiments with artificial lines were conducted to determine whether
peak current was the decisive factor. These investigations suggested that
rate of rise of current in the arc may be decisive. Ignition was obtained
with a 23-A discharge of 3.6-μs duration, 3.8 mJ energy, with a rise time of
0.42 μs. A 26-A discharge for 20 μs, 1.7-s rise time, required 34 mJ of
energy. It was concluded that high-temperature constricted arcs may
occur at currents lower than the steady-state value during the formative
stages of the arc.
Bartel's studies further indicate the influence of rate of rise of current. He

noted that high rates of arc current rise usually result in higher arc volt-
ages and, therefore, higher rates of energy release. Comparison of arc
characteristics showed the low-ignition-energy arcs to be of smaller diam-
eter and of higher temperature than the high-ignition-energy arcs. Bartels
suggests that the transition from low-energy to high-energy ignition is
related to the establishment of thermal equilibrium in the arc. The low-
ignition energies are for discharges of a time shorter than the time
required to establish thermal equilibrium in the arc, and the higher-igni-
tion energies are related to discharges longer than the time to establish
thermal equilibrium.
Effects of Electrode Material and Geometry

In connection with the experiments previously described, Allsop and
Guénault determined the effects of changing electrode geometry and mate-
rial. The data, as given in Table 9-7, do not lead to quantitative conclusions
about the pertinent characteristics of electrode material, although they do
substantiate previous statements about quenching. In all cases the voltage
required for ignition was less for two points than for a point and a rod.
Electrolytic capacitor
Electrode material and shape 110 μF 605 μF
R=0 R = 10 R=0 R = 10
Platinum point to platinum rod 25 48 20 44
Platinum points 21 43 16 40
Steel point to steel rod 25 64 21 56
Steel points 23 53 20 46
Copper point to copper rod 26 61 21 56
Copper points 21 46 18 44
Blunt nichrome to carbon 46 70 — —
Nichrome to carbon (+) 29 66 23 58
Table 9-7 Changes in Ignition Voltage (v) with Changes in Shape and Material
of Electrodes
Opening Contacts in Inductive and Resistive Circuits

When a current-carrying circuit containing resistive or inductive elements
is broken, an arc may form at the contact point. If sufficient energy is dissi-
pated in the arc, ignition will occur. The nature of the arc (the mechanism
by which it forms and ignites flammable material) is dependent on the
speed of break and the circuit constants.
If electrodes in a highly inductive circuit separate very fast, the current

attempts to fall instantaneously from its initial value to zero. Since the con-
tacts separate rapidly, the rate of change of current di/dt is very high. The
energy stored in the inductor L is used to generate a voltage E = −L di/dt,
which attempts to maintain the initial current flow. Even in circuits of
fairly low inductance it is possible to have transient voltages across open-
ing contacts of the order of kilovolts. When the inductance is large enough
and the speed of circuit break is high enough, the back voltage produced
by the inductance will be sufficient to break down the gap. Breakdown of
the gap may occur quite readily if vaporization of contact material at the
last contact point has provided a source of ions that will conduct current.
In many cases, as the contacts first separate, a short arc with a typical 10 V
to 20 V drop is formed as a result of vaporization of the last contact point.
As the contacts separate further there is insufficient voltage in the circuit
to maintain the longer arc demanded by the opening contacts; the arc dies,
and current ceases to flow. This in turn causes a high di/dt, and the stored
energy in the inductor causes a high voltage to appear across the gap,
which breaks down. This process may be repeated a few or many times,
depending on the nature of the contacts, the speed of separation, and the
amount of energy stored in inductor. The discharge in this case may be
short arc during its initial phases, may change to a glow discharge with a
drop of approximately 300 V, or may alternate between the two.
In highly inductive circuits, high breaking speeds will produce ignition

with the least amount of stored energy. For many purposes it is suffi-
ciently accurate, although somewhat conservative, to compute the energy
stored in the inductor from the equation Wl = 21 LI2 and assume that all of
the stored energy is dissipated in the arc or electrodes. Most often the igni-
tion conditions are defined by specifying current and inductance. In fast-
break and/or high-inductance circuits, breakdown of the arc and release
of the stored energy is determined by the rate of change of current. The
value of the supply voltage in the circuit is relatively unimportant in
determining the amount of energy required for ignition. Figure 9-14a
shows that for inductances above approximately 0.1 H the curves for sev-
eral voltage levels fall almost on top of one another. The voltage that can
be generated across the contacts is frequently several kilovolts. Circuit
voltage in instrument systems is usually small with respect to this value.
In circuits of low inductance and, in the limiting case, circuits with no

inductance, formation of the arc is probably almost always caused by
vaporization of the last point of contact causing the arc to form in metallic
vapor. The arc will, thereafter, exist as long as there is sufficient circuit
voltage to maintain the arc over the distance between the contacts as they
open. The arc voltage will start at the steady-state, short-arc voltage of 10 V
to 20 V and will increase as the electrodes separate. This arc will last until
the contacts are too far apart for the circuit voltage to maintain it.
In low-inductance circuits, slow and intermittent contact operations are

most efficient in producing ignition. Lower values of current are required
with slow or intermittent contact break than with a fast break. In circuits
of low inductance, if the contacts open rapidly, the voltage produced by
the change in current may not be great enough to break down the gap, and
the arc produced by vaporization of the last contact point is rapidly extin-
guished as the contacts separate. If the contacts separate slowly, the arc
persists for a longer time, and there is greater opportunity to transfer
energy to the surrounding flammable material.
Because this ignition mechanism is almost certainly a function of the life of

the short arc, the voltage in the circuit determines the contact separation
when the arc is finally extinguished. This affects the amount of quenching
by the contacts and length of time that energy is supplied to the flammable
mixture and, therefore, determines the amount of current that must be
broken to achieve ignition. Slow and intermittent break ignition in circuits
of low inductance, therefore, show a very high degree of voltage
dependence.
Test Equipment for Break-Spark Ignition

Measurements
By far the largest amount of data on ignition by break sparks has been pub-
lished by investigators in Great Britain who used the “break-spark”
apparatus, which was perfected in three stages from 1915 to 1950. The
break-flash apparatus in its earliest form consisted of a pointed resilient
strip of platinum that sprang against a platinum rod in such a fashion that
the strip broke contact with the rod at high speed. In the original apparatus
used by Wheeler (circa 1915) the platinum strip was stationary, while the
rod revolved. In a later version, called break-flash apparatus No. 1, the
moving electrode was a strip of 10% iridium-platinum alloy tapering to a
point, and at a still later date the strip was changed to platinum alloyed
with molybdenum or ruthenium. The fixed electrode was rod of platinum
alloy 1.5 mm in diameter. In break-flash apparatus No. 2, the electrodes
between which the spark was produced were of the same shape and
dimensions as in apparatus No. 1. However, the strip electrode was bent
into a smooth curve to make wiping contact with the rod, but with only the
minimum pressure necessary to ensure a good electrical contact. Rapid
break was produced by a spring mechanism and did not depend on the
elasticity of the strip. The whip of the strip still caused frequent fracture of
the strip at its fixed end. It was also somewhat difficult to set the strip at
exactly the proper curvature. The latest apparatus, break-flash No. 3, was
designed to require less frequent adjustment of contacts and to provide
easy control and measurement of the rate of electrode separation. Auxil-
iary circuits were also added to the machine to allow operation of other
devices in synchronism with the break mechanism. Break-flash apparatus
No. 3 is described in detail in Research Report No. 33 of the Safety in Mines
Research Establishment. The rate of separation of the electrodes varies
from 1 to 4 m/s. For official testing of intrinsically safe circuits, the device
was adjusted to produce a rate of electrode separation of 2.1 to 2.4 m/s.
When breaking a circuit with a 0.095-H air core inductance, at least one
ignition of an 8.3% methane-air mixture was required in 100 trials at a cur-
rent of 0.19 A, with no ignition in 100 trials at a current of 0.18 A. For
circuits of low inductance, the intermittent break apparatus was used.
The IEC standard rotating electrode holder and counter-rotating, slotted-

disc mechanism described earlier is now used for tests of inductive and
resistive circuits in all laboratories.
The author, in some unpublished experiments, used a steel disc approxi-

mately 75 mm in diameter in which a 45°-sector of the disc had been bent
down. The fixed electrode, a copper wire, was allowed to rest on the disc,
which rotated at 3,600 rpm. There was little difficulty in obtaining ignition
of methanol vapors at energy levels consistent with those obtained by the
British authorities in methane-air mixtures.
Figure 9-11 Break-Flash No. 3 Electrodes
Several investigators have used various forms of breaking-wire apparatus

in which circuits are broken by placing wires under tension. The develop-
ment of a British breaking-wire apparatus is described in IEE Conference
Report, Series No. 3 on “Flame-Proofing, Intrinsic Safety, and Other Safe-
guards in Electrical Instrument Practice.” By placing wires of small
diameter under tension it is possible to obtain ignition with break sparks
at energy levels approaching those obtainable with high-voltage capaci-

tive-discharge methods. However, these results have no direct bearing on
the problems of intrinsic safety in instrument systems because break
speeds as high as 1,000 cm/s produced by breaking wires under high ini-
tial tension are not at all representative of situations likely to occur in a
practical industrial system. Because the data obtained with such apparatus
are not pertinent to the problems of intrinsic safety, and because the appa-
ratus is suitable only for investigation of break sparks, typical designs are
not discussed further.
Figure 9-12 shows the relationship between circuit inductance and the cur-
rent required to cause ignition by break sparks, as reported by several
laboratories. The lowest curve (dotted line), from Gehm, gives the mini-
mum current for generation of sparks visible to a dark-adapted eye.
Except in circuits of very high inductance, visibility is no measure of abil-
ity to ignite. It is also noteworthy that even curves G1 and G2, derived
using the relatively sensitive PTB apparatus, are a factor of two in energy
above the levels that would be estimated based on minimum ignition data.
The other data vary up to a factor of 10 higher in energy.
Figure 9-12 Ignition Data-Inductive Circuits

The most thorough investigations of break-spark ignition and the effects

of changing parameters has been carried out by the Safety in Mines
Research Establishment, England. Many conclusions that can be drawn
from their work on break sparks are equally pertinent to electrical ignition
by closing contacts. The data are considered here because they were
obtained with break-spark apparatus.
Influence of Contact Material

In 1915 Thornton reported significant differences in ignition energy that
he attributed to the properties of the electrode material. His conclusions
were, in the main, valid; but his supporting data were crude by today's
standards, so that they are not reviewed in detail here. Wheeler, in 1926,
again related the magnitude of igniting current to electrode material.
Table 9-8 is derived from his data. The boiling points are taken from the
42nd Edition of the Handbook of Chemistry and Physics and differ greatly in
some instances from those which he used.
Igniting Current (A)

Boiling Melting
Metal First series Second series
point (°C) point (°C)
80 V 100 V 120 V 120 V
Cadmium 767 321 — — 0.22 0.23
Zinc 907 420 0.34 0.26 0.23 0.25
Silver 1950 961 0.44 0.41 0.38 0.32
Aluminum 2057 660 — — — 0.30
Tin 2270 232 0.66 0.53 0.45 —
Copper 2336 1083 — — 0.49 0.38
Gold 2600 1063 0.86 0.59 0.50 0.34
Nickel 2900 1455 0.63 0.58 0.55 0.39
Iron 3000 1535 0.58 0.52 0.49 0.42
Platinum 4300 1774 0.65 0.56 0.48 0.48
Table 9-8 Ignition Current (A) versus Electrode Material; 8.35–8.55% Methane-
Air, L = 31.75 m H
Wheeler also recognized the relationship between characteristics of a short

arc, electrode material, and ignition current required. He measured the
duration of the arc and noted an apparent relation between boiling point
of the metal and ease of maintaining the arc. Table 9-9 lists Wheeler's mea-
sured igniting currents and arc duration and, for comparison, lists the
minimum short-arc currents and voltages for these materials. These com-
parisons point inescapably to the conclusion that electrode material plays

an important role in determining ignition current through dependence of
arc characteristics on electrode material.
Material Arc duration (s) Igniting current (A) Vm(V) Im(A)

Cadmium 0.00321 0.23 11 0.1
Zinc 0.00234 0.25 10.5 0.1
Silver — 0.32 12 0.4
Aluminum — 0.30 14 —
Tin — 0.36 (est.) 13.5 —
Copper — 0.38 13 0.43
Gold 0.00070 0.34 15 0.38
Nickel — 0.39 14 0.5
Iron — 0.42 13–15 0.35–.55
Platinum 0.00081 0.48 17.5 0.9
Table 9-9 Igniting Current versus Arc Characteristics
Further investigation by Guénault, however, shows that although the

effect of electrode material is significant, the relationship is a more com-
plex one than Wheeler or Thornton had supposed. As might be expected,
in circuits of high inductance and fast break, the influence of electrode
material was nil. At lower inductances considerable differences in igniting
current attributable to electrode material were noted. These differences
were still more noticeable in slow-break tests. Figure 9-13 shows the influ-
ence of electrode material on igniting current in a slow break (12.5 mm/s),
24-V dc circuit. Guénault concluded that in circuits of high inductance the
soft, easily oxidized, low-boiling-point materials required higher ignition
current than platinum. In low-inductance circuits these materials allow
ignition at lower current levels. He attributed the differences to the fact
that in high-inductance circuits the discharge is a glow, whereas in high-
current or low-inductance circuits the discharge is a short arc, much
affected by electrode characteristics.
Widginton, using the IEC test apparatus with discs of different materials,
also found good correlation between minimum arc voltage and igniting
current in resistive circuits.
Figure 9-13 Effect of Electrode Material (from SMRE No. 41)
Zborovszky and Cotugno investigated Vm and the duration of the break

spark for many combinations of electrode materials. The data do not lead
to a simple summary, but using cadmium as a cathode material resulted in
lower Vm than would have been measured with a cathode of the same
material as the anode. The data also show the dependence of arc duration
on Vm. This dependence decreases as circuit voltage increases.
Schebsdat published curves in Ex Magazine for ignition of hydrogen-air by

circuits where one could be assured that only copper/copper contacting
could occur. In general, for 0.2 mm copper wires and a copper disk, the
ignition current levels in a resistive circuit are about 2.5 times those in the
standard apparatus. He presented data for 0.1 and 0.15 mm wires also.
0.1 mm wire caused ignition at current levels about 20% below those of
0.2-mm wire.
It has already been noted that where circuit and breaking conditions make
L di/dt large, the discharge voltage is high, and the contribution and char-
acteristics of the electrodes are relatively unimportant. Similarly, if L di/dt
is large, the circuit voltage is relatively unimportant in determining the
magnitude of the ignition current.
Figure 9-14a shows the effect of circuit voltage on the current required to
ignite 8.3% methane-air (as determined by Guénault et al.) for two differ-
ent types of test equipment. One of them separated platinum—4%
molybdenum strip contacts—at the rate of 810 mm/s; the other separated
platinum—10% rhodium rods—at 12.5 mm/s. These curves show the
expected influence of circuit voltage. When inductance is high, circuit
voltage has relatively little effect. In circuits of low inductance, voltage has
a very pronounced effect on the amount of current required—the required
current increasing rapidly as voltage is lowered. The effect of voltage vari-
ation is not remarkably different in the two sets of test apparatus.
Whether the circuit current is direct or alternating at power frequency

seems to make no difference in circuits of a few millihenries' inductance or
greater. Figure 9-14b shows ignition currents measured using 24-V dc and
15-V ac in slow- and intermittent-break apparatus, as reported by
Guénault et al.
Thomas reproduced data from Russian sources which show that in cir-
cuits of 10−4 to 10−2 H inductance the break-spark igniting current is never
lower than the dc value.
Figure 9-14a Effect of Voltage on Ignition Current-Slow and Moderate Speed

(from SMRE No. 106, CH4-Air)
Figure 9-14b Comparison of AC and DC (from SMRE No. 106)
In the range 0 to 18 kHz, one source reported igniting current to be a much

greater multiple of the dc value in circuits of very low inductance. The
ratio of ignition currents at 18 kHz and dc is only slightly greater than 1
for 10 mH, but is 6 for 0.1 mH. Another source indicates that igniting cur-
rent in a 0.1 to 1.0 mH circuit decreases at frequencies between 10 kHz and
100 kHz. The value, however, does not drop below the dc value.
Bittner is reported to have measured the voltage required in a 50-ohm cir-

cuit to ignite atmospheric mixtures of hydrogen, ethylene, propane, and
methane. For hydrogen there was little change in the ignition voltage from
20 V dc to 104 Hz. From 104 Hz to 107 Hz the igniting voltage was about
60 V. For other gases there appeared to be a gradual increase in igniting
voltage from dc to several kiloherz, increasing more rapidly to 105 Hz,
then remaining constant to 107 Hz.
Butcher et al. report spark power to ignite hydrogen in a coaxial version of

the IEC tungsten-cadmium ignition test apparatus that produced ignition
at the standard current in a 24-V, 100-mH circuit. From dc − 104 Hz igni-
tion occurred at about 2 W. Between 104 and 107 Hz, the power increased
linearly on a semilog plot to about 100 W, and then decreased to about 20
W at 5 × 108 Hz.
In resistive circuits or “circuits of negligible inductance,” the relationship

of greatest interest is that of the igniting current as a function of circuit
voltage. Figure 9-14c shows data on resistive circuits derived from several
sources. There is no easy way to rationalize differences between the
curves. That the PTB apparatus ignites at lower currents is very likely
caused by its use of cadmium for one electrode. Curve B, based on a zinc
blade against which a phosphor bronze contact is manually moved, is also
low, probably because of the properties of the zinc electrode. The British
data, derived from the investigation of Allsop, Guénault, et al. using the
intermittent break-spark apparatus, show odd and controversial cusps at
20 and 200 V. The 20-V discontinuity, in the author's opinion, may repre-
sent a change in ignition mechanism from ignition by an arc discharge to
ignition by hot particles or metallic vapor. A platinum electrode system
would have a minimum arc voltage of 15 to 20 V. The same sharply rising
ignition current is found in curves B, G, G1, and G2 obtained with different
test equipment. The discontinuity at 200 V can not be easily explained. No
similar cusp appears in other curves of Figure 9-14c or in Widginton's data
in SMRE Report No. 256.
Figure 9-14c Ignition in Resistive Circuits
Effect of Contact Separation Speed

In discussing the influence of circuit voltage it was noted that whenever
the inductive voltage is large with respect to the circuit supply voltage, the
circuit supply voltage has little influence on igniting current. This section
deals with an essentially similar matter, the effect of changing break speed
while holding the circuit supply voltage constant. This effect is, of course,
almost self-evident from previous discussion. As the speed of break
decreases, the L di/dt voltage decreases, and, in circuits of moderate to
high-inductance, the required ignition current increases. At lower values
of inductance, where the ignition mechanism depends primarily on

energy transfer from a short arc, decreasing the speed of break increases
the time available to transfer energy to the flammable material, and the
required ignition current decreases.
Figure 9-15 illustrates the effect of changing the speed of break in a circuit
of 24 V dc. The figure shows the relationship between ignition current and
inductance when the circuit is broken by break-flash apparatus No. 2, in
which the contacts separated at about 100 in./s when spring-operated and
at a few inches per second when operated manually. This comparison is of
a single electrode system broken at different speeds. The other two curves
show the relationship between inductance and ignition current for the
intermittent break apparatus previously described and an apparatus in
which the electrodes separated at 12.5 mm/s. The contacts in this appara-
tus were butting 1-mm-diameter platinum-10% rhodium alloy rods,
operated by a cam. As expected, in the higher range of inductance the fast
break is the most efficient mechanism, while for circuits of low inductance
the slow or intermittent break is most hazardous.
Lord et al. in their review of wirebreaking apparatus show that the ends of
wires breaking in tension may separate at 1,000 to 1,500 cm/s compared to
300 cm/s in the standard break-spark apparatus. This higher speed, with
the added benefit of low-mass, small-diameter wire, reduces ignition cur-
rents to a level that at 0.095 H the ignition energy approaches the
minimum energy measured with high voltage capacitive discharge.
Ignition of Materials Other than Methane

Because early interest in intrinsic safety in the United Kingdom stemmed
from mine safety problems, most work was done with methane-air mix-
tures. Table 1 in British Standard 1259, March 1963, classified 126
materials, 49 of which had been ignition-tested. The remainder were clas-
sified on the basis of similarity of chemical structure and properties to
those which had been tested. The basis for classification without testing
lies in part in the following rationale, used by a committee of experts:
(1) In a homologous series ignitability decreases as molecular weight
increases, though the order may be reversed for one or two of the
lower members of the series. In the case of the paraffin hydrocarbons
propane is the most easily ignited.
(2) Chlorination reduces ignitabilities.
(3) Esters are slightly more difficult to ignite than the corresponding
hydrocarbons.
Figure 9-15 Effect of Speed of Break (from SMRE No. 106, 24 V, CH4-Air)
Table 9-10a shows representative ignition data for materials that have
been tested, as tabulated for the author by H. G. Riddlestone of the Electri-
cal Research Association.
Most ignitable
Inductance
Compound concentration
0.102 mH 0.33 mH 0.825 mH 95 mH ~1.1 H ~15 H % mg/liter
Acetaldehyde 6.9 2.42 0.161 0.042 0.014
Acetic acid 6.2 0.44 0.086 420
Acetone 6.5 3.9 0.165 0.045 9.0
Acetylene 2.0 0.73 0.063 0.017 0.006 8.7
Acrylonite 1.95 0.128 0.034 260
Allyl chloride 2.9 0.22 0.053 207
Allylene 1.5 0.113 0.032 6.7
Ammonia 1.13 0.25 18.5
Benzene 6.7 0.165 0.041 0.0135 178
Blue water gas 1.35 0.098 0.026 0.0073 31.0
Butadiene 5.4 1.64 0.128 0.035 0.010 5.2
Butane (calor gas) 3.7 0.170 0.043 0.0125 4.5
Butene-2 2.6 0.149 0.042 5.5
Butyl acetate 2.6 0.178 0.044 220
Butylamine 2.9 0.187 0.050 230
Carbon disulfide 1.52 0.75 0.064 0.022 0.008 730
Carbon monoxide 5.7 3.3 0.155 0.043 0.0135 51.0
Chlorethane 7.6 3.0 0.195 0.051 0.017 7.25
Chlorodimethyl—ether 0.161 537
Chlorethylene 6.9 2.4 0.166 0.049 0.015 8.5
Cyclohexane 6.75 0.165 0.041 0.0135 128
Cyclohexene 3.4 0.160 0.042 0.0125 143
Cyclopropane 5.5 2.1 0.138 0.038 0.012 5.75
Dibutyl ether 2.3 0.150 0.041 175
Diethyl ether 6.7 2.3 0.132 0.046 0.012 170
Dimethyl ether 2.0 0.140 0.036 8.0
Epoxypropane (propylene oxide) 1.3 0.105 0.031 153
Ethyl mercaptan 0.148 210
Ethyl methyl ketone 2.5 0.151 0.041 200
Ethoxyethanol (cellosolve) 2.3 0.169 0.044 200
Ethylene 3.5 0.115 0.030 0.009 7.8
Ethylene oxide 3.65 1.46 0.095 0.027 0.011 11.2
Heptane 6.5 3.7 0.175 0.047 150
Hexane 6.1 3.6 0.165 0.043 122
Hydrogen 2.15 1.15 0.082 0.025 0.0067 22.0
Isohexane 4.0 0.165 0.045 0.0135 3.4
Isopropyl nitrate 2.1 0.130 0.036 400
Methane 3.9 0.165 0.043 8.3
Methane* 7.0 2.5 0.157 0.044 0.013 8.9
Methanol 6.4 2.4 0.162 0.046 0.015 215
Methyl acetate 8.3 2.59 0.178 0.043 0.015 315
Methyl acrylate 0.161 327
Nitromethane 2.6 0.151 0.045 540
Pentane 6.2 3.8 0.160 0.045 3.9
Propane 6.1 2.2 0.148 0.039 0.012 5.25
Propanol 6.7 2.4 0.151 0.048 0.014 210
Styrene 3.1 0.20 0.054 245
Town-gas 3.3 1.55 0.085 0.031 0.0075 14.0
Trioxan 1.83 0.125 0.040 415
*Industrial: >90% CH4, < 10% H2
Table 9-10a Minimum Igniting Currents (A) in 24-V Inductive Circuits (British Breakflash
Apparatus)
When minimum ignition energy is known, it might seem reasonable that

ignition current levels under less favorable conditions such as exist in the
break-spark apparatus could be predicted. Table 9-10b shows break-spark
ignition currents for a number of gases and vapors expressed as a ratio to
the igniting current for methane under the same conditions. Although the
ratio of measured currents tends to follow the pattern predicted from
known minimum ignition energy values, there are some large deviations
from the pattern, as in the values for hydrogen and propane for moderate
(0.2 to 1.0 H) size inductances. In view of the fact that ignition energies or
currents of several gases may not be in the same relationship for different
methods of generating the igniting arc, it is not surprising that when the
mechanism of ignition is radically different, as in hot-wire ignition, the rel-
ative ease of ignition of the gases may appear to be quite different.
Minimum Measured Current Ratio

ratio to Calculated
Flammable ignition ERA
methane, current ratio, APPROXIMATE INDUCTANCE, H
material energy source
W /W (Wm/WMethane)1/2
Wm (mJ)* m Methane 0.0001 0.002 0.02 0.095 0.2 0.5 1.0 14+
Hydrogen 0.019 0.068 0.26 0.26 0.23 0.28 0.44 0.42 0.49 0.49 0.45 D/T113
Ethylene 0.083 0.30 0.54 0.49 0.46 0.49 0.62 0.56 0.57 0.61 0.62 D/T106
Cyclo-propane 0.173 0.62 0.79 0.74 0.78 0.82 0.75 0.70 0.78 0.77 0.86 D/T126
Diethyl ether 0.198 0.71 0.84 0.91 0.72 0.74 0.71 0.85 — 0.94 0.83 D/T120
Hexane 0.24 0.86 0.93 0.86 0.83 — 0.89 0.78 — 0.87 — G/T232
Heptane 0.24 0.86 0.93 0.91 0.77 — 0.95 — — 0.94 — D/T232
Propane 0.25 0.89 0.95 0.82 0.79 0.76 0.80 0.73 0.76 0.80 0.86 D/T126
Methane 0.28 1.00 1.00 1 1 1 1 1 1 1 1 D/T126
*Bureau of Mines Report 500 (high voltage, capacitive discharge).
+Iron core.
Table 9-10b Comparison of Minimum Ignition Energy and Break-Spark Ignition Current
Cawley developed curves for the mean value of current or voltage in resis-
tive, inductive and capacitive circuits that would produce an ignition
under the standard test conditions using the IEC apparatus (i.e., within
400 revolutions of the wheel, or 1600 sparks). His goal was to compare the
values required for methane, propane, and ethylene to evaluate the use of
more sensitive mixtures to establish a safety factor in testing. His results
for inductive circuits show a trend similar to that of the data in Table 9-8a;
that is, the ratio of current to ignite methane to the current that is required
to ignite ethylene is higher in circuits of low inductance than in circuits of
higher inductance. The trend in ratios for propane was in the opposite
direction, but not strongly. In resistive circuits the ratio of igniting cur-
rents decreases with decrease in circuit voltage. In capacitive circuits of 10
to 1,300 μF the ratios were not well behaved for methane/propane. The
methane/ethylene ratios decreased with increasing capacitance. Cawley
concluded that using more easily ignited mixtures for achieving a safety
factor was too complicated to be recommended.
Probability of Ignition—Influence of Current and

Voltage Level
In previous sections it has been shown that the ignition process, especially
in circuits of low voltage, is not easily definable and, moreover, is depen-
dent on processes that are neither easily controlled nor entirely
understood. Ignition by a single spark is not certain. For any circuit and
test apparatus, only a probability of ignition at a stated test current can be
defined. In British practice, the so-called minimum igniting current is a
current determined by lowering the test current in steps of 5 to 10% until
no ignitions occur in 100 trials. The minimum igniting current is taken to
be midway between the no-ignition current and the previous test current
at which ignition occurred at least once in 100 trials. Other investigators
have used 1,000 to 3,000 sparks as the trial quantity.
Thomas points out that in British practice the range of real ignition proba-
bility that could yield one ignition in 100 trials is considerable. If the real
probability is 3 × 10−3, one ignition in 100 trials can occur 50% of the time.
If the real probability is 10−3, there is a 20% chance of getting ignition once
in 100 trials. If the true average probability is 10−2, no ignitions may occur
in a sequence of 100 sparks as often as once in 8 sequences.
According to Thomas, the Russian practice is to define the minimum igni-

tion current to be the current with ignition probability of 10−3. Test
sequences of 16,000 sparks are required to be 95% certain that the mea-
sured ignition probability is between 5 × 10−4 and 1.5 × 10−3. Such lengthy
testing is not routinely necessary, however, because it has been shown that
for the Russian test apparatus the relationship between ignition probabil-
ity and current is of the form:
P1 i n
------ = § ----
1·
P2 © i2¹
The data presented by Thomas shows that, for the Russian apparatus, n is
approximately 12 to 16 over a range of inductance of 10−3 to 37 H, for dc
voltages from 10 to 250 V, and for ignition probabilities less than 0.2.
In most evaluations of intrinsic safety has been customary to apply an

arbitrary factor of 1.5 to 2 when determining safe operating current levels.
Applying a factor of 2 results in permissible circuit current that is one-half
the minimum igniting current for that circuit. A factor of 1.5 permits the
circuit current to be two-thirds the minimum igniting current. If the mini-
mum igniting current has probability 0.01 of causing ignition, the circuit
with a factor of 2 will have ignition probability of 1.5 × 10−7. With an arbi-
trary current factor of 1.5, the circuit ignition probability will be 1.5 × 10−5.
These statements are based on the Russian data and assume that the same
laboratory test conditions are maintained, except for current level.
There is no evidence that the value of n is, or should be, the same for all
forms of test apparatus. One could hypothesize that the value of n might
be much higher in high-voltage capacitance-discharge ignition where the
number of variables affecting the ignition process is smaller because the
electrodes are outside the developing flame sphere.
A curve of ignition frequency as a function of current in break-flash appa-

ratus No. 3 presented by Thomas yields a calculated value of n which is
certainly greater than 20. Consideration of Rogulski's plotted data on igni-
tion with heated electrodes in an entirely different apparatus yields a
value of approximately 20 for n. Müller's data also yield a value of n of
approximately 20, for yet another apparatus. It must be emphasized that
estimates of n from these published curves are certainly not accurate to
better than ±25%.
Schebsdat and Johannsmeyer based their investigations on a probability of

ignition of 10−4 or 10−3, and 10−3, respectively. Schebsdat reported data for
a mixed circuit with 200 ohms in series with the power supply, a 100 nF
capacitor, and a 10 ohm, 5 mH inductor in series with the standard IEC
apparatus, which discharged the capacitor. He plotted the data on proba-
bility paper, which has the probability values for a normal distribution on
one axis and the independent variable on the other axis. If the data are
normally distributed, the plot is a straight line. The value of n calculated
from such a plot depends on the interval between probabilities chosen. For
the interval between a probability of ignition of 0.0001 and probability of
0.001, the effective power in the equation above was 30. If one applies a
safety factor of 2/3 to the ignition voltage at 10−3 probability, the calcu-
lated probability of ignition at the lower voltage is 5 × 10−9. The value read
from probability paper is somewhat smaller.
Johannsmeyer reported data for capacitive ignition, 18.3 volts, 1 μF, at 10−
3 probability. His data also fall on nearly a straight line on probability
paper. The value of the exponent n was 11.5 at a probability of 10−3, which
gives a calculated probability of 9.5 × 10−6 when a factor of 2/3 is applied.
The value read from probability paper is slightly lower.
Schebsdat's data for a large variety of cables also plotted as straight lines
on probability paper. The values of n for the same probability intervals as
above lay between approximately 13 and 30.
Table 5 of the 1977 Fraczek paper provided values of the exponent n from
various sources and for differing kinds of experiments. Fraczek points out
that the ratio discussed above implies that the data plot is a straight line on
a plot of log p vs. log i, which is not realistic. As current increases the cal-
culated probability can exceed 1. It is, therefore, important that one knows
the value of probability at which n was derived and uses it only over a
limited range of current values. Table 9-11 was derived from the Fraczek
table.
Probability
Test Apparatus Circuit Atmosphere n
Range
USSR V=7V Methane-air 10−3 – 0.3 12.9
Certifications L = 1 mH – 37 H
V = 70 V Methane-air 2 × 10−6 – 10−3 15.1
L < 1 mH 5 × 10−9 – 2 10−6 49.9
−3 –
SMRE V = 24 V 22% Hydrogen-air 10 1 13.9
Fraczek L = 95 mH
Type L V = 24 – 130 V 8.3 − 8.5% 10−5 – 10−1 10.5 - 120
“Barbara” Mine L = 1 mH − 10 mH Methane-air
Poland
IEC V = 10 – 12 V 22% Hydrogen-air 8 × 10−4 – 3 × 10−2 6.6 - 9.2
BELMA Co. Intrinsically safe
Poland transformers
IEC V = 20 – 24 V 8.5% Methane-air 10−4 – 10−2 6.7 - 11.2

“Barbara” Mine Intrinsically safe
transformers
IEC C = 122 – 563 μF 8.3–8.6% 10−3 – 10−1 6.4 -14
“Barbara” Mine Electrolytic type Methane-air
IEC R, L, C >10−6 31.2 - 43
USSR
USSR - prior to 1963 R, L, C 14.3 - 23
Table 9-11 Values of n from Fraczek
In R1 9183 Cawley reported the work of Matasovic, who defined a safety

factor, k:
k = I1/I2
where
I1 = current or voltage at which probability is 10-3
I2 = current or voltage at which probability is 10-8
Table 9-12 shows the current or voltage ratio reported by Matasovic in

accordance with this definition (rounded to two decimal places by this
author).
Inductive Capacitive Resistive

Methane 1.55 2.77 2.95
Propane 1.30 2.67 2.33
Ethylene 1.36 2.43 2.31
Hydrogen 1.66 2.65 2.58
Average ratio (n) 1.46 (30.4) 2.63 (11.9) 2.54 (12.4)
Table 9-12 Current/Voltage Ratios which Change Ignition Probability from 10-3
to 10-8
Approximate values of n for each mechanism of ignition, corresponding to

these current ratios and assumed probabilities, calculated by this author,
are shown in the last row. The values of n are ordered somewhat as one
would expect, based on the efficiency of the ignition process in each kind
of circuit. Inductive circuits generally are held to have lower quenching
than a resistive or capacitive circuit, and one might expect the distribution
of ignition values to have a smaller standard deviation and n to have a
higher value. Marinovic discusses this safety factor of Matasovic in his
book, but it is available only in Serbo-Croatian.
Cawley's work was motivated by the desire to verify that the probability
function is truly a straight line for low values of probability, p < 10−4, and
that it is legitimate to extrapolate on the line to probabilities as low as 10−8.
The data in Matasovic's work did not extend below a probability of 4 × 10−4.
Cawley extended the data another three orders of magnitude, or to a proba-
bility value that appears to indicate a lower ignition limit. This latter
interpretation results from the statistical design of his experiments. If no
ignition occurred in 5/p trials, where p is the probability for which the igni-
tion current or voltage is being determined, this is a statistical anomaly.
This anomaly is interpreted as a lower limit value, below which the proba-
bility of ignition is essentially zero, or the safety factor is infinite.
RI 9183 contains both plots of probability vs. current or voltage with cir-
cuit element value as a parameter, and curves of igniting current or
voltage for each of the three types of circuit with probability of ignition as
a parameter. The test mixture in all cases is 8.3% methane-air.
The work of Cawley and Matasovic emphasizes how misleading arbitrary

numbers like a safety factor of 1.5 can be. The real safety factor resulting
from changing the variable by an arbitrary 2/3 or other number depends
on the characteristics of the mode of ignition. Matasovic's data show that

in inductive circuits the reduction in probability resulting from a 2/3 fac-
tor is greater than 5 orders of magnitude. On the other hand, the same
factor in resistive and capacitive circuits results in a reduction of less than
three orders of magnitude. Either reduction is more than sufficient, but the
data illustrate:
• How silly the arguments over additional safety factor are.
• How isolated these arguments are from the reality of the ignition
process.
• How quickly people forget that the test apparatus and conditions
were designed to make ignitions happen at a level not to be found
in practice in industrial situations.
Effect of Shunt Elements on Ignition Current

In an inductive circuit ignition by a break spark is usually caused by an arc
initiated and maintained from the stored energy of the inductance. After
the circuit is broken, the circuit appears as shown in the following sketch.
The voltage e = −L di/dt is shown in a direction to strike and maintain the

arc. The current at the moment before the circuit is broken is E/(R + RL),
where R includes all the resistance in the circuit except that of the induc-
tor, RL. After the circuit is broken, the voltage, e, is driving current through
the arc. Connecting a resistor, a capacitor, or a nonlinear element such as a
diode across the inductor provides a path for current to flow in parallel
with the arc. Not only does this steal energy from the arc, but a given volt-
age e at the inductor produces less voltage at the arc because the
additional current in the shunt element increases the voltage drop in the
inductor.
A numerical prediction of the effects of adding shunt elements is not

attempted here. The exact solution of the equations depends on ade-
quately representing the V-I characteristics of the arc.
Qualitatively, though, consideration of the circuit sketched above can lead

to the following conclusions:
(1) If the shunt element draws current of significant size relative to the
arc circuit, the shunt element will significantly raise the initial current
required for ignition.
(2) If ignition without a shunt element was achieved by a high voltage
breaking down the electrode gap to form one or more glow dis-
charges with a 300-V drop in the arc, the addition of a shunt element
would seem to be an effective way of reducing the voltage appearing
at the inductor terminals.
These conclusions are not independent. They represent two facets of the
same situation and, interpreted in terms of the ignition conditions before
the shunt element was added, can be restated as follows:
(1) The effectiveness of shunt elements in increasing the circuit current

that must be broken to cause ignition will be greatest in high-speed
and/or high-inductance circuits where the arc usually starts with a
glow discharge, the currents broken are low, and the arc and circuit
resistance, R, offer a relatively high impedance path compared to that
of the shunt element.
(2) In circuits of lower inductance or slow-break conditions, where ignit-
ing currents are high and ignition is caused by short, low-voltage arcs,
shunt elements are less effective in raising the circuit current that must
be broken because the impedance of the arc path is already quite low.
It has been observed that shunt capacitors, rather than increasing the ini-
tial circuit current, can cause the required ignition current to decrease if
the circuit is broken slowly or intermittently. In Safety in Mines Research
Paper No. 106, shunt capacitance was shown to be quite effective in fast-
break circuits. In slow- and intermittent-break circuits shunt capacitance
decreased the required current to as low as 70% of the unshunted value.
Müller showed that shunting of a 4-H inductor with capacitors increased

the ignition current from 21 mA for C = 0 in a methane-air mixture to
100 mA for C = 0.25 μF. Further increasing C to 10 μF decreased ignition
current to 70 mA.
Unfortunately, the efficacy of shunt resistors and nonlinear elements in

increasing the ignition current is, as was noted above, dependent on the
fraction of the total circuit resistance contributed by the inductor resis-

tance RL. The data in Figures 9-16a–e, can be used, therefore, only to
indicate a representative degree of protection that can be attained by using
shunt elements.
Figure 9-16a shows the effect of shunting resistors and nonlinear elements
in dc circuits broken by break-spark apparatus No. 2. Resistors are effec-
tive in raising the ignition current; however, in a practical circuit, where
the current through the inductor is functional, there is a lower limit to the
size of the shunt resistor. A rectifier, however, does not shunt the inductor
except with respect to voltages generated in it when the circuit is broken.
In this case the rectifier resistance is low when protection is needed, high
to current from the circuit source, and the degree of added safety is high
compared to that obtainable using a resistive shunt of practical value.
Figure 9-16b illustrates the relative effectiveness of shunt resistors in 24-V

dc circuits with slow, intermittent, and fast break. Since the unshunted
ignition currents decrease as the break mode is changed from “slow” to
“intermittent” to “fast,” the impedance of the test circuit increases in that
order. The effectiveness of a particular shunt resistance, therefore,
increases as the rate of break is increased. This figure also shows, as does
the preceding one, that the effectiveness of a given shunt decreases rapidly
as the inductance of the circuit decreases.
Figure 9-16c shows, relative to unshunted ignition current, the effect of

using capacitive shunts across inductors. Note that in some cases ignition
current decreases.
Figure 9-16a Effect of Shunt Elements (Break-Flash No. 2, CH4-Air, from SMRE
No. 106)
Figure 9-16b Influence of Break Speed on Shunt Effectiveness (from SMRE

No. 106)
Figure 9-16c Effect of Shunt Capacitance-Variation with Break Speed (from

SMRE No. 106)
Figures 9-16d and 9-16e are derived from data presented by Berz and
Khan. They used a modified intermittent-break apparatus to measure the
effect of shunt silicon diodes on current to cause ignition. The data plotted
on Figure 9-16d are for a series of inductors with and without a Ferranti
ZR11 diode, at several circuit voltages. The curves show the same general
features which were observed in Figure 9-16a. For a given circuit voltage
the relative effectiveness of the shunt is highest in circuits of high induc-
tance. This fact is even more strikingly presented in Figure 9-16e. This
latter figure also emphasizes the reduced relative effectiveness of a shunt
diode in circuits with higher supply voltages. Shunt effectiveness can be
viewed in another way. The lower the L/RL ratio of the inductor, the less
effective the shunt diode in increasing igniting current. Increasing the
inductor resistance requires that either the value of R (see diagram on
page 100) be decreased or the voltage E be increased to maintain constant
circuit current. The effect of the latter is apparent in the illustrations. The
effect of either is attributable to the fact that the voltage drop across the arc
and resistor R, less the source voltage E must equal the diode voltage
drop. Decreasing R or increasing E raises the voltage across the arc,
increasing available energy for ignition. This is, of course, a qualitative
analysis, not a precise description of the very complex process. It is pre-
sented only as an aid to understanding.
Thomas points out that the efficacy of a diode as a shunt is dependent on

its forward recovery time. After a diode has been reverse-biased, the cur-
rent does not rise instantaneously to the normal value when the diode is
forward-biased. For a time the diode conducts less effectively and for a
given current has a higher voltage drop. The time that elapses before the
diode conducts normally is dependent not only on diode construction but
on the magnitude and rate of rise of the current pulse. Thomas suggests
that in some cases a shunt capacitor to supplement the diode and coil self-
capacitances may reduce the circuit rise time and reduce the transient
diode voltage.
Capp and Widginton noted that some zener diodes have high capacitance
and that in the work of Cartwright (see Chapter 10) they gave higher ignit-
ing current values than would diodes with lower capacitance.
Because of the important influence of both the diode and inductor charac-
teristics and the specific parameters of the circuit, it appears that the
efficacy of a shunt diode should be verified by test. In some instances the
measured effectiveness can be substantially lower than would be expected
from published data.
Figure 9-16d Effect of Shunt Silicon Diode on Igniting Current Level

Figure 9-16e Influence of Circuit Voltage on Shunt Diode Effectiveness
The data presented in this section are valid for examining the effects of
shunt elements, but they are not valid for assessing intrinsically safe cir-
cuits because the data were not taken with the standard test apparatus.
Unfortunately, the author knows of no equivalent data based on the stan-
dard apparatus. It is the author's impression that the efficacy of shunt
diodes shunting inductors is often accepted without test in Europe. In at
least one U.S. test house the efficacy is routinely tested. Some European-
certified designs have been found to ignite in U. S. tests. However, the
results may have been influenced by the method of test, such as raising the
supply voltage to obtain a 1.5 safety factor on the current.
Bass presents an analysis of the most common shunt protection circuit, a

diode that is reverse-biased to normal flow of current, and forward-biased
to current caused by the inductive voltage. He notes that after the circuit is
broken the voltage across the diode, ed, adds to the supply voltage, E in the
diagram presented earlier. The circuit will not be more ignition-capable

than one with an open circuit voltage of E + ed, and a short circuit current
of (E + ed)/R. Therefore, if the voltage across the diode is assumed to be 0.6–
0.7 V the circuit is only slightly more hazardous than the shorted supply
which has an open circuit voltage of E and a short circuit current of E/R.
This would be the case if the diode is chosen to have a low voltage drop at
the value of current passing through it.
Hot Wire Ignition

One can consider hot-wire ignition to be a limiting case of the general phe-
nomenon of ignition by a hot surface. Much work was done by SMRE to
determine limiting temperature of enclosures used in mines. These were
found always to greatly exceed the SIT of methane and were inversely
dependent on the area of the heated surface.
API Publication 2216 presents a rule of thumb that ignition of hydrocarbon

vapors in open air should not be assumed unless the surface temperature is
about 200°C above the accepted minimum ignition temperature. This doc-
ument also presents the data in Table 9-13 below on the effect of ignition
delay on the surface temperature required to ignite three hydrocarbons.
The conditions are not specified in detail; but, presumably, the surface is
exposed, and normal convection currents existed, but not wind.
Material (AIT) 100 s 10 s 1s

Pentane (243) 215 297 413
Hexane (223) 216 288 384
Heptane (204) 202 259 332
Table 9-13 Variation of Ignition temperature, °C, with Ignition Delay
Though the 100-s values of surface temperature seem to be below the

accepted AIT values, it should be remembered that surface temperature
measurements are difficult and generally err on the side of being too low
because of conduction of heat by the measuring element.
The effect of ventilation or wind is shown by the data on kerosene (210°C).

In these investigations emphasis was on ignition-capable surface tempera-
ture. In the context of the design of electrical equipment containing wires
that might reach the ignition temperature the emphasis is on the amount of
current that will cause a wire to reach the temperature needed for ignition.
The ignition mechanism of the hot wire differs from that of spark ignition
in that the energy is imparted to the flammable mixture over an extended
period of time, as compared with 10−7-10−6 s, during which energy is

released from a high-voltage capacitor, or 0.01 to 0.1 ms in inductive cir-
cuits. In few cases is the time less than tens of milliseconds, making it
relatively long compared to break-spark or low-voltage make-spark igni-
tion. Because of the greater length of time during which energy is
supplied, factors such as convection of the gas around the wire and partial
oxidation of the combustible gas or vapor play an important part in
decreasing the efficiency of the mechanism.
In the absence of convection currents, one could postulate that a wire of

diameter equal to the quenching diameter of the flammable mixture
would ignite the mixture if its temperature were raised to the spontaneous
ignition temperature of the mixture. Such a wire would, in effect, start
ignition above the critical diameter. However, the ignition temperature
usually listed in reference tables is the spontaneous ignition temperature,
determined by allowing a gas or vapor to remain in contact with the walls
of an almost completely closed 200 to 250 mL chamber for many seconds.
Conditions for ignition are highly favorable because heat is not lost from
the gas or vapor by convection or radiation.
In the case of a hot wire, however, convection currents around the wire do
not allow a particular volume of gas to remain in contact with the wire
long enough to reach ignition temperature, unless the wire is at much
higher temperature. When 114-mm-long nickel bars, 1 mm thick and 12.5
to 25 mm wide, were heated to ignite 11% methane-air, the bar tempera-
tures were 1,079°C for the 25-mm width and 1,114°C for the 12.5-mm
width. This is quite high relative to the ignition temperature of methane,
538°C. Investigations in England have shown that smaller hot surface
areas must reach even higher temperatures to cause ignition. For the most
easily ignited methane-air mixture tested, about 6% methane (because the
mechanism of ignition is different, a different balance between heat sup-
plied by burning and thermal losses is required, so the mixture is not the
same as the most easily ignited mixture for spark ignition) the surface
temperature needed to ignite the gas was about 1,550°C for a 9 mm2 area,
and about 1,150°C for a 325 mm2 area.
Bartels used glass-sheathed heaters, 3.8 cm and 5.1 cm diameter, 15 cm

long, to ignite vapors of diethyl ether, dimethoxymethane, and heptane,
whose AITs are approximately 170°C, 235°C, and 215°C. The 5.1-cm-diam-
eter heater produced ignition at temperatures of 227°C, 327°C, and 302°C.
The ignition temperature of diethyl ether was 235°C with the 3.8-cm-
diameter heater.
Silver investigated ignition by hot spots on resistors. His data show hot-
spot temperature of 600°C or higher to ignite ether-air mixtures. This tem-
perature is not the average temperature of the resistor, but the hottest spot
at the top center of a horizontal resistor. Silver references unpublished
data on ignition of various vapors by a 12-W resistor, 9 mm diameter,

90 mm long, and by a 0.9-mm wire. For carbon disulfide the temperature
of the resistor and wire were 59°C and 100°C higher than the AIT, respec-
tively. The temperatures for hydrogen were 90°C and 200°C. For other
materials, the differentials were several hundred degrees higher than the
AIT. It is not yet possible to state simple relationships about hot-surface
ignition of gases and vapors relative to the other ignition parameters such
as MESG, MIC, or MIE.
The temperature at the surface depends also on the material. If the mate-
rial is catalytic, as platinum is, the surface temperature at instant of
ignition is higher than if the wire is not catalytic. With catalytic materials,
fast reactions at the surface of the body cause a thin layer of burned gas to
insulate the hot surface from the surrounding unburned gas. Heat from
the burning layer is lost to the wire surface, raising its temperature. When
the material is not catalytic, more heat from slow reactions at the surface is
conducted to unburned gas; flame spreads faster; ignition occurs at a
lower surface temperature.
If the wire diameter is decreased, additional energy is required to bring

the expanding wave front to the critical diameter Dq, and the required
wire temperature is yet higher. The work of Stout and Jones indicated
temperatures of 2,000°C to 2,300°C. Thornton reported great difficulty in
igniting methane with any wire whose melting point is below about
1,800°C. Table 9-14 shows ignition currents for 30% hydrogen-air mixtures
as reported by Thornton.
IGNITING CURRENT (A)
Wire diameter Wire diameter Wire diameter

Material
0.1 mm 0.2 mm 0.3 mm
Molybdenum 2.1 5.2 8.7
Tungsten 1.8 5.2 8.2
Silver 1.1 3.4 6.7
Gold 1.1 3.3 5.3
Nickel 1.4 3.3 5.2
Iron 1.1 2.8 4.5
Platinum 1.1 2.5 3.8
Table 9-14 Effect of Igniting Current (A) of Wire Diameter and Material
The wire sizes Thornton tested are all larger than 38-gauge, so that they
are of a size, although of unlikely material, to be of practical interest to
instrument people. It is interesting and significant to note that he found
the smallest igniting current to be 1.1 A, which is less than the current
required for break-spark ignition in some low-voltage resistive circuits. It

is, therefore, of importance to decide whether hot-wire ignition is a mode
of ignition that should be considered in evaluating and using instrument
systems.
Stout and Jones presented data which show that some wires are so small
that they cannot become sources of ignition. Although the Stout and Jones
experiments were intended as a method of determining minimum ignition
energy by extrapolating curves to zero time, the data can be used also for
safety considerations by extrapolating to infinite time to determine the min-
imum current required for ignition. The data are plotted in Figure 9-17a.
The 1.14 × 10−3 cm diameter wire did not ignite the 11% methane-air mix-
ture, although it ignited a 20% hydrogen-air mixture. Somewhat larger
sizes caused ignition of methane-air only for relatively large currents
applied for a short time. Smaller currents over a longer period of time are
believed to have caused the wire to fuse before ignition. Figure 9-17b shows
the current required to ignite various atmospheric mixtures as a function of
wire size.
For a particular wire diameter, the current for hot-wire ignition is almost
the same for methane-air and hydrogen-air mixtures, and the igniting cur-
rents for both gases are not far removed from the currents required to fuse
the wire. This suggests that ignition in many cases may be caused by the
arc formed as the wire fused. This certainly would explain the sameness of
ignition currents for mixtures of such widely different critical ignition
energies. Most of the energy is required to melt the wire and form an arc;
only thereafter is the amount of energy related to gas properties.
The results of Stout and Jones, though interesting and applicable to any
situation where high-melting point wires may be used, still do not deter-
mine the fundamental importance of hot-wire ignition in the usual
instrument construction. The author, in the work previously cited,
reported a few measurements on copper and nickel wires that bear on this
question. Using nickel wires of 0.025 to 0.25 mm diameter, methanol vapor
ignited at ignition currents comparable to those reported for methane and
hydrogen by Thornton and by Stout and Jones. These comparisons were
made to verify an approximate equivalence in experimental technique.
Straight copper wires 0.075 and 0.25 mm in diameter were found to fuse
without causing an explosion. However, if the wire were looped, it was
possible to obtain an explosion with a current of 9 A if five loops were
made in a 0.25-mm diameter wire. If only a half-loop were made in the
wire, the wire would burn out without causing an explosion. The loops
were approximately 3 mm in diameter, of the sort that would be made in a
heating element to concentrate thermal energy.
Figure 9-17a Hot-Wire Ignition (from Stout and Jones)
Other work, reported by Shepherd and Wheeler, was undertaken to deter-

mine the nature of the hazard from miners' lamps. These typically have 2-
V, 0.65 to 1.0 A filaments. Shepherd and Wheeler found no difficulty in
igniting methane-air mixtures by heating exposed filaments or by breaking
the envelope of a lamp in a methane-air atmosphere. Ignition occurred
even when the wire merely glowed dull red because oxidation of the tung-
sten filament raised its temperature.
In order to better understand the phenomenon, they heated platinum wires

of 0.1 mm, 0.2 mm, and 0.3 mm diameter and tungsten wires of 0.065 mm
and 0.1 mm diameter in a methane-air mixture. The igniting currents they
obtained are plotted on Figure 9-17b. The plotted points represent sud-
denly applied currents. Other points plotted, except for those of Stout and
Jones, represent gradually increased currents. Shepherd and Wheeler rea-
soned that failure to achieve ignition is enhanced by slow application of
current, which sets up strong convective air currents around the wire. Igni-
tion may not result, but flameless combustion at the wire may raise wire
temperature to the fusing point. Their report counsels rapid application of
current in order to heat the surrounding gas before either strong convec-
tion currents are established or fusing occurs. They noted that as wire
diameter is made smaller the ignition current approaches fusing current,
indicating higher temperature for small sources, as reported by other
workers. The 0.1-mm platinum wire fused in air at 1.95 A. In methane-air
mixtures the fusing current decreased. In mixtures of concentration greater
than 8.25%, 1.65 A, which ignited lower concentrations, fused the wire
without ignition. These results are in apparent contradiction to those of
Stout and Jones, but may be attributable to differences in wire material,
platinum being somewhat catalytic. Shepherd and Wheeler attributed the
decrease in fusing current to combustion on the surface of the wire. It is of
more than passing interest that these authors report previous work by
Wullner and Lehmann in 1886. These investigators had found that 0.15-
mm wire would ignite mixtures less than 8.3% methane but fused without
ignition in richer mixtures. With silver wires (no size reported) no ignition
was obtained. With copper wire, ignition occurred occasionally when the
wire fused.
Shepherd and Wheeler's data show that ignition by heated wires is not
sharply dependent on concentration, as is spark-ignition. In general, rich
mixtures of methane tended to cause small platinum wires to fuse without
causing ignition. The 0.3-mm wire caused ignition at the same current
over the range from 4.3% to 15.2% methane. The smaller wires caused
ignition over a more restricted range of concentrations, but at the same
current for all concentrations. A minimum igniting current at 8% methane
concentration was observed with tungsten wire, but the igniting current
increased only about 20% when concentration was changed to 5% or 12%.
Silver, in an excellent summary report, reviewed material from the

sources already discussed, as well as from sources, including his own
reports, which include the only extensive investigation of ignition by cop-
per wires known to the author.
Figure 9-17b Ignition Current versus Wire Diameter
Silver reported inability to ignite ether-air mixtures with fuse wires from
0.5, 1.0, and 5.0 A fuses. He used currents of 3.5 to 6 A in circuits of 4 to 8 V.
Current was applied slowly. In other experiments 22, 28, and 34 gauge
(0.64, 0.32 and 0.16 mm diameter) copper wires heated for 1 s to 90 s at con-
stant voltage did not ignite ether-air mixtures except when the nos. 28 and
22 wires fused. Fusion alone was not a sufficient condition for ignition.
Ignition occurred only when the wire fused less than 15 s after voltage was
applied. Ignition currents were 12.8 A for No. 28 wire and 45.8 A for No. 22
wire.
Ether-air and pentane-air mixtures were not ignited by 34, 36, or 37 gauge
(0.16, 0.125 or 0.114 mm diameter) copper wires heated slowly (1 to 15
min) from a constant current source. The exception was a No. 37 wire
formed into a spiral and heated by 4 A.
Silver has also reported ignition of pentane-air mixtures by Nos. 28 to 42

(0.32 to 0.063 mm diameter) copper wires heated from a constant-voltage
source. These data also show that ignition is dependent on fusion of the
copper conductor. For wires finer than No. 34, current at the instant of
fusion was 4 A. For larger wires, the current increased with wire diame-
ter—to 12 A for No. 28 wire. Because these tests were made at constant
voltage, initial current flow was significantly greater than the fusing cur-
rent that produced ignition.
Zborovszky has confirmed the necessity of fusing copper wire to ignite
methane-air. Two data points for tungsten wire given in her ISA paper are
included in Figure 9-17b.
The third edition of this book summarized the state of investigations of
hot wire ignition as follows:
(1) Ignition by fine wires will occur at current levels of a few amperes
only if the wire material has a high melting point.
(2) Ignition by copper wires is possible if currents are large enough to
fuse the wire and greater than 4 A, or if the wire is coiled to concen-
trate heat.
(3) Hot-wire ignition in instruments need not be seriously considered
unless fine wires of high melting-point material can carry relatively
high currents or unless wires of copper or other low-melting-point
material are coiled to concentrate heat. Except for these cases, ignition
in circuits typical of instruments will occur because of break-sparks in
resistive circuits at current levels less than those required for hot-wire
ignition.
Dr. Schebsdat made some unpublished data available in 1980 that belies
these conclusions. He obtained ignition of 4.4% diethyl ether mixtures
using 100-mm-long copper wires. He stated that the copper wires melted
after the ignition, and indeed, because of the heating by the combustion of
the gas mixture. He reported that tungsten and steel wires exhibited this
behavior only in diameters less than 0.06 mm. His data for copper wires
are shown below. Current was applied slowly to allow thermal equilib-
rium to develop. Were the current not raised slowly the data were not
reproducible. His data are summarized in Table 9-15.
Diameter—mm Current—A Voltage Drop In Wire—V

0.035 0.7 6.2
0.060 1.35 4.8
0.100 2.70 3.1
0.150 4.00 2.25
0.200 5.30 2.4
Table 9-15 Ignition of Diethyl Ether by Hot Copper Wires (Schebsdat)
Schebsdat's data contradicts all of the conclusions stated so positively in

the third edition. However, it does not change the overall conclusion
reached in this chapter (and towards which the third edition pointed): hot
wire ignition is not a significant risk in most electrical apparatus. Long
wires of the size used by Dr. Schebsdat (the smallest were comparable to 42
gauge) are not common. Short lengths of fine wire will almost always have
considerable thermal sinking associated with them, so ignition currents
will be higher than those in Table 9-15. Even the currents in Table 9-15 are
high relative to currents that cause arc ignition, except in low voltage resis-
tive circuits. Therefore ignition by hot wires is not often a practical
concern.
Ignition by Electronic Components
For many years ignition by electronic components was not considered to

be a matter of great concern. Curves in the 1971 SMRE Annual Report sug-
gested that no material other than carbon disulfide would be ignited by a
source of temperature lower than 600°C, the temperature required for
diethyl ether, SIT of 160 °C. The data were for ignition by a 5-mm long,
5 mm diameter, 12 W resistor. Similar results were reported by Silver in
CEAL Report No. 321. In 1984 this author also confirmed the necessity for
these high surface temperatures using components of a proprietary intrin-
sically safe electronics assembly. These results were not published. Data
were being made available that showed ignition of diethyl ether at tem-
peratures in the 230°C to 250°C range to be possible in the laboratory.
Considerable interest in hot component ignition was generated by work at

the Electrical Research Association which showed that components of the
size that might be used in intrinsically safe apparatus could ignite at tem-
perature about 40% higher than the accepted SIT of materials like diethyl
ether. These results have subsequently been confirmed by work at PTB,
CERCHAR, and SMRE.
At PTB most work was carried out in a 10% concentration of diethyl ether
in air. Cold flames were observed at higher concentrations. Ignitions took
place inside a horizontal glass tube, 9 cm in diameter, and a length of
either 20 cm or 50 cm to produce a chamber of 1 liter or 2.5 liter volume.
The chamber was maintained at 40°C. Investigations were carried out
either by introducing a premixed concentration of diethyl ether into the
chamber, or by injecting liquid, which then increased the concentration as
it evaporated. The ignition sources used ranged in area from 113 mm2 to
900 mm2. Cold flames appeared at temperatures of 220°C to 280°C and
ignition occurred at 240°C to 315°C. The power required ranged from 2.3
watts to 8 watts, and the power density ranged from 4 to 18 mW/mm2. In
general, lower temperatures, lower power densities and higher powers
are associated with increasing source area.
PTB published the values of both the power flux into the components and
the ignition temperature. Thus, one can approximate the amount of power
that is accounted for by convection and conduction from the component. If
one assumes the component to be a black body, one can calculate the max-
imum power flux caused by radiation from the hot component to its 40°C
surroundings and compare it to the total power flux calculated from the
input power and the component area. Table 9-16 summarizes the data for
four components.
Input Power Radiated

Temperature Fraction
Component Area - mm2 Flux Power Flux
°C Radiated
mW/mm2 mW/mm2
Transistor 113 26 298 5.5 0.21
2W Resistor 474 15.1 296 5.3 0.35
(Vertical)
2 W Resistor 474 12 278 4.7 0.39
(Horizontal)
1 W Resistor 184 13.2 315 6.2 0.47
Thick Film 280 8.2 243 3.5 0.43
Resistor
Table 9-16 Heat Transfer and Temperature Rise of Small Components
Widginton and Tomlinson experimented in a 500-mL beaker immersed in

a 40°C bath with metallic sources heated by resistors. Source area ranged
from 402 to 1,770 mm2. Measured ignition temperatures ranged from
226°C to 260°C, higher temperature being associated with lower area.
Their work also included determining the power required in typical com-
ponents to ignite diethyl ether. In most cases surface temperature was not
measured. Except for a thermistor and two pellistors of very small area,
0.44 mm2 for the thermistor and 3 mm2 for the pellistors, all components
tested dissipated at least 1.2 W to ignite the diethyl ether mixture. Wattage
to ignite carbon disulfide ranged from 60% to 100% of the values to ignite
diethyl ether. No obvious relationship to area of source is apparent. The
ignition temperature of carbon disulfide for the large contrived sources,
area greater than 400 mm2, ranged from 113°C to 198°C, compared to the
SIT of 100°C.
Table 9-17 illustrates that the contribution of nonradiative losses follows a

pattern similar to that shown by the PTB data. The flammable vapor is
diethyl ether.
Input Power Radiated

Temperature Fraction
Component Area - mm2 Flux Power Flux
°C Radiated
mW/mm2 mW/mm2
Pellistor 3 29 210 3.6 0.12
(activated)
Pellistor 3 54 320 6.5 0.12
(poisoned)
Steel plate 402 12.4 255 3.9 0.31
Copper plate 1000 9.4 233 3.2 0.34
Brass plate 1227 4.9 233 3.2 0.65
Brass plate 1770 4.1 218 2.8 0.66
Table 9-17 Small Component Temperature Rise and Heat Transfer (from Widginton)
The change in percentage contribution of radiated energy in the two tables

above is counterintuitive. One would expect that as temperature increases
the radiative fraction of energy, which increases as the fourth power of
temperature, if one can assume blackbody radiation by the component,
should certainly increase. These components are not blackbodies and
therefore radiate less efficiently; but, in general, one would expect both the
amount and efficiency of radiation from nonceramic surfaces to increase
as temperature increases.
The work at CERCHAR focussed on measuring the thermal resistance of

various components so ignition temperature and power required to ignite
carbon disulfide could be inferred. A few tests with diethyl ether showed
the required surface temperature of small components to be about 250°C
for either material.
In studies using real components as ignition sources apparent, surface

temperatures differing by of tens of degrees may be considered to be
essentially equal. In real components the surface temperature is not uni-
form. Typical resistors have a hot spot near the center of the resistor, and
most semiconductors have one near the die.
The 1976 article in Ex Magazine on the testing of intrinsically safe circuits at

BASEEFA included a table listing the temperature rise of small compo-
nents. This table is reproduced in Table 9-18. It was shown in a figure
accompanying the table that the estimated temperature rise using these
figures would be higher than that which would be measured. As surface
temperature increases the radiative losses increase so that the net rise in
temperature increases at a slower rate than the power input.
Temperature Rise - Maximum Test

Component Size - mm
°C/W Power W
Transistors
T05 metal 8 dia. × 6 high 130 (88) [50]
Miniature 2 dia. × 2 high 600 0.03
Integrated circuits -
dual-in-line
8 pin 105 2.0
14 pin 75 2.5
Diodes
BY100 6 dia. x 9 high 55 (61)
1N4148 2 dia. x 4 long 350
BZY88 zener 3 dia. x 6 long 260 (160) 0.7
Fixed Resistors
0.03 W 1.6 dia. x 6 long 330 (120) 0.03
0.125 W 2.5 dia. x 9 long 320 0.125
0.25 W 2.5 dia. x 10 long 180 (95) 0.25
0.33 W 4.0 dia. x 13 long 120 0.33
0.5 W 4.5 dia. x 10 long 100 (85) 0.5
1.0 W 6.5 dia. x 16 long 75 (66) 1.0
2.0 W 8.5 dia. x 21 long 45 2.0
Notes: Wattage ratings taken from manufacturer’s literature. Figures in ( ) show the effect
of mounting. Figures in [ ] show effect of mounting plus heat sink.
Table 9-18 Temperature Rise of Small Components
References
“The Electric Ignition of Firedamp,” Safety in Mines Research Board,
Paper No. 20, 1926.
Shepherd, W. C. F., and R. V. Wheeler, “The Ignition of Gases by Hot

Wires,” Safety in Mines Research Board, Paper No. 36, 1927.
Morgan, J. D., Principles of Ignition, p. 15, Pitman, London, 1942.
Viallard, R., “Donnés relatives à l'inflammation des mélanges gazeux com-

bustibles par l'étincelle électrique,” Journal de Chimie Physique, 40 (1943),
pp. 101–108.
“Intrinsic Safety of Electrical Apparatus,” Safety in Mines Research Board,

Paper No. 104, 1946.
“Intrinsic Safety of Electrical Apparatus,” Safety in Mines Research Board,

Paper No. 106, 1947.
Stout, H. P., and E. Jones, “The Ignition of Gaseous Explosive Media by

Hot Wires,” 3rd Symposium on Combustion, Flame, and Explosion Phenomena,
p. 329, Williams and Wilkins, Baltimore, 1949.
“Exploders for Simultaneous Firing of Shots,” Safety in Mines Research

Board, Paper No. 107, 1950.
“The Use of Break-Flash Apparatus No. 3 for Intrinsic Safety Testing,”

Safety in Mines Research Board, Report No. 33, 1951.
“Intrinsic Safety, A Resume of Recent Progress,” Safety in Mines Research

Board, Report No. 41, 1952.
Riddlestone, H. G., “Ignition of Methane-Air Gas Mixtures by Capacitive

Spark Discharges,” Report G/T 278, The Electrical Research Association,
1953.
Riddlestone, H. G., “The Effect of Series Resistance on the Ignition of

Methane-Air Gas Mixtures by Capacitive Discharges,” Report G/T 394,
The Electrical Research Association, 1956.
Müller, K., “Eigensichere Stromkreise,” Elektrotechnishe Zeitschrift (March

1957), pp. 177–182.
Riddlestone, H. G., “Ignition of Methane-Air Mixtures,” Tech. Report D/

T105, Electrical Research Association, 1957.
Holm, R., Electric Contacts Handbook, Springer, Berlin, 1958.
Gehm, K. H., “Sonderschutzarten für den Explosionsschutz von elek-

trischen Betriebsmitteln,” Elektro Welt, 1959, Nr. 10.
Litchfield, E. L., “Minimum Ignition-Energy Concept and its Applications

to Safety Engineering,” RI 5671, U.S. Bureau of Mines, Pittsburgh, 1960.
McKinney, A. H. “Electrical Ignition of Combustible Atmospheres,” Pro-

ceedings of 1960 Symposium on Safety for Electrical Instrumentation in
Hazardous Areas, Instrument Society of America, Pittsburgh.
Devins, J. C., and A. H. Sharbaugh, “The Fundamental Nature of Electrical

Breakdown,” Electro-Technology (February 1961), pp. 103–122.
Lewis, B., and G. von Elbe, Combustion, Flames, and Explosions of Gases, 2nd
ed., pp. 323 ff, Academic, New York.
Silver, S., “Tests of the Possibility of Ignition of Ether-Air Mixtures by the

Melting of Small Fuse Wires,” Internal Report FMP-61/54-EEC, Depart-
ment of Mines and Technical Survey, Canada, 1961.
Lord, H., D. W. Widginton, and R. L. Gordon, “The Use of Breaking Wires

to Produce Incendive Electrical Discharges,” IEE Conference Report Series
No. 3, Institution of Electrical Engineers, London, 1962.
Magison, E. C., “Low-Voltage Ignition of Hazardous Atmospheres,” ISA

Journal (July 1962), pp. 29–32.
Riddlestone, H. G., and W. Nethercot, “The Effect of Discharge Character-

istics on the Ignition of Gases by Disruptive Spark Discharges,” IEE
Conference Report Series No. 3, Institution of Electrical Engineers, Lon-
don, 1962.
Rogulski, W., “Ignition of Gas Mixtures by Electric Discharges Between a

Heated and a Cold Electrode,” Nature, 194, 4831 (June 2, 1962), pp. 858–
859.
Silver, S., “A Review of the Ignition of Flammable Gases or Vapors by Hot

Wires,” Internal Report FMP-62/186-EEC, Department of Mines and
Technical Survey, Canada, 1962.
Rae, D., et al., “The Size and Temperature of a Hot Square in a Cold Plane
Surface Necessary for the Ignition of Methane,” Report No. 224, Safety in
Mines Research Establishment, Sheffield, 1964.
Silver, S., and G. K. Brown, “Ignition of Pentane-Air Mixtures by Fusing

Small-Diameter Copper Conductors in a Circuit of Low Inductance
(22 μH),” Fuels and Mining Practice Division Report FMP-64/157-EEC,
Department of Mines and Technical Survey, Canada, 1964.
Thomas, V. M., “Design of Intrinsically Safe Apparatus for Use in Coal

Mechanical Engineer (May-June 1964), pp. 295–329.
Widginton, D. W., “Ignition of Methane by Electrical Discharges,” Safety

in Mines Research Establishment, Sheffield, 1965.
Berz, I., and M. A. A. Khan, “An Extension of Existing Design Parameters

Relevant to Intrinsically Safe Electrical Circuits,” The Mining Electrical and
Mechanical Engineer (May 1966), pp. 259–263.
Cartwright, J., “Technical Requirements for Intrinsic Safety,” Electrical

Review, March 1, 1968.
Widginton, D. W., “Some Aspects of the Design of Intrinsically Safe Cir-

cuits,” Report No. 256, Safety in Mines Research Establishment, Sheffield,
1968.
Bartels, A. L., “Ignition by Hot Surfaces in Electrical Equipment,” Proceed-

ings of Conference on Electrical Safety in Hazardous Environments, pp. 109–
114, London, 1971.
Zborovszky, Z. K., “Ignition Criteria for Resistive Circuits,” IEEE Transac-

tions on Industry Applications, Vol. IA-9, No. 4 (July/August 1973), pp. 404–
409.
Silver, S., “Ignition Risks of Overloaded Electronic Components (Resis-

tors) in Intrinsically Safe Circuits,” CEAL Report No. 321, Department of
Energy, Mines, and Resources, Ottawa, 1974.
Zborovszky, Z. and L. Catugno, “Evaluation of the Cadmium Disc Break-

flash in Testing Electrical Circuits for Safety in Explosive Atmospheres,”
Contract Report HO 111585, U.S. Bureau of Mines, 1974.
Bartels, A. L., “Ignition Energies of Methane-Air Mixtures as a Function of

Electrical Discharge Types,” IEE Conference Publication No. 134, pp. 148–
152, London, 1975.
Capp, B., and D. W. Widginton, “The Intrinsic Safety of Resistive Cir-

cuits,” IEE Conference Publication No. 134, pp. 43–47, London, 1975.
Butcher, G. H., P. S. Excell, and D. P. Howson, “The Ignition Hazard Due

to Radiation from Radio Transmitters,” IEE Conference Publication No.
134, pp. 143–146, London, 1975.
Zborovszky, A., “Ignition of Methane by Hot Wires,” ISA International

Conference, Houston, 1976.
Widginton, D. W., “Intrinsic Safety Reference Curves: Some Recent Con-

siderations,” IEE Conference Publication Number 296, pp. 51–54, London,
1988.
Johannsmeyer, Ulrich, “Zündung Explosionsfähiger Gemische durch

kurzzeitige Schliessfunken in kapazitiven Stromkreisen für die Zünds-
chutzart Eigensicherheit,” Dissertation, Technical University Carolo-
Wilhelmina at Braunschweig, 1984.
Schebsdat, Fritz, “Über den Einfluss von Fernmeldeleitungen auf Schlag-

wetter-und explosionsgeschützte eigensichere Stromkreise,” PTB Report
W-3, Braunschweig, 1974.
Vogt, G., “Die Schutzart Eigensicherheit,” Gluckauf, 19 Dec. 1969, 104 (26)
pp. 1193–1206.
Schebsdat, Fritz, “Zündverhalten von Netzwerken eigensicherer

Stromkreise mit konzentriert auftretender Kapazität und Induktivität bei
Gleichstrom,” PTB Report, W-11, Braunschweig, 1978.
Markworth, D., and F. Schebsdat, “Zündverhalten Kleiner Heisser Bau-

teile in Explosionsfähiger Atmosphäre,” PTB report W-25, Braunschweig,
1985.
Widginton, D. W., and R. Tomlinson, “Ignition of Flammable Gases by

Small Heated Bodies: Determinations of Ignition Temperatures for a Vari-
ety of Heated Bodies,” Internal SMRE Report, 1985.
“Etude du Danger Reel Resultant de L'echauffement de Petits Com-

posants dans un Circuit de Securite Intrinseque,” unpublished report,
Laboratoires du CERCHAR, 1981.
Fraczek, Jerzy, “Assessing Foundations of the Safety Level for Intrinsically

Safe Control and Measuring Apparatus When IEC Breakflash is Used,”
Automatyka, Politechniki, Silesia, 1977.
Bass, H. G., Intrinsic Safety, Quartermaine House Ltd., Sunbury, 1984.
“Ignition risk of Hydrocarbon Vapors by Hot Surfaces in the Open Air,”

Publication 2216, American Petroleum Institute, Washington, D.C., 1991.
Schebsdat, F., “Zündgrenzkurven kapazitiver Stromkreise mit unter-

schiedlichen ohmschen Widerständen im Entladestromkreis,” Ex
Zeitschrift, p. 55, Oktober 1989.
Clark, K., “The Certification and Testing of Intrinsically Safe Apparatus,

Associated Apparatus and Intrinsically Safe Systems,” Ex Magazine, p. 14,
June 1976.
Schebsdat, F., “Reduction of effective capacitance in intrinsically safe d.c.

circuits,” Ex Magazine, p. 32, June, 1976.
Zborovszky, Z. K., “New concepts and experimental results on the igni-

tion limits of CR circuits,” p. 123–128, IEE Conference Publication 218,
London, 1982.
Boon, D., “The Application of Industrial Apparatus in Zone 2 Hazardous

Areas,” pp. 77–79, IEE Conference Publication 218, London, 1982.
Cawley, J. C., “A Statistical Determination of Spark Ignition Safety Factors

in Methane Propane, and Ethylene Mixtures in Air,” RI 9048, US Bureau of
Mines, Pittsburgh, 1990.
Cawley, J. C., “Probability of Spark Ignition in Intrinsically Safe Circuits,”

RI 9183, US Bureau of Mines, Pittsburgh, 1991.
Marinovic, N., Protueksplozijska Zastita Elektricnih Uredaja, Skolska Knjiga,

S-Komisija, Zagreb, 1986.
Matasovic, M., “Research into the Probability of Ignition in Intrinsically

Safe Circuit,” pp. 58–73, S-Commission Bulletin, Vol. 6, No. 12, 1977.
CHAPTER 10
INTRINSICALLY SAFE AND
NONINCENDIVE SYSTEMS
Historical Review
As a result of the efforts of ISA Committee SP12, the 1956 edition of the
NEC for the first time included a paragraph recognizing the concept of
intrinsic safety. In the 1978 NEC, Paragraph 500-1 stated: “Equipment and
associated wiring approved as intrinsically safe shall be permitted in any
hazardous (classified) location for which it is approved, and the provi-
sions of Articles 500 through 517 shall not be considered applicable to
such installations. Means shall be provided to prevent the passage of gases
and vapors. Intrinsically safe equipment and wiring shall not be capable of
releasing sufficient electrical or thermal energy under normal or abnormal
conditions to cause ignition of a specific hazardous atmospheric mixture
in its most easily ignited concentration. Abnormal conditions shall include
accidental damage to any field installed wiring, failure of electrical com-
ponents, application of overvoltage, adjustment and maintenance
operations, and other similar conditions.”
Application of the principle of intrinsic safety to electrical equipment and

wiring for use in hazardous locations represents an approach to safety that
differs significantly from the methods discussed in earlier chapters. In an
intrinsically safe system, safety is afforded by the design of the system, not
by subsequent addition of protective measures. Safety, therefore, exists
throughout the life of the equipment, during maintenance, and in spite of
maintenance. Safety obtained by application of explosionproof enclosures,
on the other hand, is safety that is readily lost as a result of carelessness or
corrosion. Should a mechanic fail to properly install the cover after a
maintenance operation, safety is lost. Corrosion or mechanical damage to
the equipment can compromise flangegaps.
Although the term “intrinsically safe apparatus” is common, the term

“intrinsically safe” properly applies to systems or to “equipment and wir-
ing,” not to devices. Except for portable, battery-powered apparatus, only
a part of any system is so energy-limited that it can be classified as intrinsi-
cally safe. Any industrial instrument loop, for example, is somewhere
connected to the power line. The power line is ignition-capable, and not
intrinsically safe. Therefore this portion of the system is designed to be
suitable for its location. Most often this latter portion is located in a Divi-
342 Intrinsically Safe and Nonincendive Systems
sion 2 or nonhazardous location, but it could be located in a Division 1

location and pressurized or housed in an explosionproof enclosure. In
either case, the system must be designed to prevent the nonintrinsically
safe portion of the system from feeding enough energy to the intrinsically
safe portion of the circuit to cause ignition, even under failure conditions.
The term “intrinsically safe apparatus,” when used, means apparatus

intended for installation in a Division 1 location that will be intrinsically
safe when connected to “associated apparatus” located in the Division 2 or
nonhazardous location, which meets specified criteria. The two devices
together constitute an intrinsically safe system. Developed in Germany,
this view of intrinsically safe systems has some advantages. These are dis-
cussed later in this chapter.
The statement “Our ΔP/I transmitter is intrinsically safe” is, therefore, a

shorthand way of saying that the transmitter, and specified interface ele-
ments located in a Division 2 or nonhazardous location, form an
intrinsically safe system in which the transmitter may be located in Divi-
sion 1.
The first application of intrinsic safety was to dc signaling circuits in Brit-

ish mines. The concept of intrinsic safety was defined by Wheeler at what
is now the Safety in Mines Research Establishment (SMRE) after investiga-
tions into ignition of methane-air mixtures. These studies were motivated
by a disaster at Senghenyold colliery in 1913 alleged by some to be caused
by sparking signaling circuits. Such circuits, powered by dry cells, con-
sisted of bare wires that were shorted by hand or shovel to ring a bell to
signal the hoist operator. They had previously been considered to be safe.
In 1936 SMRE issued the first certificate for intrinsically safe equipment
for use other than in coal mines. The second such certificate was for a
transmission system connected to a ring-balance flowmeter for measure-
ment of blast furnace gas. By the mid-1950s, certification of intrinsically
safe equipment for industrial application was commonplace.
In the United States, initial work at the Bureau of Mines on intrinsically

safe mine apparatus, (though the term “intrinsically safe” was not used)
was started at about the time that British investigations began. Rules for
telephone and signaling devices were published in 1938. Mining Enforce-
ment and Safety Administration (MESA) requirements for intrinsically
safe systems can be found in Title 30, Code of Federal Regulations, Chapter
I, Subchapter D.
During the 1950s, increased use of electrical equipment in hazardous loca-

tions led to a review of installation practices. Worldwide interest in
intrinsic safety mushroomed. By the late 1960s, almost every industrial
country had either published a standard for intrinsically safe systems or
Intrinsically Safe and Nonincendive Systems 343
had drafted one. The major industrial countries also participated in the
work of International Electrotechnical Commission (IEC) Committee
SC31G, which was established to prepare an international standard for
intrinsically safe systems.
The IEC standard, published in 1977, is important because of the influence

of IEC draft proposals on standards of Canada, the United States, and the
nations of the European Community.
Adoption of intrinsic safety as a working technique in the United States

was spearheaded by ISA with the establishment of ISA Committee
8D-RP12, now SP12, in 1949. In the early years of committee activity
progress was quite slow. There was wide divergence of opinion about
designing a safe system and equally wide variance in practice in installing
a “safe” electrical installation. However, under the chairmanship of F. L.
Maltby a great deal of progress was made. Acceptance of the principle of
intrinsic safety in the United States can be attributed directly to the educa-
tional activities of Committee ISA SP12.
The committee's goal was to promote uniform, safe, and economical prac-
tices for use of instrumentation in hazardous areas. The committee early
decided to prepare recommended practices that were both safe and eco-
nomical, based on the application of sound engineering judgment, after a
thorough consideration of available technical information. These recom-
mended practices would contain few, if any, arbitrary rules and would not
be bound by existing practices which could not be justified by a reason-
able evaluation of pertinent technical considerations.
Implicit in this position is that if recommended practices are based on

thoughtful evaluation of applicable facts they can be used by anyone who
has studied and understands the pertinent technical literature. This was a
radical departure from the concept of a central approval body, which, at
that time, was accepted everywhere else in the world.
ISA RP12.2, “Intrinsically Safe and Non-Incendive Electrical Instruments,”

published in 1965, embodied the committee's philosophy. It was the first
detailed set of requirements for intrinsically safe systems to be published
in English. Though intrinsic safety approvals of industrial apparatus had
been granted in England for more than three decades, no published crite-
ria were available. Knowledge of British requirements was based on
British input to IEC committee work, the experiences of those who had
equipment approved in England, and examination of the design of
approved systems.
ISA RP12.2 was written during a period in the United States when
third-party approval of process-control instrumentation was rare. Pur-
chasers did not demand labeling or listing, and manufacturers found that
investment of time and money to obtain such approvals was not neces-
sary. The objective of RP12.2, therefore, was to provide requirements that
could be used by anyone who had familiarized himself with the pertinent
technical facts about electrical ignition. RP12.2 emphasized basic princi-
ples, not construction requirements. More importantly, RP12.2 defined
only a unique small subset of the larger universe of intrinsically safe sys-
tems. This subset is the class of intrinsically safe systems that are so
obviously incapable of causing ignition that safety can be assessed with-
out testing. Only analysis is necessary.
NFPA 493-1967 built on the foundation of ISA RP12.2. It recognized assess-

ment of safety based on analysis and defined reference curves like those in
RP12.2. NFPA 493-1967 also recognized approval by testing, using a modi-
fied PTB apparatus (see Chapter 9) with a brass disk, to give ignition at
levels similar to those of the British break-spark apparatus. In actuality,
this modified PTB apparatus ignited at current levels about 70% to 90% of
those determined with the break-spark apparatus in inductive circuits. In
low-voltage (10 to 40 V) resistive circuits, the brass disk apparatus pro-
duced ignition at about one-third the current level of the British apparatus.
The important contribution of NFPA 493-1967 was the inclusion of specific

tests and construction specifications that defined protective components
and design features whose failure in an unsafe manner is so improbable
that the probability of failure can be ignored when doing circuit analysis.
During this same period, it became apparent that many users of intrinsi-
cally safe systems did not feel competent to assess the safety of a system.
They demanded third-party certification. Instrument manufacturers began
to obtain UL or FM listings for intrinsically safe systems. Demand for list-
ings grew also because it was recognized that some manufacturers claimed
intrinsic safety, in ignorance, for systems that were not intrinsically safe.
OSHA regulations, which initially required UL or FM listing of electrical

apparatus, forced a complete change in the United States view about the
contents of NFPA 493. In 1975 a revised edition was published, based
largely on the draft IEC standard, which contained many specific con-
structional requirements and tests in order to provide a basis for uniform
assessment by third-party agencies. Publication of the 1975 version of
NFPA 493 signaled that the United States had joined the rest of the world
in requiring third-party approvals, at least for hazardous location equip-
ment. In recognition of the changed environment. ISA withdrew RP12.2 in
1976. NFPA 493-1978 defined additional tests and design criteria to mini-
mize the number of vague rules that led to differences in interpretation
among approval body engineers and design engineers. UL 913 was
updated to agree and both documents were submitted to ANSI to be
adopted as the ANSI Standard for intrinsic safety. After several years
NFPA decided that, because NFPA 493 was a design standard, unlike
other NFPA standards, which are standards for use of apparatus, it would
withdraw the document and transfer the responsibility for maintaining an
ANSI Standard for intrinsic safety to UL. UL, on its part, established a con-
sensus committee to administer this standard, on which no organization
has more than one vote. This committee produced a revision of UL 913 in
1988, which continued the trend toward greater agreement with IEC Pub-
lication 79-11. Many of the clauses in NFPA 493 and UL 913, which were
introduced to eliminate ambiguity of interpretation by specifying design
values that are acceptable without testing and tests to validate alternative
construction, have been adopted in Publication 79-11 and by that route
have been incorporated into CENELEC EN50020. As a result, conformity
to the requirements of most standards in effect around the world can be
achieved if one designs to the CENELEC standard. If one designs to U. S.
standards, one must only adjust for the safety factor difference, the differ-
ent view about infallible film resistors and transformers, and take into
account several clauses which were added to the CENELEC document in
1995 and which are not yet in UL 913.
One possible exception to this generality is the certification of intrinsically

safe designs in Japan. It is the author's understanding that Japanese stan-
dards were aligned with IEC standards several years ago, but that
certification by TIIS (Technical Institute for Industrial Safety) still contains
some elements of past practice, which are much more conservative than
are found in the current IEC standard.
Interpreting the Definition

All definitions of intrinsic safety use words such as “incapable of releasing
sufficient electrical energy under normal or abnormal conditions to cause
ignition of a specific atmospheric mixture.” Any standard on intrinsic
safety must answer three questions:
(1) What is “a specific hazardous atmospheric mixture”?
(2) What are normal and abnormal conditions?
(3) What level of energy is incapable of is igniting the specific

hazardous mixture?
These questions have been answered on the basis of the facts concerning
electrical ignition presented in Chapters 2 and 9 and the probabilistic phi-
losophy of hazard reduction outlined in Chapter 4. If the probability that a
circuit can ignite a specific atmospheric mixture is, for practical purposes,
reduced to zero, it meets the requirement of an intrinsically safe circuit. To
make the concept of intrinsic safety useful and economic, the require-
ments for intrinsically safe systems must eliminate all probable
combinations of circumstances that could cause ignition, without being

overly conservative.
When using a document such as UL913 it is difficult to visualize the diver-

sity of opinion regarding interpretation of the definition of intrinsic safety
which existed 36 years ago. To provide better understanding of the devel-
opment of standards for intrinsic safety, several examples of extreme
views that were rejected by ISA are summarized in the following para-
graphs. It was ISA's view that these positions were unnecessarily
conservative and uneconomical. In general, the positions taken in ISA
RP12.2 are now recognized in IEC and CENELEC standards, as well as in
NFPA 493.
(1) “All intrinsically safe equipment must be intrinsically safe for hydro-
gen atmospheres.” This view found some support in the United States
and was the basis for early German intrinsic safety standards. It is cer-
tainly a safe approach, although one glance at ignition energy data
shows hydrogen to be so much more easily ignited than common
gases or vapors, most of which are typified by NEC, Class I, Group C
and D materials, that to design all equipment to be intrinsically safe in
a hydrogen atmosphere is unnecessarily restrictive and uneconomical.
(2) “All possible combinations of failures must be considered in deter-

mining whether or not energy levels might rise above a permissible
safe level under fault conditions.” This statement, though superfi-
cially plausible, would eliminate a system from being considered
intrinsically safe if any portion of the system were connected to a
power line. One could always postulate for any equipment that does
not have a self-contained power source, a possible (though not proba-
ble) train of failures extending back to an ignition-capable source.
(3) “All possible modes of failure must be considered.” This, too, is a

plausible-sounding requirement, but reduced to absurdity this provi-
sion would require designing against damage by falling aircraft, run-
away trucks, or other calamities of ridiculously remote probability.
No other method of reducing hazard must meet this requirement.
The following paragraphs summarize the development of answers to the
three fundamental questions arising from the definition of intrinsic safety.
(1) Specific Hazardous Atmospheric Mixture. A specific hazardous atmo-
spheric mixture could be of any stated composition at any tempera-
ture or pressure. Unless otherwise stated, however, it is considered to
be the most easily ignited mixture of the designated vapor or gas with
air at normal ambient temperature and atmospheric pressure.
Although the concept of intrinsic safety is applied to devices installed
within processing vessels at other than ambient pressure or tempera-
ture, these installations require special consideration. Because most
reliable experimental data are limited to measurements at normal

temperature and pressure, it seemed prudent to limit ISA-RP12.2 to
that area of application where it was most urgently needed and for
which the experimental data are supported by experience. The
inflammable mixture to be considered is the one that is most easily
ignited because it is impossible to predict what the concentration of
the hazardous atmosphere may be. The most conservative assump-
tion must be made. A specific hazardous atmospheric mixture is
denoted because there are large differences in ignition energy among
gases and vapors. A system that is intrinsically safe for methane,
whose minimum ignition energy is about 0.25 mJ, would not neces-
sarily be safe for hydrogen, whose minimum ignition energy is about
0.02 mJ. Circuitry safe with regard to arc ignition for hydrogen,
would, of course, be safe for methane.
The British grouped flammable materials into the pentane, ethylene,

and hydrogen classes (analogous to NEC Groups D, C, and B) and
defined representative test mixtures of pentane, ethylene, and hydro-
gen in air for the three classes. RP12.2 and NFPA 493-1967 used this
same grouping. NFPA 493-1975 recognized the NEC groups and
defined test mixtures for the three groups. Ignition testing for Group
A equipment is done with hydrogen-air, as for Group B.
(2) Normal and Abnormal Conditions. Normal conditions comprise normal

operation of the equipment, including the effects of maximum supply
voltage and extreme environmental conditions within equipment rat-
ings. Normal conditions also include opening, shorting, and ground-
ing of all field installed wires connecting enclosures in the
intrinsically safe portion of the system. This provision recognizes that
in most industrial systems wiring external to devices is selected and
installed by the user, and because of its exposed location is more
likely to be opened, shorted, or grounded accidentally or deliberately
during the useful life of the system than wiring inside an enclosure.
Exceptions to this provision include systems where connecting means
are specially designed and installed in protected conditions to render
grounds, shorts, and opens so improbable that they can be treated as
ordinary failures.
This pessimistic view of external wiring failure, still very common,

was influenced by early concern with safety of mine signaling sys-
tems. In such systems external wiring was not insulated or protected
from opens or grounds. Shorts between wires were the means of sig-
naling. Some experts feel that normal wiring practice in large indus-
trial plants is designed to ensure reliable operation of the plant and
that perpetuation of this requirement for intrinsically safe systems is
overly restrictive.
The author does not foresee any relaxation of the assumption during
apparatus certification that opens, shorts, and grounds in external
wiring are normal. However, most experts recognize that there is a
limit of reasonable extrapolation of this assumption. They do not
assume that in a multiconductor cable all conductors must always be
considered to short together, for example.
ISA RP12.6, CENELEC Standard EN 50039, and a draft IEC standard

based on EN50039 give guidance on a level of construction of cables
in which interconnection of different intrinsically safe circuits need
not be considered (see page 397).
Abnormal conditions include all probable failures of components and

internal wiring.
Early British practice usually considered one fault in addition to

field-wiring malfunction. RP12.2 increased the number of faults to two
to provide a greater level of safety and compensate for the fact that the
Recommended Practice would be used by individuals for analyses of
safety without testing, not by organizations with a continuity of judge-
ment and experience. The two-fault criterion is also consistent with
the assumption that in Division 1 the atmosphere is always hazardous
(see Chapter 4) requiring that there should be two independent events
lying between a safe situation and an unsafe situation.
IEC defined two levels of intrinsic safety, ia and ib, for use in Zone 0
and Zone 1, respectively. In ia intrinsic safety two faults are consid-
ered. In ib intrinsic safety only one fault must be considered. CEN-
ELEC EN50020 continues this practice, as do the standards of most
countries outside North America. The adoption in the 1996 NEC of
Zones 0, 1, and 2 to supplement Divisions 1 and 2 makes it probable,
in the author's opinion, that North American standards will eventu-
ally adopt two levels of intrinsic safety. Most process control appara-
tus will likely continue to be designed for Zone 0, but there are some
applications of intrinsic safety in Zone 1 where ib intrinsic safety
might be useful Category ib intrinsic safety permits semiconductor
current limiting, whereas ia intrinsic safety may not utilize such cir-
cuits.
(3) Safe Energy Levels. In Chapter 4 it was shown that at voltage levels
likely to be found in intrinsically safe systems (CRTs excepted) the
amount of energy required for ignition is substantially greater than
the minimum ignition energy (MIE). Although there have been occa-
sional suggestions that MIE be used as a criterion for intrinsic safety, a
more reasonable practical view has prevailed.
The British criterion for ignition-capable energy levels was a series of con-
tact mechanisms, culminating in break-spark apparatus #3, described in
Chapter 9.
ISA RP12.2 used data derived from the British apparatus, supplemented
by German and United States data on capacitive discharge ignition, to
derive a series of reference curves for use in assessing ignition capability
of a circuit. German requirements were based on the PTB apparatus, later
adopted as the IEC standard test apparatus for intrinsic safety. It is consid-
erably more sensitive; that is, it produces ignition at lower current levels
in inductive circuits and resistive circuits than break-spark apparatus #3.
The IEC apparatus is now the reference apparatus for determining ignit-
ing levels of current and voltage in all standards for intrinsic safety, except
those of the former COMECON countries (Russia and former Eastern Bloc
nations). However, some manufacturers have obtained certification of
apparatus certified to CENELEC standards in these countries without any
significant issues being raised.
Specific Requirements for Intrinsically Safe Systems

The objective of the material presented in this section is to describe and
explain the most important requirements for intrinsically safe apparatus.
This presentation is based on the 2nd edition of CENELEC EN50020
because it is now the most detailed and lengthy standard. Many of its
requirements are likely to be adopted in IEC 79-11, and will also find their
way into national standards of many countries in the coming years.
EN50020 has added detail at a faster rate than other standards. Besides
adding detail to ensure that manufacturers and certifiers have the same
understanding of the requirements, the standard drafters must also try to
ensure that there is common interpretation of the standard in all 20 or
more certifying laboratories in the European Community. Unfortunately,
the drive for harmonization of interpretation has often meant that appara-
tus using a technique not discussed in the standard can not be certified
and bear the mark of the Community. If a laboratory in one of the nations
of the EC uses independent judgment of safety outside the specific provi-
sions of the standard its certification may be binding only in that nation.
For example, the use of sintered metal flame arresters in flameproof con-
struction could not be approved and bear the distinctive Community
mark until the 2nd edition of EN50018, which gives requirements and
tests, was legally adopted by the nations of the EC. Some years ago the
author submitted data to a European laboratory that demonstrated that a
plastic carrying case had such high surface resistivity that a charge placed
on it could not be removed by contact with a metallic electrode in quantity
sufficient to cause ignition. The high surface resistivity allowed current to
flow quickly to the electrode only from a small area near the electrode.
Thus, when the case was charged to 25 kV it could be discharged only in a

series of pulses, each of which contained less energy than the minimum
ignition energy of hydrogen. There was no discussion of the validity of the
test or the data. It was dismissed solely because the test was not one
described in the standard.
Though there is a need to maintain commonality of view among testing

laboratories, the legalistic use of standards, which is developing every-
where, often results in documents containing clauses that are no more
than statements of common engineering practice. They are inserted to
ensure that a particular method of assessment can be used. This situation
is exacerbated by apparent great differences in motivation among labora-
tories. Some laboratories have earned a reputation for looking with great
zeal, but not always with proportionate common sense, for reasons to find
a design nonconforming to a standard. Many clauses stating specific inter-
pretations have appeared in standards as reactions to such instances.
The material presented in this section is intended to illustrate the kinds of

requirements found in standards for intrinsic safety and the concerns that
motivated the requirements. This material should not be used as a basis
for design. Only the latest issue of the standard used by the assessing lab-
oratory is ever a valid basis for design and assessment.
Standards for intrinsic safety do not stand alone. In North America, intrin-
sically safe apparatus must also conform to the general purpose electrical
safety standards whose objective is to ensure freedom of electrical appara-
tus from shock and fire hazards. These standards deal with matters
common to all apparatus, whether for hazardous locations or not. In sys-
tems of standards that follow the IEC and CENELEC pattern, a General
Requirements document presents the rules for all explosion-protected
apparatus with regard to grounding, strength and tightness of enclosures;
construction of terminal boxes and terminals, conduit entries; and so on.
Conformity to the relevant general purpose electrical safety standard is
now a requirement in countries of the EC, by virtue of the need to apply
the CE mark. The laboratory certifying conformity to a standard for explo-
sion protection does not assess conformity to general purpose safety
standards. At present the apparatus manufacturer may assess conformity.
All standards for intrinsic safety include the following in more or less
detail:
• definitions related to intrinsic safety
• faults to be counted
• safety factors to be applied
• specific construction requirements for all intrinsically safe
apparatus
• construction and tests of devices or elements that are “protective”

or “infallible” components. (These are presumed never to fail in a
manner that decreases safety.)
• criteria for judging when failure of a component or spacing can
be counted as a fault in an assessment of intrinsic safety
• criteria for determining when failure of a component or spacing
is presumed to occur without being counted as a fault in an
assessment of intrinsic safety
• guidelines for assessing the temperature code of apparatus with-
out performing difficult temperature measurements
• specific tests or assessment methods
Definitions
Figure 10-1 illustrates the major definitions relating to intrinsic safety.

Every standard contains many more definitions of words and phrases
used in the standard, but those illustrated in Figure 10-1 are the most
important.
Figure 10-1 Elements and Parameters of an Intrinsically Safe System
An intrinsically safe device is composed entirely of intrinsically safe cir-

cuits. It can be described in terms of the maximum voltage, maximum
current, and maximum power that can be impressed on its terminals; and
by the effective capacitance; inductance; and optionally, L/R ratio, which
can be seen looking into the terminals. Figure 10-1 shows only a single pair
of terminals. There may be multiple pairs of terminals, each of which must
be characterized by the characteristic values noted. The symbols shown are
those used in Europe. These are recognized for use in North America but
at present one is more likely to see Voc, Isc, Pmax, Ca, La, and La/Ra on associ-
ated apparatus; and Vmax, Imax, and Pmax on intrinsically safe apparatus.
Associated apparatus contains circuits that are intrinsically safe and cir-
cuits that are not intrinsically safe. If the nonintrinsically safe circuits are
protected by some other technique, the apparatus may be located in a haz-
ardous location. Otherwise, it must be located in a nonhazardous location.
The intrinsically safe terminals are characterized by the maximum voltage,
maximum current, and maximum power that can be delivered at the ter-
minals and by the maximum values of capacitance, inductance, and
optionally, L/R ratio, which may be connected safely to the terminals.
Nonintrinsically safe terminals are characterized by the maximum rated
voltage that may be applied to them. This voltage, Um, in Figure 10-1, is
usually 250 V dc or rms for equipment for connection to the power line,
but it could be 24 V or other low voltage for other equipment.
In the former case, the standard will be based on some minimum prospec-
tive current available from the power line. In the latter case, the assessment
is carried out assuming presence of the low voltage, but the certificate or
control drawing may demand that the 24 V be supplied from a protective
transformer, constructed in accordance with the provisions of the stan-
dard. The intent of this mandate is to ensure that the transformer, which is
not a part of the certified apparatus, is of a quality of construction that will
reduce the possibility of a voltage higher than 24 V appearing at the termi-
nals to an acceptably low value. In countries following the IEC pattern,
apparatus is usually certified as an entity, without regard to the specific
design of the other apparatus to which it is connected in a system. Intrinsi-
cally safe apparatus is defined by the parameters indicated in Figure 10-1.
In principle, the use of these parameters to select devices to be used in a
system is straightforward if the intrinsically safe device is a two-terminal
device. It is only necessary to ensure that Uo, and Io are equal to or less than
Ui and Ii; and that Li, and Ci, are equal to or smaller than Lo and Co. If the
intrinsically safe apparatus is a two-wire device and both wires are iso-
lated from ground, an associated apparatus (barrier) must be installed in
each wire. How the equivalent parameters are to be derived is not stan-
dardized, and there can be differences of opinion among experts about
what combinations of barriers are suitable. In North America it has now
become the practice for the manufacturer of the intrinsically safe or associ-
ated apparatus to provide a “control drawing” that provides the details of
the permissible interconnections and any special installation requirements
peculiar to that apparatus. This control drawing is assessed and verified by
the certifying agency as part of its examination of the product.
Standards following the IEC pattern define two categories of intrinsic

safety.
(1) Category ia apparatus will not cause ignition when the maximum per-
mitted values are applied to its terminals:
(a) in normal operation with application of those non-countable

faults which give the most onerous condition
(b) in normal operation with application of one countable fault and

those non-countable faults which give the most onerous condi-
tion,
(c) in normal operation with application of two countable faults and

those non-countable faults which give the most onerous condi-
tion.
Normal operation means that the apparatus conforms to the design

specification supplied by the manufacturer and is used within electri-
cal, mechanical, and environmental limits specified by the manufac-
turer. Normal operation also includes open circuiting, shorting, and
grounding of external wiring at connection facilities.
When assessing or testing for spark ignition the safety factors to be

applied to voltage or current are 1.5 in conditions a and b, and 1.0 in
condition c. (These factors should properly be called “test factors.”
The real safety of intrinsic safety is inherent in the use of the sensitive
IEC apparatus to attempt to ignite the most easily ignitable mixture of
the test gas with hundreds of sparks. This combination of conditions
is many times more onerous than any likely to occur in practice.)
(2) Category ib apparatus is assessed or tested under the conditions of a

and b above, with a safety factor on voltage or current of 1.5 in the
condition of a and b.
U. S. standards for intrinsic safety apply a test factor of 1.22 to current

and voltage (1.5 on energy) and recognize only a level of safety com-
parable to ia. The lower test factor was a reaction to the replacement
of the British breakflash apparatus by the PTB apparatus as the stan-
dard IEC test apparatus. The permissible currents and voltages in
most circuits were thereby decreased by a factor of 2 or more, because
of the higher sensitivity of the PTB apparatus compared to the British
break-spark apparatus. No compensating adjustment was made in
the IEC standard. Although few manufacturers design to the U. S. test
factors, it is difficult to change the standard to conform to the IEC val-
ues because that would result in some designs being decertified for no
reason related to safety.
When assessing either category of intrinsic safety no safety factor is

applied to voltage or current when assessing or measuring surface tem-
perature. It is recognized that there is considerable safety factor inherent
in the standard laboratory determination of the ignition temperature of a
gas or vapor. The value is much lower than the temperature of a surface
that will cause ignition in industrial conditions.
Simple apparatus must conform to the requirements of intrinsically safe

apparatus, but need not be certified, and need not bear the markings
required of intrinsically safe and associated apparatus. The following are
examples of simple apparatus.
• passive components (e.g., switches, junction boxes, potentiome-
ters and simple semiconductor devices)
• sources of stored energy with well-defined parameters (e.g.,
capacitors or inductors, whose values are to be considered when
evaluating the safety of the intrinsically safe system)
• sources of generated energy (e.g., thermocouples or photocells)
that generate no more than 1.5 V (1.2 V in some standards),
100 mA, or 25 mW; any inductance or capacitance in these
sources shall be considered as in b
In assessing simple apparatus the following shall be considered.
• Safety shall not be achieved in simple apparatus by inclusion of
voltage or current limiting devices.
• Circuits shall not increase the voltage or current available to the
simple apparatus.
• In most cases simple apparatus must withstand the 500 V isola-
tion from earth demanded of intrinsically safe apparatus, but this
requirement would not normally be demanded of thermocouples
if they are to be connected to a device which galvanically isolates
the thermocouple circuit from other circuits.
• Enclosures should meet the limitations on light metals and plas-
tics of the “General Requirements” document.
• A Temperature Code shall be assigned to simple apparatus if it is
used in a classified location.
Requirements for All Intrinsically Safe Apparatus

The requirements in the following paragraphs are in addition to those
imposed by the “General Requirements” document. It is intended that a
device must comply only if intrinsic safety depends on it. For example, if a
manufacturer uses encapsulation to provide environmental protection for
his circuitry, but does not claim any reduction of spacings or reduction in
temperature classification because of it, the requirements of the clauses of
the standard that discuss encapsulation are not relevant. It has been sug-
gested in many committees that every page of a standard for intrinsic
safety bear the caveat, “These requirements apply only if intrinsic safety
depends on them.” One of the major causes of controversy between manu-
facturers and examining engineers has been application of a clause to a
situation where intrinsic safety does not depend on conformity to the
requirements of the clause.
Enclosures
In principle the protection in intrinsically safe apparatus and associated

apparatus is provided by the circuitry itself, not by the enclosure. How-
ever, if access to conducting parts might impair safety, an enclosure of at
least IP20 per IEC Publication 529 shall be provided. The actual degree of
protection required depends on the intended application. For dusty or
mining applications IP54 might be required. In North America, outdoor
installation usually requires NEMA 4 or CSA Enclosure 4 degree of
tightness.
Provision for External Connections
Terminals for intrinsically safe circuits and terminals for nonintrinsically

safe circuits shall be separated from one another either by 50 mm spacing
or by grounded or insulating partitions. The partitions shall reach to
within 1.5 mm of the enclosure walls or shall ensure 50 mm distance
between terminals measured along any path around the partition. Non-
metallic partitions shall be at least 0.9 mm thick, and metallic partitions
shall be at least 0.45 mm thick. Alternatively a force of 30 N (Newtons)
applied with a 6-mm-diameter rod in the center of the partition shall not
deform it so that the separation function is defeated. Metallic partitions
shall be capable of conducting any fault current that may occur without
burn-through or loss of the grounding connection. Assessment of the
effectiveness of partitions shall take into account the possible position of
wires that may come loose from terminals.
The clearance between conducting parts of terminals of separate intrinsi-

cally safe circuits shall be judged in accordance with the spacing rules for
other creepage and clearances given in Table 10-1, but the metallic parts of
conductors connected to them shall be separated by a minimum of 6 mm.
All bare conducting parts of conductors connected to terminals shall be at
least 3 mm from earthed metal or other conducting parts unless intercon-
nection has been considered in the intrinsic safety analysis.
Voltage - Volts 10 30 60 90 190 375 550 750 1000 1300 1575

Clearance - mm 1.5 2 3 4 5 6 7 8 10 14 16
Separation through 0.5 0.7 1 1.3 1.7 2 2.4 2.7 3.3 4.6 5.3
casting compound -
mm
Separation through 0.5 0.5 0.5 0.7 0.8 1 1.2 1.4 1.7 2.3 2.7
solid insulation** - mm
Creepage distance in 1.5 2 3 4 8 10 15 18 25 36 49
air - mm
Creepage distance 0.5 0.7 1 1.3 2.6 3.3 5 6 8.3 12 13.3
under coating - mm
Comparative ia * 100 100 100 175 175 275 275 275 275 275
tracking index ib * 100 100 100 175 175 175 175 175 175 175
* No minimum CTI specified for voltages up to 10 V
** Solid insulation is extruded or molded, not poured
Table 10-1 Creepage, Clearances, and Separation Distances
Plugs and Sockets
Plugs and sockets for connection of external nonintrinsically safe circuits

shall be separate from and noninterchangeable with those for intrinsically
safe circuits. If intrinsically safe apparatus or associated apparatus uses
more than one plug, and socket and interchange would affect intrinsic
safety, they shall be made noninterchangeable by keying or other design
feature, or shall be marked so that interchange is obvious.
If a plug or socket is not supplied with wires installed, the normal spac-
ings for terminals shall apply. However, if a special tool is required for
fabrication, and it ensures that a strand of wire will not come free, only the
normal spacings between conductive parts in Table 10-1 apply.
Cables
Cables supplied as an integral part of apparatus shall withstand a pull test

of 30 N for 1 hour without breakage of the cable or movement of the termi-
nations that might invalidate intrinsic safety by interconnecting circuits,
breaking redundant connections, or similar means.
Dielectric Strength
The insulation between an intrinsically safe circuit and a frame or enclo-

sure that may be grounded shall withstand a test at a voltage of twice the
voltage of the intrinsically safe circuit or 500 V RMS, whichever is the
greater. The CENELEC standard prescribes 5 mA max current or the appa-

ratus must be marked with an X. This insulation requirement is to
minimize the likelihood that intrinsically safe circuitry connected to a
grounded enclosure could be subjected to invasive currents because of dif-
ferences in ground potential between the enclosure and some other
grounded point in the system. The concern arises because high-current
ground faults flowing through the resistance to earth in a power system
can elevate the potential at grounding electrodes by as much as half the
power system voltage before protective relaying or fusing deenergizes the
grounded line. This potential difference could, in principle, cause high cur-
rents to flow through intrinsically safe circuits to a ground connection at
lower potential. The standard gives no guidance to the user on the precau-
tions that one should take when a device is marked X for this reason, but
presumably one would either use galvanically isolated associated appara-
tus or ensure that power system ground faults would not cause dangerous
currents to flow. North American standards specify no maximum current;
only, “no breakdown,” as in general purpose electrical safety standards.
Relays
If the coil is connected to an intrinsically safe circuit and the contacts are in
a nonintrinsically safe circuit, the relay shall be operated within its ratings,
but shall not switch more than 5 A, 250 RMS, or 100 VA. If the normal val-
ues for creepage and clearances are doubled, the relay may be operated at
10 A or 500 VA. If higher values are to be switched, the coil and contacts
must be separated by partitions, as discussed above. However, the dimen-
sions must take into the account the ionization produced by the relay,
which would generally require greater than normal creepage and clear-
ances. Similar partitions are required between contacts that are in
intrinsically safe circuits and contacts on the same relay that are in nonin-
trinsically safe circuits. The design must be safe against the dislodging of
broken or damaged contacts.
Protection Against Polarity Reversal
Protection shall be provided within intrinsically safe apparatus to prevent

invalidation of intrinsic safety by reversal of polarity of the supply to the
apparatus. A single diode is considered acceptable. (In all forms of explo-
sion protection, it is assumed that the apparatus is installed initially in
accordance with its instructions. Polarity reversal would be inadvertent
and short term. Thus, one need not insist on redundant protection.)
Grounding Conductors, Connections, and Terminals
Any grounding conductor necessary to intrinsic safety shall be capable of

carrying the maximum possible continuous current that can flow under
normal or fault conditions. If the grounding connection is carried through
a connector, three independent connections in parallel are required for ia
intrinsic safety and two are required for ib. If a connector can be removed
at an angle, one connection shall be near each end of the connector.
Terminals for connection of grounding conductors shall be robust; not

likely to loosen under the influence of time, vibration, or temperature; and
shall provide good contact even if stranded wires are used in contacts
designed for direct clamping solid wire. Terminals intended for clamping
stranded conductors shall have a resilient intermediate clamping member.
Terminals for clamping wires up to 4 mm2 area shall also clamp smaller
wires effectively.
Terminals shall not have sharp edges that could damage conductors, be
permanently deformed or damaged by normal tightening, or transmit
pressure to conductors through insulating materials.
Type e (increased safety) terminals are presumed to meet these require-

ments, and experience shows that terminals that meet the requirements of
UL or CSA for general purpose use will also meet them.
Encapsulation
If a casting compound is used to exclude a flammable atmosphere from

contact with such components as fuses, piezoelectric or energy storage
devices with their suppression components, or relays, the requirements
below apply. If the purpose is also to lower the surface temperature of a
component, the volume and thickness must be sufficient to accomplish
that objective.
These requirements do not apply where casting compound is used for

purposes not related to intrinsic safety, such as isolation of circuitry from
the environment to improve function.
Casting compound shall have a temperature rating, as specified by its

manufacturer, which is at least equal to the highest maximum tempera-
ture reached by an encapsulated component. (Some assessors must be
reminded that the temperature rise of a component surface in encapsula-
tion is considerably lower than the temperature rise measured in free air.)
Alternatively, it may be demonstrated that a component temperature
higher than the rating does not adversely affect intrinsic safety.
The encapsulant must have a CTI (Critical Tracking Index) required by the
creepage and clearance table if bare conductors protrude from the com-
pound. Critical Tracking Index is a measure of the likelihood that a
material will “track,” that is, permit conduction across its surface between
two exposed conductors. The standard test is given in IEC Publication 112.
Only hard materials, such as epoxy, shall have an unprotected free sur-
face, thus becoming part of the enclosure. The surface shall not
permanently deform more than 1 mm or be damaged when a 6-mm-diam-
eter rod applies a 30 N force perpendicular to the surface. The surface
shall also sustain an impact of 2 J applied by the standard impact test
apparatus of the General Requirements standard without damage (except
minor surface cracks) or permanent deformation. Group I (mining) appa-
ratus shall be subjected to 20 J if the surface is part of the external
enclosure and if encapsulation is intended to exclude a flammable
atmosphere.
Requirements for Features and Components on which

Intrinsic Safety Depends
Distance Between Conductive Parts
Table 10-1 shows values of separations from which the required separa-
tion can be determined in the following specific instances:
• between intrinsically safe and nonintrinsically safe circuits

• between different intrinsically safe circuits
• between a circuit and grounded or isolated metal parts.
The values in the table are values that can be presumed not to fail to a
lower insulation resistance. Separations of one-third the tabular values, or
greater, shall be assumed to short circuit with one countable fault. Separa-
tions less than one-third the tabular value are presumed to short-circuit
without counting a fault, if that impairs intrinsic safety.
The tabulated values are values negotiated in committees over the years,
starting with the spacing requirements for Type e construction, which
were rejected as being impractically large for low voltage electronic
designs. In early standards there were no columns for voltages less than 90
volts. The present tabulated values bear no relationship to the creepage
and clearance values in IEC 1010, for example, which were based on IEC
Publication 664 values resulting from many years of research by IEC
SC28A. They are larger than the values in the latest Type e standard.
The voltage used for determining the required spacing is the sum of the
peak voltages between the conductors, based on the maximum applied
voltage specified by the manufacturer, the peak value of the nominal
mains voltage; or a voltage generated within the apparatus. If the voltage
in one circuit is less than 20% of the other, it may be neglected. Transient
voltages that may exist before a fuse opens a circuit shall be considered
when assessing clearance but shall not be considered when assessing
creepage distance. Creepage distances are assigned assuming that the
voltage is present for an extended period.
The values in Table 10-1 do not apply if a grounded conductor or partition

separates an intrinsically safe circuit from a nonintrinsically safe circuit
and if grounding of the intrinsically safe circuit does not compromise
intrinsic safety. Creepage and clearances shall be measured around an
insulating partition between conductors if it has the thickness or strength
noted on page 355. The thickness and strength of nonmetallic partitions
also must meet the requirements on page 355. Partitions that do not meet
the thickness or strength requirements shall be ignored.
“Clearance in air” and “creepage” have the same meanings as in general

purpose electrical safety standards. The CTI of the surface between con-
ducting parts exposed to air, whether it be solid insulating material,
cement, casting compound, or coating, shall have a value at least that
specified in Table 10-1. If the creepage distance between conductors
includes more than one distance, as when a conducting part is interposed,
any distances less than 1/3 the tabular value shall not be counted. Some
experts would like to see the number of adjacent spacings below the 1/3
value, which need be presumed to be shorted, as in a connector, limited to
the number that equals the total creepage distance. Where pin-to-pin spac-
ing in a connector presents a problem, common practice is to skip a pin or
ground it.
The values for creepage distance under coating apply between conductors
that are covered by two coats of a sprayed material or by a single coating
applied by dipping, brushing, or vacuum-impregnating. A solder mask
that is not damaged by soldering may be counted as one of the coatings.
Although the application of the tabular distance to flat, rigid, printed cir-
cuits is straightforward, negotiation with the assessor may be required if
the circuits are flexible. The most common difficulty with assessments of
creepage distances under coating is that the designer forgets that compo-
nent leads, which protrude from the coating either on the bottom of board
or on the top, must be judged on the basis of the rules for creepage distance
in air.
Distance through solid insulation applies to extruded materials as on a

wire, insulating sheet or tubing, or molded materials. These parts are pre-
fabricated, as compared to casting compound, which is poured.
Casting compound shall have a temperature rating at least equal to that of

any encapsulated component and a CTI per the tabular value if leads exit
the compound. It shall adhere to parts and leads, except if they are totally
enclosed by the compound. The compound shall be identified by its
generic name and manufacturer's type designation. Fault conditions
within the compound shall be assessed but spark ignition shall not be con-
sidered. (An example of an obvious fact, which is common in modern
standards to ensure that it is indeed obvious.)
The minimum distance from conductive parts and components and the
free surface of the compound shall be at least half the value in Table 10-1,
but not less than 1 mm.
The failure of an encapsulated component such as a semiconductor, which

has undefined internal spacings, is to be considered as a single countable
fault.
Insulation on wiring within encapsulation is considered to be solid insula-

tion, and the values in the table for separation apply to the sum of the
radial insulation thicknesses of two conductors, either side-by-side or in
cables. If a nonintrinsically safe circuit is in one core, an intrinsically safe
circuit is in another core, and one is enclosed in an earthed (grounded)
screen, the tabular value does not apply. In 'ib' apparatus the tabular value
may be ignored if the intrinsically safe circuit wire has insulation that will
withstand a 2,000 V dielectric strength test.
EN50020 includes useful guidance for calculating an equivalent separation

distance if the total separation consists of different kinds. One refers all
separations to the same line of the table and adds them. Spacings less than
the 1/3 tabular values are ignored. For example infallible clearance
between circuits with potential difference of 250 V RMS, 354 V peak, is
6 mm; the infallible distance through casting compound is 2 mm, and
through solid insulation, 1 mm, factors of 1/3 and 1/6, respectively. A
wire with a peak voltage of 375 V encapsulated to a depth of 2 mm can be
considered to be separated by the required distance from another conduc-
tor of voltage less than 75 V (the 20% rule). Two wires with a potential of
375 V peak between them, both insulated with 1/2-mm insulation, are sep-
arated by the equivalent of 6 mm clearance in air. If one wire has only 1/
4-mm insulation it would be necessary to ensure that the insulated wires
are spaced 1.5 mm apart; if both wires have only 1/4-mm insulation, the
spacing between insulated wires must be 3 mm.
Ratings of Components
Except for devices such as transformers, fuses, thermal trips, relays, and
switches, components on which intrinsic safety depends shall not be oper-
ated above two-thirds rated voltage, current, or power, taking into
account mounting conditions and ambient temperature specified by the
manufacturer of the component. In general, normal commercial ratings
are accepted, but testing of semiconductors in the specific mounting used
in a design may be necessary to determine the allowable power dissipa-
tion relevant to the rated junction temperature. This may be higher than
the nominal commercial rating. Components not used in accordance with
the 2/3 factor on rating shall be considered to fail without counting a fault.
Connectors
Connectors shall not be interchangeable with other connectors if a hazard

will result from interchange, or, alternatively, the connectors must be
identified so that interchange is unlikely. Failure of a connection to high
resistance is a countable fault. If a connector carries grounded circuits nec-
essary for intrinsic safety the circuits shall be redundant in accordance
with rules for counting faults.
Fuses
Where a fuse is used to protect another component the current flowing in

the fuse shall be assumed to be 1.7 times the fuse rating. (This 1.7 factor is
based on IEC fuse standards. Other standards for intrinsic safety may use
a different factor because of different ways of rating fuses.) The time-cur-
rent characteristic of the fuse shall ensure that the transient rating of the
protected component is not exceeded. If this characteristic is not available,
a type-test of the protected component is required. For zener diodes this
consists of subjecting the diode to a series of 5 rectangular 50 ms pulses of
current repeated at 20 ms intervals. The magnitude of the pulses is the
quotient of the stated Um and sum of the cold resistance of the fuse and
any infallible resistance in series. The voltage of the diode before and after
the tests, when measured using the diode manufacturer's stated reference
current, shall not differ by more than 5%. If semiconductor current limit-
ing is used, the highest voltage measured during this test shall be used as
input to the semiconductors, whose performance after the test shall con-
form to the manufacturer's specifications.
If a fuse is encapsulated the encapsulant shall not enter the fuse. Fuses that
protect components shall be accessible only by opening the apparatus
enclosure. The rating of a fuse shall be at least Um or Ui, but it need not
meet the spacings of Table 10-1. A fuse shall be rated to interrupt the larg-
est current that may flow in the circuit in which it is installed. For 250 V
mains, the prospective current shall normally be considered to be 1,500 A.
If a resistor is used to limit current to within the breaking capacity of a
fuse it shall have the following ratings:
• Current - (1.5)(1.7) Fuse rating
• Voltage - Um or Ui
• Power - 1.5(1.7 fuse rating)2(resistor value)
Primary and Secondary Cells and Batteries

Cells and batteries which may explode if short-circuited or reverse-
charged shall be used only within the safe ratings declared by the manu-
facturer, taking into account the results of applying faults to the circuit.
Cells shall not leak electrolyte unless they are enclosed to prevent damage
to components on which intrinsic safety depends. Encapsulated batteries
or batteries not declared by the manufacturer to be sealed shall be tested
by short-circuiting and charging them within the manufacturer's limits to
determine that no leakage occurs.
For assessment and test of spark-ignition capability the maximum battery

voltage is that measurable just after a full charge, following several cycles
of discharges into a short circuit and recharging. The voltage to be used
for component surface temperature assessment is the nominal voltage of
the cell specified by the manufacturer. Table 10-2 lists these voltages for
common battery types.
Peak voltage Nominal voltage for

IEC type Cell Type for assessing component temperature
spark hazard assessment
K Nickel-cadmium 1.5 1.3
Lead-acid (dry) 2.35 2.2
Lead-acid (wet) 2.67 2.2
L Alkaline-manganese 1.65 1.5
M Mercury-zinc 1.37 1.35
N Mercury-manganese 1.6 1.4
dioxide-zinc
S Silver-zinc 1.63 1.55
A Zinc-air 1.55 1.4
C Lithium-manganese dioxide 3.7 3.0
Zinc-manganese dioxide 1.725 1.5
(zinc-carbon Leclanche)
Nickel-hydride 1.6 1.3
Table 10-2 Typical Cell Voltage
Battery replacement in associated apparatus shall not affect the intrinsic

safety of the apparatus. A current limiting device need not be a part of the
battery assembly. If the battery of intrinsically safe apparatus is to be
replaced in a hazardous location, the battery and any current limiting
device shall form an integral unit. Only the intrinsically safe terminals and
the charging terminals (if provided) shall be accessible.
If it is not intended to be replaced in a hazardous location the battery may

be constructed as above, or it may be housed in an enclosure with special
fasteners. The battery shall be replaceable without impairing intrinsic
safety. If the apparatus is portable it shall be subjected to a drop test to
ensure that the battery is not ejected or does not become a hazard when
dropped. The test used in most standards is a series of drops on different
corners of the apparatus from a height of 1 m. A label shall warn against
changing the battery in a hazardous location.
An infallible resistor or redundant diodes shall be installed in series with

the charging contacts to protect against flow of ignition-capable current if
the contacts are shorted. For category ia, three diodes shall be used; two
are required for category ib. The battery charger shall be associated appa-
ratus, or a fuse shall protect the diodes or resistor (so the protective diodes
or resistor are not destroyed by a power surge from a defective charger,
for example). The fuse must be encapsulated unless it conducts no current
under normal or fault conditions.
Surface temperature and spark-ignition capability of batteries shall be

assessed at external terminals and surfaces if the batteries are sealed
(gas-tight or valve-regulated). Also judged in this manner are batteries of
equivalent construction not requiring addition of electrolyte during the life
of the battery. They shall be housed in seamless enclosures or an enclosure
with seams joined by fusion, eutectic methods, adhesives, or welding and
having elastomeric or plastic sealing devices retained by the structure and
held permanently in compression (e.g., washers or “o”-rings). Alterna-
tively the batteries may be encapsulated using a compound suitable for use
with the electrolyte. If the battery is an assembly of identifiable individual
cells, they shall be considered to be individual components for the purpose
of testing, and the failure of each cell shall be counted as a single fault.
Spark ignition tests shall be carried out at the terminals of sealed cells. If
the internal resistance is claimed as part of the protective intrinsic safety
resistance, the value shall be obtained from the battery manufacturer, or
the worst case value from tests of 10 batteries shall be used.
When testing to determine the maximum surface temperature of a battery,

all external current-limiting devices shall be shorted. If the battery con-
tains an internal fuse or other current-limiting device, special batteries
shall be obtained from the manufacturer, without these elements, to simu-
late the effect of an internal fault. (To the author's knowledge shorting of
current-limiting resistors for the purpose of assessing temperature rise has
not yet been practiced in North America.)
Semiconductors
The effects of transients generated in intrinsically safe apparatus and in its

power source shall be ignored. In associated apparatus, however, a semi-
conductor shall be capable of withstanding the peak of any applied ac
voltage divided by any infallible series resistance.
Semiconductors used as shunt voltage-limiting devices shall carry without

open circuiting, 1.5 times the current that would flow in the branch of the
circuit where they are installed if they fail to a short circuit. Transients
shall be taken into account. Confirmation for semiconductors other than
zener diodes may be from the manufacturer's data sheet if the forward
current rating is at least 1.5 times the maximum short circuit current. For
zener diodes, the power rating in the zener mode shall be at least 1.5 times
the power dissipated, and the current rating in the forward mode shall be
1.5 times that which would flow if they were short circuited.
In ia intrinsic safety controllable semiconductors (such as thyristors) tran-

sistors and regulators may be used as shunt voltage-limiting devices if
both input and output of the device are intrinsically safe, or if it can be
shown that they will not be subject to transients from the power supply
network. Two devices shall be considered to be an infallible assembly. In
associated apparatus three thyristors may be used if transient conditions
are taken into account. Such crowbar circuits shall be tested to determine
that the energy passed by the crowbar circuit as it operates does not
exceed:
Group IIC 20 μJ
Group IIB 80 μJ
Group IIA 160 μJ
Group I 260 μJ
This determination must usually be made using an oscilloscope, because

the spark-test apparatus is not appropriate for testing the let-through
energy.
Three series blocking diodes in ia apparatus are permitted, but controlla-

ble series devices shall be used as current-limiters only in ib apparatus.
The reason given for restricting the use of series semiconductor devices in
ia apparatus is that a brief transient might cause ignition in locations
where the explosive atmosphere is present continuously or frequently.
Piezoelectric Devices
If they are accessible in service, piezoelectric devices shall be tested to

determine the voltage output when subjected to the higher values of
impact specified in the General Requirements, usually 7 J. If there is a
guard, the impact shall be applied to the guard when the device is
mounted in its normal operating position. The higher voltage of two mea-
surements shall be used with the measured capacitance of the device to
calculate stored energy. It shall not exceed:
Group I 1500 μJ
Group IIA 950 μJ
Group IIB 250 μJ
Group IIC 50 μJ
These energy values are higher than those for series connected semicon-
ductors because the ignition mechanism of a piezoelectric device is known
to be a capacitive discharge, whereas the mechanism of ignition by the
pass-through energy of series semiconductors is undefined.
Impact testing shall not damage any components that limit the stored
energy.
If the user of the apparatus must take precautions to prevent the energy
from exceeding the above values these shall be stated as special conditions
and the apparatus shall be marked with an X.
Assessment of Faults in Components and Connections
A component operated above 2/3 of its rating shall be assumed to fail

without counting a fault.
A series of faults consequent to a single fault shall be counted as one fault.
Resistors are presumed to fail to any value between open and short circuit.
Semiconductors can short, open, or be subject to any condition caused by

the failure of another device. Failure to a condition that the semiconductor
dissipates maximum power shall be considered when assessing surface
temperature. An integrated circuit can be assumed to fail with a single-
fault count to any combination of open circuits and short circuits at its ter-
minals. A second fault shall not change the condition assumed in the first
fault.
Connections shall be considered to fail to open circuit. A broken conductor

is presumed to be able to connect to any part of the circuit within its range
of movement. The initial break is one fault and the reconnection is a sec-
ond fault.
Clearances and creepages are judged in accordance with Table 10-1 and
the 1/3 value rule.
Capacitors may fail to short-circuit, open-circuit, or any intermediate

value below the maximum value specified.
Inductors may fail to an open circuit and any value of resistance between
the nominal resistance and short circuit. L/R ratio can assume only values
lower than that derived from the inductor specification.
Open circuit of any wire or printed circuit track is a single countable fault.
Requirements for Infallible (Protective) Components,

Infallible Assemblies, and Infallible Connections
The 2nd edition of EN50020 addresses two constructions of transformers.
In a Type 1 transformer, the windings are on different legs of the core or
are side-by-side on the same leg. A Type 2 transformer has its windings
wound over one another, with either solid insulation, dimensioned per
Table 10-1, or a copper foil or wire winding providing a grounded shield
between the windings. The dimensions of the foil or wire are coordinated
with the fuse size in the line feeding the transformer to ensure that any
short from primary to the shield is safely grounded. Two ground leads
must be connected to the foil shield. A wound shield must consist of two
independent windings, each grounded. (Two ground connections on a
single winding would constitute a shorted turn.) The core of the trans-
former shall be grounded unless insulated cores are used. Windings shall
be impregnated or encapsulated.
The input line of mains-connected transformers shall be fused or provided

with a circuit breaker. All transformers using a grounded shield shall have
ungrounded input lines fused.
In some standards a third type of transformer is recognized. This type has

only one ground on the shield and has additional protection against fail-
ure of the primary winding to the shield by temperature cut-out or high-
temperature insulation. This construction derives from the design of Class
2 transformers used in North America for doorbells, toy trains, and many
process control instruments with low voltage RTD or thermocouple
inputs. Not all standards require fusing of the input lines because some
transformer constructions inherently prevent excessive current flow.
The type tests for infallible transformers consist of loading the transformer
to 1.7 fuse rating, taking any current-limiting resistors into account, and
checking to see that the insulation temperature does not exceed its rating
after 6 hours. Embedded thermal devices or fuses may be used to limit
temperature rise. Only one is necessary. If insulation of secondary wind-
ings from ground is not necessary for intrinsic safety, a transformer with a
shield may exceed its insulation temperature rating if it does not burst into
flames.
All mains transformers shall be routinely dielectric strength tested:

Between input and output windings the higher of 2500 V or 4X
rated voltage
All windings to core or screen the higher of 1000 V or 2X
rated voltage
Each winding supplying an intrinsically the higher of 2X rated voltage
safe circuit and all other windings plus 1000 V or 1500 V
Infallible transformers not connected to the mains must be constructed in

a fashion similar to mains transformer, but the routine testing level
between input and output windings is the same as that listed in the third
line of the table.
Damping Windings
Short-circuited turns used to reduce the effects of inductance are not con-
sidered subject to fault if they consist of seamless metal tubes or windings
of bare wire continuously shorted by soldering.
Current Limiting Resistors
Infallible current limiting resistors shall be a film type; a printed resistor

with coating accepted for printed wiring boards, or encapsulated; or wire
wound with provision for preventing unwinding of the wire if it breaks. It
shall have voltage and power ratings 1.5 times those that will occur in nor-
mal and fault conditions.
Infallible current-limiting resistors are presumed to fail only to an open

circuit condition, which is to be counted as a fault.
In North American standards the limitation on power rating of infallible

resistors has not been applied if, under 1.5 times the voltage applied in the
circuit under fault conditions, the resistance does not decrease signifi-
cantly and if the resistor does not flame. Many European experts agree
that this practice is safe, but many national committees of the IEC have
been unwilling to support the idea.
The 1975 Annual Report of the Safety in Mines Research Establishment

included a report of testing of encapsulated film resistors at two or three
times their power rating for up to 6 months. No deterioration was found.
Blocking Capacitors
Blocking capacitors making up an infallible assembly shall be of a high

reliability, solid dielectric type. Electrolytic or tantalum capacitors shall
not be used. Each capacitor shall be considered to short or open, and the
value of the assembly is the most onerous value of either capacitor. Spac-
ings external to the capacitors shall be judged against Table 10-1, but
internal spacings shall not be considered. The dielectric strength of capaci-
tors isolating intrinsically safe from nonintrinsically safe circuits shall
equal twice the sum of the voltages of the circuits plus 1,000, but at least
1,500 V. Some standards at present require only 500 V if the sum of the
voltages is less than 90 V.
Shunt Safety Assemblies
Diodes or zener diodes used as shunt components in infallible shunt

safety assemblies shall provide at least two parallel protective paths, or
the leads and tracks must be infallible wiring, or disconnection of the
diode must also disconnect the circuit or component that is being pro-
tected. The diodes shall be rated to carry the current that would flow if
they fail to short circuit. Transient passing of energy of circuits using thy-
ristors shall be measured as discussed above. If fuses protect shunt
components, the components shall be rated for continuous passage of 1.7
times the fuse rating. The current-time characteristics of the fuse shall
ensure that the transient ratings of the protected components are not
exceeded. The factor to be applied under all fault conditions is 1.5.
A shunt safety assembly is considered to be a safety shunt when it is used

to control the parameters of a specific component or circuit to a value that
does not invalidate intrinsic safety. When they are connected to power
supplies defined only by Um, transients shall be considered. Such analysis
is not required if the safety shunt is used for protecting inductors, piezo-
electric devices, or capacitors. Transient effects need not be considered if
the safety shunts are fed from an infallible transformer, a shunt diode
safety barrier, a battery, or an infallible shunt safety assembly.
An assembly of suitably rated bridge-connected diodes shall be consid-

ered to be an infallible safety shunt.
Infallible Wiring
Wires can be considered infallible against open circuit failure

• if two wires are in parallel, or
• if a single wire is at least 0.5 mm diameter (0.4 mm if is stranded
and is not flexed in service), has an unsupported length of less
than 50 mm, or is secured next to its points of connection.
On printed circuit boards with 35 μm cladding, tracks can be considered

infallible if there are two in parallel, each of at least 1-mm width; or if the
width of a single track is 1% of its length but at least 2 mm. (35 μm is
equivalent to “1 oz copper” cladding, or “1/2 oz copper” initial cladding
of the surface of a printed wiring board which is then processed to plate
through-holes by adding copper.)
Connections other than plugs, sockets, and terminals may be considered

to be infallible where:
• There are two connections in parallel.
• A solder joint passes through a printed wiring board (including
plated through holes) and is either bent over before soldering, or
if not bent over, is machine soldered.
• The connection is crimped, brazed or welded.
• There is a single screwed or bolted connection conforming to the
requirements for grounding connections.
Galvanically Separating Components
Optocouplers and similar isolating devices shall be considered to provide

infallible isolation of separate intrinsically safe circuits if the ratings of the
device take into account the required 1.5 factors, and the device can with-
stand a dielectric strength test of 1,000 V plus the sum of the circuit
voltages, but not less than 1,500 V.
If separation is between intrinsically safe and nonintrinsically safe circuits

the requirements of Table 10-1 apply, but not within sealed devices such
as optocouplers. Unless it can be shown that the nonintrinsically safe cir-
cuit cannot invalidate the intrinsic safety of the device a means of limiting
the voltage applied to the device shall be provided. This may be a single
fuse and zener diode. The fuse shall be capable of interrupting the peak
current of the supply and the diode shall have a power rating of at least
1.7X(fuse rating) X (max. diode voltage).
Diode Safety Barriers
Diode safety barriers are assemblies incorporating shunt diodes protected

by fuses or resistors, manufactured as individual apparatus rather than as
part of a larger apparatus. If there are two parallel paths in the assembly,
they may be regarded as infallible if all diodes are preconditioned for 2
hours at 150°C, then tested with the pulse current test previously
described on page 362.
Diode safety barriers shall also be routinely tested to verify proper opera-
tion of each component and to verify the fuse resistance.
The assembly shall be packaged to prevent access for repair or replace-

ment of components on which safety depends. The design shall make it
obvious (e.g., by shape or color) when a barrier is mounted incorrectly in a
group of barriers. A grounding terminal shall be provided in addition to
any terminal normally grounded for function. Alternatively, a 4-mm2
grounding lead may be provided.
Type Verifications and Type Tests
Clause 10 of EN50020 details tests to be performed on a sample or samples

of the equipment or components of the equipment. In many cases the test
values have been noted earlier in this chapter. The reader should consult
the standard used by the certifying authority for the details of each test
procedure.
Testing of ignition capability uses the standard IEC sparking apparatus

previously discussed, with the most easily ignited mixtures:
Group I (Mines) 8.3 ± 0.3% methane in air
Group IIA (Group D) 5.25 ± 0.25% propane in air
Group IIB (Group C) 7.8 ± 0.5% ethylene in air
Group IIC (Groups A and B) 21.0 ± 2% hydrogen in air
If a factor of 1.5 is to be obtained by using a more easily ignited gas mix-

ture, the following mixtures are specified. All values given are percent by
volume.
Hydrogen/Air/Oxygen Hydrogen/Oxygen
Gas Group
Hydrogen Air Oxygen Hydrogen Oxygen
I (Mines) 52 48 — 85 15
IIA (D) 48 52 — 81 19
IIB (C) 38 62 — 75 25
IIC (A&B) 30 53 17 60 40
In most cases, if testing is to be carried out, one obtains the test factor of 1.5
by adjusting circuit parameters as follows.
(a) Increase the mains supply to 110% of nominal or adjust any other
supply or battery to its maximum value. (This is an improvement
because in former years some laboratories increased all voltages
to 110% of the maximum rating so that a transmitter rated 42 V
maximum was tested with an input of 46 V.)
(b1) In inductive or resistive circuits, increase the current to 1.5 times

the fault current by decreasing values of limiting resistance, and
then, if necessary, increase the voltage further.
(b2) In capacitive circuits, increase the voltage to obtain 1.5 times the
fault voltage. If an infallible current limiting resistor is used with
a capacitor, consider the capacitor as a battery and assess the cir-
cuit as resistive. (Author's comment: This fails unless the limiting
resistor is much higher than the minimum required to make the
discharge from the capacitor safe.)
Similar adjustments are made if one is assessing the design without test-
ing, using the curves in the standard. The second edition of EN50020
specifically discusses the assessment of simple circuits, rather than testing.
Most other standards already recognize this approach, and in many test
houses almost no testing is done if it is possible to assess a circuit from
available test data on similar circuits. PTB has been especially prominent
in developing assessment tools to eliminate as much testing as possible.
Figures 10-3 through 10-10 are representative of the performance of the
standard IEC test apparatus. Do not use these curves to assess apparatus. Use
the curves in the relevant standard. The capacitive curves in most standards
differ from those in EN50020, which have been amended based on more
recent investigations. The curves for the tin disc are no longer referenced
in EN50020. They were intended for use where apparatus was designed
for installation where no cadmium or other low boiling point materials
might be found. They were used infrequently, if ever.
Routine Tests
There are few routine tests of intrinsically safe apparatus called out in the
standard. The correct operation of each barrier component shall be
checked routinely, and diodes in two-diode, ia barriers must be routinely
pulse-tested as described earlier. Dielectric strength tests on mains trans-
formers in associated apparatus shall be carried out on every unit. The
intent of the standard is that this clause applies only to protective
transformers.
Marking
In EN50020 the requirements for marking include those called out in the
general requirements document: manufacturer's name or trade name,
location of manufacture, temperature or temperature code, and so on.
Intrinsically safe apparatus shall be marked EEx ia or EEx ib, and associ-
ated apparatus shall be marked with [EEx] ia, [EEx ib] or EEX [ia] or EEx
[ib]. The relevant parameters, such as Um, Li, Ci, Lo, Co, shall be marked
wherever practicable. The name or symbol of the certifying agency and
the certificate number shall be marked on the apparatus. International
practice followed by most industrialized countries outside of North Amer-
ica is similar, except that the EEx is replaced by Ex. “Ex” signifies
“explosion-protected” apparatus to a standard pertinent to the certifier.
“EEx” signifies “explosion-protected to a CENELEC standard.”
In North America the marking includes the number of the control drawing
in addition to the marking required by general purpose standards, and
may optionally include parameters as in international practice. North
American marking, of course, must state the Class, Group, and Division
for which the apparatus is suitable. Recognition of zone terminology alone
does not permit a change from this pattern. However, if North America
adopts the international classification of materials and the Class IIA, IIB,
and IIC nomenclature, the use of international coding, such as Ex ia IIC T5
will become practical. This will be a boon to manufacturers who now find
that nameplate space is limited and the mandatory markings are difficult
to fit.
Intrinsic Safety for Dust Hazards
IEC and CENELEC standards for intrinsic safety do not refer to the haz-
ards posed by dusts. They have been written only with gas and vapor
hazards in mind, except to the extent that dust hazards influence the
Group I mining requirements. In the United States it was decided that the
frequency of application of intrinsic safety in above-ground industrial
locations made hazardous only by the presence of combustible dusts with-
out combustible gases or vapors also being present would be very low.
Projected use did not warrant the development of a standard taking full
account of the fact that ignition energies of most dusts are significantly
higher than those of most gases and vapors. Therefore, the philosophy of
U. S. standards has been that if the circuitry is enclosed in a dust-tight
enclosure (one that will pass the circulating dust test of NEMA Standard
250) and is intrinsically safe for Group C, it shall be considered intrinsi-
cally safe for Groups E, F, and G. The surface temperature of the enclosure
shall not exceed 200°C for Groups E and F, or 165°C for Group G, when
the enclosure is blanketed with dust. If the circuitry is intrinsically safe for
Group D or methane, it is suitable for Class II, Groups F and G.
If the apparatus is not in a dust-tight enclosure, all creepage and clearance

distances are presumed to be violated in the most hazardous way, without
limit to the number of connections. If current and voltage levels are safe
after this assumption and the application of the other requirements for
Class I locations, the apparatus is intrinsically safe for Class II locations, as
defined above.
The assumption of unlimited violations of creepage and clearance dis-

tances when the hazardous dust is Group G is more restrictive in some
applications than is necessary because Group G dusts are not good con-
ductors. Layers of dry dust bridging live conductors will not form a
conductive path that compromises safety. If the dust is wet and also con-
taminated by chemicals, it is not certain whether conduction will occur.
Rather than prescribe lengthy test procedures for a situation that is not
likely to be of great practical interest, the drafters of NFPA 493 chose the
oversimplified, but readily stated, rules described above. Most apparatus
will be enclosed in dust-tight enclosures to protect it against possible
adverse effects of the environment. It is not likely that the simplified rules
will be burdensome. If, at some time in the future, this assumption proves
to be wrong, the document can readily be amended, because the rationale
for the rule has been included in the appendix of the standard for guid-
ance of future committee members who may have to consider revision.
If an enclosure meets general-purpose requirements and has no openings

and if all joints are sealed by welding, brazing, soldering, or fusion or are
provided with three full thread minimum engagement, it is considered to
be dust tight without testing.
U. S. requirements for dusts are more restrictive than those in Canada to

the extent that Group C intrinsic safety is required for Group E apparatus.
In Canada, Group D considered suitable for Groups E, F, and G. The more
conservative position is taken in the United States because some of the
layer ignition energies of metallic dusts reported by the Bureau of Mines
are very much lower than the cloud ignition energies of other kinds of
dusts.
How Safe is Safe?

During the development of any safety standard there is a natural and
laudable tendency to be conservative so that there can be no doubt that the
document guarantees safety. This bias towards conservatism is especially
pronounced when the subject matter of the standard is new. It is, there-
fore, instructive to consider the probability that an explosion will occur if a

system is intrinsically safe.
Hickes-Brown Approach
An early computation is found in the Hickes-Brown paper. This analysis

was based on an early draft of an IEC standard. They computed a proba-
bility of explosion for a process control loop in Zone 1 and in Zone 0, and
compared these values of probability to those they calculated for the mine
signaling apparatus, designed to meet British standards, which had been
in use in the UK for decades. They also computed a probability of explo-
sion for an explosionproof motor controller. Unfortunately, there is some
mismatch between the figures used for calculation and the text regarding
the number of times ignition is attempted in each case, so the calculated
results are not unequivocally in explosions per year, or some other time
period. They do, however, support the thesis of the paper that there is no
rational basis for arguing that intrinsic safety rules are not safe enough,
and there is evidence that they are too conservative. The summary below
uses the numbers developed in the paper, but the author has inserted the
number of ignition attempts based on his interpretation of the paper.
Normally Nonsparking Process Control Loop in Zone 1
In this case, the following are assumed:

• Field wire breaks or shorts once per year and remains failed for 1
hour - p = 1 that there will be a spark during 1 year (actually 0.63
if MTBF is taken as 1 year).
• Ignition probability at 1.5 normal current - p = 10−5 (10−3 is the
basic probability of ignition by the IEC test apparatus and 10−2 is
their assumed effect of testing at a current 1.5 times nominal.)
• Probability of perfect mixture in Zone 1, p = 10−3 (100 hours/year
above the LEL, at most easily ignited composition 10% of that
time).
• Probability that field wire breaks at right speed, is smaller than
those customarily used, and has insulation stripped at point of
break, p = 10−2 to 10−4, depending on whether the wire breaks
inside or outside a terminal box.
The resulting probability of an ignition occurring when a loop is in its nor-

mal operating condition is therefore 10−10 to 10−12, depending on the
assumptions about the wire breakage. Assuming 1,000 loops in a plant,
the probability of an explosion in the plant during a year is then 10−7 to 10−
9
.
Because their calculations were based on a draft which called for unity fac-
tor on testing after one fault, their calculations under fault conditions are
not relevant to the present standard for ib apparatus, which requires a 1.5
test factor after one fault. However, if (as Hickes-Brown did) one assumes
that there are 100 components per loop but only one might affect intrinsic
safety, and a failure probability of each component of 10−3 is also assumed,
the probability that the critical component fails is 10−5. There was an
implied assumption that the loop operates at the maximum current per-
mitted by the standard in its normal operating condition. A component
failure cannot therefore be allowed to increase the current, so the probabil-
ity of ignition because a component has failed is less than that when no
component has failed by the probability of component failure. In the usual
case, the normal operating current would be much less than that permitted
by the standard, and the limit value of current or voltage after a compo-
nent failure would be the value Hickes and Brown assessed for normal
operation.
Normally Nonsparking Control Loop in Zone 0
In this case a flammable atmosphere is presumed to be present 100% of the

time, instead of 1%. An additional fault is allowed with the same probabil-
ity as the first fault, 10−5. After the second fault the circuit is assessed with
a test factor of 1, so the ignition probability in the test apparatus is
increased by a factor of 100. The total increase in probability by 100 × 100 is
more than offset by the low probability of the second fault, 10−5. The worst
case probability of there being an explosion, 10−11 to 10−13, is still less than
that for the loop operating at the maximum permitted values in normal
operation.
A more realistic example is that of a two-wire transmitter operating at 20

mA in normal operation from a 24-V supply in Zone 0. The probability of
this current igniting in the IEC test apparatus is 1.5 × 10−12, so the proba-
bility of ignition in normal operation in Zone 0, taking into account the
same factors given above, is of the order of 10−15 to 10−17. (Compared to
the values for normal operation in Zone 1, the probability of ignition in the
standard test is 10−7 less, but the probability is raised a factor of 100
because of continuous presence of the flammable cloud.) The probability
of an explosion after two faults is still not higher than that in normal oper-
ation if one assumes that the maximum permitted current flows after two
faults. This and the test factor of 1.0 raise the probability of ignition in the
test apparatus by 109, which is offset by a probability of 10−10 for two
simultaneous faults, yielding a probability of 10−16 to 10−18. If one assumes
the maximum permitted current flows after one fault, the probability of an
explosion is raised 105 compared to the two-fault case by eliminating one
fault, and lowered by 100 by the test factor of 1.5. The explosion has a
probability of 10−13 to 10−15 after one fault in Zone 0, so the one fault con-
dition in this example has the higher probability of causing an ignition.
Probabilities for the Mine-Signaling Apparatus
The mine-signaling apparatus motivated early work on intrinsic safety.

Two bare wires routed about the mine were energized from 24 V. Contact
between the wires operated a bell or other signal to inform the hoistman
that a load of coal was ready to be pulled to the surface. In operation a
miner shorted the two wires together. This signaling system had a long
history of safe operation after it was made intrinsically safe. The system
was evaluated based on a factor of 2 on current against a reference ignition
current based on no ignition in 100 sparks in the SMRE break-flash appa-
ratus. The field wiring was normally shorting so the arcing frequency,
based on 10 uses per hour with 10 sparks per use (from the Hickes—
Brown text), is 9 × 105 per year. The break-spark apparatus required about
twice as much current as the standard IEC break flash apparatus to cause
ignition, so this cancels out the lowered probability of the factor of two on
test current. Thus the probability of ignition can be taken to be 10−2.
Assuming the mine to be Zone 1, the probability of the most explosive
mixture being present is 10−3, as above. A factor of 10−2 was allowed for
the fact that contacting of the two wires would not be at the optimum
break speed. The resultant probability of an explosion in one year using
the mine-signaling apparatus is, thus, calculated to be 9 × 10−2. This is
much too high an estimate.
Were the estimate valid there would be an explosion in some mine using
this apparatus every year. Were the estimate of use reduced by a factor of
100, the probability of an explosion would still be unreasonably high. In
the author's opinion what is probably missing is a factor to account for
operation being terminated as soon as the presence of a flammable cloud
is suspected.
Probability of an Explosion Caused by a Flameproof

Motor Controller
Assume that the controller arcs 1,000 times a year. The probability of an
explosive cloud being present is taken as 10−2, because the controller is in
Zone 1 and the arc is presumed to be capable of igniting any mixture that
may be present. Any mixture present outside is assumed to be present
inside the enclosure. A probability of 10−3 is assigned to a cover being dam-
aged or a bolt left loose and remaining undiscovered. The resultant
probability of an explosion is, therefore, estimated to be 10−2. This estimate
is unrealistically high. The probability that the enclosure fails in a year cor-
responds to an MTBF of 107 hours, which is the value used by Benjaminsen
and van Wiechen (see Chapter 4). However, the sparking is so frequent
that the first entry of flammable vapor into the enclosure will be ignited.
The better probability value for this apparatus is, therefore, that of the
undetected apparatus failure itself, as set forth by Benjaminsen and van
Wiechen (i.e., the probability is 10−3 that there will be an explosion within a
year).
Hickes and Brown concluded that one might argue that the rules being
proposed for intrinsic safety were ultra-conservative, based on compari-
son to systems that have a long history of safe use. There is no basis for
suggesting that intrinsic safety rules are too liberal. Since the
Hickes-Brown paper was published, no changes have been made to the
rules for intrinsic safety that increase the probability of intrinsically safe
apparatus causing an explosion. The conclusion of the Hickes-Brown
paper is still valid. One can argue about the specific numbers used, but if
consistent assumptions are made for the different kinds of systems, the
intrinsically safe loop has a much lower probability of causing an explo-
sion than the other systems.
Calculated Mean Time to an Explosion

Fraczek developed the following equation for the MTBE caused by an
intrinsically safe system.
MTBE = (1/n)(I0/I)a(Tu/Pw)(Tu/Td)(1/P0)
where:
MTBE = mean time between explosions
n = number of pieces of apparatus
I0 = reference current in test apparatus for probability P0
I = actual circuit current
a = exponent applicable to probability vs. current relationship
of test apparatus (n in Chapter 9)
Pw = probability that a flammable cloud is present
Tu = mean time between failures that make apparatus a source
of ignition
Td = mean time between failure and repair, or mean time
between inspections that would discover failed apparatus
(only important if it is greater than TU).
Fraczek used the following example of historical data from a mine.
Tu/Td = unity
Tu = 0.15 years
Pw = 0.0029
Io = 0.1 A
Po = 0.001
a = 13
Tu is the historical value for the probability of an equipment failure that

could start a fire in a particular mine. Fraczek assumed that if intrinsic
safety had a MTBE 250 times that for fires the system would be safe
enough.
If no test factor is applied, MTBE = 52,000 years for one piece of apparatus
or 520 years for 100 pieces of apparatus when the atmosphere is flamma-
ble 1 day per year and the apparatus fails in an ignition-capable way every
1.8 months on the average.
If one assumes the same equipment failure rate and 100 items of equip-
ment, but assumes Zone 0, so that the flammable cloud is present always,
and 1.5 safety factor on the fault current, the MTBE is still 293 years. The
1.8-month value of Tu is many orders of magnitude higher than would be
experienced with intrinsically safe apparatus in a process control system,
so Fraczek's calculations are ultra-conservative.
The reader is invited to make assumptions and calculate other values of

MTBE. The conclusion is inescapable that there is more than enough safety
factor built into the present rules for intrinsic safety, and the technique is
many orders of magnitude safer than any other protective technique.
In Chapter 4 it was noted that one result of the works of Benjaminsen and
van Wiechen was the recognition that, if the MTBF of the equipment is
high, the frequency of the presence of the flammable atmosphere is of little
consequence in determining when there will be an explosion. The frac-
tional time the atmosphere is present is not a sufficient statement because,
when the equipment becomes an ignition source, it will presumably
remain so until the failure has been discovered by routine inspection or
other means and the failed apparatus is repaired or replaced. Therefore,
within the period between inspections the first appearance of a cloud will
cause an explosion. Where intrinsic safety is used the fractional time is rel-
atively high, greater than 0.01, and in Zone 0 would be 0.1 to 1.0, so the
Fraczek equation gives values of MTBE that are somewhat high. The per-
sistence of the ignition source is also not taken into account in analyses,
such as those by Bijl, Hickes and Brown and this author, which simply
multiply probabilities. Therefore, they yield probability values that are
low by as much as a few orders of magnitude. Because these exercises
almost always compare one technique with another, such as intrinsic
safety to explosionproofing, the numerical error is common mode and
doesn't affect the qualitative judgement about relative safety.
A very conservative way to look at intrinsic safety is to assume that after

two independent failures the output is on the reference curve, giving a
probability of ignition of the most easily ignited mixture in the IEC appa-
ratus of 10−3. If one assumes the cloud, when it appears, has the most
easily ignited concentration, then the probability of an explosion is the
probability that the equipment become an ignition source times the proba-
bility that the source will ignite the most easily ignitable mixture. This
approach ignores the fact that the real arcs will not be as effective as those
in the test apparatus and the question of what probability to assign for the
fact that the mixture isn't ideal. These liberties are justified if the calculated
probability values are much lower than those for another recognized type
of protection, such as explosionproofing, which has been safely used for
decades.
The probability of a system with two protective components of the same

MTBF, each of which must fail, becoming an ignition source is given by:
PF = (1 − ∈−t/MTBF)2
If MTBF is 106 hours the probability PF and the probability of explosion,

PE, are, for several values of t:
t-hours PF PE
104 0.0001 10−7
10 5
0.009 9 × 10−6
3 × 10−5 0.069 6.9 × 10−5
106 0.39 3.9 × 10−4
An MTBF of 106 hours is probably conservative for intrinsic safety. The

failure rates for diodes and solder joints used by Redding when discussing
reliability of barriers in his book correspond to MTBF at least two orders
of magnitude higher. If one takes into account that the failure rates used
include values relative to departure from specification rather than only to
unsafe failure to function, the real values of MTBF are even higher.
The MTBF for explosionproof enclosures assumed by Benjaminsen and

van Wiechen is 106 to 108 hours, depending on construction. The failure
probability in 104 hours for an enclosure with 108 hour MTBF is 0.0001.
The probability of explosion during that time, for the reasons discussed
above, is the same. The safety of explosionproof enclosures is understated
because of the beneficial effects of their also being protected by a level of
restricted breathing or sealing, but the extraordinary level of safety
afforded by intrinsic safety is inarguable.
Ignition by a High Speed Relay
When the German investigators designed the apparatus that is now the
IEC standard test apparatus, they did so with the intent of having a refer-
ence apparatus that is more capable of causing ignition than any real
contacting mechanism likely to be found in industry. The fact that the
apparatus is more sensitive than the British break-flash units gives cre-
dence to the assumption, as does the experience of any investigator who
has tried to obtain ignition with commercial relays, and found that the cur-
rents required are much higher than those required by any laboratory
apparatus.
Early intrinsic safety standards required that an additional safety factor be

applied if a circuit had normally operating contacts. This requirement no
longer exists. During the period when the elimination of the added safety
factor was being discussed, LCIE prepared a special high-speed relay with
a 15 mm2 fixed cadmium contact and a 2-mm-diameter tungsten wire con-
tact. Distance between contacting surfaces was 0.3 mm.
LCIE tested this relay in a standard 95-mH test circuit at 24 V. They oper-
ated the relay at 70 Hz for 5-min intervals (21,000 operations during each
interval) at currents of 90, 120, 150, 200, and 250 mA, without obtaining an
ignition of the 8% ethylene-air mixture, which in the standard test appara-
tus is ignited at 65 mA. At a current of 300-mA, ignition occurred after 6
min, 25,200 operations. The test current was then reduced and the test
interval lengthened as follows to obtain an ignition in each case.
Test current - mA Test time - min Operations
270 15 63000
250 20 84000
220 30 126000
210 30 126000
The test at 210 mA was repeated five times. In two runs ignition occurred
after 40 minutes. In three runs no ignitions occurred.
This same relay was also tested at CERCHAR (now INERIS). The investi-
gators at CERCHAR began by testing a 24-V resistive circuit in
hydrogen-air at frequencies of 5 Hz to 75 Hz, for 15 min periods at each
current, and had no ignitions up to a current of 3 A. The IEC apparatus
ignites at 250 mA in a 24-V resistive circuit. The tungsten contact became
unsoldered during the 3-A test. After resoldering the contact they changed
the circuit to a 100-mH inductor with a self-resistance of 12.5 ohms. Oper-
ation was at 30 Hz. Results are shown below.
Minimum Standard
Atmosphere Ratio of
Relay Ignition Igniting Test Time - min
Operations Currents
Current A Current A
Hydrogen 0.14 0.030 4.6 13.5
24,300
Ethylene 0.26 0.065 4 28
50,400
To determine whether any hazard exists from the breakdown of ethylene

to hydrogen and carbon as the result of sparking, the circuit was tested at
0.25 A (4% less than the igniting current) and the relay was run for 2 hr,
30 min, 2.7 × 105 operations, without ignition. The LCIE and CERCHAR
data are summarized in Figure 10-2.
Figure 10-2 Summary of LCIE and CERCHAR Relay Tests

In an intrinsically safe circuit for use in Group IIB (Group C), this relay
would be allowed to carry 43 mA, two-thirds of the igniting current for
ethylene-air, in the IEC apparatus. Based on the data plotted in Figure
10-2, the probability of ignition by this relay with the most easily ignited
mixture at this maximum permitted current is 2.2 × 10−8. This result vali-
dates the annoyance that many experts in the field have expressed for
decades at the easy assumption by testing authorities and others who
work frequently with the standard IEC test apparatus that it represents the
real world. Igniting currents in the standard IEC test apparatus are not
currents at which ignition will occur, but currents at which, in the real
world, ignition is improbable in the extreme. The data reported for the
ignition of ethylene-air at both LCIE and CERCHAR, take into account
that the relay contacts become more ignition capable as they become worn
and pitted. Initial values of igniting current were much higher than the
final values at both laboratories.
IEC Test Apparatus
Until the early 1980s, there were no officially recognized curves for the cal-
ibration of the IEC test apparatus. Such were commissioned at the Paris
meeting of SC31G in 1980. The curves presented in this chapter are repre-
sentative of the IEC apparatus calibration. Figures 10-3 through 10-6 are
adapted from Cartwright. Figures 10-7 through 10-10 are adapted from
Widginton. The work of the various investigators before the SC31G meet-
ing in Dubrovnik in 1986 for the most part confirmed these curves. The
curves in the second edition of EN50020 include some later adjustment of
the capacitance curves and some rearrangement of the inductance curves
to make them more useful for evaluation of low voltage circuits.
Figure 10-3 Ignition Characteristics, Standard IEC Apparatus, Resistive Circuits

Figure 10-4 Ignition Characteristics, IEC Apparatus with Cadmium and Tin
Disks, Capacitive Circuits, 8.3% Methane-Air
Figure 10-5 Ignition Characteristics, IEC Apparatus with Cadmium and Tin
Disks, Capacitive Circuits, 22% Hydrogen-Air
Figure 10-6 Ignition Characteristics, IEC Standard Apparatus, 24-V Inductive

Circuits
Figure 10-7 Relation Between Minimum Igniting Current and Open-Circuit

Voltage-IEC Standard Apparatus
Figure 10-8 Relation Between Minimum Igniting Current and Inductance—IEC

Standard Apparatus
Figure 10-9 Relation Between Minimum Igniting Current and Open-Circuit

Voltage—IEC Standard Apparatus
Figure 10-10 Relation Between Minimum Igniting Current and Inductance—IEC

Standard Apparatus
The calibration currents for the IEC apparatus are listed below. This cur-
rent flowing in a 24-V, 0.095-mH, air-cored coil shall cause at least one
ignition of the mixture referenced earlier in the chapter for each of the
material groups in 400 revolutions of the contact holder. The wire is at
positive polarity.
Material Group Current - mA
I (Methane) 110
IIA (Group D) 100
IIB (Group C) 65
IIC (Groups A and B) 30
Formerly there were established calibration current values for resistive cir-
cuits also, but recent consensus is that only one set of values is needed. It
should be noted that the requirement that 1 ignition be obtained in 400
revolutions is a very loose requirement necessitated by the variation in
ignition values obtained in different laboratories or in the same laboratory
at different times. Variations of 5% above and below the published curves
are common, and 10% can sometimes be experienced.
Many users of the IEC test apparatus have experienced difficulty in

obtaining ignition at the stipulated calibration currents. Cooper proposed
using a pair of hockey-stick shaped trailing wires to replace two of the
straight wires. This increases sensitivity somewhat. However, the trailing
wires did not completely eliminate the problem of unexplained sudden
changes in sensitivity.
Bossert proposed forming two spring coils in the tungsten wire electrode.
Those electrodes ignited at about the same current values as the standard
straight wires and appeared to give more consistent results and much
longer life, before breaking.
Silver investigated whether unexplained changes in calibration sensitivity

might be caused by contamination of the vapor-air mixture by sparks
early in a test series. He concluded that contamination is not a factor in
uncertain calibration values.
Thomson reviewed the IEC apparatus design from a mechanical and met-
allurgical standpoint and concluded that splintering of the tungsten wire
tip is speeded by fusion of the wire after cutting. as recommended by the
IEC. Fusion hardens the tip and causes embrittlement. He determined also
that the tungsten wires in the standard design are overstressed so that the
spring action as the wire clears the cadmium disk changes as the appara-
tus is used. This presumably alters the breaking-arc characteristic.
Overstressing at the wire holder causes permanent deformation and
breaking. One of Thomson's recommendations is to replace the tungsten

wire with steel wire.
In other investigations Silver studied the effects of electrode conditioning

on the variability of ignition sensitivity. He verified variations in igniting
current between 30 and 60 mA in a 24 V. 0.095 H circuit. After three hours
of operation the tungsten wire electrodes were shorter by 1–1.5 mm. The
tips were fractured and splintered, but the effects could not be correlated
with ignition sensitivity because even before use the tips were of various
shapes. He therefore fused the electrodes electrically in an oil bath, which
produced rounded tips of greater uniformity. When tested, these elec-
trodes became more chiseled and sharp after each explosion. Comparison
of old and new cadmium discs showed variation from 60 ma igniting cur-
rent when new to 30 mA after lengthy use, which causes deep grooves in
the surface. He also found that the presence of cadmium dust on the disc
greatly affected sensitivity. Igniting current was 32 mA to 36 mA when
dust was present and 50 mA to 52 mA when dust was not present.
Changing the length of the tungsten electrode between 10 mm and 12.5

mm decreased igniting current from 45 mA to 32.5 mA. (The standard
length is 11.0 mm.) Varying the spacing between the wire holder and the
disc surface from 8.5 mm to 11.0 mm (standard distance is 10.0 mm)
caused igniting current to change from 30 mA to 45 mA.
In later work Silver studied other wire electrode configurations and the
effects of barometric pressure, composition of the H2-air mixture, conden-
sation of water on the electrodes, ambient temperature, relative humidity,
and circuit voltage. He found that although the standard electrode-disc
combination becomes more sensitive when it is first used and tends to sta-
bilize in sensitivity, “L”-shaped electrodes caused the sensitivity to
decrease with life over a much longer period of test. At the cost of lower
sensitivity, “hockey-stick”-shaped electrodes with a new disc and L-
shaped electrodes with an old disc were more repeatable over a period of
an hour than the standard electrode construction. Average igniting cur-
rents were 40 mA and 32 mA respectively, with variation of about only 2
mA when cadmium dust was removed from the disc after each explosion.
Though the calibration current for the standard apparatus is 30mA, actual
igniting current values were frequently 10 mA to 20 mA higher, even after
much use.
Ambient pressure variations caused a change in igniting current of

approximately 3 mA/30 mm Hg change around 760 mm Hg. An increase
of 0.7 mA in the igniting current resulted from 1% deviation of the hydro-
gen-air mixture from the most easily ignited concentration of 21%.
Between 10 and 30°C, a 10° drop in temperature increases igniting current
by 1.5 mA. A light film of moisture on the electrodes increases igniting
current by 6 mA. Relative humidity changes have no significant effect on
calibration. Between 20 V and 50 V, circuit voltage does not significantly

affect igniting current.
All users of the IEC test apparatus have concluded that a worn disk is
more sensitive than a new disk. At least one other laboratory investigated
the effect of environmental changes and obtained results similar to those
reported by Silver. On the other hand, Peterson, working with meth-
ane-air mixtures, did not find the expected variations in probability of
ignition with changes of temperature, relative humidity and pressure in
the ranges likely to be encountered in mines. As a result of the widespread
concern for variations in calibration of the test apparatus many standards
now include guidance on preparing the tungsten wires and pretreatment
of the cadmium disc. The tungsten wire may be cut with sharp scissors or
it may be fused in a fixture, after which the spherical ends are removed
with tweezers. A new cadmium disc shall be aged for 20,000 revolutions
of the electrode holder in a 95 mH, 24 V, 100 mA circuit. The test gas is air.
Then provide new tungsten wires and attach the test apparatus to a 2-μF,
nonelectrolytic capacitor charged through a 2-K resistor. Using the Group
IIA test mixture, apply 70 V and run for 400 revolutions or until an igni-
tion occurs. If ignition occurs drop the voltage 5% and repeat until no
ignition occurs in 400 revolutions. Start at 60 V and repeat the test series. If
no ignition occurs at 50 V, start the series again at 70 V. If ignition occurs at
50 V, start the test series at 50 V; if no ignition occurs at 40 V start at 70 V
again.
Installation of Intrinsically Safe Systems

An installation of intrinsically safe and associated apparatus remains
intrinsically safe if it ensures that:
(a) The limiting values of capacitance and inductance, or L/R ratio
on which certification is based are not exceeded.
(b) Positive separation of intrinsically safe circuits from nonintrinsi-

cally safe circuits prevents intrusion of energy from nonintrinsi-
cally safe circuits into the intrinsically safe circuit.
(c) Power system faults and differences in ground potential do not

make the system ignition-capable.
Controlling System Parameters
In North America the primary reference for installation of associated and

intrinsically safe apparatus is the manufacturer's instructions, which are
reviewed as part of the certification process. In recent years these installa-
tion instructions have taken the form of a control drawing that details the
connections to the certified apparatus that were considered during the cer-
tification process. The control drawing may call out specific apparatus by
model number or it may give maximum values of capacitance, inductance,
and L/R ratio to be observed when making connections to the terminals of
the certified apparatus. The control drawing will also specify values of Um
that apply. In countries following the IEC pattern, an X at the end of the
certificate number indicates some special installation requirement in addi-
tion to conforming to the limiting values of capacitance, inductance, and
L/R ratio. These special conditions are usually stated at the end of the
certificate.
Certification documents and control drawings assume that the installer

will control the total of load and cable parameters so that they do not
exceed the limiting values given in the installation instructions. One
should use judgment in determining how much record keeping is needed
to ensure that the system meets this requirement at installation and con-
tinues to meet it in years to come. Much has been written about the
documentation needed to ensure that the parameters of the installed sys-
tem are known so that changes to the system do not inadvertently take it
outside its specified limits. If the system has many cable runs that are so
long that capacitance or inductance approach the limiting values, then the
recommendations for maintaining detailed data on cable lengths, types,
and parameters and standing capacitance and inductance of installed
equipment must be followed. However, if all the intrinsically safe devices
have small values of standing capacitance and inductance, and if the plant
dimensions make it extremely improbable that a cable exceeding limiting
values could ever be installed, then a less detailed set of records can be
justified.
It has been accepted for many years that the capacitance of PVC-insulated,
two-wire pairs of the type commonly used for instrument installations
will not exceed 60 pF/ft or 0.2 mH/ft. If the parameters of a specific cable
are unknown, these values can usually be used, and, in most cases, they
will give considerable safety factor. These values are sometimes not valid,
so they must be used with care. In one very large, grass-roots plant the
number of cables was so large that it was economical to order custom
cable with insulation of especially high dielectric constant. The resulting
cable insulation was thinner and the cable was lighter than standard cable,
so the cable trays were less expensive. The capacitance of this special cable
was approximately 70 pF/ft. The rule of thumb maximum values also do
not apply to cable especially installed for Fieldbus installations. Not only
is the capacitance between pairs higher, but the grounded shield increases
the total capacitance. The effect of grounding the shield of a two-wire pair
must be considered in any system. The total capacitance between a pair of
wires with a grounded shield is the sum of the capacitance between the
two wires and one-half the capacitance between one wire and the shield.
The capacitance is usually the limiting factor in systems operating in the

range 24 V to 30 V from resistive barriers, especially those designed for
Groups A and B (Group IIC). Inductance becomes a factor when the short-
circuit currents are high. Even so the L/R ratio of the cable may make the
system safe despite long lengths of cable. If the allowable L/R ratio is not
specified it can be calculated from:
L/Rmax = 4 Lmax Imax/Vmax
(valid only for resistive sources without internal inductance)
Any cable with an L/R ratio lower than this value is safe regardless of
length, as long as the capacitance does not exceed Cmax. Any coil with a
lower L/R ratio is also safe.
If the intrinsically safe apparatus has relatively high values of standing

capacitance or inductance, as is sometimes the case with electromechanical
assemblies using inductive current-to-force transducers, it may be neces-
sary to keep careful records to ensure that limiting parameters are not
exceeded when devices are replaced or interchanged. Many devices now
on the market are diode-shunted to reduce standing inductance to a negli-
gible value, so the presence of electromechanical devices is a call for
consideration of the need for more detailed record keeping, not a mandate.
Certification of intrinsically safe systems assumes that the uncertified

apparatus on the control room side of associated apparatus does not con-
tain voltages that exceed the Um rating of the associated apparatus. Um is
commonly 250 V RMS or dc, but occasionally higher. The original intent of
this limitation was to avoid unsafe situations that might arise if a barrier
certified based on use with a 250 V RMS power system were used in the
presence of a 440 V system. The fuse might not clear a short circuit fault
before the diodes open circuited, allowing ignition capable energy to pass
to the intrinsically safe output. During the 1970s and 1980s restriction to
250 V RMS caused great concern and confusion in distributed control sys-
tems containing CRTs, printers with high voltage antistatic protection,
and copiers. Some manufacturers, because of customer demand, provided
CRTs with certified transient suppression hardware, and had ferroreso-
nant supplies, whose resonant windings exceed the input line voltage, and
printers with high voltage antistatic circuitry, certified to be safe for use
with 250 V rated barriers. It was well understood by everyone that the
probability that a CRT could fail in a manner that would put high voltage
on the barrier input is extremely low, but no standard stated this fact and
no certifying agency would act on that belief without expensive testing.
The confusion was compounded by the fact that an incident was known
where a high voltage discharge in the power supply enclosure of an IEC
test apparatus ignited gas that had leaked through the bearings of the
explosion chamber. The high voltage came from the deliberate connection
of a CRT high voltage lead to the input of a barrier. The barrier clipped the
input pulse as expected, but the current pulse passing through the induc-
tance of the ground lead raised the ground lead voltage, which then
flashed over in the power supply enclosure. The situation eased, espe-
cially in Europe, after a representative from a major test house expressed
his opinion at a public conference that the high-voltage discharge from a
CRT would not be ignition-capable and that concern for CRTs was
over-emphasized. Since that time evolution of distributed control systems
has placed the CRT electrically more distant from the field interface than
formerly, and considerations of noise elimination has made most systems
immune from any danger caused by a CRT failure. It is the author's opin-
ion that the issue of low energy, high-voltage sources like CRTs is
practically dead at this time. However, any source capable of delivering
significant power, which operates at a voltage above Um, should not be
operated behind associated apparatus unless it has been shown to be safe
for use with that associated apparatus.
During apparatus certification it is assumed that intrinsically safe circuits

remain isolated, not only from nonintrinsically safe circuits, but also from
different intrinsically safe circuits. Interconnection of circuits could, in
principle, impose higher currents or voltages on a circuit than it was certi-
fied for or could increase the connected capacitance or inductance to
unsafe levels. A different intrinsically safe circuit is one that has not been
evaluated for the effect of its interconnection with another intrinsically safe
circuit and found to be intrinsically safe. In North American practice it is
assumed that intrinsically safe circuits are run in different cables, unless:
• each circuit is within a grounded metallic shield, or
• the conductors of each circuit have insulation with a minimum
thickness of 0.01 in. (0.25 mm).
The 0.01-in. value of insulation thickness was an arbitrary value intro-
duced in the first edition of RP12.6 as a recommended minimum thickness
of insulation for all intrinsically safe circuits to ensure function, not to
ensure safety. It was recognized that an intrinsically safe circuit could be
installed safely with bare wires if it were isolated from other sources, as in
the mine signaling system. Had a value of 0.008 in. (0.2 mm) been selected,
standard telephone cables would have been acceptable for IS circuits.
In the European standard, the minimum radial insulation thickness of

conductors in a multicore cable is 0.2 mm, and the insulation must with-
stand a 500 V RMS dielectric strength test. In addition, a multicore cable
carrying multiple intrinsically safe circuits shall be capable of passing
• a 500 V RMS dielectric strength test between all the cores joined
together and all the screens or armoring joined together, and
• a 1,000 V RMS test between half the cores connected together and
the other half of the cores connected together. This second test is
waived if all the individual intrinsically safe circuits are provided
with conducting shields.
No faults between different intrinsically safe circuits need be assumed if
the cable meets the above requirements and also has a shield around each
intrinsically safe circuit that covers 60% of the surface area. No faults
between cores need be assessed, if in a cable meeting the insulation and
dielectric strength requirements, no circuit exceeds an open circuit voltage
of 60 V, and if the cable is fixed and protected from damage.
Protection Against Intrusion by Nonintrinsically

Circuits
Physical separation of intrinsically safe circuits from nonintrinsically safe
circuits prevents intrusion into an intrinsically safe circuit by a nonintrinsi-
cally safe circuit. In North American practice a separation distance of 2 in.
(50 mm) is considered to be safe. This distance is derived from many years
experience in separating low-voltage, low-energy, Class 2 circuits, such as
those used in thermostats and doorbells, from power circuits. Alterna-
tively the circuits may be separated in raceways or cable trays by
nonconducting or grounded metallic partitions. Another option to spacing
is the use of MI or MC cable either for all the intrinsically safe circuits or all
of the nonintrinsically safe circuits. European recommendations parallel
these except that 50-mm separation of cable runs is not specified. No
authorities permit the intermixing of intrinsically safe and nonintrinsically
safe circuits in the same cable unless the cable has been specifically certi-
fied as part of a system.
Within enclosures and on racks and panels, spacing of terminals and the
use of separating partitions readily separates the terminations of nonin-
trinsically safe and intrinsically safe cables, but additional measures are
needed to ensure that intrinsically safe and nonintrinsically safe field
installed wiring does not become intermixed. Most manufacturers provide
for entry of intrinsically safe wiring on one side of the enclosure and entry
of the other wiring on the other side of enclosure, and provide trays or
wiring ducts to guide and restrain the wires until they are near the termi-
nals. The NEC requires that such ducts be at least 3/4 in. apart if they
must be run close to one another.
Nonintrinsically safe and intrinsically safe wiring must be installed such

that no interconnection results from a wire coming loose from a terminal.
Partitions between the row of terminals, clamping or tying the wires near
the terminals, or similar means must be used to prevent a loose wire from
making contact. This feature of a design would usually be scrutinized by
the certifying authority and covered in the installation instructions.
Controlling Effects of Power System Faults and

Differences in Ground Potential
Every conductor that enters the hazardous location and is not protected by
a barrier or other associated apparatus must be grounded. Except for iso-
lating devices in which there is negligible current flow between the input
and the output terminals, the associated apparatus also must be grounded
to provide a low impedance path to fault currents. When a fault impresses
power line voltage at the input of a barrier, the current that flows before
the fuse opens may be tens of amperes. This current flowing through the
impedance of the ground connection raises the voltage of the grounded
terminal of the barrier. In principle, if the voltage rose to a high value, it
could be a source of ignition. In North America the practice has been to
require the resistance of the ground lead from the barrier to the earthing
stake or ground mat to be less than 1 ohm. To prevent stray currents from
other sources from flowing in the ground lead it shall be isolated from the
protective grounding system and the neutral conductor except at one
point, and it shall be insulated to prevent contact with other conductors.
The requirement for insulation does not preclude the use of uninsulated
plates to collect the ground connections from many barriers or to connect
together ground leads from several busbars, as long as the mounting and
enclosure design prevent contact with other conductors. The grounding
conductor must be able to carry the fault current that might flow from the
source. 12 AWG wire is recommended in ISA RP12.6. European practice is
to call for a single conductor of 4-mm2 area or two conductors with
1.5-mm2 area. The use of two conductors is recommended by many
authorities to facilitate periodic checking of the continuity of the ground
path.
All codes emphasize that the intrinsically safe system and shields associ-
ated with intrinsically safe circuits shall be connected to the main
grounding system only at one point. Shields should be taped and isolated
from ground except where they are connected to the intrinsically safe
grounding system. Only if the associated apparatus isolates the input from
the output is it permissible to ground an intrinsically safe circuit in two
places. By preventing conduction across the associated apparatus the iso-
lating device makes the output circuit an intrinsically safe circuit that may
have a ground or not. The input circuit is separate from the intrinsically
safe circuit and may have a ground or not. European practice recognizes
that there are functional reasons for more than a single ground on a shield.
Multiple grounds may exist if the equipotential bonding system ensures
that no appreciable voltage difference between the grounding points will
be developed by the flow of currents in power system faults. In a system
of small geographical extent multiple connections to the same ground
plane or mat may be equivalent. American literature does not yet recog-
nize this option for multiple connections to a shield.
Other Installation Requirements
Intrinsically safe circuits and the cables carrying them shall be marked by
labels or by use of a light blue color. Marking alerts personnel to which cir-
cuits may be worked live without a hot permit and which cables may not
be used for nonintrinsically safe circuits. Marking also alerts workers to
observe separation requirements when installing nonintrinsically safe
cables in the area or in the same tray or raceway. Marking of raceways in
panels and enclosures containing nonintrinsically safe circuits is especially
necessary to facilitate separation of nonintrinsically safe circuits when
changes are made throughout the life of the installation.
Seals shall be installed as required for nonintrinsically safe installations of

the same cable type, or if conductors are run in conduit. Thoughtful
design should eliminate the need for most seals because the intrinsically
safe cables or conduit runs can be interrupted to prevent flow of combusti-
ble material from one classified location to another.
The requirements for grounding enclosures and other accessible metal

parts apply to intrinsically safe installations.
Inspection of Intrinsically Safe Systems

Why inspect? The objective of any inspection is to assess conformance to
accepted standards and customary practices. When inspecting an explo-
sionproof installation one would inspect the apparatus in a
“walk-around” to see by reading apparatus labels that it is suitable for the
location in which it is installed. One would ascertain that grounding and
bonding have been implemented in accordance with local codes, that seals
in conduits and cables have been installed where required, and that con-
duit joints are tight. By observing the installation one could determine that
it has been installed per plan and per accepted safe practices.
Inspection of an intrinsically safe system differs from an inspection of an

explosionproof installation in that some of the inspection must be made by
reviewing documentation. Those features of the installation similar to
those of an explosionproof installation can be checked by visual inspec-
tion, but the inspector must verify many features of the intrinsically safe
system by referring to the design documentation. A factor underlying the
tendency for standards developers to add more and more marking
requirements is an unrealistic assumption that all the necessary informa-
tion can be put on a single label so that a inspector on a “walk-around”
can judge the suitability of the installation.
The Initial Inspection
The objective of an initial inspection is to answer several important

questions.
(1) “Are the field-mounted devices compatible with the associated appa-
ratus?” This may occasionally be possible by consulting information
on labels of the field devices and the associated apparatus, but it is
usually necessary to consult the manufacturer's installation instruc-
tions. Some manufacturers supply several kinds of barriers, all of
which may be suitable for use with the same field device. Some com-
binations may be suitable for all material groups. Others might be
suitable only for Groups C and D, but not Groups A and B. A combi-
nation of field device and associated apparatus may be suitable for
Groups A and B if cable parameters are appropriately limited, but is
suitable for Group C or D if longer cables are used. It is essential to
review the installation documentation and satisfy one's self that the
associated apparatus and the field apparatus are compatible for the
hazardous location in which the field apparatus is installed.
This is not a question that should be asked for the first time during the
initial inspection of the site. During the design phase, the loop draw-
ing should, in some way, indicate the suitability of the selected combi-
nation of apparatus. The inspection should be simply a check, review,
and audit of the loop drawings.
(2) “Do the inductance, capacitance, and L/R ratio of wiring and con-
nected apparatus conform to the limits imposed by the certification or
control drawing?” As discussed earlier in this chapter, this is neces-
sary, but it need not be a difficult. It may be possible to confirm safety
in a plant with short cable runs with a minimum of detailed record
keeping. In a plant with long cable runs or cable with exceptionally
high values of the parameters, it may be essential that the installation
be documented in great detail. The question must be asked. The
amount of detail in the answer need be sufficient only to answer the
question.
(3) “Does any equipment behind the associated apparatus contains volt-
ages higher than the rated voltage?” If the answer is “yes,” this is not
prima facie evidence that the system is unsafe. It is only necessary
that someone make an engineering judgement about the system's
overall safety. This requires an assessment of the following:
• the energy level of the source

• its voltage level
• where in the system architecture the voltage is located, that is,

what sort of failures, and how many of them, must occur in order
for the high voltage to be placed on the input of the associated
apparatus.
As noted earlier in this chapter most CRTs are now located far enough
away from barrier inputs that concern for their high voltage appear-
ing as an input to associated apparatus demands an unreasonably
long chain of failures and coincidences. In the author's opinion few
systems are subject to hazard from CRTs. Although there is plenty of
room for the exercise of engineering judgment in deciding whether a
particular high-voltage source presents a risk to system safety there is
no excuse for not knowing that the source is present.
(4) “Is there any equipment connected to the loop that is not supposed to
be there?” Measurement and control systems are always changing.
There is always probability that the need for another indication has
resulted in the addition of an indicator not originally designed into
the loop. Perhaps someone decided that a particular valving arrange-
ment should be altered, and the system is now arranged in split-valve
configuration. Perhaps during installation it was found necessary to
use a transmitter of a different model than originally designed into
the system. All such changes should have been made to the loop dia-
grams as they were made, but were they? It is essential to determine
that the installed equipment is the same as that called out in the
design documentation. If there have been undocumented changes,
has the intrinsic safety of the resulting loop been verified?
(5) “Have grounding requirements been satisfied?” Observation can con-

firm whether enclosures have been grounded in accordance with local
codes and whether the barrier system has a continuous grounding
system that is isolated from the power and protective grounding sys-
tem except at one point. Did someone forget to install the grounding
jumper from a barrier bar to the collection point elsewhere in the
rack? Have shields been grounded in a safe manner—usually at one
point—with the other ends taped back and isolated?
(6) “Is the intrinsically safe wiring properly installed?” Is intrinsically safe
wiring isolated physically from nonintrinsically safe wiring? Is it iden-
tified by light blue color or labeling? Are there any intrinsically safe
cables run in the same tray or raceway as nonintrinsically safe cables?
Do spacing or partitions separate them? Have multiconductor cables
been damaged during installation in a way that might vitiate assump-
tions about isolation of circuits within the cable? If any of the system is
installed in conduit, or in cables that can conduct flammable gases
from one location to another, have seals been installed per local codes?
(7) “Are there any deviations from the manufacturer's installation

instructions?”
Some authorities recommend that it be observed whether printed circuit

boards are clean and undamaged. This author believes that this is counter-
productive and that this determination should be made only when
apparatus is in the shop. To make the determination it is necessary to open
the instrument enclosure. This exposes a field-mounted device to unrea-
sonable risk that the enclosure is not resealed effectively and that
environmental effects will damage the apparatus. In many devices, disas-
sembly of the equipment to determine whether the printed circuit boards
are clean and undamaged exposes them to greater probability of damage
than if they had not been inspected. In modern assemblies the determina-
tion that a board is “undamaged” is not one that can be made in the field.
In practice, a failure that might be in a chain of events leading to an igni-
tion is more severe than the failures that will disrupt function.
Malfunction of the board will be the cause of its replacement. Only in the
case of shunt elements incorporated in a design solely for safety and
whose opening does not cause malfunction of the apparatus would inter-
nal inspection of a device be at all justified. In this author's opinion such
an inspection should be done if the device is in the shop for some reason,
but it is not a practical inspection item in the field, being more likely to
cause problems than eliminate them. The rules for construction of appara-
tus with shunt protective elements are conservative. Failure in the field is
not likely.
It may be necessary to measure the resistance of the barrier grounding

lead, unless it can be confirmed by calculation, as in the case of very short
runs of large-diameter wire with few joints. This is probably the only mea-
surement that needs to be made. Unless the loop parameters approach the
maximum values permitted to be connected to the barrier, it is usually not
necessary to measure the parameters of the cables. In the early years it was
common to test all barriers to ensure that they would function and to
ensure that the components were grounded properly. In recent years this
practice has fallen out of favor, the opinion of many experts being that the
less disturbance of the barriers, the less likelihood of damage occurring.
Periodic Inspections
“How often should one inspect an intrinsically safe system?” The answer
to this question is similar to the answer to the question, “How often
should I see my doctor?” No answer meets every need. If a plant has been
subject to several or frequent design modifications so that the instrumen-
tation loops are likely to have been altered, then one should perform an
inspection frequently, and especially after any suspected alterations to the
system design or construction. On the other hand, if the plant has run
without difficulty since the last inspection, and any minor modifications
have been recorded and verified not to affect the safety of the system, then
a complete reinspection might be delayed for a long period of time. If the
plant operating procedures require formal updating of the records when
changes are made and reverification of the loops that have been changed,
and if these procedures are enforced, then complete inspections can
depend on the effect of other influences, like environmental effects. If the
plant has no procedures, or if they exist but are not enforced, then fre-
quent inspections to turn up the problems caused by changes to the
system of all kinds are desirable.
What can go wrong in a system that has been installed and shown to be
safe during the initial inspection? Corrosion may affect the resistance of
the connection to the ground stake. Grounding jumpers inside a cabinet
may have been disconnected during service and not reconnected. An
additional field indicator or signal processor was found to be necessary
and was inserted in the loop. A new control strategy or problems with a
measuring device demanded installation of a different device. Therefore,
those responsible for making inspections must determine the frequency of
inspections based on their knowledge of the likelihood that corrosion,
accident, or vandalism might damage the system, particularly the ground-
ing system; or that the necessity for design changes may have led to
changes in configuration of the control loops. There are no fixed rules.
What does one look for during a periodic inspection? The discussion about
determining the frequency of inspection also points to the answer to this
question. The inspector looks primarily for damage to the system, espe-
cially the grounding system, and for unauthorized changes to the system.
Maintenance of Intrinsically Safe Systems

Many intrinsically safe systems have been installed because it is univer-
sally agreed that maintenance may be performed on intrinsically safe
systems without deenergizing them and without obtaining a “hot permit”
that verifies that a flammable cloud is not present. If a control loop is mal-
functioning and a plant is producing an off-spec product because a
transmitter needs a calibration adjustment, the cost of the delay while
waiting for a safety person to reach the site and issue a hot permit can be
considerable. Though hot work is agreed-upon in principle, there is not
universal agreement on what kind of maintenance is permissible while a
loop is live. Most experts agree that the following operations in the haz-
ardous location while equipment is live are permissible:
• calibration adjustments
• replacement of items of electrical apparatus and cabling if there is

no danger of interconnection of different intrinsically safe circuits
during the operation
• use of test apparatus specifically called out in the manufacturer's
control drawing or instructions
• any maintenance operation specifically permitted by the control
drawings and instruction manual
The author believes that replacement of plug-in parts and assemblies in

the field is permissible, but some experts have reservations about doing so
because of the possibility that tools used to loosen retainers might acciden-
tally cause damage and arcing. This is mostly a moot point because such
field replacement of subassemblies is becoming increasingly rare.
Disconnecting more than one intrinsically safe circuit in a device is not

good practice unless care is taken to ensure that they cannot become inter-
connected. Similarly disconnecting or pulling cables with multiple
intrinsically safe circuits must be done only if bare ends are taped or other
precautions are taken to prevent interconnection.
Maintenance of associated apparatus may be carried out while the system

is live if two cautions are observed:
(1) The ground connection must remain intact.
(2) Protective elements shall not be disconnected from the intrinsically

safe circuit.
The safe use of CRTs and devices such as logic analyzers behind associated
apparatus is a matter of some dispute. These devices contain voltages
higher than the ratings of the associated apparatus and, in principle, could
become a source of ignition if a failure occurs. All installation rules state
that no device shall be used with associated apparatus that contains a volt-
age greater than the rating of the associated apparatus. The high-speed
current pulse of a CRT discharge can raise the potential of the ground ter-
minal of associated apparatus due to the self-inductance of the ground
loop between the CRT and associated apparatus. The energy stored may
be ignition-capable if the capacitance to ground is discharged, even though
the voltage pulse applied between the input terminals of the associated
apparatus is clamped to a safe level. However, the necessary conditions are
unlikely to exist outside the laboratory, and a significant hazard would
exist only if the intrinsically safe conductors entered a Zone 0 location. This
author's opinion is that failure of a CRT or similar device is infrequent and
is obvious to the user. More than one failure must occur to place high volt-
age on the input terminals of the CRT. Therefore it is, in the author's
opinion, safe to use such instruments behind associated apparatus if the
use is constantly attended, so that it can be disconnected in the event of a

high voltage circuit failure. Such devices should not be left unattended.
Nonincendive Equipment and Wiring

A very significant contribution of the first edition of ISA RP12.2 was a
definitive description of nonincendive equipment and wiring. Only the
name was new; the concept was old. Nonincendive equipment and wiring
is equipment and wiring that in its normal operating condition will not
ignite a specific hazardous atmospheric mixture. Such equipment had
been recognized, but not named, for several decades, in paragraph
501-3(b) of the NEC. In Class I, Division 2 areas equipment without
make-and-break or sliding contacts and without hot surfaces may be
enclosed in general purpose enclosures. In paragraph 501-3(b)(2) slidewire
contacts in potentiometers used in conjunction with thermocouples had
been specifically excepted from the prohibition of sliding contacts because
of the obviously low energy levels involved. In the 1965 NEC, the excep-
tion was replaced but “circuits which under normal operating conditions
do not release sufficient energy to ignite a specific hazardous atmospheric
mixture” were specifically recognized. The term “nonincendive” was
introduced in the 1975 Code, and the term “nonincendive circuits” is used
without the descriptive definition in the 1996 NEC.
RP12.2 also defined the contacts that must be energy limited, distinguish-
ing between contacts used during normal operation (which must be
energy-limited) and those used only during corrective maintenance,
which need not be limited. This distinction continues to this day in S12.12,
although a requirement has been added that the separation force for
plug-in components must be 15 N, and there must be a warning not to dis-
connect unless no hazard will result. (Plug in components need not exceed
an extraction force equal to three times their weight.)
In the 1975 NEC a concept peculiar to the NEC was introduced; the con-
cept of field wiring in circuits which will not cause ignition in normal
operation of the apparatus, taking into account also the effects of opening,
closing, or grounding of the circuit. This was introduced into 501-4(b) to
permit the use of ordinary location wiring in Division 2 locations. The
intent was to parallel the ability to use ordinary location wiring for intrin-
sically safe circuits in Division 1 locations. In Division 1, the intrinsically
safe circuit provides the requisite two failures to produce an explosion
because it is assumed that the probability that a flammable cloud is
present is high. In Division 2, the cloud is present only after a failure of the
plant containment system releases flammable material. An ignition source
will be present only after the electrical equipment fails. Though few users
of apparatus took advantage of the provision for use of general purpose
wiring, this philosophy was used during the mid-1970s and 1980s to jus-
tify live work on nonincendive field circuits in Division 2. It is recognized

worldwide that live work on intrinsically safe circuits may be permitted in
Division 1 or Zone 0 and Zone 1. This is because the intrinsically safe
apparatus is judged based on opening, grounding or shorting of field wir-
ing. The specification of nonincendive field circuits in the NEC provides
many users with the justification for live work on such circuits in Division
2. At present no other jurisdiction recognizes nonincendive field circuits in
a standard. The benefits this kind of circuit provides for live working in
Division 2 are not recognized outside of the United States. ISA S12.12
makes a clear distinction between nonincendive circuits, which are inca-
pable of releasing enough energy to cause ignition in normal operation of
the equipment if the circuit is interrupted by a switch or other contact, and
nonincendive field wiring, which is assessed taking opens, shorts and
grounds into account. Unfortunately the 1996 NEC defines “nonincendive
circuit” to include consideration of opens, shorts, and grounds. This defi-
nition does not encompass all the circuits that are referenced in 501-3(b).
Although there may be some confusion generated by this oversight in
defining the term in the NEC, most of the applications of nonincendive
circuits in 501-3(b) are internal to apparatus, whereas the circuits in
501-4(b) are external so there should be little practical effect.
In the early days of ISA's involvement with safety standards, the term
“nonincendive” embraced ignition-capable circuits that are not normally
sparking and are not, therefore, an ignition source until there is some fail-
ure. Currently the term is used by many to indicate energy-limited circuits
only. The other type of circuit is referred to as nonsparking.
A nonincendive component is a switch or other arcing device that will not

cause ignition when operated in a specified circuit. The voltage and cur-
rent in the circuit may or may not be related to the normal ratings of the
switch, and the circuit may contain specified values of inductance or
capacitance. To verify that a device is nonincendive it is exercised for
many cycles, typically 6,000, at its rated voltage and current to condition
the contacts. It is then actuated many times in the presence of an explosive
atmosphere to see whether ignition occurs. The test protocol is similar in
the various standards, but not identical. Some laboratories open the enclo-
sure to allow free access of the flammable atmosphere. The safety of a
nonincendive component derives from close spacings of contacts that are
massive and slow relative to the IEC test apparatus. Therefore, they
quench an incipient arc, so that higher energies are required for ignition.
In the IEC and British documents, an enclosed break apparatus is defined.

The tests are similar to those for nonincendive components, but the device
is not cycled before the test. The safety of an enclosed break device rests on
the fact that the strength of the enclosure will contain any ignition that
occurs within. Because of the small dimensions of the enclosure the pres-
sure rises are modest compared to those in an explosionproof enclosure so
normal construction is robust enough. These enclosed-break devices com-

pete with small explosionproof devices and are not presently recognized
in the ISA and CSA standards. There is insufficient understanding of the
technique and no demonstrated need for this different type of protection
in North America.
Restricted breathing, a technique discussed in Chapter 7, is not included in

the ISA and CSA standards. The use of gasketed enclosures against the
entry of gases and vapors has not been accepted by the electrical stan-
dards and certification community in North America, mostly because of
concern for continued effectiveness of gaskets over the life of the equip-
ment. Conn has used the concepts of restricted breathing, although not the
name or the defined protection technique, in several studies referenced at
the end of chapters 4, 7, and 8.
There have been widely different views in national standards circles about
nonincendive equipment, so that there has not been effective international
standardization. IEC Publication 79-15 was published only as a report.
There is, however, a strong foundation of agreement about the require-
ments. There is consensus worldwide that the standard IEC ignition testing
apparatus should be used for testing nonincendive circuits, and there is lit-
tle disagreement that the test factor should be 1.0 (i.e., no additional test
factor). The major cause for the long struggle to agree on any international
document was the different views in the U. K. and the U. S. about addi-
tional restrictions to be placed on apparatus that has been designed for
general purpose service. U. S. and Canadian experts insisted in IEC deliber-
ations that any industrial quality device could be used in Division 2 if it
were nonsparking and had no ignition-capable hotspots. British represen-
tatives pushed for additional requirements, such as more restrictive
creepage, clearances, and details, such as washers, under all terminal
screws. The difference in the two approaches was alleged to be due to the
lack of general purpose standards for electrical equipment in the U. K.,
although they have been pervasive in North America for many decades.
In the U. K. in the late 1970s, there were many who shared the North
American view, which is optimistic about the safety of nonsparking indus-
trial apparatus. Another group tended to agree in principle with the first
group, but wanted guarantees that a minimum standard of construction
would be met. Thus, Publication 79-15 became the document it is. It is
interesting to note that the group favoring a relaxed attitude towards use
of standard equipment in Division 2 also favored self-certification by the
manufacturer and evaluation with field wiring opened, shorted, or
grounded. This latter idea died. The author's opinion is that most Euro-
pean countries leaned toward the North American view. The text of the
IEC report, which was issued because consensus could not be reached on a
standard, allows most field apparatus to meet British requirements with
only the addition of washers under terminal screws. Other impediments
had been watered down for apparatus operating below 75 VDC or 60 VAC
that is in an IP54 enclosure or if the components have equivalent
protection.
Effective at the start of 1996 all electrical apparatus sold in the European
Community must conform to the Low Voltage Directive, which imposes a
responsibility for vending only apparatus that is not hazardous to the
user. In effect, all equipment must conform to some CENELEC general
purpose safety standard. In the future it may be easier for the British and
those who share their views to agree that ordinary industrial equipment
without special features is suitable for Division 2, Zone 2 use. Change may
be difficult. Years of exposure to the present IEC report and British stan-
dard have probably inured many designers to meeting unnecessary
requirements.
Another major philosophical difference between the U. K. experts and

U. S. experts was the assumption that equipment for use in Division 2,
Zone 2 must be certified. There is general agreement in most countries
(although not among all customers) that nonsparking apparatus can be
self-certified by the manufacturer or assessed by the user. It is reasonable
to expect third party assessment of other than nonsparking equipment
because few users have the capability to assess or verify a manufacturer's
claims. At present the mood in most of Europe is not to certify nonspark-
ing apparatus for use in Zone 2. Certification is becoming more essential
in North America because of customer demand and it is routinely
obtained by many manufacturers of nonsparking apparatus because it is a
no-cost extension of the approvals required for more hazardous locations.
At present much apparatus used in nonincendive field circuits in the

United States is approved for location in a Division 2 area, but it has not
been examined or marked specifically for use in nonincendive field cir-
cuits. Certified apparatus supplying nonincendive field circuits is usually
marked; uncertified apparatus is usually not marked, as in the case of
thermocouple terminals on controllers designed primarily for general pur-
pose use and only occasionally used with a thermocouple located in a
Division 2 location.
The important parameters of a device receiving energy are the same as

those of intrinsically safe equipment, Vmax, Imax, Pmax, Ci, Li, L/R ratio.
These C and L values may be smaller in the nonincendive application, but
the intrinsically safe values are usually used because they are readily
available.
It is common for devices supplying energy to use the same kind of param-
eters as for intrinsically safe apparatus; Voc, Isc, Ca, La, or L/R ratio. The
values of Ca, La, and L/R ratio are read from the curves using the values of
open-circuit voltage and short-circuit current without additional safety
factor and without considering failures in the apparatus. If the current

through the load under normal operation conditions of the loop is not
defined, this set of parameters is all that can be defined for the nonincen-
dive field circuit. If the normal current through the field device is defined,
as it is in 4–20 mA transmitters or in digital input devices, then additional
values, Ln and Cn, may be defined based on the normal voltage and current
in the loop, Vn, and Ln. The values of Ca, and La then become the maximum
values in the cable connecting the power source and the power receiver.
The values, Ln and Cn, are the maximum values for the cable plus the
receiving element. These values were defined properly in the 1986 edition
of ISA S12.12, but the 1994 edition confused the issue by calling for either
set of L and C values. This cannot be proper because the cable parameters
in a nonincendive field circuit must always be based on the open circuit
voltage and the short-circuit current. Using the cable parameters as the
maximum parameters for the loop is in practice not usually restrictive to
the application of the circuit, although it is unnecessarily conservative. Use
of the C and L values based on the normal flow of current in the loop does
not properly limit cable parameters and is therefore incorrect.
References
British Standard 1538:1949, “Intrinsically Safe Transformers for Bell Sig-
nalling Circuits.”
British Standard 1975:1953, “Primary Cells and Batteries for Intrinsically

Safe Bell Signalling Circuits in Coal Mines.”
British Standard 1259:1958, “Intrinsically Safe Electrical Apparatus and

Circuits.”
Hickes, W. F., “Intrinsic Safety,” Proceedings, 1960 Symposium on Safety for

Electrical Instrumentation in Hazardous Areas, pp. 49-64, ISA, Pittsburgh,
1960.
Haig, J., Lister, H. C., and R. L. Gordon, “The Testing of Flameproof and
Intrinsically Safe Electrical Apparatus,” IEE Conference Report Series, No.
3, pp. 39-49, Institution of Electrical Engineers, London, 1962.
Hickes, W. F., “Evaluation of Intrinsic Safety,” Instrumentation in the Chem-

ical and Petroleum Industries, Vol. 3, pp. 121-136, ISA, Pittsburgh, 1962.
Redding, R. J., “Intrinsic Safety,” Guide to Instrument Design, pp. 405-431,

Taylor and Francis, Limited, London, 1963.
Gel’fer, G. A., Ivanov, A. V., and Y. A. G. Medvedev, “Explosionproof

Electrical Equipment,” translated from Russian. U. S. Department of Com-
merce, Washington, 1964.
Thomas, V. N., “Design of Intrinsically Safe Apparatus for Use in Coal

Mechanical Engineer, pp. 295-329, May 1964.
Burgoyne, J. H., “Intrinsic Safety in the United Kingdom,” ISA Mono-

graph No. 110, pp. 95-108, ISA, Pittsburgh, 1965.
ISA RP12.2, “Intrinsically Safe and Non-incendive Electrical Instruments,”

ISA, Pittsburgh, 1965.
VDE 0170/0171d/2.65, “Vorschriften für explosionsgeschützte elektrische

Betriebsmittel,” Verband Deutscher Elektrotechniker, Berlin.
Burgoyne, J. H., “Safety in Great Britain,” ISA Monograph No. 111, pp. 52-
57, ISA, Pittsburgh, 1969.
Cartwright, J., “Application, Certification and Testing of Electrical Equip-

ment for Flammable Atmospheres,” Electrical Review, March 1968.

cuits,” SMRE Report No. 256, Sheffield, 1968.
Ministry of Fuel and Power Testing Memorandum No. 10, “Test and Cer-
tification of Intrinsically Safe Apparatus and Circuits.”
Bossert, J. A., “Safety in Canada,” ISA Monograph No. 112, pp. 32-35, ISA,
Pittsburgh, 1969.
Bossert, J. A., “Intrinsic Safety for Instrumentation,” Canadian Controls and

Instrumentation, pp. 24-26, February 1969.
Burgoyne, J. H., “Safety in Great Britain,” ISA Monograph No. 112, pp.
119-131, ISA, Pittsburgh, 1969.
Hickes, W. F., “Codes and Standards of Intrinsic Safety,” ISA Monograph

No. 112, pp. 15-18, ISA, Pittsburgh, 1969.
Magison, E. C., “Intrinsic Safety: A Review with Comment,” ISA Mono-

graph No. 112, pp. 9-14, ISA, Pittsburgh, 1969.
Bossert, J. A., “Testing and Approval of Intrinsically Safe Equipment,”

Instrumentation in the Chemical and Petroleum Industries, Vol. 6, pp. 35-41,
ISA, Pittsburgh, 1970.
Hickes, W. F., “International Standards of Intrinsic Safety,” Instrumenta-

tion in the Chemical and Petroleum Industries, Vol. 6, pp. 31-34, ISA,
Pittsburgh, 1970.
Cooper, L. R., “Experience with the Cadmium Breakflash Testing Equip-

ment,” pp. 6-10, Conference on Electrical Safety in Hazardous
Environments, London, 1971.
Bossert, J. A., “Investigation of a New Design of Electrode for the IEC Rec-
ommended Intrinsic Safety Test Apparatus,” CEAL Report 72/16,
CANMET, Ottawa, 1972.
NFPA 493, “Intrinsically Safe Process Control Equipment for Use in Haz-
ardous Locations,” National Fire Protection Association, Boston.
RP12.6, “Installation of Intrinsically Safe Instrument Systems in Class I

Hazardous Locations,” ISA, Pittsburgh, 1976.
Silver, S., “Investigation of ‘Contamination’ of Explosive Mixtures by Mul-

tiple Sparking at Ignition Energy Levels,” CEAL Report ERP/MRL 76/
100, CANMET, Ottawa, 1976.
Thomson, R., “The Mechanical Behavior of Electrodes in the IEC Cad-

mium Breakflash Apparatus,” Report MRP/PMAL-76-24, CANMET,
Ottawa, 1976.
Silver, S., “Variations of Ignition Sensitivity for V.D.E. Standard Intrinsic

Safety Apparatus and Other Factors Affecting Its Calibration,” CEAL
Report No. 355, Energy Research Laboratories, Ottawa, 1975.
Silver, S., “Effect of Electrode Conditioning on Ignition Sensitivity Vari-

ability for the Standard V.D.E. Intrinsic Safety Test Apparatus,” CEAL
Report No. 365, Energy Research Laboratories, Ottawa, 1975.
Magison, E. C., “Intrinsically Safe Systems—Trends in Design and Instal-

lation Standards,” IEE Paper PID-76-101.
Towle, L. C., “Maintenance of Electrical Instrumentation in Hazardous

Areas,” Measurement and Control, Vol. 11, London, 1978.
Hickes, W. F., and K. J. Brown, “Assessment of Explosion Probability for

Intrinsically Safe Apparatus,” Conference on Electrical Safety in Hazard-
ous Environments, p. 54, London, 1971.
Fraczek, Jerzy, “A Reliability Model for the Analysis of Hazards Caused

by Intrinsically Safe Apparatus,” Proceedings of the IFAC Workshop, The
Hague, 1986.
Fraczek, Jerzy, “The Influence of Testing Procedures on the Reliability of

Intrinsically Safe System,” Proceedings of the IFAC Workshop, Bruges, 1988.
EN 50154, “Electrical Installations in Potentially Explosive Gas Atmo-

spheres (Other Than Mines),” CENELEC, Brussels.
EN 50014, “Electrical Apparatus for Potentially Explosive Atmospheres.

General Requirements.” (BS5501, Part 1), CENELEC, Brussels.
EN 50020, “Electrical Apparatus for Potentially Explosive Atmospheres,

Intrinsic Safety 'I',” (BS5501, Part 7), CENELEC, Brussels.
UL913, “Standard for Intrinsically Safe Apparatus and Associated Appa-

ratus for Use in Class I, II, and III, Division 1 Hazardous Locations,”
Underwriters Laboratories, Northbrook.
IEC Publication 79-14, “Electrical Installations in Explosive Gas Atmo-

spheres (Other Than Mines),” International Electrotechnical Commission,
Geneva.
RP12.6, “Wiring Practices for Hazardous (Classified) Locations Instru-

mentation, Part I: Intrinsic Safety,” ISA, Research Triangle Park, 1995.
ISA S12.12. “Nonincendive Electrical Equipment for Use in Class I and II,
Division 2 and Class III, Divisions 1 and 2 Hazardous (Classified) Loca-
tions,” ISA, Research Triangle Park, 1994.
IEC 79-15, Electrical Apparatus for Explosive Gas Atmospheres, Part 15:
Electrical Apparatus with Type of Protection ‘n’,” International Electro-
technical Commission, Geneva.
CSA C22.2 No. 213-M1987, “Electrical Equipment for Use in Class I, Divi-
sion 2 Hazardous Locations,” Canadian Standards Association, Rexdale,
1987.
ISA-RP12.2.02-1996, “Recommendations for the Preparation, Content, and

Organization of Intrinsic Safety Control Drawings,” ISA, Research Trian-
gle Park, 1996.
C22.2-157, “Intrinsically Safe and Nonincendive Equipment for Use in

Hazardous Locations,” Canadian Standards Association, Rexdale.
“Excerpt from Health and Safety Research 1975,” Ex Magazine, October

1977, p. 21, Stahl, Kunzelsau.
Peterson, J. S., “Influence of Environmental Factors on Spark Ignition

Probability,” Report of Investigations 9566, Bureau of Mines, 1995.
CHAPTER 11
DESIGN AND EVALUATION OF
INTRINSICALLY SAFE APPARATUS,
INTRINSICALLY SAFE SYSTEMS, AND
NONINCENDIVE SYSTEMS
The first portion of this chapter deals with the design and evaluation of
intrinsically safe and associated apparatus. The second portion of the
chapter discusses the design of intrinsically safe systems. Although the
first part will be of direct interest primarily to those designing and manu-
facturing intrinsically safe apparatus, a quick review of the material by
those who use intrinsically safe apparatus in systems will help to establish
an understanding of the elements that make intrinsically safe circuits and
apparatus safe.
Design and Evaluation of Intrinsically Safe Apparatus

A design project for intrinsically safe or associated apparatus must include
the following tasks. Work will progress in general from one step to
another; however, the process, like any other design process, is usually an
iterative one; partially completing a step, proceeding to the next step or
two, and then looping back through the steps with more focus on detail.
Task 1—Establish the Intrinsic Safety Design Objective
Is intrinsic safety needed in all models, or should it be an add-on

option? In what material Classes and Groups must the apparatus
be used? Type ia or ib intrinsic safety? What temperature code?
Task 2—Design the Product
Task 3—Document the design for the certifying authority to save

time and money by easing the evaluation and certification process.
Task 4—Document the design in manufacturing drawings and

specifications to make it easy to control the details relevant to
intrinsic safety and its certification throughout the life of the
design.
416 Design and Evaluation of Intrinsically Safe Apparatus
Establish the Intrinsically Safe Design Objective
It is easy to write a specification for an intrinsically safe product. Just call

out that it shall be EEx ia IIC T6 and suitable for Class I, II, and III, Division
1, Groups A, B, C, D, E, F, and G. The user will be able to install the prod-
uct in any hazardous location.
But what will be the cost of such a mindless demand for the most inclusive
design specification? Intrinsic safety design increases costs. Added com-
ponents increase unit cost. Added design time increases development
costs. Alterations to circuits imposed by intrinsic safety compromise the
functions desired in the product or make it much more difficult to achieve
the desired function at the target investment, project completion date, or
unit cost. The end result of considering the available options may be that
competitive pressures in the target market demand an intrinsic safety
specification that is illogical or technically unnecessary, but the decision to
over-design should be undertaken knowledgeably, not carelessly.
In North America, most outdoor continuous process plants are classified

Division 2. The few Division 1 areas are small, and electrical apparatus is
not commonly installed in these Division 1 locations. There are exceptions,
especially in the application of level measurements. In many batch indus-
tries there is still a propensity to classify large areas around batch reactor
vessels as Division 1 and to install electrical apparatus within the Division
1 boundaries. To satisfy the requirements of both types of plant, many
manufacturers of control room interface apparatus have chosen to make a
standard design that supports nonincendive field circuits for Division 2
applications and an optional design that is intrinsically safe at somewhat
higher cost for the Division 1 applications. In much of the field-mounted
transmitter market, on the other hand, the effect of intrinsic safety on the
unit cost is small, and it is still the practice to offer intrinsic safety as a
standard feature.
Designing for Groups A and B or for IIC places more restriction on the cir-
cuit designer than designing only for Groups C and D or for IIA and IIB. If
the device being designed is a specialized device that will only be useful in
locations classified C and D, then it may be wise not to design for hydro-
gen and acetylene. Unfortunately, few products have the sharply defined
market that permits this decision, so most products need the more restric-
tive design objective. Committing to the narrower design objective may be
the only real choice if the product, like a solenoid valve, has to do a signif-
icant amount of work. In most industries there are relatively few locations
that are Division 1 or Zone 1 because of the presence of hydrogen, but
there is still a tendency to overclassify areas where hydrogen is used. If
hydrogen is used in a plant, customers frequently will specify intrinsic
safety for hydrogen even though there is little likelihood that the device
Design and Evaluation of Intrinsically Safe Apparatus 417
will ever be installed in a Division 1 location. Because the cost of designing

for hydrogen in 4 mA to 20 mA field-mounted transmitters is low, the
option of designing for C and D only is not usually a competitive option.
Picking the target temperature code is an important decision. Designing

for too low a surface temperature very quickly translates into increased
unit cost for larger components or encapsulation. There is no technical rea-
son for designing to T6. A competitive product may have a T6 rating, but
at present no gas or vapor that has been classified has an ignition tempera-
ture low enough to require T6 classification of the electrical apparatus. T5
is required only for equipment to be used for carbon disulfide. Therefore
T4 is a technically supportable design goal. The benefit of being able to
design to T4 is that if the apparatus is specified to have a Pmax of 1.2 W and
the component sizes are within the range specified in the standard, as
almost all in normal electronic designs are, the T4 rating is automatic for
an ambient temperature rating of 60°C. For an ambient rating of 80°C the
maximum power is 1.1 W, which may limit the choice of barriers some-
what. There is no need to measure surface temperature, and no need to
use encapsulation to reduce surface temperatures. Common intrinsic
safety barriers have maximum power transfer values of less than 1.2 W, so
the power specification does not limit the application of the device. If a T5
rating is essential, the same limitations on Pmax and component sizes ease
certification to T5 because the surface temperature of small components
may be 150°C instead of 100°C.
What values of Vmax, Imax, and Pmax should be specified? If a device is

intended to be connected to associated apparatus manufactured by others
then these parameters should be high enough to permit the customer a
wide selection of components in his system. For example, one would cer-
tainly design a 4 mA to 20 mA transmitter to values of at least 28 V, 100
mA, 1.2 W, unless there were a very unusual reason for doing otherwise.
These are the ratings of readily available barriers that can be used in con-
junction with a diode return barrier in a two-wire system that will not
exceed the current and power ratings of the transmitter. Many transmit-
ters have higher voltage and current ratings in North America because in
the early years of intrinsic safety some associated apparatus operated at
higher voltages, and because many experts held that the sum of the
return-barrier, short-circuit current and the supply-barrier, short-circuit
current should not exceed the transmitter rating. Unless one has knowl-
edge of the details of the intrinsically safe apparatus this is the only
position that is always safe. Although some experts have expressed the
opinion that it is highly improbable that the supply and return barriers
will be turned on at the same time, the first view is more prevalent.
If the intrinsically safe apparatus is to be used always with associated

apparatus designed by the same manufacturer, then the voltage and cur-
rent ratings should be selected to the lowest values compatible with the
desired function of the system and the desired length of cables between
the two devices. Lower values of voltage and current will almost always
ease the designer's job by permitting lower power and, perhaps, lower
cost components. The 1.2-W upper limit should be maintained, if at all
possible, for the reasons stated earlier.
What values of Ci and Li should be set as objectives for the designer? The
highest permitted values, Ci and Li, are the values of Co and Lo of the asso-
ciated apparatus it will be connected to, less the values of the connecting
cables. Until recently one could pick a value of 20% of the associated appa-
ratus Co and Lo values, or less, and forge ahead without much thought.
Now most designs must meet demanding electromagnetic interference
standards. The designer wants to have high values of Ci and Li to make fil-
tering easier. Blocking series diodes may prevent some of the filtering
elements from being visible at the input terminals, but only at the expense
of diode voltage drops. These must be traded off against the desired maxi-
mum cable length or minimum operating voltage specifications.
Design the Product
Designing intrinsic safety into a product must occur as the schematic is

drawn. It should be an ongoing consideration from the first attempts to
draw the schematic because there are always several ways to achieve
intrinsic safety. These choices should be traded off against the effect on
function, cost, and even the size or number of layers of the printed wiring
board. Intrinsic safety considerations are often taken into account only
after a nearly final functional schematic has been drawn. The designer
then passes the schematic to the intrinsic safety specialist, who adds the
components needed to ensure intrinsic safety to the schematic and writes
down the spacing rules. This procedure robs the design of the benefits of
tradeoffs between the functional design, cost, and so on, and usually is
used when the circuit design people do not want to assume any of the
responsibility for making the design intrinsically safe. The result is almost
certain to be a less-well-optimized design.
The steps in the design process of an intrinsically safe product are as

follows:
(1) Identify the circuit or circuits that must be intrinsically safe—
not an issue in intrinsically safe apparatus, but crucial in asso-
ciated apparatus.
(2) Compute currents and voltages in the intrinsically safe circuits

under normal conditions and compare them with the values
permitted by the relevant standard. Determine what adjust-
ments in component values may be necessary and assign ten-

tative values. Make tentative additions of protective elements.
(3) Compute currents and voltages in the intrinsically safe circuits

under fault conditions and compare with the permitted val-
ues. Though arcing at opens, shorts, and grounds in the nonin-
trinsically safe circuits in associated apparatus need not be
considered because they are not assumed to be exposed to a
flammable atmosphere, the effects of opens, shorts, grounds,
and component and wiring failures in these circuits on the
currents and voltages in the intrinsically safe circuits must be
considered. Complete the addition of protective elements. If
necessary, decide which circuits cannot be analyzed and must
be tested. If approximate analysis does not yield comfort that a
certifying agency will agree that the circuit is intrinsically safe,
either test, using the standard apparatus, or contract for a test
by the certifying agency early in the project.
(4) Determine whether capacitance and inductance viewed from

the input terminals of an intrinsically safe circuit conform to
the design objective.
(5) Identify constraints on layout, materials, and component rat-

ings that are implicit or explicit in the previous steps.
When evaluating apparatus to determine whether it is intrinsically safe an

engineer usually has a supporting document summarizing the reasons
why the design meets the requirements. This report would state what pro-
tective elements provide safety, what spacings are important, and how the
design meets the requirements of the relevant standard. The steps in eval-
uating an apparatus are similar to those used in design, but conducted in
different order. They are as follows:
(1) Identify the circuit to be intrinsically safe.
(2) Compute voltages and currents under normal conditions

assuming that spacings, protective component ratings, and
other construction features are as claimed. This yields famil-
iarity with the circuit topography and allows identification of
the most sensitive areas.
(3) If the circuit is not susceptible to analysis, note what ignition

testing may have to be carried out. In most cases the circuit
will have been designed with sufficient safety factor that test-
ing is unnecessary, but ignition testing may be necessary if the
circuit cannot be analyzed in terms of the few simple circuits
given in the standard.
(4) Review the layout, spacings, and component ratings to deter-

mine what faults need not be considered.
(5) Analyze the circuit under fault conditions, determine whether

the protective components and other component and material
ratings are suitable for these conditions, and determine which
faults must be simulated for test, if any.
(6) Analyze the circuit to determine whether the Ci and Li values

stated in the device specification are achieved.
(7) If the system passes ignition testing or circuit analysis review

the design with respect to other requirements of the standard,
such as enclosure tightness, marking, plastics surfaces, and
so on.
Whether one is designing or evaluating, the order of the steps is not

immutable, and the process is iterative. One should employ simplifica-
tions and approximations to seek out the major problem areas. First focus
on identifying the most significant design problem, or the circuit feature
most likely not to conform to intrinsic safety requirements. Then, repeat
the process, doing the arduous detail work.
Because the major elements of both design and evaluation are essentially
the same, the remainder of this chapter does not differentiate between the
two objectives. The principles of circuit analysis are the same for both, as
would be the discussions of the important design and construction tech-
niques for intrinsically safe design. The basis for all discussion in this
chapter will be the requirements of EN50020, as presented in Chapter 10,
and the curves in that chapter, which are not identical to those in
EN50020, but are close enough for tutorial purposes.
Design Techniques for Intrinsic Safety
The standards for intrinsic safety have now become so detailed that it is no
longer necessary to discuss the design techniques used for intrinsic safety
at great length. The standards are quite clear about the ratings required of
protective components and spacings. If the standard doesn't discuss a
technique, one should discuss its acceptability with the approving agen-
cies very early in the design process because it may not be acceptable to all
of them. The most common techniques are briefly summarized below.
Details are provided in Chapter 10 and are not repeated here.
• Separation of an intrinsically safe circuit from another circuit by
use of spacing, partitions, or shields. In a power supply, for
example, the wiring of the winding supplying the intrinsically
safe circuits might be spaced the requisite distance from nonin-
trinsically safe conductors in the primary or secondary so that no

contact can be assumed. This spacing of conductors is especially
necessary if the transformer or a feedback element is intended to
serve a protective function. At terminal connections partitions are
commonly used to ensure that the installer separates intrinsically
safe field wiring from the wiring of nonintrinsically safe circuits.
Spacing alone is permissible, but partitions are favored because
spacing may not guarantee separation if the installer is careless.
• Isolation of a low voltage winding supplying an intrinsically safe
circuit from the nonintrinsically safe mains circuit by a protective
transformer. If associated apparatus is powered from the mains,
there is almost no technique other than a protective transformer
to isolate line voltage from the intrinsically safe circuitry unless it
is feasible to use a shunt zener barrier designed to be safe against
line voltage.
• Isolation of intrinsically safe circuits from nonintrinsically safe
circuits by optical couplers. This technique is used, for example,
to isolate controller circuitry and the high level output circuit
from an intrinsically safe thermocouple measuring circuit.
• Current limiting by protective resistors to prevent voltages
applied to an intrinsically safe circuit from causing high currents
to flow in the circuit, which might violate the design target tem-
perature code. If the resistor is tested per present UL913 require-
ments, to 150% of the maximum fault voltage, it shall not
decrease significantly in resistance, and it shall not burst into
flame. This is not a trivial requirement. The enamel or varnish
coating on metal film resistors often flames if the resistor is over-
powered, as it can be under UL 913 rules. This is not an issue
under CENELEC rules, which permit operation of a protective
resistor at no more than two-thirds rating. It is important to
determine the need for current-limiting resistors early in the
design process. If the design must meet CENELEC requirements,
the resistor may have to be rated at least 1.8 (1.5 × 1.2) watts,
which uses up much more board area than the 1/4- and 1/8-watt
types that suffice for function.
• Clamping the generated emf of inductors with shunt diodes or
diode bridges to render an inductive element safe. In addition to
meeting the requirements for current rating, the diodes must be
mounted in a manner that makes opening of the connection to
the diode without removing the inductor from the circuit highly
improbable. If the inductor is a moving element, as in a force coil,
the diodes must usually be mounted on the inductor itself to sat-
isfy this requirement, and the connections to the inductor must
be robust. In Chapter 9 it was shown that shunting with a diode
with nominal voltage drop of 0.6 V would make the circuit's igni-
tion capability no worse than one with a voltage a diode drop

above the supply voltage and a short circuit current just a little
above the short circuit current of the supply.
Another way to assess the safety of the shunted inductance is to
assume that the inductance is a constant current supply. It and
the zener then form a rectangular source. If this source is safe
when breaking a resistive circuit, the shunted inductor is safe. See
the curves later in this chapter.
When testing a circuit with a diode-shunted inductance, or any
other resistive or low-inductance circuit, it is essential that any
safety factor be obtained by decreasing the series resistance, not
by increasing voltage. Assume that a 100 mH, 165 ohm coil in a
circuit supplied by 25 V is shunted by a diode and, under fault
conditions, the current is 0.1 A. An ignition test must therefore be
done with 1.5 × 0.1 = 0.15 A. Referring to Figure 10-3, it can be
seen that, in a 25 V resistive circuit, ignition of hydrogen occurs if
the short-circuit current is 0.25 A. The circuit will therefore pass
the ignition test if removing resistance from the circuit increases
the current to 0.15 A. If the test is conducted by raising the volt-
age to 37.5 V to obtain 0.15 A short-circuit current, ignition is cer-
tain because in a 37.5 V resistive circuit the short-circuit current
need be only about 0.1 A for ignition to occur. The circuit will be
rejected even though it is safe. Figure 10-9 indicates that this
caveat applies to any circuit with low inductance. One must be
careful not to use test parameters which falsely reject the circuit
when testing resistive and inductive circuits.
• Specifying the spacing between connections to protective compo-
nents to meet the tabular values that are presumed not to fail. In
modern printed circuit board designs, some surface-mounted
components, such as resistors, have solder pads that are sepa-
rated by less than the required dimension. It may be necessary to
use larger components than are required for functional reasons to
satisfy the protective spacing requirement. A similar caution is
demanded by the need to maintain conductor spacings at semi-
conductors to at least one-third of the tabular value to ensure that
shorting them is counted as a fault. This permits semiconductor-
shorting to be counted as a fault. In most intrinsically safe appara-
tus this is not yet a major issue, but it can be in associated appara-
tus, where the voltages may be much higher than in intrinsically
safe circuits. Conductors connected to the terminals must be
spaced appropriately also.
• Voltage limiting by shunt zener diodes to limit levels to safe val-
ues, for example, to permit the use of larger filter capacitor val-
ues. Determine the need for these early. They may have to be
rated at 1.8 watts, which takes up board space. In intrinsically
safe apparatus, the shunt elements to limit voltage are almost

always zener diodes, as shown in Figures 11-1 and 11-2. In associ-
ated apparatus, however, many manufacturers have used low-
power zener diodes in reference circuits that trigger semiconduc-
tor switches in crowbar circuits. This blows the supply fuse if the
reference voltage is exceeded. Gas tubes have been only infre-
quently used, as in limiting the possible voltage that could be dis-
charged from the input circuitry of a CRT should the high-
voltage drive somehow find its way to the input circuitry.
Figure 11-1 Series-Shunt Current and Voltage Limiting
• Insert a protective resistor to limit the current discharge from a

capacitor. The resistor serves no protective function after the
capacitor has short-circuited. Therefore, the resistor wattage rat-
ing can be quite small, determined only by the current that can
flow from the capacitor or the temperature rise resulting from
current flowing through it after the capacitor has shorted.
Because the total energy stored in the capacitor is low and the
resistor value is relatively low, a low-wattage resistor can usually
be used without affecting the T code.
• Series diodes to prevent internal capacitors from discharging into
field wiring. In most circuits these need have no special charac-
teristics other than a reverse voltage rating at least 1.5 times the
maximum capacitor voltage and spacing controlled so that short-
ing them counts as a fault.
It cannot be emphasized too often that the requirements for protective

components given in the standard apply only if a protective function is
necessary. Isolating transformers incorporated in thermocouple circuits to
avoid ground loops when the thermocouple is grounded need not meet
the requirements for protective transformers if they do not have an intrin-
sic safety function. Optical couplers, often used as feedback elements in a
variety of circuits where an isolating transformer is used in the forward

signal or supply path, need not be treated as protective components unless
they serve a protective function. If connecting the input and output of an
optocoupler or connecting the primary and secondary of an isolating
transformer do not affect intrinsic safety, there is no reason to design them
to be protective components. If isolation by transformer and optocoupler
is for the purpose of allowing common mode voltages to appear on the
input, the protective component specifications do not apply if the common
mode voltage is nil or very low in all applications where intrinsic safety is
required. ”Low” means that the common-mode voltage cannot charge
capacitors to ignition capable values after component failures. High val-
ues of common mode voltage in hazardous locations are not to be
expected. If they are present, then attention to intrinsic safety is usually
illogical.
If surface mounted components have spacings that are considered too

short, with or without fault count, no concern about this is necessary
unless shorting the component affects intrinsic safety of the circuit.
Revisiting Circuit Analysis
(1) Identify the Intrinsically Safe Circuit
Only battery-powered devices are likely to have all equipment and

wiring in a Division 1 location. A substantial part of almost all other
systems is located in a Division 2 or nonhazardous location and trans-
mits or receives signals over circuits leading to equipment in a
Division 1 location. Only the latter circuits need to be intrinsically
safe. The remainder of the system is called “associated apparatus”
because its operating conditions affect the energy that can be released
by the intrinsically safe circuits.
As process control systems have grown in complexity and sophistica-

tion, it has become common to find signal and control circuits
running between operating units of a plant. Although transmitting
and receiving equipment in each unit is installed in a nonhazardous
location, the signal and control circuits pass through a Division 1 loca-
tion. The apparatus at each end of the wiring run must be evaluated
for its effect on energy release in the Division 1 location.
(2) Identify the Critical Interface Location
Once the circuits to be intrinsically safe have been identified, the next
step in circuit analysis is to determine the location of the critical inter-
face; that is, identify the place where the circuit becomes intrinsically
safe. Preferably, all current- and voltage-limiting components should
preferably be grouped at the critical interface. A controller transmit-
ting 4 mA to 20 mA to a valve positioner, for example, should have
the critical interface as close to the output terminals as practical. Plac-
ing the critical interface near the terminals will reduce the number of
constraints on the design of the rest of the controller. It will also make
the safety of the design easier to understand and certify and make it
less likely that a careless design change later in the life cycle of the
design will compromise the intrinsic safety of the circuit.
(3) Analyze the Circuit under Normal Operating Conditions
The apparatus is in its normal operating condition when it conforms

with its design specification. Normal operation includes the effects of
supply voltage at maximum value, environmental conditions within
the ratings of the apparatus, the most unfavorable combination of
component tolerances, and adjustments at the most unfavorable
setting.
Normal operation also includes opening, shorting, or grounding of

connecting wiring between enclosures in the intrinsically safe portion
of the system, which is normally installed by the user, and over which
the apparatus manufacturer has no direct control. If the system is
intrinsically safe, this wiring need not be in conduit or robust cables
designed for use in hazardous locations. It may be lightly insulated
wiring designed for general purpose applications. This connecting
wiring is therefore presumed to be susceptible to damage. Its terminal
connections are assumed to be susceptible to shorting, opening, or
grounding during the expected life of the installation.
In principle, the steps to be followed when analyzing the circuit under

normal operating conditions are as follows:
(a) Draw a complete system diagram showing the value of all com-
ponents in all of the circuits connected to that portion of the cir-
cuit that is intended to be intrinsically safe. After some
experience with analyzing circuits for intrinsic safety, the reader
will find that a complete schematic of all the circuits behind the
locations of the critical interface is not always necessary and that
use of some of simplifying techniques discussed later in this
chapter will permit use of block diagrams for many portions of a
complex system.
(b) Indicate on the diagram the maximum values of circuit voltages

and currents, without considering shorts, opens, or grounds in
the wiring but assuming maximum supply voltage and unfavor-
able settings of adjustments. Component tolerances and effects of
environment must be considered, but this consideration can be
deferred until after analysis under fault conditions when the cir-
cuit design is nearly complete.
(c) Compare the current through each inductor that affects energy
release in the Division 1 location and the voltage across each
capacitor that can discharge in the Division 1 location or into a
line running into the Division 1 location with the values on the
reference curves applicable to the gas group involved in the rele-
vant standard (similar to those included in Chapter 10). Capaci-
tors in parallel or inductors in series must be summed and
treated as a single component. If any current or voltage multi-
plied by the required factor is greater than the reference value,
the circuit cannot be regarded as intrinsically safe. Make tentative
alteration of circuit constants or add a protective component. Cir-
cuit changes or constructional features to make the system intrin-
sically safe cannot be finalized until the analysis has considered
both normal and fault conditions. Further consideration of this
same portion of the circuit under fault conditions may show
additional protection to be required. It is especially helpful when
analyzing a system to record each step in a fault table, such as the
one presented later in this chapter. Recording each step in the
analysis gives greater assurance that important steps have not
been omitted. The tabular presentation also makes it easier to use
early calculations in later steps to save time, and to explain the
course of analysis to others.
(d) Compute the worst-case voltage at each capacitor and current

through each inductor, considering shorting, grounding, or open-
ing of external connection wires, separately, not in combination.
Compare these values with the reference values as in step (c).
Should any value be above the reference curve values, make a
tentative selection of protective element or alter circuit constants
and continue analysis under fault conditions.
(e) Determine open-circuit voltage and short-circuit current in resis-

tive circuits; compare against the reference curves. Adjust com-
ponent values as needed.
Analyze the Circuit under Fault (Abnormal) Conditions
When analyzing a system under fault conditions consider all probable

combinations of faults that can increase energy levels in the intrinsically
safe portion of the circuit. It does not matter whether the fault occurs in
the intrinsically safe portion of the system or in the associated apparatus.
A fault in circuitry in the Division 2 or nonhazardous location that could
connect a nonintrinsically safe circuit to an intrinsically safe circuit will
make the system unsafe.
(a) Compute voltages and currents at capacitors and inductors after
single faults and for the worst-case combination of two component
or wiring failures in addition to opening, shorting, or grounding of
external wiring. Compare these values to those determined from
the reference curves, taking into account the test factors required.
This step is the most tedious one in the analysis because a large
number of combinations of faults must be considered before the
worst case is determined.
(b) Adjust circuit configuration or component values to meet the

requirements of the standard.
(c) Fine tune the analysis by taking into account any tolerances that
have not yet been considered.
When system hardware or drawings are available, the fault analysis can
be truncated. Review the hardware or drawings to identify spacings or
other isolating constructions that eliminate some faults from consider-
ation. If the transformer is alleged to be a protective component, or if other
protective components are used, assume that they meet the pertinent
requirements and shorten the analysis. If intrinsic safety is verified at this
level of analysis, assess the suitability of the protective components.
Ways to Shorten the Analysis
Even a quick perusal of the illustrative example in this chapter will con-
vince the reader that a slavish devotion to the order of steps for circuit
analysis given above under normal and fault conditions requires so much
time and effort that simplifying assumptions must be used whenever pos-
sible to decrease the drudgery of analysis. Here are some useful
simplifying techniques.
(1) Base the first-pass analysis on the maximum available current and
voltage, taking into account the applicable test factor. Associated
apparatus based on shunt zener diodes often has the same value of
Voc and Isc after one fault and after two faults. Analyze the circuit with
one fault and the factor of 1.5 applied to the fault voltage or current,
and eliminate many steps. Repeat with two faults and no factor.
(2) Take advantage of the fact that, if there are no circuits that increase
the voltage, any resistive circuit fed by an intrinsically safe source
must also be safe relative to arc ignition. The only bar to intrinsic
safety relative to arc ignition is the presence of inductive or capacitive
storage elements.
(3) Add up all capacitances (or inductances) and assess the sum using the
highest value of voltage or current. If this assessment fails, revert to
single and two-fault analyses, but use the highest voltages and cur-
rents.
(4) Neglect the presence of resistive or capacitive shunt elements in evalu-

ating inductors. There is little data about the igniting current for paral-
lel combinations of elements. If the circuit is safe without the resistive
shunt element, it will certainly be safe with it. The same will probably
be true if the capacitive shunt is safe when assessed alone at the fault
voltage, but this is not guaranteed because some parallel L-C circuits
may be more incendive than the individual components alone.
(5) Assume all inductors are air-core. If they are analyzed to be safe, it is
almost certain that they will be safer if they have ferromagnetic cores.
If they are found not to be safe, testing or computation must be used
to verify safety. See the discussion of inductors with ferromagnetic
cores later in this chapter.
(6) Treat resistor-protected capacitors as batteries. If the current flowing

from a capacitor discharging through a protective resistor is less than
that which would be intrinsically safe in a resistive circuit operating
at the maximum voltage on the capacitor, it will certainly be safe. It
will be very much overdesigned, so this technique often leads to solu-
tions that are unacceptable because they decrease the amount of filter-
ing provided by the capacitor.
Power Supplies
Standards for intrinsic safety show the ignition characteristics of resistive
circuits by presenting a curve of open-circuit voltage, Voc plotted against
short-circuit current, Isc. This ignition characteristic is valid only for power
supplies whose V-I characteristic is a straight line; that is, the Thevenin
equivalent circuit is a voltage source in series with a resistor, as for the
unregulated supply in Figure 11-2. Of course, anyone who has studied
how such ignition data are obtained should know the significance of the
curves. However, when one is analyzing a circuit, the fact that the V-I
characteristic for a particular power supply is not linear is often forgotten.
The common 4 mA to 20 mA controller output or analog output circuit is

frequently treated as a resistive source. Because it is a constant current
source, it should not be analyzed using the curves in the standard. Curves
such as those generated by Johannsmeyer must be used to guide design
(see Figure 11-4).
Figure 11-2 illustrates the V-I characteristics of the three idealized kinds of
sources. The unregulated supply has a resistive output impedance. The
constant voltage supply has essentially constant voltage, Ez, until, at some
critical load current, the regulator no longer can operate. The voltage then
decreases along the resistive characteristic of the unregulated supply. The
supply with a rectangular output characteristic produces constant current
Ic until the voltage drop across the load reaches Ez. In practice, many
power supplies have foldback characteristics such that the output voltage
is constant until a critical level of current and then it decreases along a line
that sweeps back toward the origin so both output current and voltage
decrease as one tries to increase output current. These must be analyzed as
rectangular sources for lack of any other data, unless the specific circuit
has been tested.
Widginton emphasized the pitfalls in analyzing constant voltage regulated

power supplies. The load voltage for regulated power supplies is nearly
constant at one value until the regulator no longer can operate. The load
voltage then falls as current increases further along a line determined by
the unregulated supply voltage and the circuit resistance. If one enters Fig-
ures 10-7 and 10-9 with Ez and Isc, it is as though the supply were
unregulated with the characteristic shown by the dotted line in Figure 11-
2. The maximum power delivered to the load occurs at ½Isc, indicated by
point Pmax. It is obvious that the actual supply is capable of delivering
greater power than this.
One safe method of examining a regulated supply is suggested by the fig-

ure. If the unregulated open-circuit voltage and short-circuit currents are
in the safe region of Figures 10-7 or 10-9, a regulated supply will also be
safe. This is, of course, a very conservative way of evaluating a circuit. As
Figure 11-2 illustrates, the maximum power, Pmax, delivered by the unreg-
ulated supply may be much higher than the regulated supply can deliver.
Widginton's curves for shunt-stabilized supplies, based on the standard
IEC apparatus are shown in Figure 11-3. These may be used for any kind
of regulated supply. Plot the V-I characteristic of the actual supply.
Choose a shunt zener diode supply whose characteristic is always on or
above the plotted characteristic. If the shunt-regulated supply is safe, the
actual supply is also safe.
Figure 11-2 Voltage-Current Characteristics of the Three Idealized Power

Supplies
The dotted curves in the ignition characteristics for hydrogen were pre-
sented in the Capp-Widginton paper (1975). They stated that the igniting
current values are lower because zener diodes with much lower self-
capacitance were used in more recent investigations. Presumably the data
for methane ignition in Figure 11-3 should be adjusted also; but Capp and
Widginton did not present new values for methane. This author has,
therefore, elected to retain the older curves in order to provide some guid-
ance in design, although the user must adjust the data downward.
The analysis of ac-powered supplies derived from ferroresonant constant

voltage transformers can be considered to be similar to analysis of regu-
lated dc supplies. The V-I characteristics of the two kinds of supplies are
similar, and the safety considerations are also similar.
Figure 11-3 Relation Between Minimum Igniting Current and Stabilized Voltage
for Shunt Zener Diode Stabilized Supplies
Johannsmeyer treated supplies with a rectangular V-I characteristic in his

1979 paper. Figure 11-4 shows his ignition data for hydrogen-air and eth-
ylene-air mixtures. These were taken with the IEC apparatus at an ignition
probability of 10−3.
Figure 11-4 Ignition in R-L Circuits Using a Power Supply with Rectangular
Output Characteristic: (a) Ethylene-Air, Group IIB; (b) Hydrogen-Air
(Group IIC) [Source: Ex Magazine]
SMRE developed curves for resistive circuits with rectangular sources for
methane-air and hydrogen-air mixtures. The data for hydrogen-air agree
well with the data of Johannsmeyer. These data were developed as a way
to aid assessment of inductors shunted by zener diodes. If the inductance
is considered to have a long time constant, it can be viewed in the worst
case to be a constant current source limited in voltage by the zener diode.
If the circuit is safe relative to these curves, the actual circuit must be safe.
For ignition of hydrogen, a resistive circuit would have the ISC values
shown below, depending on the circuit type.
Circuit Type ISC - A
Resistive - 18 V 0.67
Resistive - 30 V 0.150
18 V trapezoidal (EB = 30 V) 0.18
Rectangular - 30 V 0.045
Rectangular - 18 V 0.110
The Halama and Thurlow-Green papers present an approach to power

supply design that may receive more attention in the future. The major
objective of both designs is to control the duration of any arc that occurs in
the intrinsically safe circuit to such a short interval that ignition cannot
occur. In the Halama design, the approach is to detect the spikes in the
output voltage caused by arcing. In the Thurlow-Green design, detection
of higher than normal current clamps the circuit at a low voltage.
The motivation for this approach to power supply design is to be able to

supply more power to an intrinsically safe load than is possible with con-
ventional current limited and zener diode clipped designs.
Inductors with Ferromagnetic Cores

We have always referred to the stored energy in an inductance as being
calculable from the expression
1 2
W = --- Li
2
where
W = stored energy in joules
L = inductance in henries
i = current in amperes
This expression is valid for calculating energy; however, it is valid only if

L is the same at all current levels. This is always true of air-core inductors
and is often approximately true over a narrow range of currents in induc-
tors with iron or other high permeability cores. Frequently, however, the
energy computed from measured values of L for inductors with ferromag-
netic cores is larger than the energy actually stored in the inductor. In
short, the value of inductance of a component with a ferromagnetic core
measured by conventional methods has no necessary and predictable rela-
tionship to the energy stored in the inductor at a given current level. If one
knows the geometry and constructional details of the inductor and the
magnetic properties of the core, the amount of energy stored in the induc-
tor can be estimated.
The problem of computing stored energy in a saturable core inductor can

best be understood by recognizing that the expression W = ½Li2 is a special
case of a more general expression,
B
W = V ³ H dB
0
When μ (the permeability of the material) is constant and independent of

H, then B = μH. This is the case in an air-core inductor. By definition, the
parameter L is equal to Ndφ/di, the change of flux linkages in a magnetic
circuit caused by changing the current which produces the flux.
Also,
dφ = AdB
1
di = ---- dH
N
where
A = the area perpendicular to the magnetic path
l = the length of the magnetic path
N = the number of turns through which the current I flows
B = flux density
H = magnetizing force
V = Al = the volume of the magnetic field; that is, the permeability of
the magnetic material is assumed to be high with respect to
air, so that all lines lie in the magnetic metal.
Therefore,
2
Ndφ AN dB
L = ----------- = ----------- --------
di l dH
For the air-core case where μ is constant,

2
AN
L = ----------- μ
l
B H H Al 2
W = V ³ H dB = Al ³ B dH = Al ³ μH dH = ------ μH
0 0 0 2
but
Ni
H = ------
l
so that
2 2
AN μi 1 2
W = -------------------- = --- Li
2l 2
When inductors have ferromagnetic cores, the permeability, μ, is not inde-

pendent of H, and the differential form and the general integral must be used.
A convenient form is
i
W = ³0 L ( i )i di
Since φ and B, and i and H, are related by constants in any particular case,
it is convenient to continue the discussion of the basis of the familiar B-H
curve, Figure 11-5a.
Figure 11-5a Magnetizing Characteristics

The straight line is a plot of the familiar relationship for air, B = μvH,
where μυ is the permeability of free space. The curve marked “iron” is not
drawn to scale, but is representative of core materials with nonlinear B-H
curve. At any magnetizing force, H, the ratio BI/BA defines the relative
permeability of the material, μr. For the material shown, since the B-H
curve become flat at high values of H, the relative permeability decreases
as H increases.
The slope of the B-H curve at any point is the incremental permeability,
ΔB I 2 ⁄ ΔH I2 , ΔB I1 ⁄ ΔH I1 , or dBA/dHA. For air or other paramagnetic mate-
rial, the incremental permeability is constant. In ferromagnetic materials,
the incremental permeability may be low at low-flux densities, increasing
to as much as 800,000 gausses per Oersted [1 Tesla/(A)(m)] at higher flux
densities and falling to very low values again at saturation flux density.
As was noted above, the incremental permeability, dB/dH, is proportional

to Ndφ/di, which is L, the inductance. With increasing values of H, ferro-
magnetic materials, as a class, display decreasing incremental
permeability and, therefore, decreasing inductance. In modern square-
loop materials, the inductance may change by a factor of thousands,
depending on the level of flux density.
The stored energy in the magnetic field, stated in integral form earlier in
this section, is for any value of H, proportional to the area to the left of the
B-H curve. There is no simple relationship between L at H = 0 and the
stored energy at H = H1; nor is there any apparent relationship between L
measured at H = H2 and the stored energy for that value of H. Yet most ac
bridges measure the inductance at Havg = 0. Even special incremental
inductance bridges built for transformer and filter inductor measure-
ments, which provide for measuring inductance at different levels of dc
magnetization, offer little help. The low inductance measured if the induc-
tor is operated on that the flat portion of the B-H characteristic would give
a very low estimate of the energy stored.
A certain, but often very conservative rationale, is that if the inductance

measured at Havg = 0 is used as though it were the value of an air-core
inductor, and the stored energy (or circuit current) is computed to be at a
satisfactory level, there is no need for further concern.
Energy storage may be calculated using data from observed oscilloscope

traces if the inductor is discharged through a resistor from a steady-state
current level. This method is tedious and approximate and is probably sel-
dom worth the effort. Stored energy may also be approximated by
measuring the incremental inductance at several current levels and inte-
grating L (i) i di over the desired range.
The stored energy in an inductor can be estimated roughly using conven-

tional magnetic circuit analysis if the dimensions of the core and its
magnetic properties are known. Calculate H from the number of turns, the
current for which stored energy is to be determined, and the dimensions.
Estimate the area to the left of the B-H curve and multiply by the core
volume.
The calculations are much simplified if one uses a spreadsheet program on

a computer. The area to the left of the B-H curve can be approximated by
numerical integration (see Figure 11-5b). For several levels of current in
the range of interest, calculate the related values of H by multiplying the
current by the number of turns in the coil and then dividing this product
by the mean length of the core. From the B-H curve read the values of B
corresponding to each value of H. If the values of H are close enough
together the incremental addition to the area by adding another pair of H
and B values can be approximated by treating the area between two adja-
cent values of B as a trapezoid. The area of each trapezoid is the average
value of the two values of H multiplied by the difference between the two
values of B. Sum these incremental areas to obtain the total area at the cur-
rent of interest.
Figure 11-5b Estimating Stored Energy in Inductor

The table below shows a calculation of stored energy in a transformer with

a 100 turn primary when the secondary winding is open circuited. The
mean length of the magnetic path, is 34 mm, and the core area is 9 mm2.
The core material is M-19 lamination steel, used for power transformers.
The first three columns calculate H as described above. Column 4 lists the
corresponding values of B; column 5 computes the incremental area of
each trapezoid; and column 6 sums the incremental areas to obtain the
total for any value of current. The last column is Column 6 multiplied by
the volume of the core, 0.306 × 10−6 m3, to yield the stored energy in micro-
Joules at each level of current.
These calculations show that the energy stored in the core approaches the
ignition energy of hydrogen in the IEC apparatus, 40 μJ, at a current of 100
mA. The calculated inductance of the coil at currents of 10 mA and 20 mA
is about 25 mH, the value that a bridge would probably measure. The
igniting current for hydrogen in the standard IEC apparatus for 25 mH is
about 45 mA. The magnetic properties of this core appear to permit a
higher current level than one would expect for an air-core inductor of the
same value. Because the core material for which these calculations were
made does not saturate at the low values of H investigated here, the effects
of saturation are small compared to what one would expect of a material
with higher permeability at low values of H and a lower value of satura-
tion flux density. The example does illustrate that small cores have limited
energy storage capability.
Current- Ampere B, Stored Energy -

H, AT/m Delta Area Area
Amperes Turns Webers/sq. m microJoules
0 0 0 0 0 0 0.00
0.01 1 29.41 0.25 3.68 3.68 1.13
0.02 2 58.82 0.53 12.35 16.03 4.91
0.03 3 88.24 0.75 16.18 32.21 9.86
0.04 4 117.65 0.86 11.32 43.53 13.32
0.05 5 147.06 0.95 11.91 55.44 16.96
0.06 6 176.47 1.05 16.18 71.62 21.92
0.07 7 205.88 1.10 9.56 81.18 24.84
0.08 8 235.29 1.13 6.62 87.79 26.86
0.09 9 264.71 1.16 7.50 95.29 29.16
0.1 10 294.12 1.20 11.18 106.47 32.58
If the inductor has an air gap to prevent saturation of the core by dc, the
stored energy can be estimated by assuming that the core material has infi-
nite permeability so that all the energy is stored in the air gap. In this case,
if A, I, and V pertain to the gap, then
1 2 1 2 2
W = --- VμH = --- AlμN i
2 2
All of the calculations are only approximations. Even if the calculated

stored energy based on magnetic circuit methods is exact, the amount of
energy released to the arc will be reduced by hysteresis and eddy current
losses and by the amount of energy remaining on the stray capacitance of
the coil after the arc discharge has ceased. Additionally, the amount of
energy actually dissipated in the arc depends on the speed of break and
the duration of the arc. This same factor affects the accuracy of estimates
derived from discharging through a resistor. If the L/R ratio of the induc-
tor is high, the discharge time may be much longer than the initiation time
of the ignition process, and, as shown in Chapter 9, total stored energy
may not be a sufficient measure of ignition capability. The stored energy
may appear to be ignition capable, but ignition testing may show the com-
ponent to be safe.
In SMRE Report 254 Widginton discussed ignition by inductors, with

emphasis on the issues raised by ferromagnetic core inductors. He pre-
sented a circuit for estimating the energy that a particular inductor might
discharge. The circuit simulates a showering type of discharge by a con-
stant voltage discharge at a voltage of 200 V or an arc discharge by a
constant voltage discharge at a lower voltage, say 100 V. The details of the
circuit are not discussed further here because the requirements for the
components of the circuit, the switching device that initiates the discharge,
and the time to set up the circuit and prove that its operation approxi-
mates the real discharge make it unlikely to be useful to anyone other than
a researcher. In his report he reported measurements of stored energy on a
number of relays, a transformer, and a headphone. The range of stored
energy was from 0.5 mJ to 10 mJ, but not all devices were tested over the
entire range. His data for the transformer and the headphone show the
energy stored to be proportional to the square of the current over the
range measured. Some relays followed this pattern, but others tested
showed slower than square law increase in energy because of the effect of
saturation. However, for such devices, the square law assumption will
give conservative estimates of stored energy if one extrapolates from a
measurement of stored energy at one level of current.
In small transformers for operation at tens of kiloHertz the measured

inductance may be several mH or higher, but the maximum calculated
stored energy may be less than 10 mJ when calculated as described above.
An approval house may accept the calculation without further tests.
Silver investigated the characteristics of iron-core inductors by making

measurements of the secondary windings of several transformers (prima-
ries were open-circuited). He measured stored energy in the circuit shown
in Figure 11-6.
Figure 11-6 Circuit for Measuring Stored Energy in an Iron-Core Transformer
To store energy, close switches A and B and adjust R until I is the desired
value. Diode D prevents current flow in the secondary. When I has
reached its steady-state value, open switch B. This must be a high-speed
break. The energy stored in the inductor now charges the capacitor. Read
the capacitor voltage with an oscilloscope and compute stored energy
from W = ½CV2. The computed energy is likely to be a low estimate
because of losses in the transformer, loss in the diode and switch, and
unknown shunt capacitances in the secondary. The diode D1, R1 and the
voltmeter provide a means to determine that steady state has been
reached after the switches have been closed and before switch B is opened.
Widginton stated in his report that discharging an inductor into a capaci-

tor generally yields stored energy values that do not agree with those
dissipated in a real discharge because the rate of discharge in the measur-
ing circuit differs from a real discharge.
Silver used this method to measure the stored energy in the secondaries of
three iron-core transformers over a wide range of currents. He also mea-
sured incremental inductance for many values of dc current and
integrated it to determine stored energy at each current level. Using the
break-spark apparatus, he also determined current required for ignition in
pentane-air and hydrogen-air mixtures, computed the ignition energy,
and compared it with the energy required using air-core inductors. His
data for hydrogen show the expected result: the stored energy required
for the iron-core inductor is greater than for an air-core inductor. The cal-
culated ignition energies for iron-core inductors in pentane-air were lower
than the energies for air-core inductors. It is this author's opinion that this
surprising result is only a measure of the difficulty in measuring or calcu-
lating the energy stored in a nonlinear device.
Silver's measured values of stored energy were less than those calculated
from incremental inductance for two of the three transformers. Agreement
between measured and calculated energies was only 10% to 20% for all
three transformer windings. Because any leakage currents subtract from
the measured energy and because the value of the capacitor voltage must
be read at its peak, measurements of stored energy should not be expected
to be precise.
This author feels that the calculated values of inductance may be too low
because of the method used. The incremental inductance data were plot-
ted on log-log paper. From these curves equations for L(i) were derived.
These equations were then multiplied by i and integrated. Curve-fitting
from a log function tends to minimize deviations between the equations
and the logarithm of the data rather than minimize the absolute error
between the equation and the data. A rough numerical integration of the
incremental inductance data suggests that this method might yield a
somewhat higher calculation of stored energy.
Another factor that may contribute to the difficulty in relating calculated

stored energy to that required to ignite a gas mixture is that the coil cur-
rents were considerably higher for pentane-air than for hydrogen-air. The
coil was driven much farther into saturation during the pentane-air igni-
tion tests than in the hydrogen-air ignition tests. Integration was,
therefore, carried out over a much wider range of currents, likely leading
to greater error and greater susceptibility to any experimental errors in the
incremental inductance determinations.
The most important aspect of the Silver investigation, however, is that his
data show that if one used the measured igniting current and the ignition
energy for pentane-air of an air-core indicator, the effective inductance of
the three iron-core inductors is less than the measured incremental induc-
tance at zero current by a factor of 10 to 40. This substantiates the major
point of this discussion of iron-core inductors, that incremental inductance
is not a reliable measure of igniting capability.
It is clear from these discussions that any method of assessing the stored
energy in an inductor with a ferromagnetic core is likely to be time con-
suming. The most direct determination is, in this author's opinion, an
ignition test of the circuit in which the inductor is used.
Iron-core inductors and transformers provide an excellent example of the

danger of trying to apply cookbook procedures to the design or testing of
intrinsically safe circuits. As the current level in an iron-core inductor is
pushed into the region of saturation, the rate of change of stored energy
with change in current decreases rapidly. Doubling the current may not
quadruple energy, as in an air-core inductor. Consequently, if one halves
the igniting current for an iron-core inductor, the available energy does
not drop to one-fourth the igniting value. Silver's data showed that halv-
ing the current reduced energy to about one-half to two-thirds of the
starting value. To get a factor of four on energy necessitates reducing cur-
rent to one-fifth to one-sixth the starting value.
The author feels that no rule can be set that in some circuits will not be
ultraconservative and restrictive. A factor of two on current may seem
marginal in some circuits. A factor of six might seem ridiculously conser-
vative in others. Much would depend on how sharply the iron core
saturates and the failure modes of other components in the circuit.
Transformers
Because the inductance of a transformer depends on the way in which it is

connected in the circuit, transformers represent a slightly more complex
problem than ferromagnetic core inductors. Much of what follows is
applicable in principle to air-core transformers, but these are seldom large
enough to present a safety problem. A transformer may be represented by
the equivalent circuit Figure 11-7. This equivalent circuit is the same as
that found in any elementary text. Transformers with multiple secondaries
can be represented by connecting the equivalent circuits of additional sec-
ondaries between C1 and C2.
Figure 11-7 Transformer Equivalent Circuit

Power transformers represent a particularly simple case. Usually the

power transformer inductance is only of interest when viewed from the
secondary winding. Unless the primary winding is connected to the
power line, there is no hazard. When the primary winding is connected to
the power line, the load Z in the equivalent circuit represents the source
impedance of the power line. In any practical instrument system this
impedance is usually so low that it can be presumed to be 0. For most
transformers, if the load impedance is 0, the impedance looking into a
winding is not much different from the sum of the series impedances,
n1 2 n1 2
R 1 + § -----· R 2 + j X 1 + § -----· X 2
© n 2¹ © n 2¹
The impedance represented by the center leg of the equivalent circuit is

usually much higher than the winding impedances (it represents the mag-
netizing current and core losses).
The inductance of a power transformer connected to the power line is,

therefore, the leakage inductance of the transformer windings measured
at low frequency and with direct current flowing in the windings where
applicable. Interpretation of the measurement is not subject to the same
difficulties as in an iron-core inductor, because the leakage inductance is
caused by flux lines in air. Therefore, the stored energy is more nearly pro-
portional to the current squared, as in an air-core inductor.
Coupling transformers, unless one winding is connected to a very low

impedance source, pose the same problems as saturable core inductors,
and are also subject to the influence of the reflected load impedance. The
shunt impedance representing the magnetizing current can often be
ignored because it is high compared to all others in the circuit, including
reflected load impedance. This is not always the case, however. It can cer-
tainly not be ignored if the load impedance, perhaps under fault
conditions, can become very large or infinite as the result of an open cir-
cuit. In the open-circuit case the transformer becomes an inductor, and all
the previously stated cautions apply.
In most situations ignition testing is recommended, because losses in the

transformer may make an apparently hazardous energy level quite safe.
Of course, if the inductance of the transformer is taken to be the zero cur-
rent incremental value, and the circuit is safe with an air-core inductor of
that value, no testing is required.
Transmission Lines
Since the first use of intrinsic safety it has been recognized that energy
stored in the inductance and capacitance of a transmission line contributes
to the ignition capability of a system. Until recently, however, the length
of cable between intrinsically safe and associated apparatus was usually
short enough so that little attention was given to the cable characteristics
except to specify that it must not exceed a stated length or that its induc-
tance and capacitance must be considered in computing the size of the
storage elements connected to the associated apparatus. The ignition capa-
bility of cable is usually assessed by assuming the inductance and
capacitance to be lumped, that is, to multiply inductance and capacitance
per foot of cable by the cable length and treat the resulting inductance or
capacitance as a circuit component.
Though workers in the field knew that the discharge characteristics of a

long cable are not the same as those of lumped elements and that the
lumped element approach is too conservative, there was no data on which
to base a more rational approach to assessing safety. As a practical matter,
that situation continues. The issue is now being studied by an IEC Work-
ing Group. There is hope that recommendations for handling long cable
lengths will be included in a future edition of the standard.
Virr's paper analyses the length of cable of given L/R ratio that will be most
incendive when used with several power supply configurations. He first
treats the unregulated supply with resistive output impedance. As noted
in the discussion of shunt-diode safety barriers, the most incendive condi-
tion is when the cable resistance equals the output impedance of the
supply.
If the output impedance of an unregulated supply contains an inductive

component, the total energy dissipated in a cable arc will be maximum for
a shorter cable length than if the supply inductance is zero.
In addition, he treats zener-diode regulated supplies and shows that the

most incendive length of cable depends on the ratio of the zener voltage
Vz to the open-circuit voltage of the unregulated supply Voc. If Vz ≥ Voc/2,
the most incendive cable length is the same as that for the unregulated
supply. If Vz ≤ Voc/2, the most incendive cable length is shorter; that is,
cable resistance is less than the output impedance of the supply.
The most general circuit Virr treated assumes inductance in the unregu-
lated supply and on the cable side of the zener regulator. The effect of
inductance on the supply side of the zener cannot be assessed analytically,
because only a fraction of its stored energy will pass the regulator. Induc-
tance on the output side of the zener has an effect on cable length that is
calculatable, but no generalization is possible.
Schebsdat investigated the ignition characteristics of a variety of cables,

both real and simulated. His principal conclusions are summarized in his
IEE paper.
For hydrogen-air, the ignition energy required using the standard IEC
apparatus if energy is applied in a rectangular pulse is not markedly
dependent on the length of the pulse until the pulse length exceeds 10 μs.
(The energy increases from 20 μJ to 35 μJ.) As the energy pulse duration
increases above 10 μs, the ratio of energy required for ignition to the dura-
tion of the pulse is constant.
When a long cable discharges into an arc, the amount of energy released
depends on the cable constants, length, and the difference between the
source voltage and the arc voltage. The duration of the discharge is typi-
cally 10 μs to 20 μs for 1-km long cables and 100 μs to 300 μs for 10-km
cables and is approximately linearly proportional to cable length. There-
fore, for typical cables longer than 1 km the ratio of spark energy to spark
duration is constant.
Considering the characteristics of the ignition process, together with the

characteristics of discharges from cables, leads to the conclusion that for
typical cables longer than 1 km the ignition capability is determined by the
voltage applied to the cable and that it is little affected by length. If the
lumped capacitance model is used, the ignition capability increases as
length increases.
The value of the safe voltage which may be applied to the cable is lower if
it is assumed that the arc occurs because of a fault at the middle of the
cable. Both halves of the cable then discharge into the arc, contributing the
same energy as a discharge at the cable end, but in one-half the time.
Schebsdat presented tables giving ignition voltage as a function of the

value of the current limiting resistor feeding the cable. Values are given
for 17 combinations of R, C, and L typical of cables used in Germany. One
table assumes that the cable is 1 km to 3 km long, but only 200 m at the end
of the cable is in the hazardous location. The second table assumes that the
entire cable is within the hazardous location, and the ignition voltages are,
therefore, lower than in the first table. For cables shorter than 1 km, the
permissible voltage must be determined using lumped parameters.
To use the tables, one would calculate the voltage at the cable input
needed to supply the load at the end of the cable and multiply by 1.5. Look
in the table to find a cable with ignition voltage of at least that value when
driven through a current limiting resistor applicable to the particular sys-
tem. Any cable may be used that has lower C and higher L and R, than the
one listed in the table.
Towle's paper at the 1982 IEE Conference on Electrical Safety in Hazard-

ous Environments is a good background source for information about the
difficulties posed by blind application of rules to cables, especially multi-
core cables, and the inconsistency between rules for apparatus and rules
for systems.
Until digital transmission on field buses became a reality, one could

assume in most cases that values of cable parameters of 200 nF/km (60 pF/
ft) and 0.66 mH/km (0.2 μH/ft) were conservative values to be used if the
parameters of a cable were unknown. The capacitance of most PVC-insu-
lated cables used in 4 mA to 20 mA circuits is probably less than 30 pF/ft.
In some projects the amount of cable to be used was high enough to justify
procuring a lighter design, using thinner, higher dielectric strength insula-
tion. These cables had capacitance of 60–70 pF/ft.
In 4 mA to 20 mA circuits, the open circuit voltage of associated apparatus

has typically been 28 to 30 volts, though some earlier systems used even
higher values. Whe considering the safety of cables based on multiplying
the lumped parameters/unit length by length capacitance is almost
always found to be the limiting parameter in these 4 mA to 20 mA circuits.
Inductance may be the limiting parameter in lower voltage higher current
circuits.
Cables selected specifically for use with the low frequency (31.25 kbits/s)
Fieldbus system have higher capacitance than those typically used
in dc current systems. Representative Fieldbus cable parameters may be
90-nF/km conductor-to-conductor, 600 μH/km, and 80 ohms/km. If the
cable is shielded the capacitance may be 175 nF/km. In a Fieldbus system,
the conductors are both driven and the shield is ordinarily grounded. The
worst-case total capacitance is then the capacitance between conductors
plus the capacitance of one conductor to shield, when one conductor is
grounded. In normal operation the capacitance is the sum of the capaci-
tance between the two cores and one-half the capacitance between one
core and the shield. PTB Report W-53e summarized the igniting capability
of four cables with associated apparatus suitable for Fieldbus applications.
Report W-53e is unique among PTB reports because it contains a full
description of the Fieldbus system, its function, and its intended use. Not
all the specified details are relevant to intrinsic safety. They are intended
to fully scope the system to which the report's conclusions are relevant,
and also to tell the reader how to use the data to ensure that a system will
be both safe and functional. The extensive description also justifies the
truncation of some experiments because the system would not be func-
tional with longer lengths of line than those tested.
Type R-Ω/km L-μH/km C-nF/km Comment

A 15.6 605 117 Low R, average L and C
B 12.3 380 96.3 High R, low L and C
C 83.4 530 86.7, Relatively high R,
173.95 average L, high C
D 15 1000 200 Fictitious cable; low R,
high L and C
The cables were sometimes terminated at both ends by R-C terminators of

93Ω/μF or in some tests, 92 Ω. 2.2 μF.
Initial tests were conducted on six circuits tabulated below with a 15 V,

rectangular power supply, using simulated cable type A. For convenience
in testing, Johannsmeyer, like Schebsdat, simulated cables with series-con-
nected lumped parameter sections representing 50 m to 100 m of cable
length each.
IEC Breakspark
Circuit Cable Location Terminators
1 No At supply No
2 Yes At cable output No
3 Open at output At cable input No
4 Shorted at output At cable input No
5 Yes At cable output Yes
6 Yes At cable input Yes
In no case was the igniting current for a 1-km-type A cable less than the
igniting current for the power supply alone. Oscillographs of the dis-
charges showed that the presence of the cable shortened the arcing time
compared to that of the power supply alone. The power supply by itself
sustained an arc for 1.5 mS as the voltage increased from the minimum
arcing voltage of 10 to 11 volts to the voltage of the supply as the contacts
opened. The current in the arc was determined solely by the supply. When
an open circuited Type A transmission line was attached the current was
still maintained by the supply, but the discharge lasted only 450 μS. When
the transmission line was terminated with terminators at both ends, the
line seemed to steal some of the current from the arc at the beginning of
the discharge, and the duration of the arc was cut to 390 μS. The igniting
currents for circuits 2, 4, and 5 were sufficiently higher than the power
supply alone that no further tests were conducted in that configuration.
Tests with configurations 3 and 6 were repeated with simulated 500-m
lengths of cable types B and C. 500-m lengths were chosen because longer
lengths would have not been functional in a Fieldbus system. Ignition was
found to be of the inductive-resistive type, that is, ignition occurs upon
opening a contact in a circuit-carrying current. That this was true was ver-
ified by increasing the voltage of the supply in steps up to 40 V, at each
step decreasing the current capability to ensure that ignition by opening
sparks did not occur. Ignition at closing contacts was not observed. Tests
were then made with the rectangular supply with voltage of 20 and 24 V,
cables A, B, and C, and the additional fictitious cable D. This latter cable
was simulated for length of 5,000 ft. It was concluded that an EEx ib IIC/
IIB supply could be certified as follows:
Vmax = 14 − 24 V
Imax, per PTB Report W-39
e.g., to 128 mA for Vmax = 15 V (Group IIC)
to 280 mA for Vmax = 15 V (Group IIB)
Cable parameters (per km):
R = 15 − 150 Ω
L = 0.4 − 1 mH
C = 80 − 200 nF (including shield, if one present)
Cable length up to 5,000 m
Terminators:
R-C elements with 90–100 Ω, 0–2.2 μF, one at each end of the trunk
cable
This certification pertains to a system that conforms to what is called in the

report the FISCO Model, an acronym for Field bus Intrinsically Safe COn-
cept. In this system there is only one source of power, as described above.
All devices attached to the bus, including a handheld communicator, are
sinks for current and have the following characteristics:
Vmin = 9 V
Inominal = 10 mA
Li < 10 μH
Ci < 5 nF
The first two characteristics are specifications of the IEC Fieldbus; the last
two are to ensure that the assumptions underlying the test program are
honored in the installed system.
It is also assumed that a device on the bus has Fault Disconnection Elec-
tronics operating at no higher than 20 mA more than nominal current
draw of the device. This feature protects the system from the effects of a
failed field device. It is also assumed that the bus-connected device has
reverse polarity protection so misconnection has no consequences for oper-
ation of the bus. Modulation of the average 10-mA current is ±9 mA. The
maximum current the system can be designed for is the rated current of the
supply less 29 mA. These figures are used in determining the maximum
length of cable that will function. They are not intrinsic safety related.
To investigate an EEx ia IIC system, the power supply used was a 15-V
constant voltage supply derived from a 34-V supply. The value of short-
circuit current related to ignition was determined to be 323 mA for Group
IIC, so the permitted value of short-circuit current for the supply is 215
mA. The current value permitted at the break point of the trapezoidal char-
acteristic, IOT = 120 mA.
Tests were limited to 500 m lengths of cables B and C, and a 1,000 m length
of cable A. It is assumed by Johannsmeyer that further testing would jus-
tify permitting a length of 5,000 m, as for the EEx ib supply. The specimen
certificate in the report contains the following information.
For EEx ia IIC—Supply unit with trapezoidal output characteristic

Vmax = Ez = 14 − 24 V
EB > 2 Ez
Imax, per PTB Report W-39
e.g., to 215 mA for Vmax = 15 V (Group IIC)
Cable parameters (per km):

R = 15 − 150 Ω
L = 0.4 − 1 mH
C = 80 − 200 nF (including shield, if one present)
Cable length up to 1,000 m
Terminators:
R-C elements with 90–100 Ω, 0–2.2 μF; one at each end of the trunk
cable
Example of Product Design
Figure 11-8 illustrates a circuit to be analyzed for intrinsic safety. The cir-
cuit is intended to represent the major elements to be found in most 4 mA
to 20 mA field-mounted transmitters. The functional specifications of the
transmitter are not listed, but it is typical of these transmitters that they are
able to operate with a voltage at the input terminals of 11 V minimum and
42 V maximum. Therefore, some kind of series regulator must absorb the
range of voltage at the input terminals. In the illustrative circuit this is rep-
resented by U2 operating in conjunction with a regulator circuit. U2 is
shown as a block to suggest that it makes no difference what the details of
the block are, whether there is one transistor or an integrated circuit. If
there is more than one transistor, the course of fault analysis might be
changed from that given below because failure of each transistor could
probably be counted as a fault. This analysis treats U2 as though it were a
single transistor or chip to anticipate the possibility that later it might be
desired to make it so. Similarly, U1, in combination with the 50-Ω feed-
back resistor and the Output Current Control block, controls the level of
the output current of the transmitter in the range 4 mA to 20 mA. It could
be a single transistor or more than one, but this analysis treats it as a single
electronic device.
The schematic/block diagram illustrates many important decisions that

have already been made. It has been decided to incorporate intrinsic safety
in the circuit in a way that provides the greatest possible freedom later to
change circuit constants, elements, and printed wiring board layout with-
out having to return to the approval agencies for a revised certification.
Vendors might halt production of a chip. New chips become available
with more function at the same price. Users demand more function, which
requires more memory, or more processing capability. Cost savings may
be realized from integrating several functions into a single chip. Today
many organizations are able to make major changes in the circuitry of a
transmitter and begin to ship a new design in 6 to 9 months. A shift to dif-
ferent component vendors or incremental improvement in capability of a
small part of the circuit may be accomplished more quickly. It is therefore
essential to structure the design to permit such changes, either without
amending the approval at all, or with an amendment that can be pro-
cessed by the approval house in a few months, or less.
Figure 11-8 Circuit as Intrinsic Safety Analysis Starts

To achieve this objective all the circuitry to the left of the heavy dotted line
is considered to be unspecified. Components can be chosen for function.
Spacings need only conform to the design and manufacturing standards
needed to maintain the desired level of reliability. The complexity and size
of today's integrated circuits requires printed wiring board tracks to be
spaced as close together as manufacturability and quality considerations
permit. Even in a 10-V circuit, these spacings may be smaller than those
demanded by intrinsic safety rules if shorting between tracks is to be
counted as a fault. The cost of asserting that these components and circuits
are unspecified is that failure of these circuits will be assumed to occur
without counting a fault.
Any combination of terminals maybe shorted together or opened. Failure

of components such as U1, U2, D1; the 50 Ω and 100 Ω resistors; and the
capacitors will be counted, but the cost of counting these faults is that the
components must be rated in accordance with the intrinsic safety stan-
dard, that is, the operating voltage, current, or power (whichever are
relevant to safety) must be no more than two-thirds rating. Spacings of
tracks connected to them must be at least one-third those in the spacing
table to allow a short-circuit fault of the component to be counted.
If proprietary chips are used in the unspecified circuitry, it may be easier

not to declare the device to be a component on which intrinsic safety
depends than to provide sufficient data to support a claim of conformity to
the two-thirds requirement. As noted on the schematic, the declaration
that the circuitry and layout are unspecified includes a limitation on the
total capacitance, and, if applicable, inductance. The unspecified blocks are
assumed in the analysis to be connected in the way that is least favorable
for intrinsic safety; usually, they are lumped as a single element containing
all the capacitance or inductance. The total value claimed should be some-
what higher than is actually present in the circuits to provide a cushion for
future design changes. Somewhere in the documentation for the design it
must be declared that none of the circuits in the unspecified zone increases
the voltage applied at their power terminals.
The circuit will be evaluated using EN 50020 as the reference, as summa-

rized in Chapter 10. The author has read permissible voltage values from
the curves in Chapter 10. These values are approximately in agreement
with the curves and tables in the standard. Any differences are not impor-
tant in this tutorial exercise.
The intrinsic safety description of the transmitter is shown on the sche-

matic. The value of Ci is about 15% of the value allowed by EN 50020 (2nd
ed.) to be connected to associated apparatus with Voc of 30 V. The values
of voltage and current are higher than the those of the common 28-V, 93-
mA barrier. This allows some leeway for barriers with higher ratings and
for the few milliamperes of current that some handheld communicators
may inject into the line under fault conditions. The power rating of 1.2 W
is selected to make a T4 temperature code applicable without the need for
encapsulation or a program of surface temperature measurements to sup-
port the claim.
Normal Operation
In normal operation the circuit has a nominal upper range output of
20 mA, and a maximum output of 22 mA. For terminal voltages greater
than 11 V, the regulator and Q2 control the regulator output to 8 V. This
voltage is then regulated by the power supply block to 5 V, which powers
all the other blocks. Current flowing in the 50 Ω resistor develops a volt-
age VFB, which is compared in the Output Current Control block with a
reference proportional to the desired value of current. The block adjusts
the drive on Q1 to maintain the desired value of current.
For any intrinsically safe value of voltage and current at the input termi-
nals, the currents and voltages everywhere within the circuit will also be
safe relative to resistive circuit ignition. This is because there are no cir-
cuits that increase voltage and the product of open-circuit voltage and
permissible short-circuit current increases as voltage is decreased. Only
the stored energy in inductances and capacitances is of concern. These are
discussed below. The unspecified circuitry can be assumed to fail in a way
to store the maximum amount of energy, and no fault is to be counted.
The conditions in normal operation of the circuit are as follows.
Permitted
Current or
Element value (2/3 OK? Comment
voltage
curve)
100 μH 100 mA 100 mA Yes Safe - 100 μH is essentially
resistive
0.1 μF 4V 26 V Yes Safe (.022)(100+50) + 0.7 = 4
0.01 30 V 100+ V Yes Assume Dig. Comm. block
grounds C. Permitted V from
Figure 10-5
6 nF filter 30 V Off curve Yes Worst case is one terminal
Caps grounded
12 μF 8V <8 V No Needs attention. See below.
0.5 μF 8V 18 V Yes, but -- It could be considered in
parallel with any of the blocks.
8 V is the worst case voltage,
but the capacitance needs to
be in parallel with others.
1 μF 5V 16 V Yes, but -- It is in an unspecified circuit
and can be in parallel with
another capacitor.
13.5 μF 8V 7.2 V No Add 5.6 Ω in series with 12 μF.
(12 + 1 + 0.5) It now looks like 1 μF, approx.
2.5 μF 8V 10 V Yes 5.6 Ω resistor must be a
(1 + 1 + 0.5) protective component.
The approximate equivalence of a 1-μf capacitor to a 12-μF capacitor in

series with 5.6 Ω can be estimated by drawing a horizontal line in Figure
10-5 from 12 μF to intersect the 5.6 Ω curve, then dropping a vertical to
intersect the 0 Ω curve. Read off the approximate equivalent capacitance.
These curves are not recognized in the 2nd edition of EN50020, so they
should be used with care. Estimates can also be made for capacitances less
than a few microFarads by using the curves in Johannsmeyer's 1984 work
on ignition by sparks in capacitive circuits cited at the end of Chapter 9.
The added components are shown in Figure 11-9; the values shown for
each is the value after fault analysis.
Figure 11-9 Circuit After Addition of Protective Components

Analysis Under Fault Conditions
The effects of faults and the measures taken to maintain intrinsic safety are
shown below. Remember that within the unspecified circuitry and blocks
any interconnection of circuits may be presumed without counting a fault.
Current or Allowed
Fault Element OK? Comment
Voltage Value
Short 12μF - 30 V 10 V No This short connects the input terminal
U2 pins 5.6Ω(2.5 μF) to the regulator block. This parallels
1 and 2 0.5 and 1.0 with 12/5.6 Ω as before.
Adding a voltage-limiting, 10-V zener
is marginal; so, add 10 V zener across
R-C, and increase 5.6 Ω to 15 Ω.
Anticipate second failure, opening of
this zener, and add another zener. But
the regulator block is not specified so
its failure could charge these
capacitors to 30 v without a fault count
and without shorting U2. Therefore, a
third zener is required. This mode of
charging the capacitors should have
been treated in analysis under normal
conditions. The 50-Ω resistor can be
shorted without counting a fault by an
Output Current Control block
malfunction.
Short 12μF—5.6 Ω 30 V 10 V No This short also connects the input
U2 pins (2.5μF) terminal to the regulator block. This
1 and 3 parallels 0.5 and 1.0 with 12/5.6 Ω as
before. Add a voltage-limiting, 10-V
zener from pin 3 to circuit common.
Anticipate second failure, opening of
this zener, and add another zener.
Short Same as 30 V 10 V No Add zener at Dig. Comm block input
0.01 μF above terminal to common. Anticipate
second fault and add another zener
Short 0.1 μF 30 V 26 V No Add zener at Output Current Control
U1 pins block input terminal to common.
1 and 3 Anticipate second failure and add
another zener.
Short 2.5 μF as 30 V 10 V No, Previous action resolves this issue as
U1 pins above but well.
1 and 3
Replacing the 5.6-Ω resistor with a 15 Ω resistor reduces the apparent

capacitance of the combination from 1 μF to about 0.7 μF. The sum of all
the capacitances in parallel becomes 2.2 μF, which has a permissible volt-
age a few tenths of a volt above that for 2.5 μF.
In a real design project the options exercised in the first row of the table
would have been subject to considerable discussion. A circuit such as that
for this fictitious transmitter must operate with an input current of less
than 4 mA and an input voltage of 11 V at the highest ambient tempera-
ture for which the transmitter is rated (usually in the neighborhood of
90°C). The leakage of the zeners at high temperature would be of great
concern to the designer, who also may not desire to sacrifice filtering by
increasing the series resistor to 15 Ω. Other options are to lower the capac-
itance value to give greater freedom for picking the resistor and zener
values or lower the claimed total capacitance in the unspecified circuits to
a value closer to the actual design value. It is customary to claim as large a
value as is reasonable to offer freedom of choice in later design
modifications.
Value of Ci
The foregoing steps have ensured that the spark-ignition characteristics of

the circuit conform to the requirements of the standard. Now evaluate the
circuit to determine whether the objective of Ci = 10 nF has been achieved.
A barrier with open circuit voltage of 28 v can be connected to a maximum
capacitance of 80 nF (read from Figure 10-5 at a voltage of 28 × 1.5 = 42 V).
If the design objective has not been met, the application of the transmitter
may be severely limited; if the 80 nF value is exceeded, the design is not
intrinsically safe. The analytical task is to determine whether any of the
capacitors can discharge into short-circuited input terminals, taking into
account two circuit faults. The author prefers to consider only faults of
components, assuming that in the next step spacings between tracks will
be defined so that this analysis is valid.
Shorting of the 50-Ω feedback resistor is not a countable fault because a

noncountable fault in the Output Current Control block can short VFB to
circuit common. If it is decided to accept the penalty of another diode volt-
age drop of by adding D4, then D3 and D2 are no longer necessary. An
alternative to adding D2 to protect against discharge of the 0.1-μF capaci-
tor is to place a large resistor in series with it. Because the capacitor can be
charged to 10 V, and the permissible current in a 10-V resistive circuit is
greater than 1 A, the resistor need only be more than 100 Ω to ensure that
there is no effective capacitance. There are no rules for making this kind of
judgment. One should be prepared, should the examining engineer ques-
tion the judgment, to increase the resistor value or add D2 instead or to
have the lab test for the effective capacitance.
Element Faults Comment

0.1 μF Short D1 and U1 Capacitor discharges through the
input terminals. Add D2.
0.1 + 2.3 μF Short D1 and U1 D2 prevents current in input circuit.
0.01 μF Short D1. Short zener. Add large resistor, R1, if function
permits, to reduce effective
capacitance; or, make R1 small
and add D4. (The shorted zener
can be caused without a fault
count by malfunction in Dig.
Comm. block.)
2.3 μF Short 0.01 μF and D1 If R1 can be several hundred
ohms, the discharge will be
essentially resistive. If not,
consider resistance in series with
the 1-μF capacitor; reducing the
claimed total C in the unspecified
circuitry; or add D4.
2.3 μF Short U2 either path, and D1 Add D3.
One option that has not been exercised in this example is that of rating the
zeners at the 0.1 μF capacitor and at the Dig. Comm. block at a lower volt-
age than the three zeners at pin 2 of U2, perhaps 6 V rather than 10 v. They
are in parts of the circuit where leakage is less a problem, so functional
requirements may permit this. This will reduce the power rating, which
must be specified. In some circuit configurations, but not in this example,
choosing zeners of different voltages also reduces the number of capaci-
tors that are connected in parallel across the higher voltage zener because
connecting them would also connect the lower voltage zener, reducing the
hazard.
Temperature Code
Because the input power to the circuit is defined to be 1.2 W maximum,

the design temperature code is T4, and all the components in the circuit
will likely have areas greater than 20 mm2, there is little need for concern
about measuring temperatures of components. If there are components
with areas smaller than 20 mm2, it will be necessary to show that their sur-
face temperature is not higher than 275°C.
Were the transmitter specified to have a temperature code of T5, the maxi-
mum surface temperature of any component shall not exceed 150°C.
Measurements or encapsulation may be necessary.
Component Ratings
All components on which intrinsic safety depends must be operated at not

more than two-thirds the relevant voltage, current or power rating if a
malfunction is to be counted as a fault. The table below summarizes the
selection of ratings for the components on which intrinsic safety depends.
Component Rating Comments

0.01 μF 45 V Shorting of Dig. Comm. and Output Current Control
(50 Ω) blocks applies 30 V.
0.1 45 V Shorting of U1 and 50 Ω applies 30 V.
12 μF 15 V Zener clamps to 10 V.
D1, D2, D4 150 mA, 45 V Because of 30 V on 0.01 and 0.1 μF caps, and 100
peak reverse mA current through U1 shunt circuit when U1 shorts.
Voltage
D3 150 mA, 15 V 100 mA current when U2 shorts and 50 Ω is shorted
PRV by a block short.
U1 45 V, 150 mA, 100 mA flows in shunt circuit if Output Current Control
1.8 W block malfunctions. 30 V appears across U1 if 100 Ω
and 50 Ω resistors are shorted by block malfunctions.
Max. input power is 1.2 W.
U2 45 V, 6 mA In normal operation maximum current is 4 mA and
voltage across it is nearly 30 V when it is driven to
high resistance state.
50 Ω No requirement Malfunction of Output Current Control block shorts it
without a fault count.
100Ω No requirement This resistor has no safety function. It can be shorted
without counting a fault by a block malfunction.
15 Ω Film type This is a protective resistor, so it must be film or
wirewound. The wattage can be 1/8 W or smaller
because only 120 μCoulombs will be stored on the
capacitor. After the capacitor shorts the resistor has
no protective function.
R1 Film type This is a protective resistor if D4 not present so it
must be film or wirewound. The wattage can be 1/8 W
or smaller because only 23 μCoulombs will be stored
on the paralleled capacitors (2.4 μF if U2 opens) in
the unspecified circuitry.
All zener 10 V, 150 mA, Worst case 100 mA flows in diode, so power
diodes 1.5 W dissipation is 1 W. Multiply by 1.5. If any zeners are
chosen to have lower voltage than 10 V, the current
rating remains 150 mA, but power rating decreases
by the voltage ratio.
If the design temperature code is T5, it may be necessary to select larger

resistor wattage ratings to keep surface temperature below 150°C. If this is
the case, the area must also be less than 10 cm2. One alternative to using
larger resistors is to encapsulate the components with high surface tem-
perature. The encapsulant temperature rating must be higher than the
surface temperature of the component within the encapsulation. (This
temperature will be lower than the surface temperature in air, so the tem-
perature at the interface between the component and the encapsulant
must be measured.) Another alternative is to provide heat sinking or heat
dissipation hardware for the high temperature components. The approval
agency may insist that these be attached in a way that removal of the heat
sinks is highly improbable.
Controlled Spacing
Because the device being analyzed is a field-mounted transmitter, it is

assumed that the circuit boards are coated for environmental protection
and that the coating meets the requirements for conformal coating in the
standard. It is also assumed for the purpose of this example that the
printed wiring board is a double-sided board, initially clad with 1/2 oz
copper, and processed by the additive process when plating through holes
so that the final copper thickness is 35 μm, equivalent to 1 oz cladding.
The zener diodes provide a barrier against intrusion of higher voltages into
the unspecified circuitry. PWB layout must ensure that none of the con-
trolled circuitry is close enough to unspecified circuitry that a short
between tracks could couple a voltage or current into the unspecified cir-
cuity. This spacing between the two sections of circuitry must be a
protective spacing, not considered subject to fault. For a circuit voltage of
30 V between two conducting parts in air, the spacing must be at least 2
mm and between tracks under coating, at least 0.7 mm. The dotted line A-
B in Figure 11-11 shows the location of this isolating spacing.
The spacing between leads of protective components must be the EN50020

Table 4 value of clearance in air for the maximum voltage across the com-
ponent because the leads exit the conformal coating. The distance between
coated tracks leading to the component must be the Table 4 value of creep-
age under coating. Only the 15 Ω resistor and R1 are protective
components.
For components for which a short-circuit fault is to be counted (such as

U1, U2 and the diodes, D1, D2, D3, and D4) the spacings between leads or
solder pads must be at least one-third the Table 4 values for clearance in
air, and the spacing between tracks to the leads or solder pads must be at
least one-third the value for creepage under coating. There is not yet uni-
versal agreement about whether the distance between solder pads of
surface-mounted components can be treated as creepage under coating if
the board is coated. It is safest to design based on the clearance in air val-
ues. Those who do not recognize the validity of the coating argue that the
coating may not extend under the component, or it may not seal off the
area under the component or that the solder flux cannot be completely
cleaned from under the component, so that it should be treated as a
“dirty” spacing like a clearance in air. Use the 30-V values of 0.7 mm and
0.23 mm for all these components. The table below summarizes the spac-
ings at components that the designer must control.
Between Leads or Between Coated Tracks to
Component
Solder Pads - mm Components - mm
15 Ω 1.5 0.5
R1 2 0.7
U1, U2 0.5 0.24
D1, D2, D3, D4 0.5 0.24
The spacings discussed above have assumed the design to be a double-

sided board of usual 1 mm to 1.5 mm thickness. If the board is a multilayer
board, the spacing considerations extend into the inner layers. If there are
several layers, the thickness of insulation between layers can be adjusted
as long as the overall thickness is the same as a two-sided board. To meet a
2-mm protective clearance on the top surface, one must consider the
equivalent spacing through the insulation to the next inner layer and
ensure that it maintains the isolation. In this example the distance through
solid insulation is greater than 0.5 mm and is, therefore, presumed not to
be subject to fault. This may resolve the problem. If there are several lay-
ers, it may not be practical to adjust the thickness of the insulation to
eliminate the fault. The solution then is to remove some of the copper from
the inner layer so that the distance from the conductor on the top layer to
copper on the inner layer is at least 0.5 mm for total isolation, or one-third
of that if it is only necessary to be able to count a fault. Where component
leads or feed-throughs penetrate a board, it is necessary to ensure that
they are isolated from interior layers by an annular ring of bare material of
appropriate radius.
Zener Diode Track Layout

The layout of protective zener diodes and the associated tracks must
ensure that a circuit open is required to disconnect each diode and that no
opening of a track can remove more than one diode from the circuit with-
out also removing the source of power from the protected circuitry. Figure
11-10 illustrates a correct layout and an incorrect one.
Figure 11-10 Layout of Zener Diodes to Ensure Desired Fault Count
The incorrect track routing allows a single cut to the left of point “a” to
remove all the zeners without removing the 30 V source from the block.
This kind of routing is commonly generated by automatic routing pro-
grams. A single cut in the circuit common line at “b” will remove all the
zeners, but it will also remove the return for the 30 V, if all other circuit
connections to common are made to the common track to the right of the
diode connections. Because it may be difficult to enforce that proviso over
the life of the design, multiple connections as in the correct version, are
preferred.
The correct routing ensures that a cut at a-b or to the right of a-b removes
the source of supply as well as disconnecting the zeners from the pro-
tected block. Multiple connections to the common terminal of the
protected block ensure that at least two faults are counted before remov-
ing the zener diode protection.
In the 2nd edition of EN50020 a 2-mm-wide track is assumed not to fail if

there is 35 μm copper thickness. This is another option, but it may not yet
be acceptable in all nations.
Document the Design for the Approval Agencies
Two kinds of documents must be submitted to the approval agency, a

package of documents that will be listed in the certificate, which define the
certified product, and a document that describes why the product con-
forms to the standard. The first package is mandatory. The second may be
required in a prescribed format by some approval agencies, but for most it
is an optional document whose objective is to minimize the amount of
time the examining engineer spends on the project.
Certification and Control Drawings
The first package of documentation must define the product that is being
certified to the mutual satisfaction of the manufacturer and the approval
agency. It must clearly define the portions of the design that are controlled
by the certification. This documentation may not be changed without
prior approval from the agency.
Every certificate or report of approval of a product refers to documents

that define the product. These include the model designations the agency
considers relevant to the approval, and documents, such as drawings and
specifications, that define the features of the product relevant to the
approval. Until recently it has been the practice by many North American
manufacturers and agencies to refer to manufacturing drawings and spec-
ifications in the reports issued by the approval agency. Part of the agency's
periodic audit procedure is to check to see if the drawings being used for
manufacturing have been changed from those listed in the approval docu-
ments. In the past, changes that were not relevant to the approval such as
paint color, were not held to be grievous sins; but, nevertheless, that
required explanation and official revision of the report to refer to the
revised drawings. Other agencies prepared a report that described the
product and its construction, often including vendor names and part num-
bers as well as component ratings. Changing vendors or their part
numbers are reasons for revising and reissuing the report. Some inspec-
tors, but not all, allowed leeway for implementing simple changes of no
consequence to the approval before the report was updated. As the world
grows closer to a universal ISO 9000 mentality, where all deviations from
the documented list of drawings are viewed as being equivalent sins,
regardless of the effect on the safety of the design, manufacturers have
found that they must follow the pattern established many years ago in
Europe. Europeans have been providing certification drawings to the
approval agencies that contain an overview of the product's appearance,
model numbering scheme as it applies to the approval being sought, and
all the information needed to support conformity to the relevant standard.
A certification drawing is intended to show all the relevant information in

a form that never has to be changed during the life of that version of the
product. Only a major redesign or revised model designations would
necessitate a revision of the certification drawing. The certification draw-
ing is not used for manufacturing. It summarizes the important data so the
drawings used for manufacturing the product can be changed without
referring the changes to the approval house. All the reasons for making
the unspecified zone of a design as large as possible are also reasons to use
certification drawings in the approval process The purpose is the same, to
isolate the approval process from the continuous flow of small changes to
the manufacturing drawings and to provide freedom for even major
changes if they can be implemented within the parameters specified on
the certification drawing.
Every approval house has a slightly different view of what is needed to

define the product that has been certified. The author recommends the fol-
lowing, though for some agencies it may be more information than is
needed.
(1) Pictorial view or assembly drawing of the product.
(2) Model designation table, abbreviated as much as the agency will per-
mit. Be specific only about those model designations that affect the
approval and state that the options for which there is no designation
do not affect the approval. Some agencies may insist on the complete
model number breakdown. Their field engineers compare the model
number on a device found in the field to the model shown in the list
of certified or approved apparatus or in a report. Any difference calls
to question whether the device is approved or certified.
(3) Block diagram of device. If there is more than one printed wiring
board, or if some electronic components (e. g., feed-though filter
capacitors) are mounted apart from the PWB, this may be a conve-
nient place to define the partitioning of the device and to specify
zones of unspecified circuitry. It is a useful place to state normal cur-
rents and voltages in the various parts of the circuit to help the exam-
iner understand the device.
(4) Schematics of circuitry to be certified. These schematics should con-

tain the kinds of information shown in Figure 11-9, as well as infor-
mation about the controlled spacings and controlled components. Try
to specify the controlled components in terms of electrical ratings
only and generic type numbers only. Some agencies will ask for more
information. List physical sizes, vendors’ names, or model numbers
only if an agency insists. These parameters often change even if the
electrical characteristics do not. Be as free with designations “mini-
mum” and “maximum” as practical to provide the greatest possible

freedom for later changes.
(5) Copy of artwork for printed wiring boards. Indicate zones where
spacings are not controlled. State the CTI rating of the board material.
(This could be done on the schematic also, or instead.) Indicate con-
trolled spacings if to do so will ease understanding of the design. Do
not show the revision level of the art work, because the design may be
controlled to the level that shows in the certification drawing, even
though changes should be permissible in the uncontrolled zones of
the layout.
(6) If the agency demands, show the marking to be applied to the prod-
uct. This may be an illustration of the proposed nameplate or just a
list of the markings pertinent to the approval. If a picture of the name-
plate is used, include a note stating that information other than that
applying to the approval is shown for information only and is not
controlled.
In North America the control drawing is a mandatory document, and is

part of the certification package. The control drawing describes the per-
missible connections between the certified apparatus and other apparatus.
For associated apparatus it will contain a warning against connection to a
source containing more than 250 V internally, instructions for proper
grounding, and, of course, a listing of the parameters of intrinsically safe
circuits in atmospheres of different hazard classification that may be con-
nected to it. Alternatively, the drawing may specify apparatus to be
connected by vendor name and model number. A control drawing for
intrinsically safe apparatus will specify the parameters of associated appa-
ratus to which it may be connected or model numbers and any other
limitations of the installation. A control drawing sometimes specifies the
types of cable that may be connected or the maximum permissible length
of cable. Consult ISA RP12.6 and ISA TR 12.2 for more detailed guidance
on the content of control drawings.
The control drawing will be a part of the documentation sent with the
product to the purchaser, so it should not be a part of the certification
drawing. A control drawing is not necessary for apparatus certified to
EN50020, and may, in fact, be removed from the documentation package
by some examining engineers.
Some agencies will require a statement on the certification and control

drawings that the drawing may not be changed without the agency's
approval. This is not an unreasonable demand, although it is redundant.
The numbers and revision level of the drawings will be listed on the certif-
icate or in the report and are, therefore, not to be changed without
approval. Certifying agencies often try to direct the procedures to be used
to control a design, often in small ways such as this. Agreement may be

necessary, but questioning the need is always permissible.
Some examining engineers may suggest that numbers of the drawings

supporting the certification drawing be listed on the certification drawing.
This should not be agreed to without strenuous objection. The list will be
long. It might change for many reasons which do not affect the approval.
A later version of the product might have a different mechanical design
but identical electronics, and so have a different assembly part number
because the mounting hardware has been modified. One would not like to
have to open a project to resubmit the certification drawing for approval
just because some new manufacturing drawings have been added.
Acceptable practice is for each underlying drawing to bear a note stating

that it may not be changed in any way that invalidates the certification
drawing. The best solution for the manufacturer is to negotiate agreement
that the person responsible for managing the approval will maintain a list
of all supporting drawings, will be required to approve all changes to
these drawings, and will ensure that parameters important to the certifica-
tion drawing are not changed without reference to the certifying agencies.
This method will probably acceptable only if supported by written proce-
dures and periodic internal audits.
Safety Rationale
The certification drawing and the control drawing tell the examining engi-
neer what is being certified and how it is to be installed. The safety
rationale document tells the examining engineer what he needs to know to
agree that the product described in the certification documents can be
approved. Approval agencies generally bill by the hour. Money, and more
importantly, travel time, can be saved by helping the examining engineer
agree that the design is certifiable with minimum project billing.
The author believes that the following elements should be included.

(1) Brief description of the product and how it works. Describe the oper-
ating principle of the design. Describe how major model variations
relate to one another, relative to both function and their effect on the
safety analysis.
(2) State the approval wanted. This is the design objective discussed
above. It is in the certification drawing but needs to be emphasized so
that there is no misunderstanding.
(3) Describe the strategy for making the design conform to the standard.
When submitting the design discussed in the example, one would state
that the strategy is to erect a barrier against transmission of the 30-V
maximum supply voltage clamping all points of entry to the unspeci-

fied circuitry with redundant zener diodes and isolating the controlled
circuitry from the unspecified circuitry by spacings. Be specific about
how the spacings were chosen; that is, discuss the voltage at relevant
points in the circuit and repeat the values from the table for clearance
and creepage under coating in a manner similar to the discussion
above. The discussion need not be tutorial, just a brief factual summary
of what was done, and why is all that is necessary. Lead the examining
engineer through the exercise so he or she doesn't have to figure out
the rationale independently. This description can leap directly to a
two-fault state in the circuit, and may be stated in reverse: “Because
there are redundant diodes at … after failure of … there is no way for a
voltage higher than 10 V to be transmitted into the unspecified cir-
cuitry. At U2, pin 2, three diodes are provided because …“
(4) State the purpose of D2, D3, D4, R1; that is, to prevent capacitors from
discharging into the shorted input circuit in order to maintain the
stated value of Ci after shorting of other components.
(5) Discuss the controlled spacings used in the design and shown in the
certification drawing.
(6) Justify that the ratings of the controlled components include the nec-
essary factors.
(7) If other than a T4 rating based on 1.2 W maximum power input is

requested, discuss which components will dissipate the most power
and which will have the highest surface temperature. State the results
of tests that support the temperature code requested.
Document the Design in Manufacturing Drawings,

Specifications, and Inspection Procedures
The objectives of this step are as follows:

(1) Ensure that assumptions made in the intrinsic safety analysis and the
claims made in the certification drawing are supported by the docu-
ments used to procure components and manufacture parts and
assemblies.
(2) Put in place those inspections and audits needed to ensure that the
features pertinent to the approval appear in the product, as called for
in the manufacturing drawings and specifications, and as required by
the certification documentation.
One cannot prescribe a single set of actions to accomplish these objectives

that is reasonable for every organization and product. What needs to be
done to support the certified product in addition to what is done for an
uncertified product depends on the level of control an organization main-
tains over the production of uncertified products. Every organization
needs, as a minimum, to make the manufacturing procurement drawings
and specifications reflect what is in the certification documents. For exam-
ple, if the certification is based on zener diodes with a power rating of
1.8 W, the procurement document should clearly state this value or if it is
a derived value because of the way the component is mounted in the
design, the procurement document should state the requirement that the
component mounted in a specific manner should have the 1.8-W rating.
The procurement document must go beyond the statement of the vendor's
part number. Vendors sometimes change specifications like power rating,
or the relationship between mounting and power rating, without neces-
sarily changing the part number. The voltage and tolerance shall also be
specified. If size is taken into account in the certification document, specify
the size.
Many organizations customarily include in purchase drawings almost all

the information on a vendor's data sheet for engineering reference, but
then purchase components solely by vendor and part number. Some
inspection or audit procedure must ensure that the parts match the
intended parameters. The drawing must have a reference to the certifica-
tion drawing to ensure that anyone who wishes change the drawing is
reminded to review the requirements relative to the certification. These
comments apply to any purchased component or material that underlies
the certification. Printed wiring material must be specified to have the
minimum CTI. Encapsulation or coating material must usually be called
out by vendor name and type number in the certification documentation,
so the manufacturing documentation must also restrict procurement to
these particular products.
If a certification is obtained based on a dipped coating, the manufacturing

documentation must specify application by dipping and include a caveat
that the method may not be changed without review by the person who
controls certifications. A switch to spray application may invalidate the
approval because spray-applied coatings must have two coats. Some
approval houses may consider that “two coats” implies drying between
applications. Others may accept the fact that the purpose is to ensure com-
plete coverage and that two nozzles applying coating sequentially to the
same area of the board meets that objective.
In the certification world today there is much diversity in the means used
to determine whether a manufacturer is controlling his production within
the limits imposed by certification. At one extreme there are bodies who
accept the assurance by the manufacturer that he will comply, but make
no audits of products or procedures. Others periodically audit products

for conformance to the certification documents, usually on the basis of a
report detailing construction features and components, but pay no atten-
tion to the manufacturer's quality assurance methods. Others periodically
audit the client's quality assurance procedures and check the documenta-
tion called out in the certification documents to see if there are any
discrepancies, but seldom audit the product. This is not the place to debate
the pros and cons of how best to control a product to its certification. Both
of the latter two methods work and have worked reasonably well for
many years. However, during the 1990s there has arisen a strong interest
in ISO 9000 procedures for quality assurance. Some approval houses now
require that manufacturing plants have a quality assurance system that at
least parallels ISO 9002 and superimpose additional features they consider
necessary for hazardous location products. As this chapter is being written
one can be confident that during the next decade more certifying agencies
will demand accreditation of manufacturing plants for certified apparatus
to ISO 9002. This may be required by some of the certification/quality con-
trol modules supporting the EC EXAT Directive.
The basic tenet of ISO 9000 standards is this: write down how you will
operate and also how you will determine whether you operated that way.
For products with controlled components this means that the manufac-
turer may have to institute audit or inspection procedures to ensure that
the components and processes called out in the documentation have been
incorporated in the product.
Design of Intrinsically Safe Systems

General Principles
This section aims to summarize the principles underlying the design of an

intrinsically safe system, and the techniques available to implement the
system. This material addresses itself to those whose objective is to assem-
ble an intrinsically safe system from commercially available components.
These components are the intrinsically safe apparatus in the Division 1,
Zone 0/Zone 1 location and the associated apparatus in the Division 2,
Zone 2 or nonhazardous location.
The controlling demand of all intrinsically safe system design is that every
ungrounded conductor entering the Division 1 location (or Zone 1/Zone 0
location) must be protected against the inflow of nonintrinsically safe volt-
age, current, and power levels from the Zone 2, Division 2, or
nonhazardous location.
If associated apparatus protects every ungrounded line so that application

of line voltage to the associated apparatus does not cause nonintrisically
safe voltage and currents to flow in the protected lines, the apparatus on
the nonhazardous side of the associated apparatus may be chosen with
regard to function only. It need not be certified. The only restriction on
selection of this apparatus is that there be no voltage in it that exceeds the
voltage rating of the associated apparatus. Most associated apparatus is
rated for intrinsically safe outputs after application of line voltage, usually
250 V RMS.
When one is designing an intrinsically safe system, the major obstacles to

intrinsic safety fall broadly into one of two categories.
(1) The field-mounted equipment is not certified to be intrinsically safe.
(2) The control house equipment is not certified as associated apparatus.
If the field-mounted apparatus and the control house interface apparatus

have been certified, separately or together, the installation limits are
defined by the control drawings and instructions for the equipment. One
chooses the alternatives that best fit the plant situation.
Field-Mounted Apparatus Not Certified
If the field-mounted device is a simple apparatus, it need not be certified

when connected to certified intrinsic safety barriers. Simple apparatus
includes passive devices, such as switches, junction boxes, resistance tem-
perature detectors, potentiometers, and simple semiconductor devices;
sources of stored energy, such as inductors or capacitors with well-
defined characteristics, which are to be taken into account when assessing
intrinsic safety; or sources of generated energy, such as solar cells and
thermocouples. These sources shall not generate more than 1.5 V, 100 mA,
or 25 mW.
In some standards, the stored energy sources are not mentioned, and sim-
ple apparatus is defined as apparatus which neither generates nor stores
more than 1.2 V, 100 mA, 25 mW, or 20 μJ. These standards tacitly assume
that inductive or capacitive storage elements will be taken into account as
part of the L or C which may be connected to the associated apparatus.
Simple apparatus must be separated from nonintrinsically safe circuits and

apparatus to ensure that nonintrinsically safe energy levels will not be
injected into the intrinsically safe circuits. Most experts assume that 50-mm
spacing or partitions (as described in Chapter 10) provide sufficient protec-
tion against intermixing of the two circuits. Sometimes it has been stated
that the simple apparatus shall not be in the same enclosure as nonintrinsi-
cally safe circuits or apparatus, but, in the author's opinion, this is too
restrictive a translation of reasonable intent into a rule.
Simple apparatus must be investigated for the applicable temperature

code. If the barrier supplying it can deliver no more than 1.2 W maximum
power, then a T4 rating is usually reasonable and defensible. If a T5 rating
is desired it is necessary to investigate the temperature rise when exposed
to the maximum power delivered from the barriers used in the system.
If the field-mounted apparatus is not certified, and does not conform to the
limitations imposed on simple apparatus, the apparatus cannot easily be
used in an intrinsically safe system. Few, if any, system designers can
afford to make the investment required to self-certify a design and,
because the user cannot control the details of the design, the assessment
applies only to the specific design of the piece of apparatus investigated. In
the current regulatory environment few managements would risk using
apparatus that has not been certified by a third party, so the need to con-
sider self-certification is unlikely to arise. It may sometimes be necessary to
permit temporary use of a product that has been submitted to a third party
for certification. The supplier should provide a written assessment of the
intrinsic safety of the device and a schedule for the availability of the certi-
fication. The device should not be installed unless the user has sufficient
knowledge and resources to review the vendor's self-certification docu-
ment and assess that certification is nearly certain on the schedule stated.
Control House Apparatus Not Certified As Associated Apparatus
For the same reasons that it is not practical to self-certify an item of field-
mounted apparatus, it is not sensible to consider self-certifying or accept-
ing a vendor's self-certification of control room apparatus, and there is
seldom any need to consider such an option. In principle, one needs only
to interpose between the control house apparatus and the field apparatus
a certified associated apparatus, such as a barrier. The barrier must be
compatible with the intrinsic safety parameters of the field device and
must also allow the system to function as intended when it is connected
between the field device and the control room device.
Uncertified Control Room Apparatus Connected to Uncertified

Field-Mounted Apparatus
A system designer may occasionally encounter a special case of uncerti-

fied field-mounted apparatus connected to uncertified control room
apparatus. It arises usually because the system designer wants all inputs
to a particular I/O device to be intrinsically safe. If this can be assured, the
designer avoids the problems of maintaining isolation between intrinsi-
cally safe and nonintrisically safe circuits connected in the same terminal
compartment. One solution is to provide the field-mounted device with

another type of protection, usually an explosionproof or flameproof enclo-
sure or pressurization. Barriers can be installed in the signal lines to the
control room. These are usually mounted in the same enclosure as the
apparatus generating the signal, but could be installed in a separate enclo-
sure suitable for the location. Barriers also should be installed at the
control room apparatus. These two sets of barriers prevent ignition capa-
ble energy from entering the Division 1 location on the signal lines from
either the field apparatus or from the control room apparatus.
Barriers must be selected so that the current drawn into a fault caused by
shorting and grounding of the signal lines in the Division 1 location
remains at a safe level; that is, the sum of the short circuit currents of the
barriers must be safe. This scheme is seldom applicable to a transmitter
powered over the signal lines, as in two-wire, ungrounded, 4 mA to 20 mA
transmitters. It is difficult to specify barriers that meet the functional
requirements of the circuit and also have safety descriptions with low
enough values of current and voltage to permit summing four barrier cur-
rents into a ground fault in the line between the barriers. The technique is
more frequently used with field-mounted transmitters that have their own
source of power, so that the signal from the transmitter to the control
house is a voltage or a current of relatively low level, perhaps 1 V to 5 V, or
4 mA to 20 mA. Barriers with safety descriptions (such as 9 V, 100 mA) can
be used in such signal lines to ensure both safety and function of the
circuit.
Examples of Basic Intrinsically Safe System Architecture
Figures 11-11 through 11-14 exemplify the use of single protective blocks
in circuits typically found in process measurement and control applica-
tions. The examples are drawn assuming that the protective blocks are
fuse-protected, zener diode barriers, as described in the standards for
intrinsic safety, which must be grounded. This type of barrier is discussed
in more detail later in this chapter. In Zone 1 applications the barrier
might be a nonisolated barrier with active series regulating components.
Vendors of barriers make available a wealth of application information,

including specific recommendations of barrier models for use in the appli-
cation being discussed. The reader would be wise to use this free support.
In all of the illustrations, the hazardous location to the left is identified as

Division 1, Zone 1. In many applications the configuration illustrated may
be acceptable for Zone 0, but some experts, and some national installation
codes, recommend that lines entering Zone 0 be protected by blocks that
galvanically isolate the Zone 0 circuitry from the circuitry in the Zone 2 or
nonhazardous location. Galvanic isolation is isolation by an element such
as a transformer or optical isolator so that no significant current can flow
across the element into the Zone 0 location. Galvanic isolation prevents
current flow between the input and output circuits of the isolator such as
could flow in a nonisolated system with field and control house grounds
at different ground potentials. If galvanic isolation is provided, the circuit
in the Zone 0 location may be grounded or not, depending only on the
desired function of the circuit. In some jurisdictions if the intrinsically safe
circuit is not galvanically isolated from the nonintrinsically safe circuit,
which is the case in all the illustrations in Figures 11-11 through 11-14, the
intrinsically safe circuit must not be grounded. This is the reason that stan-
dards for intrinsically safe apparatus mandate isolation of the circuitry
from ground (or enclosure), and a 500-V dielectric breakdown test to
ensure it.
Figure 11-11 illustrates simple approaches to temperature measurement

using resistance temperature detectors, RTDs, and thermocouples. The
uppermost configuration of a bridge is seldom used today. It was once
commonly used when a d'Arsonval type galvanometer or a self-balancing
potentiometer provided an indication or record of temperature. The mid-
dle diagram shows the resistance element powered by a constant current
supply and the voltage signal across the resistance element being ampli-
fied to be used for indication or control, as is common in digital indicators
and controllers. The amplifying block also does digital computations that
compensate the output for changes of resistance in the leads to the ele-
ment. Although it is not necessary that the resistances in all three of the
barriers in the two diagrams be equal, most barrier manufacturers provide
barriers for this application which are matched in resistance and which
have low change in resistance as ambient temperature changes. Which
two barriers need matched characteristics depends on the way in which
the instrument compensates for changes in lead resistance. If a four-wire
connection to the element is used, then four barriers are needed, one in
each lead to the bulb, unless the measuring circuit maintains the fourth
wire at ground potential.
The connection of barriers in thermocouple circuits connected to electronic

instruments such as the one illustrated poses little problem in selecting the
barriers. A low-voltage barrier, 10 V, 200 mA, provides for longer leads
than higher voltage barriers, and yet is of high enough voltage so common-
mode voltages are unlikely to cause currents to flow in the lead and barrier
resistances and cause measurement errors. In those jurisdictions where a
grounded thermocouple is not permitted, or in applications where a
grounded thermocouple causes measurement problems, this arrangement
is replaced by an isolating type T/C voltage or current converter.
Figure 11-11 Temperature Measurement with RTDs and Thermocouples
The upper illustration in Figure 11-12 shows digital input circuits pro-
tected by individual barriers. Circuits protected by a common supply
barrier and individual return barriers are illustrated at the bottom. Most
digital input I/O circuits operate on a current of 10 mA to 15 mA and a
supply voltage of 24 V. The return barrier can be a low-voltage, low-resis-
tance barrier, such as a 10 V, 200 mA barrier. The supply barrier must
protect against the 24-V supply voltage, so a 28-V, 93-mA barrier would
probably be chosen. A single supply barrier could supply a number of
switches, depending on the resistance of the field wires and the actuation
current of the sensing units.
Figure 11-12 Digital Inputs

Figure 11-13 illustrates a strain gauge bridge with one of the supply lines
grounded. Diode leakage flowing in the resistance of the common return
to the supply can, in some cases, introduce an error in the measurement.
The supply barrier must be selected to have a working voltage higher than
the supply voltage. The barriers on the signal lines may be lower voltage
units because the maximum signal will be much less than the supply
voltage.
Figure 11-13 Strain Gauge Bridge
Figure 11-14 shows the most usual design of an analog output module
such as a 4 mA to 20 mA controller. The single barrier protects the output
because the negative side of the controller is grounded. This barrier usu-
ally must have a rated working voltage equal to the power supply voltage.
Figure 11-14 Transmitter Input and Controller Output
Passive-Shunt Diode Safety Barriers
In many applications the use of passive-shunt-diode zener barriers is the

most cost-effective way to make the circuit intrinsically safe. Figure 11-16
illustrates an early version of the shunt diode safety barrier. In modern
practice the cap over the terminal to the intrinsically safe circuit is no
longer considered necessary. Distance alone provides safe separation
between intrinsically safe and nonintrinsically safe circuits. Some models
still use a busbar mounting arrangement similar to that shown, but many
other designs are now available. Some clamp on the standard DIN rail;
others mount on printed circuit boards supplied by the user or in cabinets
supplied by the barrier manufacturer. There are many proprietary physi-
cal designs and many designs of mounting hardware intended to conserve
space in systems using hundreds of barriers. Barriers similar to that shown
in Figure 11-16 had very low packing density when installed in cabinets
and racks; so the installed cost, including the cost of racks and cabinets,
was very high, as was the consumption of precious floor space in the con-
trol room or marshalling area. Advances in high-density packaging by
barrier manufacturers have been aimed at reducing the total installed cost
of barriers.
Figure 11-15 illustrates common configurations of passive-shunt-diode

barriers. Only fuse-protected models are shown because these are the
most common. They are flexible in application because they are designed
to be intrinsically safe after application of 250-V line voltage. In resistor-
protected barriers, a resistor replaces the fuse. This configuration is not
practical for barriers rated for application of 250 V RMS because the resis-
tor value and/or its wattage rating are impractically high.
In use, the working voltage applied at the nonintrinsically safe terminals is

less than the voltage rating of the diodes so they do not conduct. Applica-
tion of a higher voltage causes current to flow through one or both diodes
to ground, as indicated by the dashed line in sketch (a). The voltage drop
across any resistance in the path to ground is added to the output voltage
of the barrier, so this resistance must be very low. Most codes require it to
be 1 ohm or less. The safety rating of the barrier is the highest voltage that
exists at the intrinsically safe terminals when rated input voltage is
applied, only one diode (at the high end of its voltage tolerance) remains
in the circuit, and series resistors are at their minimum values. The current
rating is the maximum voltage divided by the minimum value of R1. R2
has no safety function and is not included in all barriers. When R2 is
present it provides a means to check whether the diodes are intact by ener-
gizing the barrier from both ends. R3 may consist only of the fuse
resistance or may be an additional resistor to limit surge current through
the diode. The diagrams show only two diodes. This would be the case if
the diodes are routinely tested as described in Chapter 10; if not, three
diodes provide the required redundancy.
In barriers in the 20 to 30 volt range intended for general use, the value of
R1 is usually determined by the maximum permissible current permitted
by the resistive ignition curve, taking account of the 1.5 factor at the maxi-
mum open circuit voltage. In barriers with lower open circuit voltage, the
current output is usually determined by considering the amount of induc-
tance that may be connected. Barriers with open-circuit voltages above
30 V have limited application because in Group IIC atmospheres the per-
mitted load and cable capacitance is very small.
The barrier shown in sketch (a) is called a positive barrier because it is

intended to be connected to a supply with the negative side grounded.
Sketch (b) shows a negative barrier. Both are intended for connection to dc
sources. For connection across an ac source a zener voltage must be pro-
vided in both directions as shown in sketch (c).
Figure 11-15 Typical Passive Barrier Circuits

Figure 11-16 Typical Barrier Construction and Mounting
Zener diodes with low voltages are not readily available so barriers with
outputs of several volts or lower use one or more conventional diodes in
series. A second set must be provided, as shown in sketch (d) to afford
protection for faults of both polarities, so these barriers are inherently
nonpolarized.
The dual channel barrier illustrated in sketch (e) provides two barriers in a
single package. Some vendors provide more than two channels in one
package. A common application of two-channel barriers is in two-wire
floating circuits such as are common with 4 mA to 20 mA transmitters.
One barrier has a high voltage rating (e.g., 28 V) for use in the supply lead
(+), and the other has a low voltage rating (e.g., 10 V) for use in the signal
return lead, which will never be at greater than 5+ volts to ground. Triple
channel barriers with matched resistance are available for use with resis-
tance thermometer detectors.
A dual-channel, nonpolarized barrier has the disadvantage that if oppo-

site polarity faults are applied to the two ungrounded input terminals, the
voltage between the output terminals will be the sum of the two barrier
outputs, which limits the amount of capacitance that may be connected.
Sharing diodes as in the star-connected barrier of sketch (f) limits the volt-
age across the load to the voltage of one barrier.
The diode return barrier in sketch (f) makes application of barriers to

ungrounded two-wire circuits much more flexible. In the dual-channel
barrier discussed above, unless the transmitter were rated for the maxi-
mum voltage of either barrier and the sum of the short circuit currents of
the two barriers, one could not apply the dual barrier unless one could
ensure that subjecting the device to the sum of the short circuit currents of
the barriers would pose no hazard. In many cases this is true, but a user
would not usually have sufficient information to make the decision. Using
triplicated diodes in the return path barrier reduces its contribution to the
fault current to essentially zero; so the voltage, current, and power of the
supply barrier is, for practical purposes, the voltage, current, and power of
the pair of barriers in parallel.
The advantages of passive barriers such as described here are as follows:

(1) relatively low cost per barrier package
(2) high packing density in modern packages so hundreds of barriers can
be installed in a small space, giving relatively low installed cost
The following are major disadvantages of passive barriers:

(1) the need for a dedicated low resistance connection to the grounding
point of the ac supply system
(2) the need to isolate loads in hazardous location from ground, as
determined by a 500-V dielectric strength test, both initially and
during service.
(3) the measuring error that leakage through the diodes may introduce in
some applications
(4) the inability to use in Zone 0 applications in some jurisdictions
because of code of practice requirements for isolating constructions
Although most passive-shunt-diode barriers are designed for Groups A

and B (Group IIC) to allow maximum flexibility in application, a number
of barriers are certified only for Group C (Group IIB) in order to take
advantage of the higher permitted output current. These are useful espe-
cially for driving paralleled transmitters or high-current solenoids.
Some barriers are available with replaceable supplemental fuses. These

fuses are quick acting and of lower rating than the internal fuse of the bar-
rier. They will open before the internal fuse, thus avoiding the
replacement of an entire barrier if a field lead short or ground, or a noise
pulse, causes a high enough current to blow the internal fuse.
Active Barriers
In an attempt to alleviate the disadvantages of passive barriers many

kinds of active barriers have been introduced.
Shunt Diode Barriers Augmented by Active Circuits—Current Mirrors
Probably the first electronic ancillary function added to the basic shunt-
diode barrier was a current mirror. The most common function of a cur-
rent mirror is to make a low impedance measurement of the current
flowing in a floating two-wire, 4 mA to 20 mA loop and reproduce it in an
output circuit that will drive a much higher load impedance. The need for
current mirrors arose in the 1970s in 4 mA to 20 mA transmitter loops. The
supply barrier in a transmitter loop at that time was usually a 28-V, 300 Ω
barrier with an end-to-end resistance of about 340 Ω. The return barrier
was perhaps a 10-V barrier with end-to-end resistance of 230 Ω. Maximum
working voltage of the supply barrier was typically 26 V.
The voltage drops around the loop at 21 mA to permit overranging so as

to be able to indicate the difference between a full-scale reading and maxi-
mum possible current in the loop can be summed up by the following:
Supply barrier 7.14 V
Transmitter 11–12 V
Return barrier 4.83 V
250 Ω controller input resistor 5.25 V
Total drop around the loop 28.22–29.22, greater than the
supply voltage
Even today, when the supply barrier may have a total resistance of only
280 Ω and the drop across the return barrier is only 2 V, the total drop
around the loop is 24.4 V to 25.4 V. The drop around the loop can be
reduced by placing the 250-Ω measuring resistor on the hazardous loca-
tion side of the return barrier so the return barrier drop need not be
counted, but the drop is still 22.4 V to 23.4 V, with no allowance for line
drop. The drop in the example in the table can be reduced by the same tac-
tic, but the design is still marginal. For this reason, most transmitter loops
are fed from barriers incorporating a current mirror. Usually a regulator
circuit is included so that the transmitter supply need not be tightly regu-
lated. The regulator provides a voltage of 15 V to 20 V to power the
transmitter despite wide swings of supply voltage, such as might result
from using only a ferromagnetic regulating transformer on the supply, or
an unregulated supply. The mirror repeats the transmitter current to a
load of at least 250 Ω, but typically greater.
Overvoltage/Overcurrent Protection
If a barrier is accidentally short-circuited, the fuse may blow if it has not

been designed to be proof against short circuits. Before the introduction of
replaceable fuses in series with the encapsulated fuses in the barrier, this
meant replacing the barrier. A troublesome plant startup might mean
sales of hundreds of replacement barriers for the barrier vendor. Thus,
models were introduced that incorporate electronic current and/or over-
voltage sensing and automatic limiting of current to a value below the
value of the safety fuse.
Neither current mirrors nor automatic current limiting necessarily isolate

the barrier input circuit from its output. Intrinsic safety of the loop is still
provided by a shunt-diode barrier. The ancillary electronics are on the
nonhazardous side of the barrier. Protection still requires that the shunt-
diode barrier be grounded to a low-impedance grounding system.
Barriers with Electronic Limiting or Regulation
Barriers with electronic limiting have been available for many years. Some
used SCRs triggered by low-power zener reference circuits to avoid the
high wattage zeners needed in passive barriers but used a resistor for cur-
rent-limiting. Others used linear current-regulating circuits in addition to
zener-steered voltage-limiting circuits to achieve a more rectangular or
foldback type of V-I output characteristic. These latter types of barrier typ-
ically have lower short-circuit currents for the same open-circuit voltage
than passive barriers, which may be an advantage if the load has a low
current rating.
Loop-Powered Isolated Devices
These devices operate from the 4 mA to 20 mA current signal in a loop and

replicate the signal for driving devices such as solenoids or electropneu-
matic transducers or valve positioners. Because the devices are isolated,
the load need not be isolated from ground. Other devices produce only an
indication of the value of the loop current, but no output signal.
Barriers with Isolated Input and Output Circuits
There is a barrier that provides complete isolation of the input circuitry

from the output for almost every application. Many were designed to sat-
isfy the preference in some markets for isolated barriers for loads located
in Zone 0 locations. Others are used because it is difficult to ensure that
sensors such as thermocouples remain isolated from ground during their
lifetime. Using isolated barriers frees the user from concern that two
grounds may develop in a system, thus compromising function, and vio-
lating the basis for certification of a nonisolated barrier. For discrete
signals, such as those used for relay or LED actuation, the isolation may
use an optical-isolator open loop, without feedback. For analog signals
where high accuracy is unnecessary, as in the protection of a controller
output driving an electropneumatic valve positioner or transducer, a
transformer may couple a chopped dc signal to a rectifier that drives the
load, without feedback of the rectified signal.
For use with two-wire transmitters with floating outputs, isolating barri-
ers provide a regulated source of voltage to operate the transmitter and
current-mirror circuitry to measure the transmitter current and replicate it
in an isolated circuit to the controller or other load.
Some isolated barriers provide overvoltage/overcurrent protection. Isola-

tion of intrinsically safe switch contacts, located in the hazardous location,
from a load in the nonhazardous location is often provided by a relay.
Closing the intrinsically safe switch causes an amplifier to energize a relay
coil, the contacts of which control the load and which may operate at line
voltage.
The major advantage of all isolated barrier devices is that no grounds are
needed, but all grounds are tolerated. The system is safe regardless of how
it is grounded, so the user may ground the system to ensure best function.
Intrinsically Safe Multiplexers
Many varieties of intrinsically safe multiplexers are available to sequen-

tially collect the signals from many sensors or transmitters in the
hazardous location and transmit them over a pair of intrinsically safe
wires to a receiver. Another wire or wires, also intrinsically safe, provide
power for 4 mA to 20 mA transmitters. The wires carrying the multiplexed
signals may be intrinsically safe, by virtue of grounded barriers at one end,
or they may be driven by isolating barriers. In some designs reed switches
provide isolation of low-level sensors such as thermocouples and RTDs, so
grounding of the sensor does not contravene the certification if grounded
barriers protect the transmission line for the multiplexed signals.
Choosing Associated Apparatus and Barriers
There are many logical approaches to selecting a barrier device. No matter

how the selection is attacked, the same decisions must be made. The
author suggests making the choice between a passive barrier and an iso-
lated barrier first. This decision immediately reduces the number of
options that must be investigated; but, if the decision is to use isolated bar-
riers, it makes many of the other steps easier to complete. Isolated barriers
tend to be designed with features needed for specific target applications.
Vendors of isolated barriers have already considered the problems pecu-
liar to that application and have incorporated the needed features in their
designs. Vendors also have preengineered recommendations for applica-

tion of passive barriers, but the user must be more careful to ensure that
the system will function in addition to being safe. The user must:
(1) Decide whether the barrier must be an isolating device or not. Making
this decision first immediately reduces the number of alternatives that
might otherwise be investigated. In the discussions earlier in this
chapter, isolation was discussed primarily in the context of being
unable to procure an isolated field device; being unable to maintain
isolation from ground over the useful life of the installation, as in
thermocouple or RTD applications; or local codes that require isola-
tion for Zone O systems. Another compelling reason for choosing an
isolating barrier is to reduce installation cost. If a low-resistance
grounding conductor must be installed and run for a great distance to
connect the dedicated barrier ground conductor to the supply system
ground, the cost can be high. If one installs only a few intrinsically
safe loops, the average installation cost per loop may be so high that
choosing more expensive isolating barriers rather than passive barri-
ers may reduce total system cost. In most petrochemical plants, if the
hazardous location apparatus is located in the same unit as the con-
trol room where the barriers are located, the installation of a suitable
barrier ground for passive barriers may not be costly. This is because
most such plants have a very low impedance ground grid, which
allows inexpensive connection to the ground point, “the ground
stake,” of the supply system. If the hazardous location equipment and
the associated apparatus are located in different units, with different
ground reference planes, it may be wise to use isolating barriers to
eliminate vexing questions about the effect on function of ac potential
differences between the two ground planes and the consequences of a
grounded cable.
The cost of running a special grounding conductor for passive barri-

ers is most likely to come into question in plants operating inside
buildings, especially older ones, because the building steel cannot be
relied upon to provide the same kind of low-resistance ground mat
that concrete-enclosed reinforcing rod with connections to the struc-
tural steel of the plant provides in the outdoor plants of the continu-
ous process industries One must run a dedicated conductor to the
service entrance ground.
(2) Define the safety characteristics of the load. Describe the device by
vendor and model number and state the intrinsic safety parameters
relevant to this application (i.e., from the certification that is applica-
ble in the jurisdiction of the plant). At present the values of standing
capacitance and inductance may be slightly different depending on
the laboratory whose certification is being quoted. The values of per-
missible applied voltage, current, and power may also differ among
laboratories. If the application is not Group A or B (Group IIC) state

the gas group that applies. If the intrinsically safe device is a solenoid
or other load that requires significant power, designing for only the
gas group that is essential can make available greater power capabil-
ity from the associated apparatus, and may make the difference
between having a functioning system, or not.
(3) Define the operating electrical characteristics of the device, such as

minimum operating voltage and current requirements; internal resis-
tance; load impedance, if any; and estimated cable resistance. Record
the expected range of output voltage of the loop or system power sup-
ply under use conditions.
(4) Consult vendor catalogs to determine what system configuration the

vendor suggests for the application at hand, and choose a candidate
barrier or barriers.
(5) Check to ensure that the power-supply voltage does not exceed the
maximum working voltage of the passive barrier chosen or remains
within the design range of an active barrier.
(6) Check to see that the voltage drops around the loop are within the
capability of the power supply or the active barrier output.
(7) Check to ensure that the load and cable parameters do not exceed
those permitted by the barrier specifications, and that the barrier,
field device and mounting location all are certified for the material
group of the location.
Systems with Multiple Sources
The assessment of a system with barrier outputs that may be in parallel is

relatively straightforward if both barriers have the same ground reference
and the same polarity with respect to ground. The maximum current is the
sum of the two short circuit currents, and the voltage is the Thevenin
equivalent of the two barriers with output terminals in parallel. There are
no simple approaches if the two barrier devices have different kinds of
output characteristics (e.g., a rectangular source in parallel with a trape-
zoidal supply). Goldner et al. provided a series of figures, and guidance
for using them, in PTB Report W-39 and in Ex Magazine. The material can-
not easily be summarized here. The interested reader should consult the
original report or the presentation in Ex Magazine.
Designing a Nonincendive System
In the United States, designing a nonincendive system almost always

means ensuring that the system has nonincendive field wiring, so that it
may be serviced while energized. Nonincendive field wiring, not recog-
nized yet outside the United States, is incapable of causing ignition when
the field wiring is opened, shorted or grounded. In keeping with all other
standards for nonincendive apparatus, the judgement is based on the
equipment being in its normal operating condition and used within its
stated ratings.
Although it is necessary in an intrinsically safe system to protect every

ungrounded line entering the Division 1, Zone 1 location from carrying an
ignition capable energy level, in most measurement and control instru-
mentation the normal operating condition of the apparatus limits the
current to levels that will not cause ignition. This is obvious in the case of
RTDs and thermocouples If one opens, shorts, or grounds the leads to
these sensing elements, currents in the milliampere range may flow, and
the open circuit voltage will be only a few volts.
The output of a 4 mA to 20 mA A/O block or controller typically will not

exceed approximately 30 mA under short-circuit conditions, and the open
circuit voltage will be 24 V to 30 V. Even though this device is a constant
current source and must be judged according to the data provided by
Johannsmeyer, this level is not ignition-capable in a resistive circuit as
well as in most valve positioner or I/P transducer loops. If the load is
inductive, ask the vendor for a maximum inductance value and check it
against the Johannsmeyer data. In some cases it may be necessary to use
the vendor's intrinsically safe model, which has diode shunting of the coil,
to give low effective inductance.
Digital input circuitry normal operates with a current through the switch
when it is closed of 10 mA to 20 mA, so the switched loop itself is not igni-
tion-capable. Analog inputs from 4 mA to 20 mA transmitters operating
from a 24 V to 30 V supply are also not normally ignition capable.
The parts of a process measurement and control system that need special
attention because they are likely to violate the conditions for nonincendive
held circuits are the following:
(a) A power supply that delivers power to one or more transmitters or
status switches without any impedance to limit the current that flows
if the positive wire to a field device is grounded. Because there is nor-
mally no large capacitance or inductance involved in these circuits, a
resistor can usually be sized from the resistive ignition curves. It
should be installed at the power supply terminals so that any dis-
charge must pass through it.
(b) A 4 mA to 20 mA controller that drives an inductive load, such as an

electromagnetic valve positioner or transducer, is usually safe; but in
some cases it is necessary to ensure that a vendor's diode-shunted
model is used to keep the effective inductance at a safe level.
(c) Any inductive load such as a solenoid that is being driven by a digital
output must be investigated for the effect of opening the circuit, as
well as for the effects of grounding the supply lead.
The salient difference between intrinsically safe systems and nonincendive

systems is that the user cannot assess the safety of uncertified apparatus in
an intrinsically safe system, with the exception of simple apparatus, but
the user can assess the safety of a nonincendive system. Although it is cus-
tomary today to want a certificate for any apparatus that is safety-related,
situations will arise where a certificate is not available.
The user of the apparatus can learn all that is needed about the system by
making simple measurements when the apparatus is in its worst case nor-
mal operating condition and by asking a few questions of the
manufacturer of some pieces of apparatus.
If a device such as a controller or a thermocouple-to-current converter is

not certified, measure the short-circuit current between the input termi-
nals and between the input terminals and ground, switch the meter to the
voltage range and measure the open-circuit voltage between the terminals
and between the terminals and ground. Assess the safety by looking at the
resistive ignition curves, applying a test factor of 1. Use the Johannsmeyer
curves if the device has a constant current output.
The same measurements can be taken with any other source of energy to
the field wiring. However, in the case of an unprotected transmitter power
supply, the fuse may blow; a warning that protection is needed that can-
not be ignored.
The fuse, as in intrinsically safe circuits, is not considered to provide pro-

tection against arc ignition.
References
“Code of Practice on Zener Barriers,” British Industrial Measuring and
Control Apparatus Manufacturers' Association.
ISA RP12.2, “Intrinsically Safe and Non-Incendive Electrical Instruments,”

Instrument Society of America, Pittsburgh, 1965.
McKinney, A. H., “Circuit Design,” Monograph #110, pp. 105–108, Instru-

ment Society of America, Pittsburgh, 1965.
Towle, L. C., “The Use of Safety Barriers on Hazardous Plants,” Instrument

Engineer, pp. 60–64, October, 1965.
Redding, R. J. and L. C. Towle “Barrier Method of Ensuring the Safety of

Electrical Circuits in Explosive Atmospheres,” Proc. IEE, Vol. 113, No. 12,
pp. 2070–2074, December, 1966.
Silver, S., “Energy Safety Factors of Air and Iron-Cored Coils with Refer-
ence to Intrinsic Safety,” Report FMP66/96-EEC, Department of Mines
and Technical Surveys, Ottawa, 1966.
Maltby, F. L., “Design of Intrinsically Safe Circuits,” Instrumentation in the

Chemical Petroleum Industries, Vol. 3, pp. 134–137, ISA Pittsburgh, 1967.
Widginton, D. W., “A Method For Assessing the Effective Inductance of

Components Used in Intrinsically Safe Circuits,” SMRE Research Report
No. 254, 1967.

cuits,” SMRE Report No. 256, Safety in Mines Research Establishment,
Sheffield, 1968.
Magison, E. C., “Design of Intrinsically Safe Systems,” Instrumentation in

the Chemical and Petroleum Industries, Vol. 6, pp. 42–47, ISA, Pittsburgh,
1970.
Morgan, M. J., “Intrinsically Safe Barrier Variations,” Instrumentation in the

Chemical and Petroleum Industries, Vol. 6, pp. 48–55, ISA, Pittsburgh, 1970.
Redding, R. J. Intrinsic Safety, McGraw-Hill, London, 1971.
Schebsdat, F. “Über den Einfluss von Fernmeldeleitungen auf schlagwet-

ter- und explosionsgeschützte eigensichere Stromkreise,” PTB Report W-
3, Braunschweig, 1974.
Capp, B. and D. W. Widginton, “The Intrinsic Safety of Resistive Circuits,”

IEE Conference Publication No 134, pp. 43–47, London, 1975.
Halama, S., “Suppression of the Ignition Hazard by Control of the Break-

ing Arc of a Power Source,” IEE Conference Publication No 134, pp. 38–42,
London, 1975.
Schebsdat, F. “The Influence of Transmission Cables On Intrinsically Safe

Circuits With Direct Voltage,” IEE Conference Publication No 134 pp. 48–
53, London, 1975.
Thurlow, R. and P. J. Green, “Progress in the Design of Intrinsically Safe

Power Supplies,” IEE Conference Publication No 134, pp. 32–37, London,
1975.
Virr, L. E., “Methods for Assessing the Inductive Effects of Cables Used In
Intrinsically Safe Systems,” SMRE Technical Paper P19, Sheffield, 1976.
Göldner, H. D., Johannsmeyer, U., Schebstat, F. and H. Storck, “Zusam-

menschaltung nichtlinearer und eigensicherer Stromkreise,” PTB Bericht
W-39, Braunschweig, 1989.
Johannsmeyer, U., “Zünderverhalten von Netzwerken eigensicherer

Stomrkreise mit elektronischen Strombegrenzungseinrichtungen,” PTB
Report W-16, Braunschweig, 1979.
Johannsmeyer, U., “Investigations to the Intrinsic Safety of Field Bus Sys-

tems,” PTB Report W-53e, Braunschweig, 1994.
ISA-RP12.6, Wiring Practices for Hazardous (Classified) Locations Instru-

mentation, Part I: Intrinsic Safety, Instrument Society of America,
Research Triangle Park, 1994.
Bass, H. G., Intrinsic Safety, Instrumentation for Flammable Atmospheres,

Quartermaine House Ltd., Sunbury, 1984.
“Introduction to Intrinsic Safety”, ELCON Instruments, Inc., Annapolis,

1988.
Johannsmeyer, U., “The Ignition Characteristics of Nonlinear Current and

Voltage Sources in the Testing of the Intrinsically Safe Circuits,” Ex Maga-
zine, p. 45, December 1988.
Safety in Mines Research Establishment - 1974 Annual Report, “Inductive

Components Shunted with Zener Diodes,” Ex Magazine, p. 28, October
1977.
Goeldner, H. D., Johannsmeyer, U., Schebsdat, F., and H. Stork, “Combi-

nation of Non-Linear and Linear Intrinsically-Safe Circuits,” Ex Magazine,
pp. 11–21, December 1990.
Towle, L. C., “Cable Parameters for Intrinsically Safe Circuits,” IEE Con-
ference Publication 218, pp. 228–233, London, 1982.
ISA TR12.2, “Intrinsically Safe System Assessment Using the Entity Con-
cept,” Instrument Society of America, Research Triangle Park, 1995.
CHAPTER 12
IGNITION BY OPTICAL SOURCES
Overview
Until early in the 1980s it was assumed that a simple answer to many
safety issues was to use optical fibers to transmit information about the
plant. Optical coupling of control house equipment to field equipment
with appropriate electric-to-light and light-to-electric transducers at each
end would eliminate the safety issue of an electrical apparatus failure cou-
pling ignition-capable energy into a low-level circuit. In this way one
could avoid the complications of intrinsic safety, as well as gain increased
immunity from noise and interference.
Early investigators of the possibility of ignition by optical sources con-

cluded that direct ignition of gas or vapor clouds by radiation is not likely
to be a hazard worthy of concern at the power levels used in instrument
systems for measurement and control. The experimental data did, how-
ever, indicate a mechanism of ignition that needs serious consideration at
levels of radiation that can be found in industrial applications. Heating a
particle or surface with radiation from a laser or an optical fiber can ini-
tiate ignition of a flammable gas or vapor.
Mechanisms for ignition that are not considered needful of concern with
current applications of optical sources are as follows:
• Thermal ignition of a gas volume—Gases absorb only in narrow
bands, so the total amount absorbed is a small fraction of the inci-
dent energy. Thus, the efficiency of this mechanism is very low.
• Photochemical ignition due to photodissociation of oxygen mole-
cules by radiation in the ultraviolet region, setting the stage for
further chemical reactions.
• Direct laser-induced breakdown of the gas at the focus of a strong
beam, producing a plasma and sparking.
All these mechanisms require much more energy than is likely to be

present in most current applications of optics and optical fibers.
From December 1990 to May 1994, a collaborative program in the Euro-

pean community undertook research to better bound the phenomenon.
On the basis of this work, and other work done earlier, the U. K. Optical
Sensors Collaborative Association (OSCA) issued a guidance note in Feb-
494 Ignition by Optical Sources
ruary 1995. This note presented the conclusion that a continuous wave
device, radiating in the visible and near visible portions of the spectrum, is
not hazardous if either,
(a) the radiated power is less than 35 mW, or
(b) the peak radiation flux is less than 5 mW/mm2
No position was taken on pulsed or intermittent sources or for sources
smaller than 0.00003 mm2. No guidance was given about safe radiated val-
ues in the presence of gases or vapors that are more difficult to ignite than
those for which the limit values were derived. Excluded also are “sensitive
materials” (i.e., pyrophoric materials or materials containing their own
oxidizer).
This conclusion is helpful to an extent because many applications of opti-

cal fibers fall below the very conservative guidelines recommended by
OSCA. However, the methods by which one is to judge whether a value of
radiation that is higher than the OSCA limit values poses a hazard in a
particular application, or how to specify procedures to ensure a safe instal-
lation with respect to explosions, are not addressed by this document. It is
OSCA's position that other bodies must answer these questions.
The mechanism of ignition underlying the OSCA recommendation is the

ignition of gases and vapors by hot particles and surfaces. Some of the
data included in the OSCA summary are PTB's data on ignition by small
components, which is discussed in Chapter 9 of this book. Thus, the mat-
ter could reasonably be discussed along with the discussion of ignition by
hot surfaces, components, and small wires in Chapter 9.
The author has chosen to devote a separate chapter to this subject for the
following reasons:
(1) The occurrence of an explosion is dependent on three elements being
present:
• a flammable mixture
• a source of radiated energy of the duration and intensity needed
to cause ignition
• an appropriate target particle or surface
Safe practices for electrical installations have been discussed elsewhere
in this book in terms of reducing the probability of the simultaneous
presence of an ignition source and a flammable atmosphere. The need
for a third factor for an explosion to occur opens up an expanded set of
alternatives. One should certainly exercise these alternatives before
concluding what protective measures are appropriate in any specific
case.
Ignition by Optical Sources 495
(2) The technical basis for developing rational responses when radiation
from fibers or lasers may be hazardous is not well understood. There
is not yet strong consensus on whether the groupings of materials that
are useful for responding to the dangers of spark ignition when using
the techniques of explosionproof construction, intrinsic safety, and
nonincendive circuits are relevant to this hazard. Hot-surface ignition
by components has been dealt with outside of such groupings
because the hazard is not directly related to the phenomenon of arc
ignition. There is no consensus yet on when a hazard from radiated
energy may exist. The values quoted by OSCA are lower limit values
with a divisor of about 2 already factored into the lowest experimen-
tal values. (PTB ignition of carbon disulfide at 50 mW with a
0.002-mm2 fiber and Carleton and Weinberg's worst-case irradiance
value for ignition of coal, 9.34 mW/mm2, are not shown in Figure
12-1.) They are based largely on work with two materials with low
ignition temperatures, carbon disulfide and diethyl ether. This is sen-
sible if one's objective is to quote values below which one can feel con-
fidant that no hazard results. Statement of such limits is a useful
conclusion, as long as it is understood that it is analogous to applying
a factor of safety to minimum ignition energy values. It is useful
because it takes the first step, providing a rationale for eliminating
from concern a large number of applications from the universe of
applications that might conceivably pose a hazard.
There are no data relating probability of ignition to strength of source,

as exists for electrical sources. Safety analyses such as those discussed
in Chapter 4 are less definitive because this knowledge is lacking.
Figure 12-1 Data from EC Research Program and OSCA-Recommended “No Risk”
Limits
(3) To separate this discussion from the long history of dealing with elec-
trical ignition issues. A new look should be taken at this new family of
hazards. Only poor solutions will arise from attempts to apply rules
established for one hazard to another of a different type. This
approach yielded a clause in the 1996 NEC, for example, about “opti-
cal fiber cables and fiber-optic devices approved as an intrinsically
safe system...” a subject no standard has ever addressed, and one that
is misleading to the extent that the reader might conclude erroneously
that a protective technique related to intrinsic safety is required to
ensure a safe installation.
(4) The author believes that an orderly analysis of the applications that
seem likely to present a hazard will:
(a) lay the basis for technically sound guidelines for reacting to the
potential hazard
(b) help focus on those facets of the phenomenon of ignition by radi-

ation that deserve highest priority for further study
Ignition by Radiating Sources

The purpose of this section is to summarize the salient facts about ignition
by sources of radiation, emphasizing the visible and near-visible wave-
lengths. In most systems, broader-band emissions are attenuated by the
optical elements in the path of the beam unless they are made with infra-
red transmitting materials. The principles discussed are equally applicable
to systems operating in the infrared regions of the spectrum. The treat-
ment of experimental results is deliberately oversimplified because the
objective of this section is to present enough information to help scope out
boundaries for action by standards developers and regulators. The reader
interested in a more complete understanding of the underlying phenom-
ena is encouraged to refer to the original sources.
Data on ignition by optical sources, which has been studied for about 15
years, is more limited than that on ignition by electrical sparks, which has
been studied for 85 years. Much of it has been obtained to define a lower
limit value, below which the risk of ignition is negligible.
Data for ignition by radiation, like any combustion data, are experiment
dependent. The results depend on the following:
• size of the irradiating beam
• size of the particles being irradiated
• spectral properties of the source
• spectral properties (reflectance and absorbency) of the target

material
(Transmittance of targets other than dust clouds is small enough
to be neglected.)
The ignition energy or power values depend also on the way in which
measurements are reported. Some investigators report “thermal diame-
ter,” the diameter of a white circle traced by the beam on a smoked
Kaowool target, as the beam diameter. Kaowool is a proprietary, low-den-
sity insulating mat of ceramic fibers. Though white at visible wavelengths,
its fibrous nature provides a multitude of cavities that increase its emit-
tance in the infrared. Its absorption coefficient is near 1 at the wavelength
of a CO2 laser, 10 μm. Coatings of SiC, coals, and lacquer can increase the
absorbency to 0.8–0.95 at 1.06 μm, the wavelength of the Nd:YAG lasers
that are used for some investigations.
Other investigators consider beam diameter to be the diameter of a circle

that contains 63% of the beam energy. These diameters vary directly with
changes in beam configuration by the optics of the system, whereas the
thermal diameter depends on the amount of power in the beam as well as
the optical configuration. For a given optical configuration, the thermal
diameter increases as power in the beam increases. Other investigators
have reported the maximum power flux on the axis of the beam.
The phenomena of concern can be categorized as follows:

• ignition of combustible dust clouds by direct illumination of the
cloud
• ignition of dust clouds by heating or igniting particles collected
on a surface in contact with the dust cloud
• ignition of gas or vapor by irradiation of a dust cloud within the
vapor cloud
• ignition of a gas or vapor cloud by heating isolated particles or a
particle-covered surface in contact with a flammable cloud
Ignition of Combustible Dust Clouds by Direct

Illumination of the Cloud
Proust, of INERIS (French National Institute for Industrial Environment
and Risks), reported inability to ignite coal dust-air mixtures with beam
power up to 50 W from an Nd:YAG laser. Concentration of the mixtures
was 0–400 g/m3. Particle diameter was 23 μm. (Typical LELs for coal dust
are 60 to 100 g/m3.) Devolatilization was observed at about 10 W beam
power, and burning of individual particles was observed above 30 W
beam power. At 50 W, the temperature rise in the beam was a few tens of
°C. Beam diameter was less than 3 mm. Power in the beam is attenuated
by the dust particles in the path so the amount of power in the cloud
depends not only on the beam power but also on the particle concentra-
tion. In experiments with suspensions of 23 μm coal, 28 μm starch, 40 μm
sulfur, 30 μm lycopodium, and 15 μm aluminum powders, no ignitions of
the organic materials were obtained for beam powers up to 45 W. (The
sizes given are mean particle diameters.) Ignition of a cloud of 15 μm alu-
minum particles was obtained when the concentration of particles was
12,000 particles/cm3. In this series of experiments, ignition of starch, sul-
fur, and lycopodium was attempted with both focussed and unfocused
beams. Only unfocused beams were used with coal and aluminum. The
diameter of a focussed beam is nearly zero at the focal point so the power
density is very high.
Ignition of Dust Clouds by Igniting or Heating Particles

Collected on a Surface in Contact with the Dust Cloud
Proust investigated the ignition of starch clouds by 45-μm particles of sili-

con carbide dusted on a vertical Kaowool mat to the maximum depth that
would adhere. His results were not highly reproducible, attributed by
Proust to the fact that the incipient flame was highly stretched by convec-
tive currents. He felt that flame stretch was responsible for no ignitions
except when the beam was enlarged by a factor of 2, decreasing the con-
vective currents. The minimum beam power for an ignition was 25 W. The
coefficient of transmissivity through the cloud under conditions when
ignition occurred was probably about 0.1 to 0.25, so the beam was strongly
attenuated by the cloud itself. He estimated that perhaps 5 W would be
required at the point of ignition.
Proust also investigated ignition of clouds of lycopodium, starch, and sul-

fur dusts by a Kaowool mat coated with the dust present in the cloud. He
concluded that using a combustible dust layer on the Kaowool did not
demonstrably sensitize the ignition process compared to a noncombusti-
ble layer. He also noted that the process is sufficiently complex that
considerably more data and study would be needed to verify whether the
expected increase in sensitivity when irradiating combustible dust is real-
ized or not.
Proust observed that ignition of a cloud proceeds as follows:

• decomposition/vaporization of irradiated particles in the layer
• thermal plume development in particles above a glowing point
on the target
• flame kernel development above the glowing point
and, finally, though not necessarily,

• flamespread in the cloud
The time between turning on the laser beam and development of the ther-
mal plume averaged 2.5 seconds, but ranged from 0.6 to 4.7. The time
depends on the density of the dust cloud, the characteristics of the dust
layer, and the beam power, and, this was not easily reproducible. Another
variable in these experiments is the density of the cloud, which typically
dropped 40–50% between the excitation of the laser and the formation of a
glowing area in the layer. Though the glowing area in the layer is neces-
sary for an ignition to occur, its presence does not always produce an
ignition. As observed in the previous dust cloud ignition studies, the
incipient flame is stretched by convection currents and may be quenched
in its infancy.
One certain conclusion from Proust's work is that ignition of a dust cloud
is much more likely when one irradiates a surface with an accumulated
dust layer than if one irradiates the cloud directly.
Table 1-16 in the BBC Explosion Protection Manual indicates ignition of a

number of gas-air mixtures in which was suspended metallic, coal, or
graphite dust. The nature of the ignition source is not stated, but the data
is at least qualitative support for Proust's work. Table 1-17 in the same
book reports ignition of dust-air mixtures by commercially available flash-
bulbs. PF100 bulbs, the most powerful, ignited a variety of dusts. Weaker
bulbs ignited metallic and coal dusts. This author has not yet researched
the original sources in order to relate this information to the work of later
investigators.
Ignition of Gas or Vapor by Irradiation of a Dust Cloud

Suspended in the Gas or Vapor Cloud
Proust, in connection with his work discussed earlier, attempted to ignite

7% methane-air mixtures in which he irradiated clouds of carborundum
particles of 45-μm, 60-μm, and 99-μm diameter over a range of concentra-
tions. He was unable to obtain ignition with beam power as high as 50 W,
the power limit of the laser. The laser was operated in the multimode
regime, and the beam was neither concentrated nor expanded (statistical
diameter 1.98 mm). He was not able to obtain ignition with 45 μm parti-
cles, whether the beam was focussed into the cloud or the beam was
expanded by a factor of two.
One reason that ignition by irradiating dust particles is difficult is that

most particles pass through the small region irradiated by the laser beam
before the ignition process can get started. This may not apply to particles
less than 25 μm, but such small particles tend to agglomerate and fall out
of suspension unless prevented from doing so by additives. SiC was cho-
sen as the experimental material because in attempts to ignite 7%
methane-air mixtures by irradiating a layer dust on Kaowool, the SiC
layer was more sensitive than a layer of the organic dusts, iron oxide, alu-
minum oxide, or silicon oxide. Only aluminum particles layered on the
Kaowool yielded lower values. These values were lower by a factor of 4.
Ignition of Gas or Vapor Cloud by Heating Isolated

Particles or a Particle-Covered Surface in Contact with a
Flammable Cloud
The investigations under the EC/OSCA program, and earlier work of
other investigators, demonstrate that the phenomenon of ignition
involved is a thermal phenomenon akin to ignition by small hot wires,
small surfaces, and small components. It is a thermal phenomenon that
differs from these modes of ignition of flammable gases and vapors only
in the way the body or surface is heated. Within the experimental variabil-
ity inherent in investigating any predominantly thermal ignition process,
it makes no difference how a particle or body is heated. As Tortoishell
observed, the gas or vapor cloud neither knows nor cares how the heat got
there, only that it is there.
What facts support this allegation? Figure 12-1 is the summary of those
data from the EC program that are near the limits suggested as “no risk”
values. The basis of the data points is summarized in the legend. Some of
the PTB data for electrically heated objects appear to be the same data dis-
cussed in Chapter 9 under the subject of ignition by small components.
The data points form a consistent pattern, one consistent with our knowl-
edge of hot-surface and hot-component ignition when the heat comes
from nonradiating sources. The data for large objects or targets form a
continuum with the data for small targets and objects. Data for electrically
heated objects are consistent with data for similar objects heated by
irradiation.
Ignition by a thermal source is characterized by ignition-delay time that

increases as one approaches a limiting value of the experimental variable.
The ignition delay corresponding to handbook autogenous ignition tem-
perature (AIT) values may have been several minutes. Carleton and
Weinberg showed that at the lower limit values of irradiance, such as
those plotted in Figure 12-1, the ignition delay may be tens of seconds. The
radiant flux, I mW/mm2, required for ignition at a delay of 1/2 second is
more than twice the value of the radiant flux for a 1-second delay. A plot
of It vs t yields a straight line with an intercept of (It)min at t = 0. Stout and
Jones investigated ignition of methane and hydrogen by heated wires of
15 μm to 43 μm diameters and found also that a plot of I2Rt vs t approxi-
mated a straight line. For hot wires, their interpretation was that this
indicated constant losses at the wire temperature required for ignition.
The intercept of this line with the axis at zero time can be used to estimate
minimum ignition energy for that mode of ignition.
Though the linearity of this expression at very short times can be debated,
the intercept determined from the linear assumption is certainly lower
than it would be for any other realistic assumption.
In order to provide a context for thinking about ignition by radiation

impinging on an object or surface, Carleton and Weinberg discussed the
following expression:
E = Emin + Pmint + (It)minA + IminAt
E is in Joules, P in Watts, t in seconds, it is in Joules per unit area, A is area,

and I is irradiance or power flux in watts per unit area; t and A are the
variables. The other parameters are experimentally determined constants
for a particular material.
If both t and A approach 0, the value of Emin should approach the mini-
mum arc ignition energy if the target is a black body. The investigators
chose 0.01 mJ as the limit value because the minimum ignition energy of
CS2 is 0.015 mJ. The value of this term is usually not important because it
is much smaller than the other three terms.
Dividing the equation by t yields an equation for igniting power, and

dividing by A yields an equation for irradiance.
E min ( It ) min A
- + P min + ---------------------
P = ---------- + I min A
t t
E min P min ( It ) min

- + ----------- + ----------------- + I min
I = ----------
tA A t
As t approaches infinity
P = Pmin + IminA
P min
- + I min
I = ----------
A
For large values of A

( It ) min
I = I min + ----------------
-
t
The physical significance of these terms is that for very small target areas,
especially those with dimensions smaller than the quenching distance of
the gas or vapor involved, the controlling measure of ignitability is the
power absorbed by the target that causes ignition. The minimum value is
measured after a long ignition delay.
For targets that are large compared to the quenching distance, the control-
ling parameter is the irradiance required for ignition after a long ignition
delay time.
If a large area is illuminated for a very short time, as with a super-power-

ful flashbulb, the controlling parameter is the energy density (It)min.
For conditions relevant to the use of optical instruments and fibers, the
important equation is:
P = Pmin + IminA
Carleton and Weinberg measured values of P for small areas, 100-μm

diameter, which determined Pmin; and for large areas, 15-mm diameter,
which determined Imin. Because the objective was to define safe limit val-
ues, CS2 was the test vapor. These values are shown in Table 12-1. The
numbers in the parentheses after the name of some of the materials is the
absorption coefficient at 1.06 μm as determined by the PTB investigators.
The numbers in parentheses, after the values of Power and Flux for these
materials measured with the Nd:YAG laser, are calculated values for an
absorption coefficient of 1. These values are less diverse than the raw data
and indicate the significance of the target's spectral properties. Similar cal-
culations were not made for the argon ion laser values because this author
does not know the coefficients for the wavelength of 0.488 μm. In general,
the coefficients should be higher at the shorter wavelength because of the
geometry of the layer of dust particles, but the spectral properties of the
target material may in some cases offset that increase.
If one treats the power value as one measurement and the flux density
value multiplied by the area of the 15-mm-diameter target as a second
measurement, one can solve a pair of equations for values of Pmin and Imin.
The calculated value of Pmin is not different from the measured power.
The few calculations of Imin made by this author are lower than the mea-
sured value of I by 10–15%. The author's calculations were made to see if
the value of Imin might be close to the emitted power flux of a black body
at the AIT of the vapor (i.e., 1.1 mW/mm2 for 373 K, the AIT of carbon dis-
ulfide). It seems to be much higher, perhaps influenced by the relatively
small 15-mm-diameter target.
Nd -YAG Argon ion

Material Appearance Power Flux Power Flux
mW mW/mm2 mW mW/mm2
Match head Dark blue 169 25.23 26 11.41
Fe O (PTB) Black(0.86) 296.7 15.563 216.0 10.37
Charcoal Black 368.9 15.18 257.1 12.47
Coal Black (0.95) 368.9 14.73 (14) 216.4 9.34
(350)
SiC (PTB) Dark 530.2 18.79 (16) 70.4 13.571
grey(0.85) (451)
SiC (I. C.) Beige (0.6) 619.4 29.8 (18) 158.0 17.05
(372)
Cork dust Brown 2,600 258.1 299.3 14.68
Wood Tan 3,389 253.4 299.3 25.86
Lycopodium Yellow 5,166 201.2 485.1 27.72
Kaowool Off-white (0.1) 16,760 277.3 905.0 >311
Cotton wool White >24,000 >300 >950 >311
Table 12-1 Values of P and I for Small and Large Targets, Respectively (from
Carleton and Weinberg)
The final equation for safe limits given in the EC report is:
E = 0.01 + 35 t + 2 A + 5 At
where the units of the constants are respectively, mJ, mW, Joules/mm2
and Watts/mm2. As noted above the value 0.01 was assigned based on
pulsed ignition experiments, which are unlikely to be representative of sit-
uations outside the laboratory. The value is lower than the value for any
material relevant to this study, and an additional safety factor was
included. In this author's opinion a value of ignition energy appropriate to
the mixture used, and related to the IEC standard test apparatus, is proba-
bly more appropriate. Tortoishell reported ignition energy of 5.5 mJ for 3%
carbon disulfide in air. However, the value of this term is not likely to
affect the application of the equation significantly. It is recognized that
because targets investigated during this study were not black bodies, and
the sources emitted in narrow bands, that values of Emin calculated from
experimental data are orders of magnitude greater than the value
assumed in the limit equation.
The value of 2 mJ/mm2 in the energy density term was calculated by Car-
leton and Weinberg. This author also believes that the limit equation
should always be written as “Elc = …” so that those who see it are always
prompted to remember that lc stands for “limit case,” and to “Think

before using, just as one should think before using minimum ignition
energy values in connection with practical applications!”
Ignition of Gases and Vapors Other than Carbon

Disulfide
The effect of material characteristics should ideally be accessible through

the general equation:
E = 0.01 + 35 t + 2A + 5 At
The first term is the minimum ignition energy and would be replaced by
the relevant value for the new material. Carleton and Weinberg assume
the third term to have the same value for all materials. The last term is a
temperature-related term that one might think should be related to the
AIT of the material. At present the values determined experimentally are
much higher than the value of I corresponding to the AIT of carbon disul-
fide, about 1.1 mW/mm2.
The second term is a power term. In their experiments with ignition of

methane and hydrogen with hot wires, Stout and Jones interpreted the
constant I2R losses as the power required to make up for convective trans-
fer of energy from the wire when at the temperature required for ignition.
One could interpret Pmin in the general equation as the power supplied to
the system to make up for all losses to the surroundings when the target is
at the temperature required to ignite the flammable material. How one
should adjust the value for each material is not self-evident.
Hot wire ignition studies show that the temperature of thin wire for igni-
tion of diethyl ether (AIT = 433 K) is close to the melting point of copper
(1,356 K). The wire temperature for ignition of hydrogen (AIT = 823 K)
and that required for ignition of methane (AIT = 810 K), is near the melt-
ing point of nichrome (1,625 K). Two points do not give guidance about
the shape of the function, but perhaps there is some theoretical guidance
to be found in the more recent literature. Reassessment of some of the hot-
wire ignition results from the point of view of the temperature required of
a wire of given size, and the slope of the energy vs time plots might pro-
vide insight into ignition by irradiated small objects. Meanwhile, the
experimental results using optical sources do not give a strong clue, other
than that higher AIT means a higher value of power for ignition using
small targets. Figure 12-1 contains the points of Zhang for hydrogen and
methane and points of n-pentane from PTB.
Adler, Carleton, and Weinberg reported results from targeting a 1-mm-

radius sphere of Kaowool with a CO2 laser. At 10.6 μm, the wavelength of
the radiation the white Kaowool has nearly unity absorbency. Their
results are shown in Table 12-2 (multiplication by 3.14 gives total power).
This author added MIE, AIT, MIC, and MESG values.
MIC-mA
Irradiance - MIE
Material (Break- MESG-mm* AIT-K*
mW/mm2 μJ
spark #3)*
Methane 167 280 195 1.1 810
Pentane 112 — 164 0.93 531
Propane 110 250 146 0.97 766
Diethyl ether 90 198 145 0.84 433
Ethylene 87 83 108 0.65 708
Hydrogen 80 19 75 0.28 823
Carbon disulfide 22 15 — 0.20 363
*From Bossert, Table 9-1
Table 12-2 Comparison of Measured Irradiance Values for Ignition with Other
Combustion Parameters
Power values calculated from this data are not the coefficients of the sec-
ond term of the general equation because the target is much too large. The
target is too small to consider the irradiance values to be values of the irra-
diance coefficient in the fourth term of the general equation. Though the
other combustion parameters can be used to give some idea of the
expected ordering of irradiance values, none correlates exactly with
irradiance.
Hills et al. illuminated individual, 38 μm to 45 μm particles of coal to ignite

mixtures in air of the four intrinsic safety test gas mixtures, with the
results shown in Table 12-3. The power reported is the total power exiting
from a cleaved 50/125-μm-multimode fiber, not the total power absorbed
by the particle, which is less than that. The ratio of power for ignition of
the several gases does not agree with the ratios of other combustion
parameters given in Table 12-2.
Gas % Gas—v/v Minimum power—mW

Methane 8 300
Propane 5 240
Ethylene 8 200
Hydrogen 21 140
Table 12-3 Power to Ignite IS Testing Mixtures
Dubaniewicz et al. investigated ignition of 15% hydrogen-air and 10%

methane-air mixtures by coal and rock dust particles of various sizes
attached to the end of a 200/230-μm fiber. The hydrogen mixture was
ignited by a 150 μm to 212 μm coal particle at a level of 250 mW total
power emitted from the end of the fiber; 500 mW were required for rock
dust. No ignitions of methane-air were observed with power of 500 mW
using either coal or aluminum particles.
The Nature of the Hazard

This section of the chapter addresses the nature of the hazard by first
attempting to define prototypical applications that need consideration,
then drawing the fault tree that describes events leading to an explosion. It
is possible to assign probabilities of occurrence to the salient events on the
fault tree. Using these probabilities, one can estimate the probability of an
explosion in typical installations, using the pattern established by Ben-
jaminsen and van Wiechen, discussed in Chapter 4. That step provides a
basis for making judgments about which installations need additional
safeguards and what steps might be necessary to ensure safe installations.
This step should point also to technical areas whose further study would
most benefit the definition of safe practices.
Prototypical Applications
The author believes that the prototypical applications of radiation dis-

cussed below represent the types of applications likely to be encountered
in or associated with measurement and control applications. These
devices are not intended to have all the features of real devices for mea-
surement and control. They represent only the essential elements of such
devices.
(1) Heaters or lamps—Heaters and lamps are typical of sources that
deliver energy to a target through a path whose environment is not
controlled and from which there is no feedback information to the
source to indicate any change in the radiant flux or radiated power
(2) Interrupted beam counting devices, position detectors, and open path
colorimeters—These devices transmit a beam through a path that is
not necessarily controlled relative to the presence of small particles
and may also provide an illuminated surface on which layers of dust
can accumulate. In counters and position detectors interruption or
reflection of the beam produces a signal that is used to operate a sim-
ple counting device, or perhaps a more complicated positioning
device. In colorimeters the reflected or transmitted signal may be
detected by an array of sensors that provide input for sophisticated
computing. In all such devices the function of the device provides a
relatively easy path to the addition of a manual or automatic capabil-

ity to detect whether the transmitted beam has changed. In the more
sophisticated instruments the beam intensity or power is regulated to
improve stability, so a device failure is a prerequisite to changing
from a safe level to an unsafe level.
(3) Fiberoptics links for communications or to power sensors or transduc-

ers—These links accept a signal of known maximum power level and
transmit it to another device. The receiver uses the signal to take
action or perform tasks. Lack of signal is usually easy to detect. No
hazard can exist until after the fiber breaks. A variety of means is
available to detect a break and remove power from the fiber input.
(4) Closed path analytical devices where the beam may be transmitted
through a sample cell that in normal operation has no solid targets
other than the cell walls or sensor—In these devices a failure is
needed to provide a target of unknown properties.
The essential elements of these prototypes are included in the fault tree
shown in Figure 12-2.
Figure 12-2 Fault Tree for Explosion Initiated by Radiation-Heated Hot Particles
or Surfaces
Discussion of Factors Requisite to an Explosion and

Their Assumed Probabilities
Presence of Flammable Atmosphere
Nothing new needs be defined. The definitions of zones already in use are
adequate and reasonably well-understood. They should be used to avoid
defining any new terms unnecessarily. For analysis this chapter will use
the values of frequency and duration of the presence of a flammable atmo-
sphere used in Table 4-4.
Presence of a Target Particle or Surface to be Heated by the

Radiant Energy
(1) Is a dust layer present that might be illuminated by radiation? Unless

measures are taken to avoid the accumulation of a layer, the probabil-
ity, Pd = 1 (MTBF = 0).
(2) Are there particles that provide a suitable target suspended in the
path of the radiation? Pd = 1 (MTBF = 0) in an open-path device.
(3) Does the sample filter in the analytical device maintain the radiant
path free of targets until it fails? Assume MTBF = 104 hr, equivalent to
failure once a year.
(4) Does pressurization keep the radiant energy path free of targets? Use
106 for the MTBF of the basic pressurization source. For operation of a
backup alarm or shutdown system, assume the shutoff times used in
Table 4-4 for the time delay protective devices used in increased
safety systems.
(5) Is the source of radiant energy ignition-capable? Assume Ps = 1 if it is
above the 35 mW or 5 mW/mm2 values quoted by OSCA. This
assumption, although unrealistic, forces attention on those aspects of
the phenomenon that need study to provide the technical basis for
setting standards for safe installations. If there is an interlock system
actuated by failure of a fiber or a monitor on beam strength, assume a
basic MTBF of 106 hr for the protective monitoring or interlocking sys-
tem and shut-down times similar to those used in Table 4-4 for the
shutdown elements in an increased safety system.
(6) If a fiber is run above ground and is not protected from accidental
damage by conduit or other raceway, or is not contained in robust
cable construction, assume MTBF = 106 hr. The probability of a bro-
ken fiber, PFB = 0.01 after 104 hr. If the fiber is protected, assume
MTBF = 108 hr, PFB = 0.0001 after 104 hr.
MTBE Calculations
Table 12-4 summarizes the assumptions and the computed values of

MTBE using the assumed probabilities and MTBF values discussed above.
Following the practice of Benjaminsen and van Wiechen, the explosion-
proof enclosure and unprotected sparking source are included as
references for comparison of the safety of other protection techniques. Not
all possible combinations were calculated. The objective is to elucidate the
nature of the hazard, and additional lines can be calculated if there is a
need. In the derivation of the equations for MTBE an inspection interval of
10,000 hours was assumed. This inspection would detect failures of any of
the apparatus or protective systems, except intrinsic safety, where the fail-
ure of a protective component is assumed to be undetectable during an
annual inspection.
Comments on the Calculations
The calculated probability of explosion for sparking sources without pro-

tection and those in an explosionproof enclosure are the same as those in
Table 4-4. The values for explosionproof enclosures with MTBF of 106 also
apply to nonincendive apparatus. The entries in Table 4-4 are identical for
both.
For lighting equipment and open-path devices without a protective ele-

ment, it is assumed that a target particle or surface is present and that the
beam strength is sufficient to cause ignition.
Calculations for open-path devices with interlock on beam strength are

patterned after the calculations for nonsparking devices with shutdown
devices. The equation is the simplified one presented in Chapter 4.
MTBE for optical fiber links with shutdown on loss of transmission was
calculated using the equations in Chapter 4 for nonsparking power appa-
ratus in a flameproof enclosure because two independent failures are
required to provide an ignition source.
Ignition by Optical Sources
Area Classification Zone 0 Zone 1 Zone 2 Unclassified
Total Hr/Yr 10,000 1,000 100 10 1 0.1 0.01 0.001 0.0001 0.00001
Frequency and 1× 1,000 × 100 × 100 × 10 × 1× 0.1 × 0.01 × 0.001 × 0.0001 ×
Duration 10,000 hr 1 hr 1 hr 0.1 hr 0.1 hr 0.1 hr 0.1 hr 0.1 hr 0.1 hr 0.1 hr
Protective
Equipment Time Delay MTBF g 10-1 10-2 10-2 10-3 10-4 10-5 10-6 10-7 10-8
Element
τm τg 1 1 0.1 0.1 0.1 0.1 0.1 0.1 0.1
Sparking(2) None -∞ 1.0 2.0 2.0 3.0 4.0 5.0 6.0 7.0 8.0
Sparking(2) Flanged 106 6.0 6.0 6.0 6.0 6.0 6.4 7.3 8.3 9.3 10.3
joints
Flameproof Threaded 107 7.0 7.0 7.0 7.0 7.0 7.4 8.3 9.3 10.3 11.3
joints
Lighting equip- None -∞ 1.0 2.0 2.0 3.0 4.0 5.0 6.0 7.0 8.0
ment
Open path None -∞ 1.0 2.0 2.0 3.0 4.0 5.0 6.0 7.0 8.0
counters and
instruments
Optical fiber link Break in 106 6.0 6.0 6.0 6.0 6.0 6.4 7.3 8.3 9.3 10.3
fiber
108 8.0 8.0 8.0 8.0 8.0 8.4 9.3 10.3 11.3 12.3
Open path Interlock 100 hr 106 6.0 6.0 6.2 7 8 9 10 11 12 13
counters and with beam 10 hr 6.0 6.2 7 8 9 10 11 12 13 14
instruments strength 1 hr 6.0 6.8 7.8 9 10 11 12 13 14 15
0.1 hr 6.0 7.0 8.0 9.7 10.7 11.7 12.7 13.7 14.7 15.7
Optical fiber link Shutdown if 100 hr Break—106 8.3 8.3 8.5 9.3 10.3 11.3 12.3 13.3 14.3 15.3
beam inter- 10 hr Shutdown - 8.3 8.5 9.3 10.3 11.3 12.3 13.3 14.3 15.3 16.3
rupted 1 hr 106 8.3 9.1 10.1 11.3 12.3 13.3 14.3 15.3 16.3 17.3
0.1 hr 8.3 9.3 10.3 12 13 14 15 16 17 18
Optical fiber link Shutdown if 100 hr Break—108 10.3 10.3 10.5 11.3 12.3 13.3 14.3 15.3 16.3 17.3
beam inter- 10 hr Shutdown - 10.3 10.5 11.3 12.3 13.3 14.3 15.3 16.3 17.3 18.3
rupted 1 hr 106 10.3 11.1 12.1 13.3 14.3 15.3 16.3 17.3 18.3 19.3
0.1 hr 10.3 11.3 12.3 14 15 16 17 18 19 20
(1) Assumes beam strength and presence of target satisfy ignition requirements.
(2) Known applications against which to compare other calculated results.
Table 12-4 Log10 MTBE Values for Radiating Systems(1)
511
Discussion of the Results
Because any calculation of the probability of an explosion requires that

one must make a number of assumptions, estimates, and judgments, it is
good practice to compare the results in some way to similar results for
which there is considerable experience. The entries in this table suggest
that a reasonable estimate of a log MTBE that indicates a safe system
might be 6 to 8, based on the results for an open spark in an unclassified
location. The probability is determined by one's estimate of the interval
between gas clouds in an unclassified location where ignition capable
apparatus is used—100 years, 1,000 years, or 10,000 years?
To help choose a value, consider that nonincendive apparatus and non-

sparking apparatus without shutdown systems have a long history of safe
use in Zone 2, Division 2. The assumptions made in Table 4-4 yield the
same MTBE values for nonsparking apparatus as for an explosionproof
enclosure with flanged joints. The calculated log MTBE values for Zone 2
are 6 to 7.3. Most quantitative definitions of Zone 2 have used a borderline
value for presence of a flammable cloud of either 0.1 or 1 hr/yr. This
points to a log MTBE value of 6 to 6.4.
A third reference value might be the demonstrated safety of flameproof

enclosures in Zone 1. The range of values for log MTBE in Table 12-1 is 6 to
7, an average time between explosions of 10 to 100 years. These values are
understated. The actual MTBE is much longer for enclosures for use out-
doors because this analysis takes no account of the restricted breathing
properties of the enclosure. The impediment to entry of a flammable
vapor into the enclosure is certainly important for dwell-times of the flam-
mable cloud of 1 hr or less.
This author estimates that the difference is at least an order of magnitude

for 1-hr dwell-time and 2 or 3 orders for shorter dwell-times (see Chapter
7).
However, it is the comparison between the MTBE values that is important,

so it is not necessary to be bogged down discussing the absolute accuracy
of the values. Two observations about this table nearly leap out of the
numbers:
(1) Lighting equipment and open-beam devices with no added protec-
tion have much lower MTBE values than explosionproof enclosures.
No one would seriously argue that an MTBE of 100 hr is acceptable in
Zone 1, or even 1,000 to 100,000 hr in Zone 2. An MTBE of 100,000 hr
implies a 0.095 probability of an explosion within a year.
(2) The systems with some kind of interlock have MTBE values equal to
or greater than those of explosionproof enclosures.
(3) A beam carried by a fiber with an MTBF of 106 hr will have the same
MTBE as an explosionproof enclosure in Zone 2.
The calculations summarized in Table 12-1 were based on the beam being
strong enough to cause ignition. They also assumed the presence of a par-
ticle or surface that would provide a source of ignition when irradiated.
Nothing was assumed about the properties of a beam of radiation or its
target that make them a source of ignition.
The first observation is at odds with experience. This author knows of no

history of ignitions in hazardous locations by beams of radiation from
lighting fixtures and illuminating equipment. Open-beam counters in haz-
ardous locations are less numerous, probably used only in Zone 2 in
facilities packaging flammable products. The significance of the numbers
in the table must be that the beams in lighting installations are not strong
enough to cause ignition if particles of the requisite size are present; or,
alternatively, there are no particles of appropriate size.
The first possibility is relatively easy to check by looking at a catalog of

lighting fixtures and doing some arithmetic. The total number of lumens
(1 lumen = 0.00147 W) in the beam for typical incandescent fixtures using
up to 500 W lamps is several thousand, which yields a beam density at the
lens of less than 1 mW/mm2. A few tenths is typical. Beam nonuniformity
might double or triple these values at the center of the lens; thus, illumi-
nating beams are not ignition-capable. Other light sources may deliver
more lumens in the beam, but, in general, have greater areas at the lens, so
the power density is not much different. Illuminating fixtures, therefore,
deliver only a fraction of the power flux, 5 mW/mm2, named by OSCA as
a level at which there is essentially no likelihood of ignition. The conver-
sion efficiency is so low that a 500-W lamp delivers a beam with a total
power of only 5 W. It is reasonable to conclude that open-beam systems
using much smaller sources are also not likely to be a hazard. This is con-
firmed by Proust's work on ignition of dust clouds and vapor clouds by
irradiating a dust cloud.
The presence of a suspended particle in the path of a beam is not quite so

easy to judge. Weinberg et al. reported that coal dust particles 80 μm in
diameter are required to ignite carbon disulfide, and particles 200 μm in
diameter are required to ignite the higher hydrocarbons. Smaller particles
do not contribute sufficient heat of combustion. Particles of other materi-
als, which contribute less heat of combustion than coal, can be assumed to
have a larger minimum diameter for ignition to occur. Most Bureau of
Mines experimentation with dust explosions was carried out with
“through 200 mesh” dust, diameter less than 74 μm. Although larger parti-
cles may help propagate an explosion if they are temporarily suspended in
a cloud, particles larger than those used in the Bureau of Mines investiga-
tions tend to settle out. Fine particles tend to agglomerate and settle out
also. The terminal settling velocity for 70-μm flour dust in air is about
20 cm/s. Particles of 100-μm would settle at twice that velocity. Within
closed enclosures without ventilation or stirring, or in closed raceways,
the presence of dust particles in suspension is, therefore, highly
improbable.
Presence of a dust layer is somewhat more probable. However, if the

apparatus is in a location classified hazardous because of the presence of
dust, installation rules mandate dust-tight enclosures and raceways, so
dust layers are also improbable inside apparatus and raceways.
The MTBE for an explosionproof enclosure containing sparking apparatus

in Zone 0 is not much different from that of the same apparatus in Zone 1
because the MTBE is only slightly affected by the frequency and duration
of the flammable cloud. When the apparatus fails, an ignition will occur
the first time a cloud appears, and it is not really important whether it
appears early in the year or late in the year if the average time between
explosions is 100 years. However, the consensus is that only intrinsically
safe apparatus or apparatus with two types of protection should be used.
The last two families of values in the table are not descriptive of any prac-
tical system, but represent systems with two lines of defense and a
shutdown capability. The MTBE is increased by redundancy, but the prob-
ability of an explosion is reduced by a much larger factor to an
exceedingly low value. In Zone 0 the time delay before shutdown does not
affect the MTBE because ignition will occur at the first presence of a cloud,
and Zone 0 allows that to occur even before a 0.1 hr (or any shorter time
one can name) shutdown delay can expire. For two levels of protection
with identical MTBF, say 106, the combined MTBF is 1.5 times as large.
The probability that one protection will fail in 10,000 hr is 0.01. When there
are two levels in series, the probability of both failing in 10,000 hr is 0.0001,
so redundancy in practice greatly increases the time before the probability
of an explosion equals that of the nonredundant system. The fact that the
MTBE in the table is less than for the system without time delay is attribut-
able to the approximate expressions that Benjaminsen and van Wiechen
derived to ease the burden of calculation.
This discussion has not addressed the question of what power or power
density levels are hazardous for different groups of materials, whether
they be grouped by autogenous ignition temperature or by some other
parameter related to the material's combustion properties. The discussion
has assumed that the beam is ignition-capable. The following conclusions,
drawn from this discussion, show that safety can be assured in many sys-
tems without resolving the question of how ignition parameters are
related to material characteristics.
(1) For any hazard to exist the location of the beam must be classified
because of the potential presence of flammable gases or vapors.
(a) If the location is also classified hazardous because of the potential

presence of clouds or layers of combustible dust, the interior of
raceways and enclosures containing electrical apparatus or wir-
ing will not have either a cloud or a layer.
(b) If the location is not classified hazardous because of the potential

presence of dust clouds or layers, one should assume, in the
absence of any precaution being taken, that a layer of dust could
accumulate. This layer might then serve as a suitable target
should an optical fiber break.
(2) Open beams from illumination or simple counting or position-moni-

toring systems will pose a hazard only if the equipment design is
quite different from conventional designs or the application of con-
ventional equipment is carried out in an unusual fashion. The illumi-
nation levels in these special cases must be more than an order of
magnitude higher than in typical applications. A system to shut down
the beam if it exceeds its design power or intensity affords sufficient
protection.
(3) Breakage of an optical fiber is of sufficiently low probability that no

additional level of protection is needed in Zone 2. In Zone 1 no addi-
tional protection is needed if the fiber is afforded extra protection
against damage, as by running it in raceway. In Zone 0, should such
an application exist, the system should have an interlock to shut
down power if the fiber breaks. The interlock must be designed to act
quickly, taking into account the relationship between the power car-
ried in the fiber, the power density required for ignition, and the igni-
tion delay time. Most people will not trust the probabilities, and in
Zone 0 applications would probably want to control beam strength to
a safe level before the fiber breaks. The availability of the rational
design option should not, however, be removed by careless or
thoughtless standards writing.
(4) The foregoing analysis shows that whether the amount of radiation in
the beam is able to ignite the vapors likely to be present becomes a
concern only if:
(a) the beam is open to an uncontrolled environment and is of much

higher strength than normal illuminating beams, or
(b) the beam is ignition capable and a break in the fiber carrying the
beam will occur in Zone 1 where both a vapor cloud and a suit-
able target are assumed to be present
Standards Actions to Address Hazards of Radiation -

Opinions of the Author
(1) Ignition by radiated energy is not primarily an electrical hazard, and it
should not be addressed in the NEC or CEC. Installation rules for opti-
cal fibers likely to be intermingled with electrical wiring make sense.
Guidance is necessary to protect those who are likely to work with
them from shock hazard and some guidance is required for the use of
optical fiber cables with conductive elements in hazardous locations.
The 1993 NEC rules are sensible for any installation of optical fiber in
any location. But the possible initiation of explosions by radiation
from broken fibers or from analytical devices is not primarily an elec-
trical problem. It is not a problem that will be addressed by installa-
tion rules similar to those of the NEC. It is an issue for some other
committee of NFPA that can deal with complex issues, or for ISA.
(2) As the discussion of MTBE shows, the potential hazard can be dealt
with to a considerable extent by taking reasonable measures in the
design of apparatus. Optical fiber links are routinely monitored in
most process control systems. If transmission is lost another path for
transmission is found immediately. Too much information passes
along the fiber to permit otherwise. Simple approaches can reduce the
hazard to an acceptable level even if the equipment failure is pre-
sumed to provide an ignition source, as in the calculations discussed
above.
Many optical fiber systems, in fact, operate at radiation flux or power

levels below those needed for ignition. Ignition by radiating sources is
a subject for concern, and it needs to be addressed, but precipitous
action is unnecessary.
(3) The use of electrical apparatus incorporating optical systems that do

not exit the apparatus is already fully circumscribed by existing stan-
dards for electrical apparatus in hazardous locations. No additional
safeguards are needed if the optical beam is treated like any other
ignition source in an enclosure.
(4) For larger objects the guidelines for ignition by hot surfaces are con-
servative, but useful, because they already are well known.
A surface less than 10 cm2 in area may reach a temperature above the
specified limit for its marked temperature class if it can be shown that
no ignition will occur with a safety margin of 25 K for T1, T2, or T3; or
50 K for T4, T5, or T6.
To avoid measuring the temperature of small objects specific conser-

vative criteria have been used for many years for apparatus to be
coded T4 or T5 (normal maxima of 135 °C and 100 °C, respectively).
With T4 rating:
Area < 20 mm2: Maximum surface temperature < 275°C (Equiva-

lent to I = 4.6 mW/mm2 at 40°C ambient)
Area > 20 mm2: Power dissipation < 1.3 W at 40°C ambient, 1.2 W
at 60°C ambient, 1.0 W at 80°C ambient
20 mm2 < Area < 10 cm2: Maximum surface temperature < 200°C
(I = 2.3 mW/mm2 at 40°C ambient)
For T5 marking, components less than 10 cm2 area may have a surface
temperature of 150°C.
(5) A Recommended Practice or Code of Practice that is tutorial and sug-

gests engineering approaches to the problem, not rules, is the most
helpful document that could be made available at this time. There are
too many holes in the database to run to rule-writing at this stage. It is
not necessary to write rules at this stage because only unusual appli-
cations in Zone 0 are likely to pose a serious hazard. These need to be
engineered in accordance with risk assessment practices already used
in industries processing flammable materials. Education about igni-
tion by radiating sources is needed, not arbitrary rules. Many poten-
tial hazards of ignition by radiating sources can be dealt with using
existing practices. There is no compelling present danger that pre-
vents taking time to learn.
Need for Further Study and Research

Ignition by radiating sources is a subset of the subject of ignition by small
hot objects. If one considers being able to extrapolate from one set of con-
ditions to another without experiment to be a measure of knowledge, then
not a great deal is known about this subject. In the absence of this kind of
knowledge, a number of “limit” rules or guidelines have been developed,
especially for use in the field of intrinsic safety. “Limit” guidelines specify
criteria, which, if satisfied, permit one to proceed with no further investi-
gation. They are typically ultraconservative and are designed to substitute
an easy dimensional measurement or a simple calculation for the prime
criterion, a time-consuming test.
In general, the same questions need to be answered that the investigations

into ignition by radiating sources have addressed, and only partially
answered.
The primary question from an application standpoint is the relationship of

the composition of the gas or vapor cloud to minimum ignition parame-
ters, particle size, wire size, and so on. It is possible that some elucidation
is possible just by reviewing and analyzing papers on hot-wire ignition,
for example. Emphasis in some past reviews was on determining what
current might be safely permitted. A reassessment of the literature from
the point of view of the temperature required for ignition might be useful.
More experimentation with laser sources, because they are relatively con-
venient ways of delivering energy to a small body, especially with black
bodies to filter out the variations due to spectral characteristics of the tar-
gets, will certainly be necessary. In some industries there may exist
applications of economic importance that justify specific study to establish
practical limits of operation of the system.
Any further experimentation or study must have as its objective a better

understanding of the phenomena involved. Alternatively, any study
should be tailored to be representative of the conditions found in a partic-
ular industry or application so that the results are a practical basis for
establishing recommended safe practices.
References
OSCA Doc. No 94/56.
McGeehin, P., “Optical Techniques in Industrial Measurement: Safety in

Hazardous Environments,” Report EUR 16011 EN, Luxembourg, 1994.
Annex I: Carleton, F. B., and F. J. Weinberg, Summary Report,
Imperial College.
Annex II: Welzel, M. M., Bothe, H., and H. K. Cammenga, Final
Report, PTB.
Annex III: Tortoishell, G., Summary Report, SIRA Test and Certifi-
cation Ltd.
Annex IV: Scott, K.S., University of Leeds.
Annex V: Proust, Ch., Final Report, INERIS.
Hills, P. C., Zhang, D. K., Samson, P. J., and T. F. Wall, “Laser Ignition of
Combustible Gases by Radiative Heating of Small Particles,” Combustion
and Flame 91:399–412 (1992).
Adler, J., Carleton, F. B., and F. J. Weinberg, “Ignition of Flammable Atmo-

spheres by Radiation-Heated Fibrous Agglomerates,” Proc. R. Soc. Lond. A
440: 443–460, (1993).
Olenick, H., Rentsch, H., and W. Wettstein, BBC Explosion Protection Man-
ual, 2nd ed., Brown Boveri & Cie, Mannheim, 1983.
Dubaniewicz, T., Cashdollar, K., Green, G., and G. Cucci, “Ignition Tests
with a Fiber-Optic Powered Instrument,” Proceedings of the 41st Annual ISA
Analysis Division Symposium, Vol. 29, pp. 175–184, ISA, Research Triangle
Park, 1996.
Bossert, J., and R. Hurst, Hazardous Locations, 1st ed., Canadian Standards
Association, Rexdale, 1986.
Magison, E. C., “Are Optical Beams an Explosion Hazard?,” In Tech, Sept.

1997, pp. 61-66.
CHAPTER 13
DUST HAZARDS
Nature of the Dust Hazard

The hazard caused by the presence of combustible dusts is more compli-
cated than hazards caused by flammable gases and vapors. Although the
possible mechanisms of ignition are the same for both kinds of flammable
material, their relative importance is different for dusts than for gases and
vapors. However, despite the more complicated nature of the combustion
properties of dusts, the philosophy of protection against dust explosions
initiated by electrical apparatus is, in this author's opinion, much simpler
in principle, and more straightforward in practice, than the philosophy of
protection against electrical equipment initiating gas or vapor explosions.
When considering ignition of gases and vapors by electrical apparatus, it

is easy to simplify the discussion to two subjects, arc ignition at low volt-
age and ignition by hot surfaces. In most applications arc ignition need be
considered only at low voltages. Static electricity is a hazard in a few
applications. The relevant literature is relatively easy to identify, and
safety standards are available for most hazards related to electrical equip-
ment. Harmonization has progressed worldwide. The subject of dust
hazards is more wide ranging. Arc ignition is a relatively unimportant
issue. Ignition by high voltage sources, particularly by static electricity, is
an important concern in industries that process dusts or transport dusts on
conveyors or in pneumatic systems. The ignition temperature of layers of
dusts affect the hazard of electrical apparatus, but is also important to
those who store dust in bulk.
The parameters of dust explosions, such as pressure rise and rate of pres-
sure rise, are not very pertinent to reducing the hazard of ignition by
electrical apparatus. They are, on the other hand, of priority interest to
those designing facilities for handling dusts, especially in the design of
venting systems that reduce the loss of property and life when a dust
explosion occurs.
Standards for protection against dust explosions are now at about the
same point standards for explosionproof apparatus were three decades
ago. An American practice and a European practice are published
together in an “international” standard. One might hope that this victory
of history over need will not persist. Agreement on test methods is pro-
gressing, so agreement on equipment design and selection of apparatus
522 Dust Hazards
may be forthcoming in the future, driven by international market

demands.
This wide spectrum of concerns about dust hazards has produced an

extensive literature on dust explosions. The reader interested primarily in
the hazards due to electrical apparatus must sift carefully to find the infor-
mation of most importance to his problem.
The development of an explosion in a dust cloud differs in important

details from the development of an explosion in a gas or vapor, but the
major concepts presented in Chapter 2 apply generally to both types of
explosions.
Like gas or vapor mixtures, dust suspensions have a minimum ignition

energy and a lower explosive limit. In practice the concept of an upper
explosive limit for a dust cloud is not useful, though the UEL for coal has
been measured in the laboratory. Pressures developed in dust explosions
are of the same order of magnitude as those developed in gas or vapor
explosions.
The salient differences between dust explosions and gas or vapor explo-
sions arise from the physical differences between the two systems. A dust
cloud is a nonhomogeneous suspension of solid particles in a suspending
medium, usually air. Ignition characteristics and the characteristics of a
dust explosion depend on the chemical composition of the dust; the shape,
size, and concentration of the dust particles; and the chemical composition
of the suspending medium.
The following paragraphs summarize and illustrate the most important

features of dust hazards. The reader who is seriously concerned with dust
hazards should consult the reference documents. The Bureau of Mines lit-
erature, in particular, contains a wealth of data too voluminous to
reproduce in this chapter. These documents are a ready reference library
of information on dusts.
The publications of the National Materials Advisory Board of the National

Academy of Sciences also provide insight into a range of subjects related
to classification of dusts and prevention of dust explosions. Among the
references listed at the end of this chapter are the proceedings of several
symposia on dust hazards. Such symposia typically include a number of
papers that summarize limited aspects of the subject. A reader interested
in pursuing a particular subject may find these references to be a conve-
nient way to take his knowledge to the next level.
Dust Hazards 523
Influence of Chemical Composition of the Dust

Chemical composition is the prime determinant of minimum ignition
energy, minimum explosive concentration, pressure developed by the
explosion, rate of pressure rise, and ignition temperature. However,
though the range of values for these characteristics within a group of dusts
is wide, there are not substantial differences between groups. Whether one
is considering metallic dusts, carbonaceous dusts, plastic or chemical
dusts or agricultural dusts, the values in Table 13-1 are typical of those
measured by the Bureau of Mines. Pressures and pressure rises are those
measured in the Hartmann apparatus, a 1.2 L chamber, which has been
used in many laboratories around the world. Modifications to the appara-
tus have been made in recent years to improve the reproducibility of data
among laboratories. Rate of pressure rise is a crucial parameter in the
design of venting systems, direct explosion quenching systems, and explo-
sion chamber isolating systems. If the flame width of a combustible
mixture is small relative to the size of the chamber in which the explosion
takes place, and one assumes that the flame velocity is a constant, the rate
of pressure development obeys the “cubic law” that
(dP/dt)max V1/3 = K
For gases, the constant K is KG, and for dusts, KSt. St is the abbreviation for
Staub, German for dust. The units normally quoted are bar-m/s. The pres-
sure rise depends only on the concentration of the combustible material,
not the volume of the vessel. Therefore the rate of pressure rise is lower in
a larger chamber, such as the 20-L chamber now being proposed as an IEC
standard, and the 1-m3 chamber used by Bartknecht and his colleagues at
Ciba-Geigy. Both the rate of pressure rise and pressure depend on the
ignition source used. At Ciba-Geigy, a 10 J igniter gave lower pressure and
lower values of K for both gases and dusts, compared to those measured
when ignited by capacitive discharges just above the minimum ignition
energy or by a 10,000 J pyrotechnic igniter.
Cloud ignition energy 5 mJ and higher

Minimum explosive concentration 0.015 oz/ft3 and higher
Maximum pressure developed 30–150 psi
Rate of pressure rise less than 15,000 psi/s
Ignition temperature—cloud 200°C and higher
Ignition temperature—layer 150°C and higher
Table 13-1 Typical Dust Parameters

524 Dust Hazards
This table points up the most significant difference between Class I gas or
vapor hazards and Class II dust hazards. Though other characteristics are
similar, the ignition energy of even the more easily ignited industrial dusts
is 20 times greater than that of typical Class 1, Group D, materials. Only a
few dusts such as zirconium and thorium hydride ignite at energies below
10 mJ. These dusts ignite spontaneously at room temperature under some
conditions.
The ignition energy given in Table 13-1 refers to values such as those
reported for dust clouds by investigators at the Bureau of Mines. These
determinations were made by discharging a capacitor into the primary of
a step-up transformer. The arc occurred between electrodes in the second-
ary circuit. The stored energy on the capacitor required for ignition was
reported as minimum ignition energy. An undetermined amount of
energy was undoubtedly lost in the transformer, so the reported energies
are somewhat higher than the actual minimum ignition energy.
Investigators at Ciba-Geigy found that of 38 dusts tested, of all families,

the minimum ignition energy by capacitor discharge was approximately
equally distributed in the four ranges, 1 to 10, 10 to 100, 100 to 1,000, and
>1,000 mJ. The pressure rise in the 1m3 chamber were in the same range as
those in the table above, but tended to be higher, especially for metallic
dusts. Their data led them to the position that data from work in chamber
volumes less than 20 L do not correlate well with larger scale testing.
The data from Bureau of Mines testing was used to develop two compara-
tive measures about a dust, its ignition sensitivity, and its explosion
severity. The reference material for these comparisons was Pittsburgh coal
dust.
{ ( Cloud ignition temp. ) ( Min. ignition energy ) ( LEL ) }
Ignition sensitivity = --------------------------------------------------------------------------------------------------------------------------------------1
{ ( Cloud ignition temp. ) ( Min. ignition energy ) ( LEL ) } 2
{ ( Maximum pressure rise ) ( Maximum rate of pressure rise ) }
Explosion severity = ----------------------------------------------------------------------------------------------------------------------------------------------------1-
{ ( Maximum pressure rise ) ( Maximum rate of pressure rise ) } 2
Subscripts 1 and 2 refer to the material tested and the reference Pittsburgh
coal, respectively.
Those using KSt values classify dusts by dust explosion class.

Dust Explosion Class KSt
St 0 0
St 1 > 0 - 200
St 2 > 200-300
St 3 > 300
Dust Hazards 525
Of 600 dusts tested, the majority fall into ST 1, according to Bartknecht.

(This author estimates that the percentage is about 75% in ST 1, 16% in ST
2, and the remainder in ST 3.)
Less is known about the minimum ignition energy of dust clouds than of
gases and vapors. Eckhoff discussed the work of earlier investigators who
found that the stored energy, ½ CV2, required to ignite dust clouds
decreased by a factor of 10 to 20 when the capacitor was discharged
through a resistance of 104 ohms to 105 ohms. Since much of the energy
was dissipated in the resistor, actual spark energy to cause ignition was
only on the order of 1 mJ. From the data, one can calculate initial voltage
of about 5 kV to 6 kV on the capacitor. The data, therefore, are interest-
ingly similar to those presented by Riddlestone for methane-air (see
Chapter 9).
Eckhoff presents evidence to support the hypothesis that very short capac-
itor discharge times, about 1 μs, cause pressure waves around the arc that
may force dust particles away from the arc. Slower discharges, on the
order of a few tenths to several milliseconds, correlate with a reduced
stored energy for ignition and do not disrupt the dust cloud.
Eckhoff also measured spark-ignition energies much lower than those

reported by the Bureau of Mines. He used a capacitive discharge circuit in
which an auxiliary capacitor provided a short trigger pulse to break down
an air gap, which then discharged the capacitor storing the “ignition
energy.”
Eckhoff's work does not necessarily mean that Bureau of Mines data are
not reasonable measures of ignition energy in practical cases. Unless one
can imagine a circuit charged to 5 kV to 6 kV, which discharges through
104 ohms to 105 ohms into a dust cloud, they probably are a measure of
ignition energy in practical situations, even if they are not minimum val-
ues. In the author's opinion Eckhoff's work is primarily of interest because
it helps elucidate dust-ignition phenomena.
The ignition energy of dust layers was measured at the Bureau of Mines
by discharging a capacitor charged to 400 V into a 1/16-in.-thick layer of
dust on a steel platen. The platen was connected to the negative terminal
of the capacitor. The positive terminal was connected to a pointed elec-
trode that was manually lowered until a spark discharge occurred.
Capacitor size was changed to vary ignition energy.
Relatively little layer ignition data were reported for materials other than
metals. A few data were reported for explosives. No consistent relation-
ship between cloud and layer ignition energy was apparent. In many cases
the layer ignition energy reported was considerably higher than the cloud
ignition energy, but for numerous samples the layer ignition energy was
526 Dust Hazards
lower than the cloud ignition energy by a factor of 10 or more. One can
speculate, based on Eckhoff's work, that, in a layer, the effect of the pres-
sure wave might be negligible compared to the effect in a cloud.
Nevertheless, most of the layer ignition energies reported were substan-
tially above those for gases and vapors. The exceptions were for samples
of finely ground metals. Metals are different from other dusts in that it is
generally accepted that if one can produce fine enough dust, almost any
metal will be pyrophoric, and may ignite spontaneously due to oxidation.
For example, in RI 6516 a number of dusts are reported to ignite spontane-
ously in quantities greater than 1 g.
The layer ignition temperature of dusts depends on the thickness of the

dust layer. The author took the data in Table 13-2 from curves in the BBC
Explosion Protection Manual. The figures in parentheses are values read
from the curves of permissible surface temperature. In another table in
that book the authors indicate the T code for each dust listed. All are T4,
T3, or T2. In his discussion of dust explosion-related standards Greiner
shows a figure giving the maximum permissible surface temperatures,
depending on the ignition temperature of a 5-mm layer. His figure shows
the permissible apparatus surface temperature to be 75 K below the layer
ignition temperature of a 5-mm layer if the layer on the apparatus is 5 mm
thick or less. For thicker layers the permissible surface temperature
decreases, reaching about 40% of the 5-mm value at 50 mm. Layer ignition
temperature is determined using a 5-mm layer on a hot plate.
Layer Ignition Temperature (°C)

thickness-mm Cork Low volatile coal High volatile coal
5 325(267) 285(220) 230(160)
10 300(200) 263(170) 210(130)
20 270(170) 240(135) 190(100)
30 250(135) 225(110) 180(75)
40 230(115) 210(90) 175(60)
50 213(100) — 170(40)
Table 13-2 Dependence of Layer Ignition Temperature on Layer Thickness
Consideration of chemical composition provides some guides for subdi-

viding broad classes of dusts. Among carbonaceous dusts ignition hazard
is closely related to volatile content. If volatile content is below 8%, there
may be a fire hazard, but there is essentially no explosion hazard. Most
carbons, charcoals, and cokes are below 8% volatile content. Lignite, pitch,
soft coals, and asphaltic materials typically are 30% to 40% volatile and
present severe dust explosion hazards. Exceptions to the general rule are
Dust Hazards 527
modified materials, such as chromate-treated charcoals and low-mois-

ture-content palladium activated carbons. The United States Bureau of
Mines tests showed that these samples, though low in volatile content,
could be ignited as a cloud in a furnace; untreated samples could not.
Studies of the combustion of dust particles have shown that burning of the
volatile components occurs first. The solid fraction then burns.
Structural similarity as a basis for estimating explosion hazard, so useful

with vapors, is applicable also to dyes, drugs, plastics, and pesticides.
Bureau of Mines Reports RI 5971 and RI 7132 discuss, at some length, the
influence of structure on hazard. Solid materials in these categories pos-
sess much more complex chemical structures than the vapors which have
been classified. For this reason, although some broad guides have been
formulated, the effects of even simple substitutions and modifications of
structure cannot yet be succinctly stated.
The ignition figures quoted above are typical of dry (less than 5% mois-
ture) materials without appreciable inert material. Addition of moisture or
inert material provides heat-absorbing mass without increasing the
amount of energy released. (Many metal dusts react with water, and mois-
ture may, on the contrary, increase explosion severity.) Moderate amounts
of moisture or inert material do not markedly change ignition energy,
maximum pressure, or rate of pressure rise, but Bureau of Mines tests
showed that the effects increase rapidly as the amount of moisture or inert
material approaches the value that quenches the explosion. Typically, the
moisture level must be 15% to 50% of the dust concentration to prevent
ignition by a strong inductive spark. The percentage of inert material, such
as fuller's earth, required to prevent spark ignition is frequently as high as
90%. In general, the efficacy of inerting material is related to its heat capac-
ity. However, the alkali salts are much more effective than their relative
heat capacity would suggest. Minimum concentration for ignition
increases linearly with either moisture or dry inert addition.
Influence of Shape, Size, and Concentration

Shape, size, and concentration of the suspended dust has a significant
effect on ease of ignition and the properties of the explosion. Irregularly
shaped particles produced by milling and grinding operations have a high
area-to-volume ratio. They are more easily ignited and represent a more
severe explosion hazard than spherical particles produced by operations
like spray drying.
Experimental data verify the expected relationship between particle size

and explosion hazard. The finer the dust, the more homogeneous the
cloud, and the greater the surface area for reaction.
528 Dust Hazards
The relationship between particle diameter and the important hazard

parameters is summarized below. The conclusions are for a range of parti-
cle sizes of 0.001 in. to 0.012 in. (24 μm to 300 μm).
In some cases the values of the parameters increase sharply for particle
sizes above 0.003-in.-diameter.
Ignition temperature Flat or somewhat increasing
Maximum pressure in closed chamber Slight increase
Minimum concentration (LEL) Little dependence below 0.003 in.
Increases with size above 0.003 in.,
sometimes strongly
Minimum ignition energy Increases with larger diameter
Rate of pressure rise Decreases with increasing particle
size
Particle sizes larger than those for which conclusions are drawn above are
not important in determining explosion hazard. It has been reported that
0.015-in. cornstarch or fuller's earth particles are the largest that affect the
development of an explosion and that the average cornstarch particle size
of 0.028 in. is the largest that will ignite in air.
Metal dusts in particular may react quite differently if particle size is

smaller than the above range. Very small particles, approximately 40 × 10−
6
in. (1 μm) in diameter, will ignite solely because of exothermic reaction
with oxygen, sometimes even in a layer.
Schwab and Othmer confirmed the influence of particle size reported by

the Bureau of Mines and further emphasized that rate of pressure rise is
determined by the amount of oxygen present. There is, therefore, an opti-
mum surface area for the dust particles in a given chamber volume. As
particle size increases, the concentration in weight per unit volume must
thus increase. They also show that in a vented chamber the inverse depen-
dence of rate of pressure rise on particle diameter causes the maximum
pressure attained to increase as particle size is decreased. The effect is, of
course, dependent on the amount of venting.
In Chapters 2 and 9 it was shown that for spark ignition the ignition
energy of gases and vapors changed markedly with concentration, and
there exists a well-defined, most easily ignited concentration. This is not
true of dusts. Ignition energy drops from a relatively high value at the LEL
to the minimum value at several times the LEL. Further increase in con-
centration does not materially change the energy requirement. Cloud
ignition temperature also drops to a low value and remains essentially
constant as concentration increases.
Dust Hazards 529
The maximum rate of rise in pressure and the maximum pressure reached
during an explosion occur at different concentrations, but both occur at
concentrations well above the LEL. Excess dust is not an efficient inerting
material.
Another distinctive feature of dust hazards is that there is no well-defined

UEL. For practical purposes it does not have significance. The LEL for coal
dust is about 0.05 g/l. The UEL of a cloud appears to lie in the region of
2-5 g/l. At these high concentrations flame does not propagate through
the cloud.
Chemical Composition of the Suspending Medium

The effect of reducing the oxygen content is similar to the effect of adding
moisture or dry inert material. The explosion characteristics change slowly
at first, and then very rapidly as the oxygen concentration approaches the
limiting value for the particular dust composition and geometry.
The diluent used to reduce oxygen content is effective in proportion to its

molar heat capacity. Carbon dioxide is a more effective diluent than nitro-
gen. At high temperatures water vapor is as effective as carbon dioxide.
Argon and helium are preferred diluents for metal dusts. Many metals
react with carbon dioxide or nitrogen. The hydrides of thorium, uranium,
and zirconium, however, are preferably inerted with carbon dioxide.
Table 13-3 lists properties of some typical dust samples. The references
listed at the end of this chapter include a wealth of other data; the table is
merely illustrative of these. Sample-to-sample differences are large, even
for the same material. It is important to note that the cloud ignition tem-
perature determined by blowing a dust cloud against hot furnace walls is
frequently several hundred degrees higher than the layer ignition temper-
ature. The latter temperature is determined by raising the temperature of a
dust layer in an oven or on a hot plate. It is probably time-dependent for
many materials. Dehydration and changes in composition occurring over
long exposure to elevated temperatures below the reported layer ignition
temperature may result in ignition at a lower temperature than that
reported.
530 Dust Hazards
Ignition temp. (°C) Cloud ignition Maximum explosion Max rate

Type of dust LEL oz/ft3
Cloud Layer energy, (mJ) pressure (psi) (psi/s)
Metal Powders (RI6516)

Aluminum, atomized 650 760 50 0.045 73 20,000
Iron, carbonyl 320 310 20 0.105 41 2,400
Magnesium, atomized 620 490 40 0.040 90 9,000
Manganese 460 240 305 0.125 48 2,800
Tin 630 430 80 0.190 37 1,300
Titanium 330 510 25 0.045 70 5,500
Uranium Hydride 20 20 5 max. 0.060 43 6,500
Vanadium 500 490 60 0.220 48 600
Zinc 680 460 960 0.500 48 1,800
Dow metal 430 480 80 0.020 86 10,000
Zirconium 20 190 15 max. 0.045 55 6,500
Plastics (RI5971)
Acrylamide Polymer 410 240 30 0.040 85 2,500
Casein 520 — 60 0.045 73 1,000
Cellulose Acetates 470 400 25 0.045 135 6,000
Methyl methacrylate 440 — 15 0.020 101 1,800
Phenolic resin 580 — 10 0.025 89 8,500
Polyethylene resin 450 — 10 0.025 83 2,500
Polysterene 490 — 120 0.020 62 1,500
Synthetic rubber, hard 320 — 30 0.030 93 3,100
Urea molding 460 — 80 0.085 89 3,600
compound
Vinyl butyral resin 390 — 10 0.020 84 2,000
Agricultural products (RI5753)
Alfalfa 470 220 370 0.160 76 800
Cocoa 510 200 100 0.045 68 1,400
Coffee 720 270 160 0.085 38 150
Cornstarch 380 330 40 0.055 58 2,200
Cotton linters 520 — 1,920 0.500 44 400
Egg Whites 610 — 640 0.140 58 500
Garlic 360 — 240 0.100 57 1,300
Milk, skim 490 200 50 0.050 95 2,300
Nut shells, pecan 440 210 50 0.030 112 3,500
Pectin 410 290 35 0.075 132 7,000
Rice 440 240 50 0.050 105 2,700
Sugar 370 400 30 0.045 109 5,000
Wheat Flour 440 440 60 0.050 97 2,800
Table 13-3 Explosion Characteristics of Typical Dusts

Dust Hazards 531
In Chapter 2 it was pointed out that reported ignition energy data for
gases and vapors are conservatively biased because conditions in the labo-
ratory are contrived to be more favorable to ignition than conditions to be
expected in the field. Dust-ignition data are also conservative for the same
reason. Experimenters use well-maintained electrode systems whose
geometry is highly favorable to ignition, and they use the most easily
ignited concentration. In addition, the effects of particle size and moisture
content also are weighted in favor of ignition. In Bureau of Mines work,
particle size was 200 mesh or finer (less than 74 μm) in almost all cases,
and moisture content was reduced, with few exceptions, below 5% before
testing. As noted earlier, minimum ignition energy increases with particle
diameter. In a real cloud it is unlikely that the dust particles would be as
fine as through 200 mesh. Although larger particles will settle out, tending
to leave only fine particles in suspension, the fine particles tend to agglom-
erate. Reduction of moisture content of course reduces the energy
required to evaporate the moisture during ignition.
Distinguishing Features of Dust Hazards

Though the small-scale ignition processes in dust explosions do not differ
greatly from those of gas and vapor explosions, the gross nature of dust
hazards is sufficiently different from gas and vapor hazards to require dif-
ferent techniques of hazard reduction. Dust hazards are cumulative; gas or
vapor hazards are dispersive. If process equipment releases a combustible
vapor or gas the hazard is usually only transient. Even heavier-than-air
vapors trapped in low spots are eventually dispersed or diffused. Should a
second release of combustible material occur the effects are usually inde-
pendent of the first release. Dust hazards, on the other hand, are
conservative. Dust settles out of the air and is stored for future combustion
on beams, windowsills, and equipment. Fortunately, the dust hazard is vis-
ible and obvious from the moment the hazard first exists. The LEL of
wheat flour, 0.05 oz/ft3, is equivalent to about one-half teaspoon per cubic
foot, not an unnoticeable concentration by any means. Before the concen-
tration reaches the LEL, visibility is reduced to a few feet and breathing is
difficult.
Although air currents can carry a gas or vapor hazard to a great distance
from the source of combustible material, primary dust hazards exist
mostly at processing machinery where air currents keep dust continu-
ously in suspension, or the processing equipment continuously feeds
more dust into the air to replace that which has settled. A true Division 1/
Zone 21, dust area is in most cases quite limited. However, because of the
conservative nature of dust hazards, a very real hazard may exist in a
large area surrounding a Division 1 area. The extent of this Division 2/
Zone 22, area depends on the following factors: frequency of housekeep-
ing, height at which dust is released, velocity and direction of release,
532 Dust Hazards
particle size and density, velocity and direction of drafts, and presence of
confining walls or barriers. In some plants the extent of the Division 2 area
is considered to be about 10 ft from a point of open handling of dusts,
assuming that there is nothing unusual about any of the factors listed
above.
When a cloud of gas or vapor ignites, the entire supply of combustible

material burns in a single explosion. The combustion wave does not travel
farther than the extent of the combustible gas or vapor. On the other hand,
if a small cloud of dust ignites, the resulting puff, harmless in itself and
causing no damage, may dislodge additional dust from nearby equipment.
This new and larger cloud ignites, causes a larger blast, shakes more dust
into the atmosphere, and the resulting series of explosions may demolish a
structure, even though the initiating explosion was harmless by itself. The
only certain means of stopping multiple explosions of this sort is a pro-
gram of equipment maintenance and good housekeeping practice in
buildings whose design discourages the accumulation of dust layers.
Kauffman's work suggests that a layer thinner than the minimum thick-
ness to support burning may propagate an explosion when layered dust is
entrained by the moving wave front.
Schwab and Othmer note that multiple explosions of a different variety

can occur and that these are the reason why industrial dust explosion
damage is often puzzling; windows, doors, and walls collapse inwardly.
They state that after the first puff has shaken down a large dust cloud that
ignites, the resulting explosion consumes all the oxygen, but not all the
dust. After this explosion air rushes into the building, causing inward col-
lapse of walls, and the remaining unburned dust ignites to cause another
explosion. This process may repeat.
The conservative nature of the dust hazard also leads to another problem
not found in Class I locations, the accumulation of thermally insulating
blankets of dust on equipment. Surface temperatures of equipment must
be below the layer ignition temperature of the dust, even after a dust layer
accumulates and interferes with conduction of heat away from the
equipment.
In industries that process large amounts of finely ground materials, explo-

sions that damage equipment and plant are frequent. They may do limited
damage, and not be as newsworthy as an explosion in a large export grain
elevator, but even a small initial explosion demands measures to minimize
the damage and prevent spread to nearby facilities. One method is to vent
the explosion to a safe area, as is done with gas and vapor explosions in
process vessels. This demands considerable forethought. One of the distin-
guishing feature of a vented dust explosion is that the flame extends a
much greater distance from the vent than a gas explosion of the same
Dust Hazards 533
magnitude. The cause is the slow burning of particles carried out of the
vessel in the vented jet.
Area Classification
Division 1
Figure 13-1, adopted from T. W. Moodie's paper, illustrates the factors that
must be considered when classifying a Class II location, as well as the cri-
teria for deciding whether the area is Division 1, Division 2, or
nonhazardous. The diagram shows that a location must be classified Divi-
sion 1 if low-resistivity dusts are present; if a combustible cloud
concentration is present frequently, or periodically; or if process malfunc-
tion both releases a combustible dust cloud and causes electrical failure
that provides a source of ignition.
The 1996 NEC presently classifies Group E (metallic) dusts as electrically

conductive, and locations in which these dusts are present must be classi-
fied Division 1. A presumed rationale for Division 1 classification is that
conductive dust that infiltrates equipment may bridge between two live
parts, providing a conductive path to initiate an arc. The hazardous mate-
rial, therefore, perhaps can provide its own mechanism for ignition.
Moodie presented data (see Figure 13-2) that show that Group F dusts
tested by him are not conductive, even when wet with highly conductive
coal mine drainage water. The samples did not break down under pro-
longed voltage stress of 6,000 V/cm, nor did vibration affect resistivity
significantly. Moodie suggested that a conductive dust is one whose resis-
tivity is less than 100 ohm-cm, or one that breaks down under prolonged
application of a voltage stress of 1,000 V/cm.
In the author's opinion there is substantial question whether the severity

of metallic dust hazards is as real as is assumed by the mandate to classify
as Division 1 those locations where metallic dusts may be present (see
Chapter 3). It is generally accepted that metallic dust particles exposed to
air oxidize rapidly. Therefore, many metallic dust layers will not have low
resistance. It seems to the author that the ability of a metallic dust to pro-
vide a path for arcing is likely to be voltage-dependent. Perhaps the
problem exists at powerline voltage, but not at 10 V, for example.
534 Dust Hazards
Figure 13-1 Area Classification of Class II Locations

Dust Hazards 535
Figure 13-2 Resistivity of Typical Dusts

536 Dust Hazards
The minimum explosive concentration of many dusts is tabulated in

Bureau of Mines reports. A rule, often presented jocularly, but based on
fact, is that, if you can see your outstretched hand in front of your face, the
dust cloud is not above the LEL. Moodie reported plant measurements to
determine the actual dust cloud concentration at points of presumed
severe hazard. He measured gravimetrically to get a 3 to 5 min average
value and photodensitometrically to determine short-term concentration
variations. The gravimetric measuring equipment pulls an air sample
through a filter disk for a preset sampling period. The weight gain of the
filter disk and air flow through it yield the average cloud concentration.
The photodensitometer measures the decrease in light transmitted by a fil-

ter tape that has had dust drawn through it. This instrument provides a
measure of short-term variation. The gravimetric measurement provides a
more accurate average measurement. The ratio of peak to average reading
of the densitometer, when multiplied by the average concentration deter-
mined gravimetrically, yields a measure of the peak cloud concentration.
Because tens of densitometer readings can be taken during the period a
weight sample is taken, standard statistical techniques can be used to esti-
mate the probability that the cloud concentration instantaneously
exceeded a predetermined value, the LEL, for example.
Measurements reported by Moodie showed that even in locations consid-

ered in the milling industry to be especially hazardous, the cloud
concentration did not exceed 7% of the LEL concentration and the proba-
bility that the LEL would be exceeded in a 1-s interval was 10−12. One
cannot generalize from Moodie's data about all dusty locations, but the
value of measuring to determine the real hazard rather than guessing is
well illustrated in Moodie's reports.
A dust cloud with concentration higher than the LEL is likely to exist fre-
quently only in locations where energy has been added to the dust.
Typical locations are at pouring spouts, near conveyors, or near milling
machinery. In modern plant design the objective is to keep the dust inside
processing machinery.
Any location where process malfunction releases an ignitable concentra-

tion of dust and simultaneously causes electrical equipment to spark or
overheat is Division 1. Moodie cites the case of a bag dust collector. When
the filter bag ruptures and releases dust the fan motor may overheat. Bag
rupturing is a common enough occurrence that it is not considered
unusual, so only the electrical overload protection on the motor prevents
overheating. Such a facility is, therefore, Division 1, unless secondary pro-
tection is provided. A differential pressure sensor that deenergizes the fan
motor when the filter bag breaks provides a second defense against explo-
sion, and the location can then be classified as Division 2.
Dust Hazards 537
Division 2
If consideration of resistivity, cloud concentration, and process malfunc-

tions do not warrant a Division 1 classification, the data about the location
must be examined relative to Division 2 classification.
Division 2 classification is assigned if the dust layer thickness exceeds the

minimum depth required to propagate a fire, or if a combustible dust
cloud may be present infrequently. (The NEC specifies a Division 2 loca-
tion also to be where dust accumulation may interfere with safe heat
dissipation of electrical apparatus, but this requirement is implicit in the
specification of minimum depth to support flame.)
The practice in some companies is to classify any location with a layer

deeper than 1/8 in. as Division 1. OSHA has specified 1/8 in., as the action
criterion for initiating housecleaning. If the depth of a dust layer exceeds
1/8 in., no matter how much the total amount of dust present, clean-up
shall be initiated.
Moodie has shown that 0.044-in. (1.1-mm) layer of grain elevator dust will
not continue to burn after the ignition source is removed, but a 0.066-in.
(1.7-mm) layer will propagate flame. The critical layer thickness of other
dusts is unknown and cannot be inferred from these data, but experimen-
tally determining the critical layer thickness, is a powerful criterion for
deciding what areas should be classified Division 2 or for effecting a
house-cleaning program to ensure that the thickness required for a Divi-
sion 2 classification is never reached.
A layer 0.001 in. (0.025-mm) thick just obscures the color of a painted sur-
face. A 0.004-in. layer yields about 1 quart of sweepings from a 100 ft2 area.
In his book, Palmer discusses the minimum layer thickness to sustain

smoldering. The required layer thickness depends approximately linearly
on particle size. He notes minimum depths of 2 mm for fine grass dust
(through 240 mesh) and 47 mm for cork dust of 1.8-mm mean particle
diameter. However, air flow rate over the layer is also important. For
beech sawdust of 0.5-mm mean particle diameter, the minimum layer
thickness to support smoldering dropped from 13 mm in still air to 1.6
mm with air flow of 2.2 m/s. The data suggest to this writer that any layer
deeper than 1 mm should be considered a fire hazard unless there is posi-
tive evidence that the layer cannot support continuous combustion under
the conditions in which it exists.
If process equipment fails infrequently and releases a cloud of dust above

the LEL, the location is Division 2. Where dust is transported in closed sys-
tems, wear may occasionally cause a break in a duct, releasing a heavy
538 Dust Hazards
dust cloud. If such occurrences are rare, perhaps once per year, the area
may be classified Division 2.
If the facts about a location do not justify Division 1 or Division 2 classifi-

cation, the area is nonhazardous.
Electrical Apparatus For Dusty Locations

Approach to Reducing Hazard
Design of electrical apparatus (other than intrinsically safe systems) for

use in Class II locations is based on two objectives:
(1) Keep the dust away from ignition sources.
(2) Prevent ignition of dust that accumulates on the apparatus.
The methods most commonly used are the following:

• dust-ignition-proof enclosures (Division 1)
• dust-tight enclosures (Division 2)
• pressurization (Division 1 and 2)
• sealing (Division 2)
• intrinsic safety (Division 1)
Dust-Ignition-Proof Enclosures
Dust-ignition-proof (DIP) enclosures are recognized by the NEC as suit-

able for use in Division 1 locations for all kinds of electrical equipment.
The principle features of a dust-ignition-proof enclosure are tight joints to
prevent entry of dust and escape of hot particles from arcing contacts,
robust construction for protection against mechanical damage and to min-
imize hot spots in the event of electrical fault to the enclosure, and external
surface temperature low enough to prevent ignition of external dust lay-
ers. UL and CSA requirements are similar, but not identical. Enclosures
must be of substantial construction; customarily they are metallic.
Metal-to-metal joints at least 3/16 in. wide are preferred. Spacing between
mating surfaces must not exceed 0.0015 in., with additional allowance of
0.001 in. per 1/8 in. added width to a maximum of 0.008 in. clearance. A
gasket is permitted (by CSA) if it is mechanically attached to protect it
from abuse. UL requires 3/8-in. minimum gasket width. CSA permits
minimum width of 1/8 in. if the maximum gasketed dimension is 12 in.,
increasing to 3/8 in. for gasketted dimension over 36 in. Gaskets of cork,
natural rubber, or other materials likely to age or be easily damaged are
Dust Hazards 539
not satisfactory. UL prohibits glued gaskets. CSA will accept adhesive

attached gaskets subject to special investigation to ensure long life.
Dust-ignition-proof enclosures are tested in a chamber full of swirling

dust, with intermittent loading of the equipment inside the enclosure so
that cycling temperature will increase the likelihood that the enclosure
will breathe in dust. Equipment is examined to determine whether dust
has entered. There must not be sufficient dust present to suggest that after
several years' operation in a dusty atmosphere enough material could
accumulate to interfere with operation or create a fire or explosion hazard.
There must, of course, be no ignition. There shall be no charring of a blan-
ket of dust on the surface of the enclosure.
Surface temperature of a DIP enclosure is measured when the apparatus is

blanketed with as much dust as will be retained on the surface. The limits
are given in Table 13-4. These were formerly the maximum surface tem-
peratures for all enclosures in Class II locations.
Equipment Subject to Overload

Equipment Not
Group (motors, transformers, etc.)
Subject to Overload
Normal Operation Abnormal Operation
E 200 200 200
F 200 150 200
G 165 120 165
Table 13-4 Maximum Surface Temperature (°C)
In the 1996 NEC, the temperature limit is stated to be the ignition temper-
ature of the specific dust to be encountered. If the dust may dehydrate or
carbonize, the temperature shall not exceed the lower of the ignition tem-
perature or 165°C.
Dust-Tight Enclosures
Dust-tight enclosures are suitable for use in Class II, Division 2, locations
and in Class III, Division 1 and 2, locations. The prime requirement is tight
construction to exclude dust and contain arcs and sparks.
If a dust-tight enclosure is listed by a third-party agency, it is tested like a

dust-ignition-proof enclosure, except that only one cycle of dust exposure
is necessary.
540 Dust Hazards
Pressurization
Purging to reduce hazard in Class II locations is recognized in the NEC,

which refers to NFPA 496 for design and installation details (see Chapter
6). Requirements for Class II locations are, in principle, similar to those for
Class I locations.
Internal pressure must not be less than 0.1 in. of water (25 Pa) if the spe-
cific particle density of the dust is 130 lb/ft3 (2.1 × 103 kg/m3) or less. If the
particle density is greater than 130 lb/ft3, the internal pressure must be
maintained at 0.5 in. of water (125 Pa) or greater.
Electrical interlocks on doors are not required if a key or tool is needed for
entry. Warnings against opening are required unless the area is known to
be nonhazardous.
Alarm or indication of pressurization loss must be provided. Either audi-

ble or visual devices are acceptable.
Motors, transformers, and other apparatus subject to overload shall be

provided with automatic means to deenergize them if temperature
exceeds the design limits.
The 1996 NEC does not mention sealing with regard to Class II hazards,
but does permit the use of dust-tight enclosures to enclose arcing contacts
inside a general-purpose enclosure, in both Division 1 and Division 2
locations.
Intrinsic Safety
Though ignition energies of dusts are much lower than those of gases and
vapors, no special rules for intrinsic safety have been drafted specifically
for Class II hazards. Taking into account that the market for intrinsically
safe apparatus specifically designed for Class II hazards is limited, devel-
opers of North American intrinsic safety standards have taken the view
that minor modifications to the rules for Class I apparatus will satisfy
needs for Class II apparatus.
In UL 913, if the apparatus is intrinsically safe for Group D and the enclo-
sure is dust-tight, the circuits leading to it are considered intrinsically safe
for Groups F and G. If intrinsic safety is desired for Group E, the appara-
tus must be intrinsically safe for Group C. In Canada intrinsic safety for
Group D is considered satisfactory for all dust groups.
If the enclosure is not dust-tight the usual intrinsic safety rules are
applied, using the assumption that dust will short out all or any spacings
between conductors in whatever way is most hazardous. This decision
Dust Hazards 541
was made because otherwise it is necessary to consider the possibility of

nonconductive dusts becoming conductive in the presence of moisture
and atmospheric pollution. The commercial need for specialized rules for
Class II seemed not to justify the effort that would have been required to
sort out the needed creepage and clearance guidelines.
Ways to Reduce Hazard

It is beyond the scope of this book to consider plant and equipment
design. The principles of hazard reduction are straightforward, but much
still needs to be learned about implementation.
(1) Prevent accumulation of dust. If dust is not there it cannot ignite as a
layer or be dispersed as a cloud. Maintain handling equipment to
keep dust inside. Promptly remove any dust that escapes.
(2) Eliminate sources of ignition by using electrical apparatus of proper
design, grounding metal parts of dust-handling machinery to reduce
static charges, eliminating tramp metal, properly selecting conveyor
materials, enforcing safety rules about open flames, welding, and so
on. (This is a simple summary, but no one knows how to do some of
these things. Eliminating static charges that may ignite dusts in han-
dling systems is an especially difficult problem.)
(3) Treat the dust or handle it in a way that makes it difficult to ignite.
Inerting with rock dust is common in mines, but obviously impracti-
cal for foodstuffs. Carbon dioxide and nitrogen are sometimes used in
grinding and milling operations.
(4) Design machinery and plant to minimize damage if an explosion
occurs. Use flame arresters to prevent flame spread and vents to
relieve pressure and reduce structural damage.
(5) Detect the early pressure rise when an explosion occurs in a closed
system and quench it with an inerting material. Typically, this may be
sodium or potassium bicarbonates, fluorocarbons, ammonium phos-
phate, or bromine compounds.
IEC and CENELEC

The development of IEC standards related to dust hazards has been
underway for many years. The effort has been difficult because at the out-
set there were two conflicting philosophies. North America already had
practices for design and use of apparatus in dusty locations, described
above, which had developed since 1929. European, especially German,
philosophy was that the dust-tightness of apparatus should be specified in
terms of the degrees of protection defined in IEC 529 (see Chapter 15). Sur-
542 Dust Hazards
face temperature should be measured in the absence of a dust layer, and

the user should make appropriate adjustments for the presence of dust
layers thicker than 5 mm.
IEC standards for the design and selection of apparatus therefore are writ-
ten in terms of “Practice A,” the European practice, and “Practice B,” the
North American practice. It is quite possible that when the IEC standards
are used as a basis for a CENELEC standard that Practice B will be deleted
from the CENELEC documents. Just as a location with thick dust layers
raises some issues with regard to Division 1 or Division 2 classification, it
raises some controversy in Europe. Some want to take into account the
presence of deep dust layers that might be thrown into suspension by
allowing classification of such locations as Zone 21, which would usually
be used for locations with dust clouds above the LEL in normal operation
of the plant. Others do not wish to confuse the definitions of zones in this
fashion. The possibility of a layer becoming entrained is not especially a
function of depth, as noted earlier in this chapter. In fact, a given volume
of dust in a thinner layer may be more dangerous than the same volume of
dust in a deeper layer, that is, the exposed surface area of the layer may be
more important than the depth.
Publication 1241-3 gives Zone definitions as follows (summarized by this

author). Zone classification is also summarized in Table 13-5.
• Zone 20 is a location where explosive concentration present con-
tinuously or frequently, and/or where layers of uncontrollable or
excessive thickness can be formed. Zone 20 occurs typically
inside equipment.
• Zone 21 is a location, not classified Zone 20, where explosive
clouds of dust are likely to occur in normal operation; for exam-
ple, near filling spouts.
• Zone 22 is a location not classified Zone 21 in which explosive
clouds exist infrequently and for short duration, or “in which
accumulations or layers of combustible dust may be present
under abnormal conditions and give rise to combustible mixtures
of dust in air.” If removal of dust deposited after an abnormal
event is not ensured, then the location shall be classified Zone 21.
Presence of Dust Dust Layers of Controlled Thickness

Grade of Source
Cloud Disturbed often Disturbed rarely
Continuous 20 21 22
Primary 21 21 22
Secondary 22 21 22
Table 13-5 Summary of Zone Classification

Dust Hazards 543
Releases are classified as continuous, primary, and secondary. In essence,

a primary release is one that occurs in normal plant or equipment opera-
tion and a secondary release is one that is not expected in normal
operation, is infrequent, and when it occurs, is of short duration.
Specification of apparatus in Publication 1241-1-1 is limited to Zones 21

and 22. Electrical apparatus would not normally be located in Zone 20,
that is, inside conveying and processing equipment. The common require-
ments for apparatus of both Practice A and B are given in terms of the
necessity to comply with clauses of the General Requirement document.
The most salient differences between apparatus for Zone 21 (dust-tight)
and that for Zone 22 (dust-protected) is that the latter need not comply
with the insulation resistance limitations of plastics enclosures and the
impact and drop tests.
The special requirements of Practice A are that the enclosure used in Zone
21 must have Degree of Protection IP6X per Publication 529. For Zone 22
the requirement is IP5X.
The special requirements of Practice B for Zone 21 apparatus include:

• six cycles of heating at rated load and cooling, in the presence of a
circulated dust cloud
• temperature rise test while blanketed with dust
• dimensional requirements for width and clearance of joints, gas-
kets, etc., similar to those for dust-ignition-proof enclosures in
North America
Enclosures for Zone 22 are to be tested for dust entry after only two ther-
mal cycles and for temperature rise while blanketed with dust. No
dimensional requirements are given.
All apparatus must be marked with DIP (A or B)(21 or 22) and tempera-
ture or temperature code. DIP in IEC terminology stands for Dust Ignition
Protection.
Publication 1241-1-2 gives rules for selection, installation, and mainte-

nance of apparatus protected by enclosures.
In Installation Practice A, apparatus conforming to apparatus Practice A is

specified. The maximum surface temperature shall not exceed two-thirds
of the cloud ignition temperature of the dust involved or 75 K less than the
dust layer ignition temperature determined for a 5-mm-thick layer. For
thickness greater than 5 mm, the maximum permitted temperature is
given by a set of curves. These are the same as those discussed by Greiner.
544 Dust Hazards
In Installation Practice B apparatus conforming to apparatus Practice B

shall be used. The surface temperature shall not exceed two-thirds the
cloud ignition temperature of the dust involved or 25 K less than the igni-
tion temperature for a 12.5-mm layer of dust.
Laboratory tests shall be conducted on the following:

• Practice A apparatus (if the ignition temperature of a 5-mm layer
is less than 250°C)
• Practice B apparatus covered by dust layers in excess of 12.5 mm
• Apparatus to either Practice A or B blanketed by a layer more
than 50 mm thick
In Practice A, DIPA21 apparatus shall be used in Zone 21 for conductive

or nonconductive dusts, as well as in Zone 22, if conductive dusts are
present. Either DIPA21 or DIPA22 apparatus may be used in Zone 22 if the
dust is nonconductive.
In Practice B, DIPB21 apparatus shall be used in the presence of noncon-

ductive dust in Zone 21 and in the presence of conductive dust in either
Zone 21 or 22. In Zone 22 either DIPB22 or DIPB21 apparatus may be used
if the dust is nonconductive.
Optical radiation into Zone 21 shall not exceed an irradiance of 5 mW/mm2.

Radiation into Zone 22 shall not exceed 10 mW/mm2 (see Chapter 12 of this
book).
Ultrasonic power used in Zone 21 or 22 shall not exceed 0.1 W/m2 or

0.1 W/cm2, respectively, and a frequency of 10 MHz. The circuit must
ensure that the piezoelectric elements often used as transducers are free of
ignition-capable charges.
References
C22.2-157, “Intrinsically Safe and Nonicendive Equipment for Use in Haz-
ardous Locations,” Canadian Standards Association, Rexdale.
Schwab, R. F., and D. F. Othmer, “Dust Explosions,” Chemical and Process

Engineering, pp. 165–174, April 1964.
Olenik, H., Rentsch, H., and W. Wettstein, BBC Explosion Protection Man-
ual, 2nd Revised Ed., Brown Boveri, & Cie, Mannheim, 1983.
Conti, R. S., Cashdollar, K., Hertzberg, M., and I. Liebman, “Thermal and
Electrical Ignitability of Dust Clouds,” Bureau of Mines, RI 8798, Pitts-
burgh, 1983.
Dust Hazards 545
“Classification of Dusts Relative to Electrical Equipment in Class II Haz-

ardous Locations,” NMAB 353-4, National Academy Press, Washington
D.C., 1982.
Proceedings of the International Symposium on the Explosion Hazard Classifica-

tion of Vapors, Gases, and Dusts, NMAB 447, National Academy Press,
Washington D.C., 1987.
“Test Equipment for Use in Determining Classifications of Combustible

Dusts,” NMAB 353-2, National Academy Press, Washington D.C., 1979.
Proceedings of International Symposium on Grain Dust Explosions, Grain Ele-

vator and Processing Society, Minneapolis, 1977.
“Brenn- und Explosions-Kenngrossen von Stäuben,” Hauptverband der

gewerblichen Berufsgenoussenschaften e. V., Bonn, 1980.
Griffith, W. C., “Dust Explosions,” Ann. Rev. Fluid Mech. 10:93–105, 1978.
Hertzberg, M., “A Critique of the Dust Explosibility Index: An Alternative

for Estimating Explosion Probabilities,” RI 9095, Bureau of Mines, 1987.
Hertzberg, M., Cashdollar, K., and J. J. Opferman, “The Flammability of

Coal Dust-Air Mixtures,” RI 8360, Bureau of Mines, 1979.
Verkade, M., and P. Chiotti, “Literature Survey of Dust Explosions in

Grain Handling Facilities: Causes and Prevention,” Project 400-25-04,
EMRRI, Iowa State University, Ames, 1976.
Conti, R. S., Cashdollar, K., and R. Thomas, “Improved 6.8L Furnace for
Measuring the Autoignition Temperatures of Dust Clouds,” RI 9467,
Bureau of Mines, 1993.
“Classification of Combustible Dusts in Accordance with The National

Electrical Code,” NMAB 353-3, National Academy of Sciences, Washing-
ton, D.C., 1980.
“The Investigation of Grain Elevator Explosions,” NMAB 367-1, National

Academy of Sciences, Washington, D.C., 1980.
“Pneumatic Dust Control in Grain Elevators, Guidelines for Design, Oper-

ation and Maintenance”, NMAB 367-3, National Academy of Sciences,
Washington, D.C., 1982.
“Prevention of Grain Elevator and Mill Explosions,” NMAB 367-2,

National Academy of Sciences, Washington, D.C., 1982.
546 Dust Hazards
“Guidelines for the Investigation of Grain Dust Explosions,” NMAB 367-4,

National Academy of Sciences, Washington, D.C., 1983.
International Symposium on Grain Elevator Explosions, Vol. 1, Preprints, Vol. 2,

Proceedings, NMAB 352-2, National Academy of Sciences, Washington,
D.C., 1978.
Proceedings of the International Symposium on Grain Dust, Kansas State Uni-

versity, Manhattan, 1979.
Proceedings of First National School on Explosibility of Industrial Dusts, Kar-

pacz, Poland; English Translation, Kansas State University, Manhattan,
1979.
First International Colloquium on Explosibility of Industrial Dusts, Polish

Academy of Sciences, Warsaw, 1984.
Bartknecht, W., “Gas, Vapor, and Dust Explosions,” Proceedings of the

International Symposium on Grain Elevator Explosions, NMAB 352-2, Wash-
ington, D.C., 1978.
Aldis, D. F., and F. S. Lai, “Review of Literature Related to Engineering

Aspects of Grain Dust Explosions,” USDA Publication 1375, 1979.
Greiner, H., “Standardization Work on Dust Explosion Protection,” Ex

Magazine, Stahl, Kunzelsau, October 1994.
Kauffman, C. W., “Needs in Dust Explosion Testing,” Proceedings of the

International Symposium on the Explosion Hazard Classification of Vapors,
Gases, and Dusts, NMAB 447, pp. 41–64, National Academy of Sciences,
Washington, D.C., 1987.
IEC Publication 1241-1-1: “Electrical Apparatus for Use in the Presence of

Combustible Dust: Part 1: Electrical Apparatus Protected by Enclosures:
Section 1: Specification for Apparatus,” Geneva.
IEC Publication 1241-1-1-2: “Electrical Apparatus for Use in the Presence

of Combustible Dust: Part 1: Electrical Apparatus Protected by Enclosures:
Section 2: Selection, Installation, and Maintenance of Apparatus,” Geneva.
IEC Publication 1241-1-3: “Electrical Apparatus for Use in the Presence of

Combustible Dust: Part 3: Classification of Areas Where Combustible
Dusts Are or May be Present,” Geneva.
CHAPTER 14
HUMAN SAFETY
Introduction
Intrinsically safe equipment and wiring provide plant safety. They will
not cause an explosion. Most intrinsically safe systems can be worked on
live without danger of releasing sufficient energy to cause ignition. The
plant, therefore, is protected. But what about the person working with the
instruments? The goal of this chapter is to help answer this question by
reviewing the important parameters relative to the effects of electric shock
on human beings.
The concern of this chapter is macroschock, that is, shock caused by con-
tact with current-carrying electrodes outside the body. In recent years the
problem of microshock, that caused by current-carrying electrodes intro-
duced into the body, has been subjected to extensive debate and study.
Microshock hazard levels are higher than those of macroschock because
insertion of catheters, electrodes, and other artifacts into the body may
cause current flow directly through the heart. The normal protection
afforded by shunt paths through other parts of the body is lost. Industrial
hazards due to contact with live parts and conductors are macroshock
hazards.
Effects of Electrical Shock

Table 14-1 lists the important effects of electrical shock and the current lev-
els generally associated with them. The values for current listed in Table
14-1 are not absolute. They are approximate response levels of an adult to
current flow from hand to foot. As with ignition data, reported levels
depend not only on the experiment performed but also on the objective of
the experiments. If the objective is to establish reference levels of relatively
low hazard, the 0.5 percentile value may be quoted. At this value it is pre-
dicted that the response will occur in only 5 of every 1,000 subjects at
current levels below the reference value. If average values are reported,
the numerical values can, of course, be significantly higher.
There is substantial variation in adult response to electric currents. The

figures given in Table 14-1 are biased very heavily toward the minimum
response. Other sources may quote somewhat different values.
548 Human Safety
60 Hz Current Effect
0–1 mA Imperceptible
1–3 mA Perceptible, mild
3–5 mA Annoyance
6–9 mA "No-let-go" or freezing
30 mA Asphyxation
80 mA Ventricular fibrillation
5–10A Cardiac arrest, burns
Table 14-1 Effects of Electrical Shock
Dalziel presented perception current data for males. The average level is
1.1 mA. The 0.5 and 99.5 percentile values are 0.4 and 1.8 mA, respectively.
The average for women is about 0.7 mA.
Reaction current is the smallest value that will cause an unexpected invol-
untary reaction, which might result in an accident. Dalziel reports on a
study conducted in 1967 at U.L. On the basis of this study, an ANSI stan-
dard established 0.5-mA maximum leakage current for two-wire, 120-V
appliances. For large cord-connected appliances the level set was 0.75 mA.
Although many discussions of electrical shock phenomena deal with

effects at current levels above the "no-let-go" level, the reaction current
level is of considerable relevance to industrial situations. Although the
primary effect of a mild shock of a few milliamperes might only be tran-
sient discomfort or annoyance, the resultant involuntary reaction could
cause fall from a ladder, cuts or abrasions from violent contact with the
equipment, or perhaps severe shock from another source.
Dalziel and Lee reported "let go" current distribution curves for men and
women at 60-Hz frequency. This is the largest current value at which the
subject is able to release a live electrode. At higher currents muscular
paralysis occurs and the victim cannot release his hold. They found the
average "let go" current for 28 women to be 10.5 mA; the average for 135
men was 16 mA. The range around the average was approximately ±4 mA
for women and ±6 mA for men. The half-percentile value for women was
found to be 6 mA; the half-percentile level for men, 9 mA. The "let go" cur-
rent depends very much on the subject's physiological development.
especially in the arms and wrists, but it is also affected by the subject's
motivation. Subjects competing against one another tend to measure
higher than average "let go" currents.
Although it is generally assumed for safety's sake that current levels above
the "let go" value will freeze the victim to electrodes, this is not always the
Human Safety 549
case. If the muscles affected by the current are extensor muscles, the victim
may be thrust violently away from the current-carrying surface. Instances
have been reported where two individuals came in contact with the same
electrically hot object—one individual could not let go and died: the other
was thrown violently away from contact and survived.
The level at which most people can release an energized electrode is

important because just above the "let go" level the subject loses muscular
control. Breathing may be arrested during the period of shock. Although
breathing will usually be restored when current ceases, the "let go" current
level represents a lower boundary of dangerous current levels.
Ventricular Fibrillation
In ventricular fibrillation the heart ceases its normal rhythmic pulsing and
flutters wildly. Normal operation is completely destroyed. In human
beings the heart does not regain normal operation naturally after shock is
removed. For this reason ventricular fibrillation is fatal unless artificial
respiration is applied almost immediately or unless fibrillation is stopped
by application of a defibrillator.
When the heart fibrillates it stops pumping blood. The brain suffers irre-
versible damage unless artificial respiration or some other means of
providing the patient with oxygen is used.
Dalziel and Lee summarized the data used to arrive at threshold current
levels for ventricular fibrillation and concluded that in humans the maxi-
mum current that would not cause fibrillation in 99.5% of the population
is given by the relationship
i = 116 ⁄ T mA
where T is the time of application of the current in seconds. The relation-

ship is presumed valid over periods of application from 8.3 msec to 5 sec.
The minimum current that will cause fibrillation in 0.5% of the population
is given by the equation
i = 185 ⁄ T mA
Both equations are based on a weight of 50 kg (110 lb). The current-caus-

ing fibrillation is linearly dependent upon body weight, so for children or
for heavier adults the current level will be smaller or larger in proportion
to the deviation of their weight from the assumed 50 kg.
It is dangerous, of course, to apply currents greater than the "let go" cur-
rent to human beings, and it is impossible to establish relationships such as
550 Human Safety
those given in the previous paragraphs on the basis of human experimen-

tation. Dalziel and Lee drew their conclusions from reports of experiments
on animals. Several hundred dogs, pigs, sheep, and calves were included
in the experiments, with weights from 5 to 80 kg (11–176 lb). Since there is
good correlation of current to weight over a very wide range of body
weights, it is assumed that the data are also applicable to humans.
Kouwenhoven noted that very brief shocks can greatly exceed the
accepted hazardous values without serious effect. The heart is particularly
sensitive to electric shocks during the T phase of the heart cycle, which
lasts about 20% of the time between beats. If a shock that is short relative
to the time between beats occurs outside the T phase there will be no
effect; although the same current applied during the T phase may cause
fibrillation. He stated that there have been a number of cases in which two
linemen straightened a line by jerking it from the ends, causing the center
to fly up and momentarily contact an energized conductor. One man was
uninjured, because the brief shock did not occur during his T phase, but
the other was knocked to the ground unconscious and pulseless.
High-Current Shocks
Shocks of 5 to 10 A burn tissue, and may cause complete cardiac arrest.

This differs from fibrillation, in that heart action is completely stopped.
This effect is used deliberately to revive a patient who is in ventricular
fibrillation. The fibrillating heart is out of control: It is not working syn-
chronously as it should. A high current applied to the heart paralyzes it.
On removal of the current, the heart may resume normal action. Defibril-
lators are commonly used in hospitals because electric current is only one
of the several common causes of fibrillation. Use of salty, greased elec-
trodes prevents burns.
The Body as a Circuit Element

The thresholds of shock effects have all been described in terms of current
passing through the body along a particularly disadvantageous circuit
path, such as arm to arm or arm to opposite leg, so that the current path
passes through the trunk. Since current level is the important parameter, it
is important to understand how the human body functions as a circuit
element.
The body offers lowest resistance to current flow at places where there are
no callouses or thick skin. These are the crotches of the knees and elbows,
the soles of the feet, the palms of the hands, a band around the waist, and
the face. These are illustrated in Figure 14-1.
Human Safety 551
Figure 14-1 Normal Skin Resistance Patterns
The outer layer of skin, the epidermis, has very high resistance when it is
dry because it lacks capillaries. It is a horny dry material without conduc-
tive solutions. It is a nonlinear resistor. At low voltage, its resistance is
typically 10,000 to 100,000 ohms. The resistance drops to approximately
1,000 ohms if the voltage is raised to 500 to 1,000 volts.
552 Human Safety
The derma, the red layer of skin directly beneath the epidermis, is rich in
capillaries and body fluids and has very low resistance. In a typical
human, if the epidermis is pierced, the resistance of an arm or a leg can be
considered to be 500 ohms. The resistance arm to arm, leg to leg, or arm to
leg, is, therefore, about 1,000 ohms. In safety literature one often finds ref-
erences to “a 500-ohm man,” the resistance of a man when one connection
is made to both hands and another to both legs.
At powerline frequencies the body is a volume conductor. When current

flows through the body from arm to arm or arm to leg, it spreads out
through the trunk. The portion of current that actually flows through the
heart, approximately 10% of the current that flows from hand to foot,
causes extreme damage.
The body can be considered to be primarily resistive. Relative to fibrilla-

tion, 60 Hz is the frequency of highest sensitivity, but the curve is
relatively flat in the range 20 to 100 Hz. Approximately twice the current is
required at 5 and 2,000 Hz as is required in the 20 to 100 Hz range.
The actual resistance the body offers as a circuit element depends on the
amount of contact area provided. Figure 14-2 gives typical values for
human circuits of various types. The left-hand column lists typical circuit
resistances assuming that the epidermis is dry. The center column gives
values for wet epidermis. The third column gives typical values of resis-
tance of floors, shoes, and gloves, which offer the first line of defense
against electric shock.
Human Safety 553
Figure 14-2 Typical Resistances of Human Circuits
What is a Safe Level of Circuit Voltage?

One could determine from Figure 14-2 and the data presented previously
a level of voltage considered to be without significant probability of lethal
danger. However, in almost all industrial countries of the world, it is rec-
ognized that voltage sources up to 30 V RMS or 42 V peak, or 60 V dc, can
be considered nonhazardous in dry locations. Voltages above these levels
may be permitted if the current does not exceed 3.5 mA RMS, 5 mA peak,
or 15 mA dc.
Generally speaking, no particular pains must be taken to safeguard opera-

tors and technicians from contact with a circuit at these and lower
voltages. It is worthwhile to consider, on the basis of previous discussion,
what the probability of lethal contact may be. A 30-V RMS source with a
500-ohm man would allow a current of 60 mA to flow. If it flowed for 4 s,
it would exceed the 58 mA which would be calculated from Dalziel and
Lee's minimum current equations. The calculated 58-mA current is the
current that might cause fibrillation in 5 out of 1,000 adults who weigh 110
554 Human Safety
lb. More importantly, in assuming a 500-ohm man, we have assumed that

contact to the 30-V circuit has been made through both hands and both
feet and that the epidermis has been pierced at all four extremities. One
can hardly suggest that such contact is likely. Therefore, although the 30-V
RMS limit is in theory not absolutely safe, it recognizes that most people
wear shoes, especially in industrial environments, and that the kind of cir-
cuit contact required to cause lethal exposure is nearly impossible to
achieve accidentally.
Conclusion
Many intrinsically safe systems operate at voltages below 30 V RMS (42-V
peak) and therefore can be considered to pose essentially no personnel
hazard. Intrinsically safe circuits can be designed to operate at potentials
up to several hundred volts and may operate at current levels potentially
capable of causing fibrillation, although an improbably low resistance
contact is required.
Injury to an individual contacting either circuit could result from involun-

tary surprise reaction. The worker might suffer serious injury from a fall.
A worker could also suffer a serious cut or laceration from sharp edges of
the equipment when withdrawing the contacting arm from the apparatus.
References
Lee, R. H., "Human Electrical Safety," ISA Monograph No. 110, pp. 27–34,
Instrument Society of America, Pittsburgh, 1965.
Dalziel, C. F., and W. R. Lee, "Lethal Electric Currents," IEEE Spectrum

(February 1969), pp. 44–50.
Kouwenhoven, W. B., "Human Safety and the Electric Shock," Electrical

Safety Practices, ISA Monograph No. 112, pp. 91–97, Instrument Society of
America, Pittsburgh, 1969.
Friedlander, G. D., "Electricity in Hospitals," IEEE Spectrum (September

1971), pp. 40–51.
Dalziel, C. F., "Electric Shock Hazard," IEEE Spectrum (February 1972), pp.
41–50.
CHAPTER 15
DEGREE OF PROTECTION BY
ENCLOSURES
Readers of this book are most likely to find references in standards and lit-
erature to two kinds of descriptors of the protection offered by enclosures:
NEMA enclosure type numbers and the IEC/CENELEC IP code.
The NEMA System

NEMA (National Electrical Manufacturers Association) Standards Publi-
cation No. 250 describes the NEMA system of defining the characteristics
of enclosures and the tests required to verify conformance to the require-
ments. Enclosures for use in unclassified locations are designated for use
indoors, outdoors, or indoors and outdoors. Enclosures for indoor hazard-
ous locations are identified by a number for that particular type of
protection. If, in addition, the enclosure must be used outdoors, it must be
identified also by the identifying type number appropriate to the expected
service conditions.
Tables 15-1 and 15-2 indicate in approximate terms the intended applica-
bility of enclosures for indoor and outdoor unclassified locations.
Provides some protection against these Enclosure Type

environmental conditions 1 2 4 4X 5 6 6P 12 12K 13
Incidental contact with enclosed equipment * * * * * * * * * *
Falling dirt * * * * * * * * * *
Falling liquids and light splashing * * * * * * * * *
Circulating dust, lint, fibers, and flyings * * * * * * *
Settling airborne dust, lint, fibers, and flyings * * * * * * * *
Hosedown and splashing water * * * *
Oil and coolant seepage * * *
Oil or coolant spraying and splashing *
Corrosive agents * *
Occasional temporary submersion * *
Occasional prolonged submersion *
Table 15-1 Enclosures for Indoor Unclassified Locations

556 Degree of Protection by Enclosures
Provides some protection against these Enclosure Type

environmental conditions 3 3R 3S 4 4X 6 6P
Incidental contact with the enclosed equipment * * * * * * *
Rain, snow, and sleet * * * * * * *
Sleet *
Windblown dust * * * * * *
Hosedown * * * *
Corrosive agents * *
Occasional temporary submersion * *
Occasional prolonged submersion *
Table 15-2 Enclosures for Outdoor Unclassified Locations
Type 7 enclosures are explosionproof enclosures that conform to UL stan-

dards and are intended for application in indoor Class I locations. Type 8
enclosures are explosionproof to UL standards for outdoor hazardous
locations. Type 9 enclosures are dust-ignition-proof in accordance with
UL standards. Type 10 enclosures meet MSHA requirements.
Enclosures for use in nonhazardous locations, which have passed the cir-
culating dust test or the wind-blown dust test, as appropriate, may be
used in Class II, Division 2, locations or in Class III, Division 1 or 2
locations.
Table 15-3 gives a brief overview of the verification test for each character-
istic of an enclosure defined in the NEMA system. This approximate
tabulation is presented to aid understanding of how the NEMA system
compares with the IEC method of coding. There is not a one-for-one corre-
spondence between the enclosure type and the verification test specified.
For example, an enclosure that must pass the hose-down test is not
required to be subjected to a wind-blow dust test also. Experience shows
that the former test confirms the latter capability.
Degree of Protection by Enclosures 557
Enclosure Property Verification Test

Incidental contact with enclosed equipment 0.125 in. rod shall not enter a nonventilated
enclosure, or 1/2 in. rod shall not enter a
ventilated enclosure, unless live parts are
greater than 4 in. from opening; then, 1/2 in.
- 3/4 in. rods may enter.
Falling dirt Inspection.
Falling liquids and light splashing Exposed to dripping water for 30 minutes
with no significant entry (Type 2); or no
entry (Types 5, 12, 12K).
Rust resistance 24-h salt spray.
Corrosive agents 200-h salt spray.
Circulating dust, lint, fibers, and flyings 5 min in circulating dust chamber or expo-
sure to atomized water spray; no entry of
either.
Settling airborne dust, lint, fibers, and Similar to above with lower dust concentra-
flyings tion and lower water flow. Dust circulation is
15 s on, 30 s off.
Hosedown and splashing water 1-in.-diameter nozzle, 65 gal/min water
stream at a distance of 10 to 12 ft for 5 min,
minimum.
Oil or coolant seepage 2 gal/min water with wetting agent from 3/8
in. nozzle at 10 to 12 ft distance for 30 min.
Oil or coolant spraying and splashing Same as above.
Occasional temporary submersion 6-ft submersion for 30 min.
Occasional prolonged submersion Internal pressure of 6 psi shall not drop
more than 2 psi in 24 hr; or 6 ft submersion
in water for 24 hr, without entry.
Rain, snow and sleet Water spray of 18 in./hr at 5 psi for 1 hr.
Windblown dust Jet of 90 to 100 psi compressed air and
Portland cement against all seams and
joints; or stream from 1-in. nozzle delivering
45 gal/min at 10 to 12 ft. No entry.
Sleet Water spray for 1 hr, then temperature
dropped to 20°F until 3/4-in. ice forms on
reference bar. No damage to enclosures
and external operating mechanisms must
function.
Table 15-3 Verification Tests for NEMA Enclosure Properties

IEC 529
IEC 529, also published as CENELEC EN 60529, states the degree of pro-
tection afforded by an enclosure in terms of a code; that is, IPXX,
optionally followed by one or more letters. The first numeral pertains to
the ingress of solid foreign objects, and, protection against access to haz-
ardous moving or live parts. The second numeral pertains to the degree of
protection against entry of water with harmful effects. The latest edition of
Publication 529 also provides for an optional additional letter designation
pertaining to access to hazardous parts and a supplementary letter per-
taining to a specific characteristic of the apparatus. The use of an
additional letter code permits the designation of the degree of protection
of an enclosure that may have relatively little protection against ingress of
solid objects, but which, by virtue of baffles or other construction, pro-
vides a higher degree of protection against contact with dangerous live or
moving parts. Most commonly, only the basic IP code is used. Table 15-4
summarizes the meanings of the first and second numerals and the addi-
tional and supplementary letters. Tables 15-5 and 15-6 summarize the test
conditions for each of the characteristic numerals.
First characteristic Against ingress of solid Against access to

numeral foreign objects hazardous parts with:
0 Unprotected Unprotected
1 ≥50 mm diameter back of hand
2 ≥12.5 mm diameter finger
3 ≥2.5 mm diameter tool
4 ≥1 mm diameter wire
5 dust-protected wire
6 dust-tight wire
Second characteristic Against ingress of water
numeral with harmful effects
0 Unprotected
1 vertically dripping
2 dripping (15° tilted)
3 spraying
4 splashing
5 jetting
6 powerful jetting
7 temporary immersion
8 continuous immersion
Optional additional Against access to
letter hazardous parts with:
A back of hand
B finger
C tool
D wire
Optional Supplementary
supplementary letter information specific to:
H High voltage
M Moving parts in motion
during water test
S Moving parts stationary
during water test
W Additional protective
features provided for
protective against specific
weather conditions
Table 15-4 Meaning of IEC Publication 529 IP Code Numerals and Letters
Numeral Degree of Protection Type Test

0 No protection against contact or entry of None
solids
1 Protection against accidental contact by No touching of live or moving parts by a 50
hand, but not deliberate contact. mm-diameter sphere.
Protection against large bodies.
2 Protection against contact by fingers. No admission of 12.5-mm-diameter ball. No
Protection against medium-size foreign contact with a test finger of standardized
bodies. design, roughly 80-mm long by 12-mm
diameter.
3 Protection against contact by tools, No entry by 2.5-mm-diameter wire.
wires, etc. Protection against small
foreign bodies.
4 Protection against contact by small tools No entry by a 1-mm-diameter wire
and wires. Protection against small
foreign bodies.
5 Complete protection against contact with Tested in dust chamber with vacuum applied
live or moving parts. Protection against to inside of enclosure, 200-mm H2O
harmful deposits of dust. approximately, to draw 80 enclosure
volumes through enclosure. Dust may enter,
presumably not enough to affect function.
6 Complete protection of live or moving Same test as above, but no entry of dust
parts. Protection against ingress of dust. permitted.
Table 15-5 Degree of Protection against Contact and Entrance of Solid Foreign Bodies
Numeral Degree of Protection Type Test

0 No protection None
1 Protection against drops of Similar to that for Numeral 2, with enclosure
condensed water. Condensed on turntable for 10 minutes exposure in nor-
water falling in enclosure shall mal mounting position. Flow rate 1 mm/min.
have no effect.
2 Protection against drops of Tested with dripping reservoir for 2.5 min. in
liquid. Drops of falling liquid each of 9 positions of 15° tilt. Water shall
shall have no effect when not interfere with function or collect of cable
enclosure is tilted to 15° from entry. Drip rate 3 mm/min.
vertical.
3 Protection against rain. No Oscillating spray head sprays enclosure for
harmful effect from rain at angle 10 min. from angle between ± 60°. Water
less than 60° from vertical. shall not interfere with function or collect at
cable entry.
4 Protection against splashing from Same test as above, but angle is ± 180°
any direction. from vertical. Water shall not interfere with
function or collect at cable entry.
5 Protection against water jets from any 6.3 mm nozzle, 12.5 l/min., 2.5-3 m from
direction. equipment from all directions for 1 min./m2
area, 3 min. minimum. Water shall not inter-
fere with function or collect at cable entry.
6 Protection against conditions on ships' Same as 5 except 12.5 mm nozzle 2.5-3 m
decks. Water from heavy seas will not from equipment, 1001/min, and no water
enter. shall enter.
7 Protection against immersion in water. Immersion under head of 1-m for 30 min.
Water will not enter under stated condi- No entry of water.
tions of pressure and time.
8 Protection against indefinite immersion Test to be agreed between manufacturer
in water under specified pressure. and user.
Table 15-6 Degree of Protection Against Ingress of Liquid

Correlation of NEMA and IP Codes

Each NEMA type number refers to an enclosure with a specific list of prop-
erties, including properties such as corrosion resistance, ice resistance, and
so on, which are not relevant to enclosure tightness or access to live parts.
The IP code classifies primarily with respect to ingress of solid objects,
ingress of water, and protection from contact with live or hazardous mov-
ing parts. Thus, there is no direct correlation between NEMA type
numbers and IP codes. It is possible to state the NEMA type number with
protection against ingress of objects or water equivalent to certain IP codes.
Table 15-7 lists the IP code describing the degree of protection against the
entry of solid objects and water afforded by NEMA enclosure types.
NEMA Enclosure Type Number IEC/CENELEC IP Code

1 IP10
2 IP11
3 IP54
3R IP14
3S IP54
4 and 4X IP56
5 IP52
6 and 6P IP67
12 and 12K IP52
13 IP54
Table 15-7 Conversion of NEMA Type Numbers to IEC IP Codes (Do not use to
convert from IP Code to NEMA Type Number.)
References
Standards Publication No. 250, “Enclosures for Electrical Equipment
(1,000 Volts Maximum),” NEMA, Washington, D.C., 1991.
Publication 529, “Degrees of Protection Provided by Enclosures (IP

Code),” International Electrotechnical Commission, Geneva, 1989.
ACKNOWLEDGMENTS
The first edition of this book was based largely on work done by ISA RP12.
The author drew freely on concepts, attitudes, and conclusions developed
during committee deliberations. Although specific references were cited
whenever possible, the most important contributions could not be so
acknowledged because the patterns of thought, ideas, and concepts were
developed by the committee during its many years of activity.
The committee's program of study, teaching, and standards writing

demanded substantial personal commitment from many individuals and
continued support by their sponsoring organizations. Without that com-
mitment there would have been no basis for this book.
In recent years my sources of information and understanding have multi-

plied. Some who have read an earlier edition have sent me copies of their
own works so that my knowledge base would be expanded. Most have
been recognized by reference in the bibliographical references at the end
of each chapter. It has now become almost impossible to identify to
myself, much less credit in print, all those who in some way have contrib-
uted to my understanding of the subjects covered in this book. I can only
acknowledge the contributions of many dedicated colleagues in national
and international committees on which I have been fortunate enough to
have served and learned.
I owe thanks to Bill Lawrence for helping me make Chapter 3 more cur-
rent and to Dave Bishop, Tom Dubaniewicz, and other members of ISA
SP12.21, who made many helpful suggestions about Chapter 12.
As I get older, however, I am ever more aware of the contribution that

Fred Maltby made to the work of ISA in the field of safe use of electrical
equipment in hazardous locations. Fred insisted that his committee follow
these steps:
• Understand the technical background of the subject of the stan-
dard.
• Agree on what the standard should say. If this disagrees with other
standards, establish a plan to educate people and change the erro-
neous standard, if at all possible.
• Write the standard.
Unfortunately, many committees have not been trained to act in this fash-
ion and leap to the pen or word processor before they understand their
subject.
Thanks, Fred!
APPENDIX A
DERIVATION OF INITIAL PURGE
VOLUME REQUIREMENT
Consider a case of volume VT divided into n compartments of equal vol-

ume with communicating passages so that all partial volumes are in
series:
If the concentration of combustible material (gas or vapor) in the first com-

partment is C1 and if a volume of purge gas ΔV of combustibles
concentration CP is introduced, assuming perfect mixing in the first vol-
ume (this is assumed for subsequent sections also), the resulting change in
concentration in the first chamber ΔC1 is
C 1 V T ⁄ n + C P ΔV
ΔC 1 = ------------------------------------------
- – C1
V T ⁄ n + ΔV
In differential form,
dC –( C1 – CP )
--------1- = --------------------------
-
dV VT ⁄ n
Using the Laplace operator method and defining C10 as the concentration
in chamber 1 before any purge gas is added,
C1 ( s ) CP
SC 1 ( s ) – C 10 + -------------
- = -----------------
-
VT ⁄ n SV T ⁄ n
C P + S ( V T ⁄ n )C 10
C 1 ( s ) = --------------------------------------------
-
[ S ( V T ⁄ n ) + 1 ]S
564 Derivation of Initial Purge Volume Requirement
For the second section the purging input will be the volume ΔV with con-
centration C1, and
C 1 ( s ) + ( V T ⁄ n )C 20
C 2 ( s ) = -----------------------------------------------
-
( S + n ⁄ V T )V T ⁄ n
where C20 is initial concentration.

C P + S ( V T ⁄ n )C 10 + S ( V T ⁄ n ) [ ( V T ⁄ n )S + 1 ]C 20
C 2 ( s ) = ------------------------------------------------------------------------------------------------------------------------
-
2
S [ ( V T ⁄ n )S + 1 ]
after substituting the expression for C1(s).
The concentration in the nth and last section, the one of interest in deter-
mining the amount of purge volume required, is
n–1
C P + S ( V T ⁄ n )C 20 [ ( V T ⁄ n )S + 1 ] + … + S ( V T ⁄ n ) [ ( V T ⁄ n )S + 1 ] C no
C n ( s ) = ------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------
-
n
S ( [ V T ⁄ n ]S + 1 )
For the usual situation the concentration of combustible gas or vapor can
reasonably be assumed to be sensibly uniform throughout the enclosure
so that C10 = C20 = C30 = Cno and
2 n – 1½
C P + S ( V T ⁄ n )C 10 ® 1 + [ ( V T ⁄ n )S + 1 ] + [ ( V T ⁄ n )S + 1 ] + … [ ( V T ⁄ n )S + 1 ] ¾
¯ ¿
C n ( s ) = -----------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------
n
S [ ( V T ⁄ n )S + 1 ]
Solving this equation and transforming back to the original variable Cn,
the concentration of combustible in the nth section, and V, the total vol-
ume of purge gas, the general solution is
k =n–1
– nV ⁄ V T 1- § nV k
C n = C P + ( C 10 – C P )ε ¦ --- --------·
k! © V T ¹
k=0
If the purge air is clean, as it should be, Cp = 0, and the concentration as a

function of purge volume for specific numbers of sections of equal volume
is given by
–V ⁄ VT
C 1 = C 10 ε
– 2V ⁄ V T 2V
C 2 = C 10 ε 1 + -------
VT
Derivation of Initial Purge Volume Requirement 565
– 3V ⁄ V T 3V 1 3V 2
C 3 = C 10 ε 1 + ------- + --- § -------·
VT 2 © VT¹
– 4V ⁄ V T 4V 1 4V 2 1 4V 3
C 4 = C 10 ε 1 + ------- + --- § -------· + --- § -------·
VT 2 © VT¹ 6 © VT¹
– 5V ⁄ V T 5V 1 5V 2 1 5V 3 1 5V 4
C 5 = C 10 ε 1 + ------- + --- § -------· + --- § -------· + ------ § -------·
VT 2 © VT¹ 6 © VT¹ 24 © V T¹
The curves in Figure A-1 show the concentration in the last section in
terms of the initial concentration C10, plotted against total purge volume
relative to total case volume VT. Even in the worst case, that of a single
undivided volume, four volumes of purge air will reduce the concentra-
tion to less than 2% of the initial value.
The lowest LELs listed in NFPA pamphlet No. 325 are about 0.6%. Four
volumes of purge are, therefore, adequate, even when the initial concen-
tration of such vapors is at an unusually high level of 30%. Any
subdivision of the case into serially connected volumes improves the
effectiveness of the purge.
Since the requirement for initial purge of four case volumes is safe for
even a single section enclosure, it follows that any subdivision, even if the
subdivisions are of unequal size, is in the direction of improving the purge
effectiveness.
566 Derivation of Initial Purge Volume Requirement
Figure A-1 Effectiveness of Purge

APPENDIX B
EXAMPLES OF DERIVATIONS
Derivation of an Equation for Loss of Pressure in a

Sealed System
Pt is the pressure inside system (atm); V is the system volume (cm3); Pa is
the pressure outside the (atm);
Pt, V Pa
l is the seal length (in.); Pt0 is the initial system pressure (atm); K is the leak
rate (cm3/(yr) (in.) (atm2); and t is the time (yr).
2 2 V dP t
– Kl ( P t – P a ) = ----- --------
P a dt
The left-hand side of this equation is the conventional equation for flow
through a small restriction, taking density changes into account (see Bed-
well and Meyer). The right-hand side of the equation is the change in
volume of gas in the system referred to pressure Pa.
This equation has the solution

2
– 2P ( Kl ⁄ V )t
Pa ( 1 + C2 ε a )
P t = ------------------------------------------------------
2
-
– 2P ( Kl ⁄ V )t
( 1 –C2 ε a )
At
t = 0, P t = P t0
Therefore,
P t0 – P a
C 2 = -------------------
-
P t0 + P a
568 Examples of Derivations
So
2
– 2P ( Kl ⁄ V )t
P t0 + P a + ( P t0 – P a )ε a
P t = P a ------------------------------------------------------------------------------
2
-
– 2P ( Kl ⁄ V )t
P t0 + P a – ( P t0 – P a )ε a
Solving for t in the above equation,

Pt
§ ----- P t0 ·
+ 1· § ------- –1
V © Pa ¹ © Pa ¹
t = --------------- ln -------------------------------------------
2
2P a Kl § ----- Pt P t0
– 1· § ------- + 1·
©P ¹© P ¹
a a
If t is in days and V is in cubic inches,

Pt
§ ----- P t0 ·
+ 1· § ------- –1
2985 V P a © ¹ © Pa ¹
t = -----------
- -----
-ln -------------------------------------------
P a Kl § ----- Pt P t0
2
– 1· § ------- + 1·
©P ¹© P ¹
a a
In typical hermetically sealed devices V/l is in the range 0.2 to 10 in.3/in.
If Pt is taken to be 1.1 and Pa is taken to be 1.0, for the extremes of the V/l
range,
597 2.1 P t0 – 1 29850 P t0 – 1

t = --------- ln ------- ----------------- or t = --------------- ln 2l -----------------
K 0.1 P t0 + 1 K P t0 + 1
If seal is Grade A and K = 1 and initial pressure Pto is 2 atm,

t = 1160 days t = 58200 days
or
3.18 yr 159 yr
If seal is Grade C, K = 104 and Pto = 2 atm,

t = 2.8 hr or 5.82 days
Examples of Derivations 569
Derivation of Expressions for Seal Breathing

We first derive the equation for half-day flow through a seal caused by
ambient temperature and barometric pressure variations.
For a small restriction the flow equation is

Kl 2 2
Q = --------- ( P 2 – P 1 )
365
where
Q = flow rate (cm3/day)
K = leak rating [cm3/(yr) (in.) of seal/atm]
l = length of seal
P2 and P1=pressures on opposite sides of seal (atm)
If both P2 and P1 are near 1 atm this expression can be simplified by

replacing (P22 − P21) by (P2 − P1) (P1 + P2) ≅ 2ΔP.
Kl Kl
Q = ------------- ( P 2 – P 1 ) = ------------- ΔP
182.5 182.5
For considering breathing of a seal caused by daily pressure cycling, we

use the following model:
P2 + AP2 sin φ
Vo V P1 – BP1 sin φ
where
P2 = the pressure inside the enclosure (atm)
P1 = the pressure outside the enclosure (atm)
A = the fractional change in internal pressure caused by ambi-
ent temperature changes
B = the fractional change in external pressure
Vo = the system volume (cm3)
V = the volume of gas flow in or out of the system
φ = 2πd, where d is in days
P 20 φ
P 2 = P 20 – -------- ³ Q dφ + AP 20 sin φ
Vo 0
P 1 = P 10 – BP 10 sin φ
Note that it is assumed that A and B are acting in the direction to simulta-
neously increase or decrease the pressure drop across the seal. This is a
very conservative assumption.
Assuming that P20 and P10, the initial values of P2 and P1, are both 1 atm,
and noting that
dV
V = ³ Q dφ and Q = -------
dφ
the flow equation can be written as
------------- dV
182.5 V
- = ( A + B ) sin φ
------- + -----
Kl dφ V o
The solution to this equation is

(A + B) Kl Kl Klφ ⁄ 128.5V o
V = ----------------------------------- ------------- -------------------- sin φ – cos φ + ε
Kl 182.5 182.5V o
1 + § --------------------·
© 182.5V o¹
For a practical range of variables the exponential term can be neglected.

This expression can be written
(A + B) Kl
V = -------------------------------------------------- ------------- cos ( φ + θ )
Kl 2 1 ⁄ 2 182.5
1 + § --------------------·
© 182.5V o¹
where
θ = Arctan Kl/182.5Vo
The maximum flow volume during the day will be when cos (φ + θ) = 1:
(A + B) Kl - ( A + B )V o
V max = --------------------------------------------------
⁄
------------ = --------------------------------------------------
Kl 2 1 2 182.5 182.5V o 2 1 ⁄ 2
1 + § --------------------· 1 + § --------------------·
© 182.5V o¹ © Kl ¹
If K → ∞ then Vmax = (A + B)V0, as used in the illustration in the text.

Having calculated the half-day flow for a given enclosure, we are now in a
position to determine how long it will take for the concentration inside the
enclosure to reach a specified concentration, if the outside concentration is
specified.
In the following C0 is the concentration outside the enclosure; Ci is the

concentration inside; Cio is the initial concentration inside; d is the time in
days. The development of the required equation can be carried out by
recording the amount of contaminant material in the system after V vol-
umes are breathed in and the resulting concentration in the system before
breathing out. After breathing out the concentration in the system, of
course, remains the same, but the amount is decreased.
Amount of Contaminant in System

Concentration Before Breathing Out
after Breathing In
C 10 V 0 Initially C 10
C 10 V 0 + C 0 V 1st cycle C 10 V 0 + C 0 V
--------------------------------
-
V + V0
2 2nd cycle 2
C 10 V 0 + C 0 VV 0 C0 V C 10 V 0 + C 0 VV 0 ( C 0 V )
---------------------------------------- + C 0 V - + -------------------------------------------------------
---------------- -
V + V0 V + V0 V + V0
3 2 3rd cycle 3 2
C 0 VV 0 C 10 V 0 + C 0 VV 0 C0 V C 0 VV 0 C 10 V 0 + C 0 VV 0
------------------ + ---------------------------------------- = C 0 V - + -----------------------
---------------- - + ---------------------------------------
-
V + V0 (V + V )
2 V + V0 ( V + V ) 2
(V + V )
3
0 0 0
At the end of the portion of a cycle during which volume flows into the
system,
V0 d V V0 V0 2 V0 3 V0 d
C i = C 10 § -----------------· + C 0 ------ ----------------
- + § -----------------· + § -----------------· + … § -----------------·
© V + V 0¹ V 0 V + V 0 © V + V 0¹ © V + V 0¹ © V + V 0¹
The expression in the brackets is of the form

2 3 d
γ+γ +γ +…+γ
The general form is 1 + γ + γ2 + γ3 ⋅⋅⋅ γn−1, where n = d + 1.
Σnterms = [(1 − γn)/(1 − γ)] − 1, where γ = V0/(V + V0).

V0 d V0 d
C i = C i0 § -----------------· + C 0 + 1 – § -----------------·
© V + V 0¹ © V + V 0¹
Solving for d,
C0 – Ci
ln § --------------------·
© C 0 – C i0¹
d = -------------------------------
V0
ln § -----------------·
© V + V 0¹
Derivation of Expressions for Pressure Testing Sealed

Enclosures
The equation presented earlier in this chapter for change in concentration
in an enclosure caused by diffusion can be written in terms of A, the total
area through which diffusion occurs, the sum of the areas of all the leaks
through the enclosure seal.
lV C 0 – C i0
t DH = -------- ln § --------------------·
ΔA © C 0 – C i ¹
Assume Cio, initial concentration inside the enclosure is 0 and that C0, the
concentration outside, is twice LEL. At t = 5.78 × 104 s (16 hr) Ci, the con-
centration inside the enclosure shall be no higher than the LEL. The
equation becomes:
4 lV lV
5.78 × 10 = -------- ln 2 = 0.693 --------
ΔA ΔA
where Δ is the diffusion coefficient of the diffusing gas or vapor into air.
The time response of an enclosure to pressure changes was derived in the

first section of this appendix. For our purposes here it is preferable to start
with the flow equation presented earlier in the chapter, and write it in
terms of total area of all leaks.
As written earlier, r and l in inches;

4
17 r 2 2 3
Q = 5.78 × 10 ---- ( P 2 – P 1 ) std cm ⁄ yr
l
4
10 r 2 2 3
= 1.83 × 10 ---- ( P 2 – P 1 ) std cm ⁄ s
l
If r and l are in centimeters,

4
9r 2 2 3
Q = 1.12 × 10 ---- ( P 2 – P 1 ) std cm ⁄ s
l
Expressing this equation in terms of area also:

2
8A 2 2
Q = 1.13 × 10 ------ ( P 2 – P 1 )
l
If P2 and P1 are nearly equal,
2 2
( P 2 – P 1 ) = ( P 2 – P 1 ) ( P 2 + P 1 ) ≅ 2P 1 ( P 2 – P 1 )
and the equation can be written:

2
8A 3
Q = 2.26 × 10 ------ ( P 2 – P 1 ) std cm ⁄ s
l
Using a derivation paralleling that in Appendix B-1, the expression tpH,
the time for P2 to drop to one-half its initial value, is
0.693lV – 9 lV
t pH = -------------------------------- = 3.1 × 10 -----2-
8 2
2.26 × 10 A A
If area A is assumed to be made up of n holes of area An the half-time due

to one hole alone would be
– 9 lV
t pH = 3.1 × 10 ------2
An
Because of n holes in parallel:

– 9 lV
t pH = 3.1 × 10 ---------2-
nA n
The aim is to develop an equation for a pressure half-time test, so solve for
A = An in the diffusion half-time equation and substitute into this
expression.
From the diffusion equation:

lV
A = nA n = 0.693 -------------
Δt DH
–9 –9
3.1 × 10 lV 3.1 × 10 lVΔt DH – 9 Δt DH
t pH = ------------------------ ---------- = ------------------------ -------------------- = 4.43 × 10 -------------
An nA n An 0.693lV An
If tDH is 5.78 × 104,

–4 Δ
t pH = 2.56 × 10 ------
An
Derivation of Flow Test Equation

If a test limit tpH has been established based on an assumed lower limit An,
the maximum flow limit can be established.
Starting with the equation for pressure half-time, solve for An2.
– 9 lV
t pH = 3.1 × 10 ---------2-
nA n
2 – 9 lV
A n = 3.1 × 10 -----------
nt pH
In terms of n holes of area An, the flow equation is
8 A n 8 2n
2
Q = 2.26 × 10 § ----· --- ( P 2 – P 1 ) = 2.26 × 10 A n --- ( P 2 – P 1 )
© n¹ l l
8 –9
( 2.26 × 10 ) ( 3.1 × 10 )lVn ( P 2 – P 1 )
= --------------------------------------------------------------------------------------------
-
nlt pH
( P2 – P1 )
= 0.7V -----------------------
t pH
This equation gives the allowable flow per second for a chosen differential
pressure and tpH. The P2 − P1 must be small so that P2 ≈ P1 as in the origi-
nal derivation of the flow equation.
Example: If tpH is 30 s, and P2 − P1 is 2 in. of water, (0.0052 atm)

0.7V ( 0.0052 ) –4
Q = --------------------------------- = 1.21 × 10 case volumes / s
30
= 0.44 case volumes / hr
INDEX
Index Terms Links
abnormal conditions 341

AC and DC
comparison 304
active barriers 482
AIT 20 78
American Petroleum Institute 3
analog output circuit 429
analog output module 476
analyzer rooms 197
aqueous solutions of flammable liquids 75
arc characteristics 270 300
arcing
in capacitive circuits 274
area classification 43 49 118 533
dusts 62
gases and vapors 60
geometric approaches to 49
practice of 62
special cases of 58
Arrhenius equation 19 21
Article 500 44
This page has been reformatted by Knovel to provide easier navigation.

Index Terms Links
associated apparatus 351 352

choosing 484
design and evaluation 415
maintenance of 405
authority having jurisdiction 49
autoignition temperature (AIT) 20
automatic current limiting 483
barrier construction 480

barriers 472
batteries 261
primary and secondary 363
Benjaminsen 122 379
Bijl 120
blocking capacitors 369
bolt spacing 170
break spark 315
breakdown voltage 269 270
break-flash apparatus 278
break-spark apparatus #3 349
break-spark ignition current 310
breathing 223
British break-flash 381
BS 4683 104
BS 6941 104 219 244
bubble tests 235
Bureau of Mines 342

Index Terms Links
cables 106 356

ignition characteristics 445
MC or ITC 100
MI, MC, MV 100
sealing 187
Type MI 99
cable parameters 395
cage rotors 257
calibration currents
IEC apparatus 392
Canadian Electrical Code (CEC) 6
Part 1 43 102
Canadian Standards Association (CSA) 6
capacitance 288
capacitive circuits
ignition characteristics 385 386
test equipment 278
capacitor discharge 285
casting compound 361
category ia apparatus 353
category ib apparatus 353
CE Mark 12 350
cells
primary and secondary 363
CENELEC 11 162
CENELEC EN50016 204
CENELEC EN50019 251
CENELEC EN50020 345 349

Index Terms Links
certification 463
certification documents 395
certification drawing 464 466
characteristic values 351
chemical compatibility testing 237
circuit analysis 424
circuit voltage
safe level of 553
Class I 44 46 48
Group A 44
Group B 44
Group C 44
Group D 44
guides for 55
Class I hazards 49
Class I liquids 53
Class I, Division 1 99
Class I, Division 2 100
seals 100
wiring in 100
Class II 44
Group E 45
Group F 45
Group G 45
Class II liquids 53
Class II locations 47 57
area classification 534
pressurized enclosures 184 193
Class II, Division 1 47 89 101
Class II, Division 2 47 89 101

Index Terms Links
Class III 44
Class III locations 48
Class III, Division 1 48 102
Class III, Division 2 48 102
Class IIIA liquids 53
Class IIIB liquids 53
classification 45
Class I
guides for 55
Class I locations 51
dusts 89
Group E 89
Group F 89
Group G 89
gases and vapors 77
mixtures 70 74
classification of a location 50
classification of dusts
historical review 89
classification system 76
clearance 355
clearance in air 360
clearances 254 355 356
closing contacts 284
cloud
ignition temperature 523
cloud ignition energy 523
code-enforcing authority 49
combusion zone 19

Index Terms Links
combustible dusts 521

Class II 44
combustion 18
combustion wave 17 22
combustion zone 21
Committee on Electrical Equipment in
Hazardous Chemical Atmospheres (EECA
committee) 78
comparison of NFPA and IEC classifications
methods 81
comparison of U. S. and IEC classifications 80
component ratings 459
components
ratings 362
computing stored energy 434
concentration
influence of 527
conductive dusts 90
Division 1 classification 91
conductive parts
distance between 359
conduit 99 106 107
sealing 187
connections 171 358
connectors 362
constant current source 429
contact material
influence of 299
contact separation speed
effect of 306

Index Terms Links
continuous dilution 59 185 194 199

control drawing 352 394 463 465
control house apparatus not certified as
associated apparatus 471
controllable semiconductors 365
controllable series devices 365
controlled spacing 460
controller output 429 477
controlling concentration 113
creepage 254 355 356 360
critical diameter 25
critical flame sphere diameter 22
critical ignition energy 22
critical interface location 425
CRTs 396 405
CSA 6 161
CSA C22.2 #213 244
CSA C22.2 No. 213 219
current limiting 421
current limiting resistors 368
current mirrors 482
current transformers 261
curves 383
damping windings 368

degree of protection by enclosures 555
design criteria 159
detonation wave 154
developing flame sphere 27
Index Terms Links
dielectric strength 356 368

diffusion 222 223 227
diffusion half-time 241 242
diffusion tests 229
digital input circuits 474
digital inputs 475 487
diode bridges 421
diode return barrier 481
diode safety barriers 371
diodes
series 365
diode-shunted inductance 422
DIP 90
discharge capacitors 284
discharge path 291
divergence 23
Division 0 61 64
Division 1 61 63
Division 1 locations 55 116
Division 2 61 63
Division 2 locations 56 116
divisions 46
doubly sealed and vented installation 59
dual channel barrier 480
dust
chemical composition of
influence of 523
explosion characteristics of 530

Index Terms Links
dust cloud
explosion in 522
ignition energy of 525
dust explosions
parameters 521
protection against 521
dust hazards 521
features 531
IEC standards related to 541
dust layers
ignition energy 525
dust parameters 523
dust-ignition-proof (DIP) 90
dust-ignition-proof enclosures 538
dust-tight enclosures 538 539
EECA 10
EECA committee 78
effect of increasing pressure 31
effect of increasing temperature 31
effect of test conditions 32
ejection of light-alloy particles 150
electric arcs 267
electric shock 547
effects of 548
electrical apparatus for dusty locations 538
Electrical Apparatus in Chemical Atmospheres
(EECA) 10

Index Terms Links
electrical ignition
low voltage 267
electrode conditioning 393
electrode geometry 39
effects of 265 266
electrode material
effects of 273 294 301
electronic components
ignition by 332
electronic limiting 483
EN50014 63
EN50018 349
EN50020
curves 383
encapsulant 362
encapsulation 104 217 244 245
358
enclosed break apparatus 407
enclosed break devices 175
enclosure 187
enclosure design 139
enclosure shape
effect of 145
enclosure volume 142 143
enclosures 161 256 355
enclosures for indoor unclassified locations 555
enclosures for outdoor unclassified locations 556
enclosures with internal source of release 194
energy-limited circuits 407
equipment classifications 118

Index Terms Links
equivalent separation distance

calculating 361
Ex d 103
Ex e 103
Ex i 104
Ex m 104
Ex n 104
Ex p 105
Ex s 105
explosion pressure 146
explosion severity 524
explosion testing 140
explosionproof enclosures 137
installation of 173
maintenance of 174
exponential probability function 122
external connections 253
provision for 355
Factory Mutual Research Corporation (FM) 4 5

fault conditions 427 456
fault tree 132 507 508
faults
assessment of 366
ferroresonant constant voltage transformers 431
Fieldbus cable 446
field-mounted apparatus not certified 470 471
fire triangle 51
flame propagation 21
Index Terms Links
flame velocity 28
flameproof 103
flameproof apparatus 126
flameproof motor controller 377
flammability limits
effect of increasing pressure 31
effect of increasing temperature 31
flammable gases and vapors
Class I 44
flange gap
distortion of 147
obstacles in 152
flange material
effect of 150
flange width
effect of 144
flashpoint 53
flow testing 233
flyings
Class III 44
FM 4 5 163
fog 54
Fraczek 378
frictional sparks
ignition by 150
fuses 362
future for classification of materials
material classification
future for 64

Index Terms Links
gas mixture 371

gas or vapor
pressure of 35
temperature of 32
general purpose electrical safety standards 350
geometric approaches to area classification 49
geometry
effects of 294
ground potential
differences 399
grounding conductors 358
grounding enclosures 400
Group A 44
Group B 44
Group C 44
Group D 44
Group E 45 89
Group F 45 89
Group G 45 89
Group IIA 49
Group IIB 49
Group IIC 49
grouping of dusts 45
guides for class I area classification 55

Index Terms Links
hazard of mixtures 53
hazard reduction 108
principles of 541
heat-tracing systems 263
hermetic sealing 219
Hickes-Brown approach 375
high voltage 274
high-current shocks 550
highly inductive circuits 295
high-voltage capacitive discharges 32
holes in enclosure walls 171
hot component ignition 332
hot wire ignition 324 505
hot work 404
how groups are determined 45
human body
as a circuit element 550
human circuits
resistances of 553
human safety 547
humidity
effect of 152
hydrostatically test 165
IEC 10
IEC 1010 359
IEC 1241 62

Index Terms Links
IEC 20-ml apparatus 84

IEC 529 541 558
IEC 529 IP Code Numerals and Letters 559
IEC apparatus 349
calibration currents 392
IEC area classification 60
IEC Committee SC31G 343
IEC Publication 664 359
IEC Publication 79-10 61
IEC Publication 79-11 345 349
IEC standard apparatus 388
IEC standard test apparatus 381
IEC test apparatus 383
curves 383
ignition 21
by frictional sparks 150
by hot surfaces 516
by optical sources 493 497
by radiating soruces 497
by thermal source 501
combustible dust clouds 498
gas or vapor 498
gases and vapors
by hot particles and surfaces 494

Index Terms Links
ignition (Cont.)
irradiance values for 506
modes of 266
test equipment 296
typical test results 284
ignition capability 371
ignition characteristics
capacitive circuits 385 386
inductive circuits 387
resistive circuits 384
ignition data-inductive circuits 298
ignition delay 20 501
ignition energy 23 27 28 265
effects of electrode geometry 266
pressure effects on 35 36
ignition parameters
establishing 283
ignition probability 311
ignition process
time scale 40
ignition sensitivity 524
ignition source location
effect of 148
ignition testing 233
ignition-capable equipment 117
immersion 217
impact testing 167 366
increased safety 103
increased safety apparatus 127
inductance 291 389 391

Index Terms Links
inductive circuits
opening contacts in 294
inductors with ferromagnetic cores 433
inert gases
effect of 38
infallible (protective) components 367
infallible current limiting resistors 368
infallible isolation 370
infallible transformers
type tests 368
infallible wiring 370
inflammability limits
effect of pressure on 35
initial inspection 401
initial pressure 151
inspections
initial 401
periodic 403
Institute of Electrical and Electronic Engineers 3
internal connections 253
internal fans 256
internal release 194
International Electrotechnical Commission
(IEC) 10
International Society for Measurement and
Control 6
international standard test apparatus 280

Index Terms Links
intrinsic safety 104 341 342 538

540
definitions 351
design techniques 420
general requirements 350
ia and ib 348
interpreting the definition 345
standards 350
intrinsic safety for dust hazards 373
intrinsically safe apparatus 341 342 351
design and evaluation 415
requirements 349 354
routine tests 372
intrinsically safe circuit
different 397
separation 398
terminals 355
intrinsically safe design objective 416
intrinsically safe equipment 117 341
certification 342
intrinsically safe multiplexers 484
intrinsically safe product
design process 418
intrinsically safe system architecture 472
intrinsically safe systems 341
design 415 469
inspection of 400
installation 394
maintenance of 404
parameters 351

Index Terms Links
intrinsically safe systems (Cont.)

safety administration (MESA)
requirements 342
IP code 555
iron-core transformer 440
ISA 6
ISA accreditation 9
ISA Committee SP12 6 7
ISA RP12.2 343 406
ISA S12.12 219 244
ISA S12.4 184
ISA SP12 343
ISA TR12.2-1995 8
ISO 9000 463
isolated barrier 484
isolating barriers 484
isolating the ignition source from the flammable
material 113
isolation 421
ITC cable 100
jet entrainment 138

joints 167
KSt values 524

Index Terms Links
L/R ratio 396 444

layer ignition temperature 523 526
leak tests 235
leakage compensation 205
leakage tests 236
LEL 29
let go current 548
LFL 30
light-alloy particles
ejection of 150
live work 406
loop-powered isolated devices 483
low inductance circuits 295
Low Voltage Directive 409
low-boiling-point materials in water 54
lower explosive limit (LEL) 29
lower flammability limit (LFL) 30
luminaires 260
magnetic field
stored energy 436
Manufacturing Chemists Association 3
marking 172 210 373 400
material classification 60 69
not already classified 66
in the United States 75

Index Terms Links
materials 161
other than methane 307
Maximum Experimental Safe Gap 77 139
maximum pressure developed 523
MC cable 100
mean time to explosion 378
mechanical energy 23
MEIC 31
MESG 77 84 139 157
correlation between spark ignition energy
and 157
correlation with MIC ratio 81
methods based on eliminating the source of
ignition 111
methods based on permitting ignition 110
MI cable 100
MIC 79
MIC ratio 79
correlation with MESG 81
MIE 32
mine-signaling apparatus 377
minimum current 270 272
minimum explosive concentration 523
minimum igniting current 79 388 389 390
391
minimum igniting voltage 288
minimum ignition energy (MIE) 32 274 310
minimum layer thickness 537
minimum voltage 270 272
misclassification of a mixture 73

Index Terms Links
mist 54
mixed wiring 107
mixture
mislassification 73
mixture classification 70 74
most easily ignited concentration (MEIC) 31
MTBE 126 378 510 516
MTBE equations 127
MTBE values for radiating systems 511
MTBF 122
multiple grounds 399
MV cable 100
National Academies of Science and

Engineering 10
National Academy of Sciences 522
National Electrical Code (NEC) 3 43 99
recognition in 183
National Fire Protection Association (NFPA) 3 9
National Materials Advisory Board 522
National Research Council 10
NEC 3 43
negative barrier 478
NEMA and IP codes
correlation of 562
NEMA enclosure properties
tests for 557
NEMA enclosure type numbers 555
NFPA 3 9
Index Terms Links
NFPA 493 9 344

NFPA 493-1967 344
NFPA 496 184
requirements for pressurization systems 187
NFPA 497 10 63
NFPA 497-1997 73
NFPA 497M-1983 13
NFPA 497M-1997 68
NMAB 353-1 68
nonincendive 406
nonincendive circuits 406 407
nonincendive component 407
nonincendive equipment 117 406 408
nonincendive field circuits 406 409
nonincendive field wiring 487
nonincendive system 487
nonpolarized barrier 480
nonsparking 407
nonsparking apparatus 127 409
normal and abnormal conditions 347
normal operating conditions 425
normal operation 353 453
Occupational Safety and Health Administration

(OSHA) 3 12
oil immersion 105 217
open-circuit voltage 388 390
optical couplers 421
optical fibers 493 516
Index Terms Links
optical instruments and fibers 503

optical radiation 544
Optical Sensors Collaborative Association
(OSCA) 493
optocouplers 370
OSCA 493
OSHA 3 12
OSHA regulations 344
oxygen content
reducing 529
partitions 355 357

passive barrier circuits 479
passive-shunt-diode zener barriers 477
periodic inspections 403
permissible gaps 160
permitted gaps 167
piezoelectric devices 366
plugs 356
polarity reversal 357
positive barrier 478
potting 244
power supplies 430
power system faults effects 399
pressure
effect of increasing 31
of gas or vapor 35
pressure half-time 242
pressure piling 153
Index Terms Links
pressure rise 523

rate of 146
pressure testing 231
pressurization 59 105 183 538
540
pressurization of an enclosure housing a
containment system 206
pressurization systems
NFPA 496 requirements for 187
pressurization with continuous flow 204
pressurization with leakage compensation 204 205
pressurized apparatus
assessment of 208
pressurized control rooms 193
pressurized enclosures
Class II locations 184 193
pressurizing system 187
prestressing gaskets 236
probability 115
probability of explosion 375 377
probability of ignition 375 377
probability values 121
product design
example 450
protection against intrusion by nonintrinsically
circuits 398
protection required 114
protective components 423
protective gas 59
protective gas system 188

Index Terms Links
protective grounding system 399

protective resistor 423
protective transformer 421
PTB apparatus 298
PTB Report W-39 448
PTB Report W-53e 446
PTB test apparatus 281
purging 183
quantifying the risk 114

quantitative area classification 65
quenching 139
quenching diameter 25 27
quenching distance 25 27 274
radial air gap 257

Raoult’s Law 71
reaction current 548
rectangular sources 433
rectangular V-I characteristic 432
reducing explosion hazard 110
reduction of area classification by the use of
pressurization 58
regulated power supplies 429
regulation 483
relays 357

Index Terms Links
resistance 291
effect of 286
resistance elements 262
resistance temperature detectors 473
resistive circuits
ignition 306
resistivity guidelines 91
resistivity guidelines for Groups E, F, and G 90
restricted breathing 244 408
restricted breathing enclosures 243
results of comparison of U. S. and IEC
classifications 80
risk 109
rotating electrical machines 256
routine testing 165
RP12.1 7
RP12.2 8
RP12.4 8 184
S12.16 252
safe energy levels 348
safe radiated values 494
safety factors 353
Safety in Mines Research Establishment
(SMRE) 342
safety rationale 466
sand filling 105
sand immersion 217
Index Terms Links
saturable core inductor 434

sealed devices
requirements for 229
sealing 217 219 538
conduit or cable 187
seals 100 166 173 400
quality of 219
semiconductors 365
separation 420
separation distances 356
series diodes 423
series resistance
effect of 291
shape
influence of 527
shape, size, and concentration
influence of 527
shields 399
shunt diode 321 421
shunt diode barriers 482
shunt elements
effect of 315
shunt safety assemblies 369
shunt silicon diode
effect of 322
shunt voltage-limiting devices 365
shunt zener diodes 422
simple apparatus 354 470
assessing 354
simplifying techniques 427

Index Terms Links
singly-sealed systems 58
SIT 20
size
influence of 527
skin resistance 551
small component temperature rise 333 334
SMRE 342
sockets 356
solid insulation 361
distance through 360
source current
effect of 291
spacing
specifying 422
spacing requirements 359
spark ignition energy 157
correlation between MESG and 157
sparking signaling circuits 342
special cases of area classification 58
special fasteners 172
specific hazardous atmospheric mixture 346
spontaneous ignition temperature (SIT) 20
static pressurization 204
strain gauge bridge 476
strength 165
stress in the walls 148
Swiss standards 240
systems with multiple sources 486

Index Terms Links
T code 78 99 255
Technical Institute for Industrial Safety (TIIS) 345
temperature
effect of increasing 31
of gas or vapor 32
temperature code 417 458
temperature marking 189
temperature profile 19
temperature rise
small components 333
terminals 358
test conditions
effect of 32
test factor 372
testing 173
precautions in 281
thermal shock testing 167
thermocouples 473 487
TIIS 345
time scale of the ignition process 40
transformers 368 442
transients
effects of 365
transmission lines 444
transmitter input 477
turbulence outside an enclosure 153
Type e 251
Type e terminal boxes 252

Index Terms Links
Type e terminals 358

Type MI cable 99
Type N 104
Type of Protection e 251
Type of Protection m 245
type tests 371
Type X pressurization 186 190
Type Y pressurization 186 190
Type Z pressurization 186 189
Types of Protection 103
d 103
e 103
i 104
m 104
n 104
o 105
p 105
q 105
s 105
UEL 29
UFL 30
UL 4 163
UL 913 345
ultrasonic power 544
Um 397
uncertified control room apparatus connected
to uncertified field-mounted apparatus 471
unclassified locations 54
Index Terms Links
unclassified or nonhazardous location 117

Underwriters Laboratories, Inc. (UL) 4
Underwriters’ Laboratories Standard No. 698 217
upper explosive limit (UEL) 29
upper flammability limit (UFL) 30
van Wiechen 122 379

variation in igniting current with pressure 37
ventricular fibrillation 549
voltage
effects 303
voltage limiting 422
voltage-current curve 268
volume
effect of 142 143
Westerberg apparatus 84
and MESG values 84
why inspect 400
wind pressure 222
windows 165
wiring 100 341 406
wiring insulation 361
wiring methods 105
Woinsky and Chamlee 69

Index Terms Links
zener diode stabilized supplies 431

zener diode track layout 461
Zone 0 48 63 64
Zone 1 48 63
Zone 2 48 63
Zone 20 62
Zone 21 62 102
Zone 22 62 102
zone classification 61 542

Electrical Instruments in Hazardous Locations PDF

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Electrical Instruments in Hazardous Locations PDF

Uploaded by

Copyright:

Available Formats

ELECTRICAL

All rights reserved.

Printed in the United States of America.

ISBN 10: 0-9792343-1-X ISBN 13: 978-0-9792343-1-6

No part of this work may be reproduced, stored in a retrieval system, or

Library of Congress Cataloging-in-Publication Data - In Progress

The prominence of ISA RP12 in matters of electrical safety was evident in

With growing sophistication about electrical safety also came increased

The second edition recognized the maturation processes of that period.

The general statements of principle supplied by RP12.2 no longer sufficed

It was even more necessary to focus on fundamental principles in the third

Although the third edition recognized these changes in the environment

I have also added and deleted full chapters. Specifically, chapters on

I have included my personal opinions for two reasons:

CHAPTER 4 PRACTICE AND PRINCIPLES OF HAZARD REDUCTION

CHAPTER 9 IGNITION OF GASES AND VAPORS BY ELECTRICAL

CHAPTER 11 DESIGN AND EVALUATION OF INTRINSICALLY SAFE

APPENDIX A DERIVATION OF INITIAL PURGE VOLUME

Growing Interest in the Safety of Electrical Instrument

Such installations demanded reconsideration of the methods used to

When accessibility was not a prime factor in generating interest in new

The Role of Underwriting and Standards Developing

Many users of industrial measuring and control instrumentation devel-

Any interested person or organization may propose a change to the Code.

Underwriters Laboratories, Inc., and Factory Mutual

The NEC frequently refers to “approval” by “the code-enforcing author-

UL was founded in 1894 as a testing laboratory. Its purpose was to exam-

UL has prepared standards for a broad range of safety devices, consumer

Factory Mutual Research Corporation is a division of the Factory Mutual

Until about 1960, the industrial instrument industry only infrequently

During the late 1960s, many users began to require UL listing or FM

The demand by users for UL or FM labeling of industrial instruments

Canadian Electrical Code and Canadian Standards

Committees of manufacturers, inspectors, and users prepare CSA stan-

An Approvals Council for all electrical equipment has authority to accept

CSA is an NRTL and its certification to ANSI standards for electrical

Although onsite inspection services are available for unique installations

Influence of the International Society for Measurement

The Instrument Society of America (ISA) Committee SP12 (formerly 8D-

During this period, SP12 produced three recommended practices. RP12.1,

available to reduce electrical hazards. The revised document also attempts

ISA issued RP12.4, “Instrument Purging for Reduction of Hazardous Area

RP12.2, “Intrinsically-Safe and Non-Incendive Instruments,” published in

In recent years the subcommittee structure has been expanded to include

The aim of each of the five subcommittees is to make available recom-

SP12 members have had continuing influence in the standards activities of

standards for electrical devices in hazardous locations. Several serve on

Recognition of intrinsic safety and nonincendive circuits in the NEC, intro-

Two of the Recommended Practices written by SP12 served as the founda-

Accreditation of ISA as a voluntary consensus standards writing organiza-

National Fire Protection Association

The NEC is only one of hundreds of standards published by the NFPA in

In general, NFPA standards have tended to emphasize recommendations

NFPA committees often provide a forum for consideration of important

National Academies of Science and Engineering

The National Academies of Science and Engineering operate through the

During the past 25 years a number of agencies, including the U. S. Coast

International Electrotechnical Commission

industrialized nations adopt IEC standards, and it is common practice for

A particularly important influence of IEC standards arises from the com-

CENELEC is made up of members of the EC and the European Free Trade

In general, if an IEC standard is available and is technically acceptable,

is better than no standard at all. CENELEC, a regional organization with