Professional Documents
Culture Documents
Electrical Instruments in Hazardous Locations PDF
Electrical Instruments in Hazardous Locations PDF
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
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.
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.
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 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
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.
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.
(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
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.
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
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.
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.
4 Historical Background and Perspective
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.
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
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.
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.
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.
Burklin, C. R., “Safety Standards, Codes and Practices for Plant Designs,”
Chemical Engineering (September, October, November 1972).
Historical Background and Perspective 15
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.
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.
COMBUSTION ZONE
TEMPERATURE
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
20 Combustion and Explosion Fundamentals
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.
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.
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.
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.
Combustion and Explosion Fundamentals 23
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.
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.
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.
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.
26 Combustion and Explosion Fundamentals
Figure 2-3 Ignition Energy versus Quenching Distance (in Calcote data taken as
electrode spacing at which ignition energy was twice minimum)
Combustion and Explosion Fundamentals 27
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
28 Combustion and Explosion Fundamentals
have been carried out with low voltage arc ignition. Some of the results
are discussed in Chapter 9.
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)
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)
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.
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.
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-
Combustion and Explosion Fundamentals 33
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 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.
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.
Figure 2-6 Effect of Pressure on Ignition Energy (data on hydrogen from Table 10, from
Lewis and von Elbe, first edition)
If W = kD2, then
2 2
W1 p D
-------- = ----0- = ------1-
W0 p1
2
D0
2
Combustion and Explosion Fundamentals 37
and
D1 p
------- = ----0-
D0 p1
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.
38 Combustion and Explosion Fundamentals
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
40 Combustion and Explosion Fundamentals
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.
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.
Lewis, B., and G. von Elbe, Combustion, Flames and Explosions of Gases, 2nd
ed., Academic, New York, 1961 (Chapter V).
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.
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.
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.
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.
NFPA 497 and NFPA 499 contain the complete list of materials that have
been grouped by NFPA.
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
46 Classification of Hazardous Locations and Combustible Materials
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
Class I Locations
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 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.
Classification of Hazardous Locations and Combustible Materials 47
Class II Locations
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 Class III, Division 2 location is one in which easily ignitable fibers are
stored or handled, except in the process of manufacture.
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
50 Classification of Hazardous Locations and Combustible Materials
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.
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.
(4) Specify NEC group classification of the material at the point of release.
leg emphasizes its importance and makes more obvious that one approach
to preventing an ignition is to prevent the presence of the oxidizer.
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.
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.
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.
wind at grade, but vapor was not detectable 18 in. (0.45 m) above grade,
30 ft (9.1 m) away.
Division 2 Locations
Class II Locations
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.
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
60 Classification of Hazardous Locations and Combustible Materials
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.
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
Classification of Hazardous Locations and Combustible Materials 61
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.
It is highly probable that the IEC document will be revised to accept these
concepts.
Material Classification
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
64 Classification of Hazardous Locations and Combustible Materials
The second, first proposed by Italian experts, and which appears now to
be gaining acceptance, is as follows:
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.
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.
(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.
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.
68 Classification of Hazardous Locations and Combustible Materials
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
Classification of Hazardous Locations and Combustible Materials 69
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
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
70 Classification of Hazardous Locations and Combustible Materials
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.
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.
(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
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
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
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.
74 Classification of Hazardous Locations and Combustible Materials
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.
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.
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
Classification of Hazardous Locations and Combustible Materials 75
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.
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.
* Some of the concerns expressed by the author have been addressed in NFPA 497-
1997, which was published after this manuscript was completed.
76 Classification of Hazardous Locations and Combustible Materials
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.
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 Hazardous Locations and Combustible Materials 77
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.
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-
78 Classification of Hazardous Locations and Combustible Materials
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
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.
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.
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
80 Classification of Hazardous Locations and Combustible Materials
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
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.
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.
Table 3-4 MESGs and MIC Ratio Data for 12 Selected Materials
Classification of Hazardous Locations and Combustible Materials 81
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.
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.
82 Classification of Hazardous Locations and Combustible Materials
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
84 Classification of Hazardous Locations and Combustible Materials
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.
Classification of Hazardous Locations and Combustible Materials 85
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.
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.
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
Classification of Hazardous Locations and Combustible Materials 87
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.
enough to greatly exceed the 10 hr/yr figure, a plant would likely be vio-
lating EPA or OSHA regulations.
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.
Classification of Hazardous Locations and Combustible Materials 89
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.
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).
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.
References
NFPA 70, National Electrical Code, National Fire Protection Association,
Quincy, MA.
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.
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.
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.
Ex e
Ex i
Ex m
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.
Practice and Principles of Hazard Reduction Practice 105
Ex p
Ex s
Wiring Methods
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.
Outer sheaths of cables that are not laid in earth or sand-filled conduits
shall be flame retardant, unless otherwise protected from fire.
Practice and Principles of Hazard Reduction Practice 107
Conduit Systems
The rules for conduit systems are similar to those of North American
practice.
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
Zone 1 Locations
Zone 2 Locations
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.
bility of fire or explosion above that which existed before the equipment
was installed.
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.
110 Practice and Principles of Hazard Reduction Practice
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.
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.
Practice and Principles of Hazard Reduction Practice 113
Controlling Concentration
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.
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.
120 Practice and Principles of Hazard Reduction Practice
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
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.
Practice and Principles of Hazard Reduction Practice 121
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.
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
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
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
1 ( MTBF 1 ) ( MTBF 2 )
MTBF = ---------------------- = ----------------------------------------------------
-
( λ1 + λ2 ) ( MTBF 1 ) + ( MTBF 2 )
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
124 Practice and Principles of Hazard Reduction Practice
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.
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.).
Practice and Principles of Hazard Reduction Practice 125
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.
Table 4-3 Area Classification Compared with Frequency and Duration of Explosive
Mixture
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.
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 = 0.01 hour for failures that will be removed by short circuit or earth
fault protection.
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:
• Non-sparking apparatus:
log ( MTBE ) = log ( MTBF ) apparatus + P g ( τf = τm + τg )
128 Practice and Principles of Hazard Reduction Practice
• Pressurized apparatus:
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
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
Practice and Principles of Hazard Reduction Practice 129
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.
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.
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
Practice and Principles of Hazard Reduction Practice 131
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.
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.
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.
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.
References
Hickes, W. F., “Intrinsic Safety,” Proceedings of 1960 Symposium on Safety for
Electrical Instrumentation in Hazardous Areas, Instrument Society of Amer-
ica, Pittsburgh.
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.
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.
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.
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.
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.
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.
Explosionproof Enclosures 143
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.
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.
would then have relatively greater effect, through both heat loss and jet
characteristics, than when exit velocity is high.
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.
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.
146 Explosionproof Enclosures
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.
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.
Explosionproof Enclosures 149
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.
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.
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.
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.
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.
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.
152 Explosionproof Enclosures
Effect of Humidity
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.
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.
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.
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.
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.
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-
Explosionproof Enclosures 157
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.
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.
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.
Design Criteria
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
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.
162 Explosionproof Enclosures
CENELEC
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
FM
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
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
Explosionproof Enclosures 165
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
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 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
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.
168 Explosionproof Enclosures
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
Permissible Gap-mm
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
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
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.
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
Class 2 fit ≤ 20 7
Class 3 fit ≤ 20 6
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
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.
Explosionproof Enclosures 171
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.
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
Special Fasteners
Marking
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.
Explosionproof Enclosures 173
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.
The material in this section is based on the requirements of the NEC and
the CEC.
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.
have a softening or melting point 20K higher than the maximum operating
temperature of the component.
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.
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.
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.
178 Explosionproof Enclosures
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., “Ignition in a Transient Turbulent Jet of Hot Inert Gas,” Com-
bustion and Flame, 19, 1972, pp. 187–195.
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.
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
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.
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.
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.
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
188 Reduction of Hazard by Pressurization
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.
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.”
Type Y Pressurization
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.
Markings
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)
192 Reduction of Hazard by Pressurization
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.
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
194 Reduction of Hazard by Pressurization
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.
(2) The system must prevent ingress of flammable material from outside
the enclosure if the enclosure is in a classified location.
The type and magnitude of internal release is defined under normal and
abnormal conditions. “Normal” and “abnormal” conditions refer to the
Reduction of Hazard by Pressurization 195
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.
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, Div. 1 None None None
Class I, Class I, Div. 2 None Z None
Division 2 Nonclassified Z X Inert
Class I, Div. 1 None None None
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 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
Reduction of Hazard by Pressurization 197
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.
Analyzer Rooms
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.
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.
Figures 6-2, 6-3, and 6-4 show the decision trees from which Table 6-2 was
derived.
200 Reduction of Hazard by Pressurization
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)
Reduction of Hazard by Pressurization 201
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.
Static Pressurization
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.
(3) Purging precedes the subsequent connecting of the supply to the elec-
trical apparatus in the opened enclosure.
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.
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%.
Marking
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,
212 Reduction of Hazard by Pressurization
panel components that interrupt smooth flow, panel leakage, and exhaust
port location.
Reduction of Hazard by Pressurization 213
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.)
They calculate the exhaust port area from the orifice equation:
Q
A = ------------------
K 2gh
where
A = effective port area (ft2)
214 Reduction of Hazard by Pressurization
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.
Reduction of Hazard by Pressurization 215
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.
Lee, R. H., “Fuses,” Electrical Safety Practices, ISA Monograph No. 112,
Instrument Society of America, Pittsburgh, 1969, pp. 58–60.
All three methods of hazard reduction impede contact between the source
of ignition and the flammable atmosphere and/or quench any incipient
flame.
(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.
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.
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.
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
Encapsulation, Sealing, and Immersion 221
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.
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.
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.
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.
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.
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.
dC i C o – C i πd 2
-------- = Δ ----------------- ---------
dt V0 l 4
226 Encapsulation, Sealing, and Immersion
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.
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.
Encapsulation, Sealing, and Immersion 227
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.
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.
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.
Encapsulation, Sealing, and Immersion 229
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.
230 Encapsulation, Sealing, and Immersion
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.
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
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
For several typical devices the required grade of seal is shown in Table 7-3
assuming 16-hr diffusion half-time.
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.
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.
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
Flow Testing
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
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.
234 Encapsulation, Sealing, and Immersion
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.
One could derive many other test procedures by combining or altering the
methods noted above. All ignition tests performed at the most easily
Encapsulation, Sealing, and Immersion 235
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.
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:
Leakage Tests
(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).
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.
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.
238 Encapsulation, Sealing, and Immersion
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
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.
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.
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.
240 Encapsulation, Sealing, and Immersion
Swiss Standards
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.
Encapsulation, Sealing, and Immersion 241
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.
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.
– 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
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
Encapsulation, Sealing, and Immersion 243
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.
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-
244 Encapsulation, Sealing, and Immersion
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-
Encapsulation, Sealing, and Immersion 245
ever, many devices are now potted or encapsulated using epoxies, PVC, or
other plastics which do not require additional support.
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.
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.
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.
Encapsulation, Sealing, and Immersion 247
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.
References
Fowle, F. E., Smithsonian Physical Tables, Smithsonian Institution, Washing-
ton, D.C., 1933.
Seals and Seal Testing Procedure, Mil-S-8484 (USAF), June 25, 1954
(obsolete).
Test Methods for Electronic and Electrical Parts, Method 112C, Seal, MIL-
STD-202 F.
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.
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.
Internal fans shall comply with the clearance requirements and material
restrictions for external fans, as stated in the general requirements stan-
Increased Safety, Type of Protection e 257
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.
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.)
Increased Safety, Type of Protection e 259
Figure 8-1 Temperature Rises During a Short Circuit Test - Class F Insulation
and Uninsulated Rotor Bars
260 Increased Safety, Type of Protection e
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.
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-
Increased Safety, Type of Protection e 261
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.
Batteries
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.
262 Increased Safety, Type of Protection e
Resistance Heaters
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.
(3) A protective system, separate from any control system, which discon-
nects the heater from the supply if the temperature exceeds a preset
value.
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.
Olenik, H., Rentzch, H., and W. Wettstein, Explosion Protection Manual, 2nd
edition, Brown, Boveri & Cie, Mannheim, 1983.
264 Increased Safety, Type of Protection e
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
(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.
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
Ignition of Gases and Vapors by Electrical Means 269
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.
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.
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.
and initial current that are incompatible with the arc characteristics of the
electrode material.
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.
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.
The electrodes did not remove energy during the initial growth of the
flame sphere. Because there are only small thermal losses from the spark
Ignition of Gases and Vapors by Electrical Means 275
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.
276 Ignition of Gases and Vapors by Electrical Means
Figure 9-5 Bureau of Mines High-Voltage Test Apparatus; (a) Apparatus for
Capacitance > 100 pF; (b) Apparatus for Capacitance < 100 pF
Ignition of Gases and Vapors by Electrical Means 277
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)
278 Ignition of Gases and Vapors by Electrical Means
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.
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.
Precautions in Testing
Although the testing apparatus used in different laboratories differs in
detail, all investigators observe the same precautions.
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.
geometry or contamination of the contact surface has not altered the igni-
tion capability of the test apparatus.
(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.
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.
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
284 Ignition of Gases and Vapors by Electrical Means
first ignition for each value of the variable. At least 16 runs must be made
for each value of the variable.
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.
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.
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.
Figure 9-10b Relation Between Minimum Igniting Voltage and Capacitance (Widginton)
Ignition of Gases and Vapors by Electrical Means 289
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
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.
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
292 Ignition of Gases and Vapors by Electrical Means
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.
the certification laboratory with all other factors adjusted to maximize the
probability of an explosion.)
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
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.
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.
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.
Ignition of Gases and Vapors by Electrical Means 301
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.
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.
Ignition of Gases and Vapors by Electrical Means 303
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.
(3) Esters are slightly more difficult to ignite than the corresponding
hydrocarbons.
308 Ignition of Gases and Vapors by Electrical Means
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.
Ignition of Gases and Vapors by Electrical Means 309
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)
310 Ignition of Gases and Vapors by Electrical Means
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.
Ignition of Gases and Vapors by Electrical Means 311
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.
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.
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.
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-
Ignition of Gases and Vapors by Electrical Means 313
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
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.
(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:
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.
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 9-16a Effect of Shunt Elements (Break-Flash No. 2, CH4-Air, from SMRE
No. 106)
Ignition of Gases and Vapors by Electrical Means 319
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.
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.
322 Ignition of Gases and Vapors by Electrical Means
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.
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
Ignition of Gases and Vapors by Electrical Means 325
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.
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
326 Ignition of Gases and Vapors by Electrical Means
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.
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
Ignition of Gases and Vapors by Electrical Means 327
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.
328 Ignition of Gases and Vapors by Electrical Means
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 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.
Ignition of Gases and Vapors by Electrical Means 331
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.
Ignition of Gases and Vapors by Electrical Means 333
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.
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.
References
“The Electric Ignition of Firedamp,” Safety in Mines Research Board,
Paper No. 20, 1926.
Lewis, B., and G. von Elbe, Combustion, Flames, and Explosions of Gases, 2nd
ed., pp. 323 ff, Academic, New York.
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.
Vogt, G., “Die Schutzart Eigensicherheit,” Gluckauf, 19 Dec. 1969, 104 (26)
pp. 1193–1206.
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.”
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.
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 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.
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
344 Intrinsically Safe and Nonincendive Systems
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.
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.
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.
(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.
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.
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.
Intrinsically Safe and Nonincendive Systems 349
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.
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
Intrinsically Safe and Nonincendive Systems 351
Definitions
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.
Intrinsically Safe and Nonincendive Systems 353
(1) Category ia apparatus will not cause ignition when the maximum per-
mitted values are applied to its terminals:
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
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
Dielectric Strength
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.
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.
Encapsulation
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.
Table 10-1 shows values of separations from which the required separa-
tion can be determined in the following specific instances:
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.
360 Intrinsically Safe and Nonincendive Systems
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 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.
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.
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
Fuses
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-
Intrinsically Safe and Nonincendive Systems 363
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)
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.
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
Piezoelectric Devices
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.
Resistors are presumed to fail to any value between open and short circuit.
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.
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.
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.
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.
Blocking Capacitors
Infallible Wiring
Diode safety barriers shall also be routinely tested to verify proper opera-
tion of each component and to verify the fuse resistance.
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
372 Intrinsically Safe and Nonincendive Systems
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.)
(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.
Intrinsically Safe and Nonincendive Systems 373
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.
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
374 Intrinsically Safe and Nonincendive Systems
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.
Hickes-Brown Approach
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.
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.
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.
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
378 Intrinsically Safe and Nonincendive Systems
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.
Io = 0.1 A
Po = 0.001
a = 13
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.
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.
380 Intrinsically Safe and Nonincendive Systems
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.
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.
382 Intrinsically Safe and Nonincendive Systems
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
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.
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.
384 Intrinsically Safe and Nonincendive Systems
Figure 10-4 Ignition Characteristics, IEC Apparatus with Cadmium and Tin
Disks, Capacitive Circuits, 8.3% Methane-Air
386 Intrinsically Safe and Nonincendive Systems
Figure 10-5 Ignition Characteristics, IEC Apparatus with Cadmium and Tin
Disks, Capacitive Circuits, 22% Hydrogen-Air
Intrinsically Safe and Nonincendive Systems 387
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.
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.
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
Intrinsically Safe and Nonincendive Systems 393
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.
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.
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.
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.
396 Intrinsically Safe and Nonincendive Systems
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.
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.
• 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.
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.
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.
400 Intrinsically Safe and Nonincendive Systems
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.
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:
(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?
(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?
Intrinsically Safe and Nonincendive Systems 403
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.
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
404 Intrinsically Safe and Nonincendive Systems
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.
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
406 Intrinsically Safe and Nonincendive Systems
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-
Intrinsically Safe and Nonincendive Systems 407
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.
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
Intrinsically Safe and Nonincendive Systems 409
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.
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
410 Intrinsically Safe and Nonincendive Systems
References
British Standard 1538:1949, “Intrinsically Safe Transformers for Bell Sig-
nalling Circuits.”
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.
Burgoyne, J. H., “Safety in Great Britain,” ISA Monograph No. 111, pp. 52-
57, ISA, Pittsburgh, 1969.
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.
Burgoyne, J. H., “Safety in Great Britain,” ISA Monograph No. 112, pp.
119-131, ISA, Pittsburgh, 1969.
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.
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.
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.
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.
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
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.
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.
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-
Design and Evaluation of Intrinsically Safe Apparatus 421
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
(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.
(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.
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
428 Design and Evaluation of Intrinsically Safe Apparatus
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.
(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.
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
Design and Evaluation of Intrinsically Safe Apparatus 429
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.
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.
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.
Figure 11-3 Relation Between Minimum Igniting Current and Stabilized Voltage
for Shunt Zener Diode Stabilized Supplies
432 Design and Evaluation of Intrinsically Safe Apparatus
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]
Design and Evaluation of Intrinsically Safe Apparatus 433
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
where
W = stored energy in joules
L = inductance in henries
i = current in amperes
434 Design and Evaluation of Intrinsically Safe Apparatus
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
Design and Evaluation of Intrinsically Safe Apparatus 435
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
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.
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.
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.
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.
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
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.
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
Design and Evaluation of Intrinsically Safe Apparatus 441
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.
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.
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
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.
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.
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-
Design and Evaluation of Intrinsically Safe Apparatus 445
tance on the output side of the zener has an effect on cable length that is
calculatable, but no generalization is possible.
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.
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.
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-
446 Design and Evaluation of Intrinsically Safe Apparatus
tem. Any cable may be used that has lower C and higher L and R, than the
one listed in the table.
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.
Design and Evaluation of Intrinsically Safe Apparatus 447
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
448 Design and Evaluation of Intrinsically Safe Apparatus
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
R = 15 − 150 Ω
L = 0.4 − 1 mH
Terminators:
R-C elements with 90–100 Ω, 0–2.2 μF, one at each end of the trunk
cable
Vmin = 9 V
Inominal = 10 mA
Li < 10 μH
Ci < 5 nF
Design and Evaluation of Intrinsically Safe Apparatus 449
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.
Terminators:
R-C elements with 90–100 Ω, 0–2.2 μF; one at each end of the trunk
cable
450 Design and Evaluation of Intrinsically Safe Apparatus
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.
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.
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.
454 Design and Evaluation of Intrinsically Safe Apparatus
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 added components are shown in Figure 11-9; the values shown for
each is the value after fault analysis.
Design and Evaluation of Intrinsically Safe Apparatus 455
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
Design and Evaluation of Intrinsically Safe Apparatus 457
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
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
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.
Design and Evaluation of Intrinsically Safe Apparatus 459
Component Ratings
Controlled Spacing
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 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 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.
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.
(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.
(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.
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.
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.
(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
Design and Evaluation of Intrinsically Safe Apparatus 467
(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.
(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.
468 Design and Evaluation of Intrinsically Safe Apparatus
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
Design and Evaluation of Intrinsically Safe Apparatus 469
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.
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.
470 Design and Evaluation of Intrinsically Safe Apparatus
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.
cally safe circuits or apparatus, but, in the author's opinion, this is too
restrictive a translation of reasonable intent into a rule.
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.
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.
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.
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.
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.
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.
Design and Evaluation of Intrinsically Safe Apparatus 475
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 11-14 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.
Design and Evaluation of Intrinsically Safe Apparatus 477
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.
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.
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.
Sharing diodes as in the star-connected barrier of sketch (f) limits the volt-
age across the load to the voltage of one barrier.
Active Barriers
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.
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.
Design and Evaluation of Intrinsically Safe Apparatus 483
Overvoltage/Overcurrent Protection
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.
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.
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.
(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
486 Design and Evaluation of Intrinsically Safe Apparatus
(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.
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.
488 Design and Evaluation of Intrinsically Safe Apparatus
(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 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.
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.
References
“Code of Practice on Zener Barriers,” British Industrial Measuring and
Control Apparatus Manufacturers' Association.
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.
Virr, L. E., “Methods for Assessing the Inductive Effects of Cables Used In
Intrinsically Safe Systems,” SMRE Technical Paper P19, Sheffield, 1976.
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.
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.
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).
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.
Figure 12-1 Data from EC Research Program and OSCA-Recommended “No Risk”
Limits
Ignition by Optical Sources 497
(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
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
498 Ignition by Optical Sources
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.
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.
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.
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.
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.
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.
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.
As t approaches infinity
P = Pmin + IminA
P min
- + I min
I = ----------
A
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.
For conditions relevant to the use of optical instruments and fibers, the
important equation is:
P = Pmin + IminA
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.
504 Ignition by Optical Sources
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
Ignition by Optical Sources 505
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.
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.
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.
Prototypical Applications
(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
508 Ignition by Optical Sources
(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.
Ignition by Optical Sources 509
Figure 12-2 Fault Tree for Explosion Initiated by Radiation-Heated Hot Particles
or Surfaces
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.
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
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
512 Ignition by Optical Sources
(2) The systems with some kind of interlock have MTBE values equal to
or greater than those of explosionproof enclosures.
Ignition by Optical Sources 513
(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.
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.
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.
Ignition by Optical Sources 515
(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:
(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
516 Ignition by Optical Sources
(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.
(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.
Ignition by Optical Sources 517
With T4 rating:
References
OSCA Doc. No 94/56.
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).
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.
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
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.
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.
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.
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.
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'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.
Studies of the combustion of dust particles have shown that burning of the
volatile components occurs first. The solid fraction then burns.
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.
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.
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.
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
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.
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.
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.
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.
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.
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.
Division 2
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.
dust cloud. If such occurrences are rare, perhaps once per year, the area
may be classified Division 2.
Dust-Ignition-Proof Enclosures
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.
Pressurization
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.
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
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.
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.
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.
References
C22.2-157, “Intrinsically Safe and Nonicendive Equipment for Use in Haz-
ardous Locations,” Canadian Standards Association, Rexdale.
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
Griffith, W. C., “Dust Explosions,” Ann. Rev. Fluid Mech. 10:93–105, 1978.
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.
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.
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
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.
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.
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
i = 185 ⁄ T mA
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
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
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
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.
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
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.
References
Lee, R. H., "Human Electrical Safety," ISA Monograph No. 110, pp. 27–34,
Instrument Society of America, Pittsburgh, 1965.
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.
Tables 15-1 and 15-2 indicate in approximate terms the intended applica-
bility of enclosures for indoor and outdoor unclassified locations.
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
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.
Degree of Protection by Enclosures 559
Table 15-4 Meaning of IEC Publication 529 IP Code Numerals and Letters
560 Degree of Protection by Enclosures
Table 15-5 Degree of Protection against Contact and Entrance of Solid Foreign Bodies
Degree of Protection by Enclosures 561
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.
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.
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.
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.
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
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
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
– 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
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.
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
If Pt is taken to be 1.1 and Pa is taken to be 1.0, for the extremes of the V/l
range,
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)
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
570 Examples of Derivations
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
------------- dV
182.5 V
- = ( A + B ) sin φ
------- + -----
Kl dφ V o
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
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.
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
Solving for d,
C0 – Ci
ln § --------------------·
© C 0 – C i0¹
d = -------------------------------
V0
ln § -----------------·
© V + V 0¹
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 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.
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
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.
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
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
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
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
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
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
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
inductive circuits
ignition characteristics 387
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
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
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
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
resistance 291
effect of 286
resistance elements 262
resistance temperature detectors 473
resistive circuits
ignition 306
ignition characteristics 384
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
This page has been reformatted by Knovel to provide easier navigation.
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
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
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
This page has been reformatted by Knovel to provide easier navigation.
Index Terms Links
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