Fine Coal: Challenges in

Challenges in Challenges in
Challenges in FINE COAL Processing, Dewatering, and Disposal

FINE COAL
Processing, Dewatering, and Disposal
Edited by Mark S. Klima, Barbara J. Arnold, Peter J. Bethell FINE
Coal mining and preparation have had a long history in the United States
and the world, serving as the engine of growth for many industries. Today,
new sources of energy, increased environmental awareness, and more
stringent regulations from the U.S. Environmental Protection Agency and other
organizations are changing the way coal is found, extracted, and used. As
a result, fine coal cleaning, dewatering, and refuse disposal are now at a
COAL
Processing, Dewatering,
major crossroads.
and Disposal
The increased level of fines, and near-density material in the inferior seams
being mined today, necessitates the development of more efficient fine coal
cleaning devices. This in turn requires improvements in traditional dewatering
techniques to address the need for acceptable moisture levels in plant products.
Moreover, the larger volume of fine refuse being generated, coupled with
harsher disposal regulations, requires upgraded treatment options.
This book is a compilation of information presented at the 2012 Fine Coal

Symposium, sponsored by the Coal Preparation Society of America; the
Pittsburgh Section of the Society for Mining, Metallurgy, and Exploration,
Inc.; and the Pittsburgh Coal Mining Institute of America.
Provided by international coal companies, major research organizations,

technology developers, and industry leaders, the information includes both
general knowledge and in-depth discussion on the current challenges
facing the industry, techniques for designing more efficient plants, and new
cleaning and dewatering technologies. The book is a practical yet cutting- Mark S. Klima
edge resource for plant designers, engineers, and other practitioners, and Barbara J. Arnold
for university students and faculty. Peter J. Bethell
The Society for Mining, Metallurgy, and Exploration,

Inc. (SME), advances the worldwide mining
and minerals community through information
exchange and professional development. SME
is the world’s largest association of mining
and minerals professionals.
!SME_FineCoalPDD_FULL_CV_L2.indd 1 9/10/12 11:18 AM

Challenges in
FINE
COAL
and Disposal
Mark S. Klima
Barbara J. Arnold
Peter J. Bethell
Published by the
Society for Mining, Metallurgy, and Exploration, Inc.
© 2012 by the Society for Mining, Metallurgy, and Exploration. All rights reserved.
Society for Mining, Metallurgy, and Exploration, Inc. (SME)
12999 East Adam Aircraft Circle
Englewood, Colorado 80112
(303) 948-4200 / (800) 763-3132
www.smenet.org
SME advances the worldwide mining and minerals community through information
exchange and professional development. SME is the world’s largest association of
mining and minerals professionals.
Copyright © 2012 Society for Mining, Metallurgy, and Exploration, Inc.

Electronic edition published 2012.
All Rights Reserved. Printed in the United States of America.
Information contained in this work has been obtained by SME from sources believed to be
reliable. However, neither SME nor the authors guarantee the accuracy or completeness of
any information published herein, and neither SME nor the authors shall be responsible
for any errors, omissions, or damages arising out of use of this information. This work is
published with the understanding that SME and the authors are supplying information
but are not attempting to render engineering or other professional services. It is sold with
the understanding that the publisher is not engaged in rendering legal, accounting, or
other professional services. If such services are required, the assistance of an appropriate
professional should be sought. Any statement or views presented here are those of the
authors and are not necessarily those of SME. The mention of trade names for commercial
products does not imply the approval or endorsement of SME.
No part of this publication may be reproduced, stored in a retrieval system, or transmitted

in any form or by any means, electronic, mechanical, photocopying, recording, or otherwise,
without the prior written permission of the publisher.
ISBN: 978-0-87335-363-2
Ebook: 978-0-87335-375-5
Library of Congress Cataloging-in-Publication Data

Challenges in fine coal processing, dewatering, and disposal / edited by Mark S. Klima,
Barbara J. Arnold, and Peter J. Bethell.
pages cm
Includes bibliographical references and index.
ISBN 978-0-87335-363-2 (print) -- ISBN 978-0-87335-375-5 (ebook)
1. Coal washing. I. Klima, Mark S., editor of compilation. II. Arnold, Barbara J., editor of
compilation. III. Bethell, Peter J., editor of compilation.
TN816.C375 2012
662.6’23--dc23
2012035131
Contents
Preface. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ix
Part 1 Perspective on Coal Fines

Milestones in Fine Coal Cleaning Development. . . . . . . . . . . . . . . . . . . . 3
Dealing with the Challenges Facing Global Fine Coal
Processing. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 33
Coal Properties and Challenges for Fines Processing. . . . . . . . . . . . . . . 47
Maximizing Fine Pyrite Rejection at the Arch Coal Leer Plant. . . . . . 67
Part 2 Technology Developments and Plant Installations
StackCell Flotation—A New Technology for Fine Coal
Recovery. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 81
Fine Coal Processing Developments in Anglo American Thermal
Coal South Africa. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 95
Fine Coal Processing—Technical Developments in Australia. . . . . . 123
Part 3 Beneficiation Technologies
Fine Coal Processing with Dense-Medium Cyclones. . . . . . . . . . . . . . 139
Development of the Reflux Classifier. . . . . . . . . . . . . . . . . . . . . . . . . . . . 159
Column and Nonconventional Flotation for Coal Recovery:
Circuitry, Methods, and Considerations . . . . . . . . . . . . . . . . . . . 187
Mechanical Cells for Fine Coal Flotation—An International
Perspective. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 211
Gravity Separators for Ultrafine Coal Cleaning. . . . . . . . . . . . . . . . . . 227
Design and Operating Guidelines for Combined Water-Only
Cyclone and Spiral Circuits. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 241
Part 4 Moisture Reduction and Special Topics
Performance, Operation, and Maintenance Experience of
Coal Ultrafines Filtration with Modern High-Speed
Disc Filters. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 261
Dewatering Fine Coal and Tailings with a Filter Press. . . . . . . . . . . . . 279
Deep Cone Thickener at Lone Mountain Processing Plant . . . . . . . . 293
Centribaric Operation Update. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 303
Belt Filter Press in Coal Tailings Dewatering—A Comprehensive
Economic, Design, and Process Analysis . . . . . . . . . . . . . . . . . . . 309
Fine Coal Drying and Plant Profitability. . . . . . . . . . . . . . . . . . . . . . . . . 329
iii
iv CHALLENGES IN FINE COAL PROCESSING
Nano Drying Technology—A New Approach for Fine Coal

Dewatering. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 345
Development of a Fine Coal Recovery Operation at the
Centralia Mine Coal Slurry Impoundment Structures. . . . . . . 361
Pressurized Fluidized-Bed Combustion Technology for
Fine Coal Utilization. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 377
Index . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 391
Contributors
H. Akbari Randy Corder
Southern Illinois University Arch Coal, Inc.
Department of Mining & Mineral Grafton, West Virginia
Resources Engineering
Carbondale, Illinois Van Davis
Alpha Natural Resources
Zulfiqar Ali Bristol, Virginia
Virginia Tech
Department of Mining & Minerals Greg DeHart
Engineering Arch Coal, Inc.
Blacksburg, Virginia Teays Valley, West Virginia
Barbara J. Arnold G.J. de Korte

PrepTech, Inc. CSIR
Apollo, Pennsylvania Pretoria, South Africa
Peter J. Bethell Andy Dynys

Marshall Miller and Associates Taggart Global, LLC
Taggart Global, LLC Canonsburg, Pennsylvania
Arch Coal, Inc. T. Estes
Virginia Beach, Virginia Decanter Machine, Inc.
R. Bland Johnson City, Tennessee
Nano Drying Technologies, LLC Jeff Euston
Beckley, West Virginia FLSmidth
Robert Bratton New South Wales, Australia
Virginia Tech Michael Fenger
Department of Mining & Minerals PFBC–Environmental Energy
Engineering Technology, Inc.
Blacksburg, Virginia Monessen, Pennsylvania
Cory Chafin Matt S. Fenzel
Arch Coal, Inc. Phoenix Process Equipment Company
Teays Valley, West Virginia Louisville, Kentucky
Lance Christodoulou Bruce Firth
Eriez Flotation Division CSIRO
Erie, Pennsylvania Brisbane, Queensland, Australia
vi CHALLENGES IN FINE COAL PROCESSING
J. Franklin Gerald Luttrell

Walter Energy Virginia Tech
Brookwood, Alabama Department of Mining & Minerals
Engineering
K.P. Galvin Blacksburg, Virginia
University of Newcastle
Centre for Advanced Particle Processing B. McDaniel
and Transport Nano Drying Technologies, LLC
Newcastle Institute for Energy and Beckley, West Virginia
Resources
Callaghan, New South Wales, Australia M.K. Mohanty
Southern Illinois University Carbondale
B.K. Gupta Department of Mining & Mineral
Lone Mountain Processing, Inc. Resources Engineering
St. Charles, Virginia Carbondale, Illinois
Jürgen Hahn Robert Moorhead
Bokela GmbH FLSmidth Krebs
Karlsruhe, Germany Salt Lake City, Utah
C. David Henry Graham O’Brien
Coalview Recovery Group, LLC CSIRO
Somerset, Pennsylvania Brisbane, Queensland, Australia
Rick Honaker Mike O’Brien
University of Kentucky CSIRO
Department of Mining Engineering Brisbane, Queensland, Australia
Lexington, Kentucky
Dave Osborne
Michael Kiser Xstrata Coal Queensland
Virginia Tech Brisbane, Queensland, Australia
Department of Mining & Minerals
Engineering Esko Polvi
Blacksburg, Virginia PFBC–Environmental Energy
Technology, Inc.
Jaisen Kohmuench Monessen, Pennsylvania
Eriez Flotation Division
Erie, Pennsylvania G. Prat
Técnicas Hidráulicas S.A.
Deborah A. Kosmack Mungia, Spain
CONSOL Energy, Inc.
Research & Development W. Schultz
South Park, Pennsylvania Decanter Machine, Inc.
Johnson City, Tennessee
CONTRIBUTORS vii
Fred Stanley Asa Weber

Alpha Natural Resources FLSmidth
Bristol, Virginia Midvale, Utah
Chris Swanepoel Keith Wilkes

Anglo American Thermal Coal FLSmidth
Witbank, South Africa Rugby, United Kingdom
Larry Watters Eric Yan

Taggart Global, LLC Eriez Flotation Division
Canonsburg, Pennsylvania Erie, Pennsylvania
Preface
new sources of energy, increased environmental awareness, and more stringent
regulations from the U.S. Environmental Protection Agency and other organi-
zations are changing the way coal is found, extracted, and used. As a result, fine
coal cleaning, dewatering, and refuse disposal are now at a major crossroads.
techniques to address the need for acceptable moisture levels in plant prod-
ucts. Moreover, the larger volume of fine refuse being generated, coupled with
Symposium, sponsored by the Coal Preparation Society of America; the Pitts-
burgh Section of the Society for Mining, Metallurgy, and Exploration, Inc.;
and the Pittsburgh Coal Mining Institute of America.
general knowledge and in-depth discussion on the current challenges facing the
industry, techniques for designing more efficient plants, and new cleaning and
dewatering technologies. The book is a practical yet cutting-edge resource for
plant designers, engineers, and other practitioners, and for university students
and faculty.
ix
Perspectives on
Coal Fines 1
Milestones in Fine Coal
Cleaning Development
Dave Osborne
ABSTRACT
Effective cleaning of fine coal (i.e., via the removal of non-coal material and subse-
quent reduction in moisture to an acceptable level) is mostly dependent on favor-
able economics (i.e., value of the product component obtained), capability, and
subsequent performance of the preparation equipment, and the extent to which
beneficiation of the total coal can be optimized.
The success of such an approach will be influenced by many other factors, not
the least being the proportion of fines in the raw coal. Hence, as mining has become
progressively more mechanized, the proportion of fines has increased and the jus-
tification for fine coal cleaning has tended to increase. However, the conundrum
associated with more fines is the added risk of increased moisture and the accompa-
nying need for improved and cost-effective dewatering equipment.
In fine coal cleaning, gravity methods such as jigs, tables, spirals, upward cur-
rent separators based on the forerunning Stokes classifier, as well as autogenous
cyclones are featured. Other non-gravity methods include froth and column
flotation and fines aggregation approaches including (oil) agglomeration and bri-
quetting. Perhaps the most desirable approach, dry cleaning, with its tremendous
appeal of avoiding the moisture issue altogether, is almost unachievable via conven-
tional methods, but some sorting, magnetic, and electrical techniques have offered
some promise over the years but are still far from being commercialized.
Not surprisingly, much of the development that has occurred in fine coal cleaning
has taken place with higher-value coal, often metallurgical, and almost all techniques
for both cleaning and dewatering have undergone some form of improvement or
development during the past decade as a result of better design or improved materials
of construction and wear resistance. Other focus areas have been process control and
monitoring, automation, and of course sampling and analysis.
4 Perspective on Coal Fines
This chapter reviews the development pathway for fine coal beneficiation and
covers all aspects in a more general way, leaving the details for others to describe in
subsequent chapters.
INTRODUCTION
To provide some context for this chapter, several pertinent questions are first
addressed:
1. What is fine coal cleaning? The term coal fines is almost always taken to
mean raw or cleaned coal below 1 mm from which ultrafine material
sized below ~50 µm has been removed by some form of desliming
(classification) step. Cleaning of this “fraction” can be achieved by
application of any one of a variety of separators, predominantly in
water, to recover a valuable component(s). In some cases, this outcome
is achieved via a single stage and others in two or more stages.
2. What are its main drivers? Economics is the main driver and this can
be influenced by the prevalent financial climate. In many cases in past
eras, there has been no such justification for recovering “value” from
this fraction, and this coal was wasted only to be recovered at a later
stage when the financial climate became favorable. Other drivers can
include sustainable development, environmental impact, and new
technical developments.
3. When did fine coal cleaning start? Probably the earliest commercial
developments date back to the flowing-film concentration devices
introduced in the early 19th century, but serious coal cleaning whereby
a fine coal cleaning circuit was designed into a coal preparation plant
occurred in the 1930s. By this time, most of the gravity concentra-
tion devices (i.e., jigs, tables, cyclones, etc.) were in use, many having
emerged from mineral processing applications for which they were
originally developed.
4. What has its development pathway been? Fine coal beneficiation whereby
both cleaning and dewatering were incorporated probably emerged
as an integral part of flowsheet design during the 1940s and 1950s.
Knowledge transfer from the minerals industry applications undoubt-
edly contributed much to the development of both cleaning and dewa-
tering equipment, but there were some processes specifically introduced
for coal cleaning such as the Dutch State Mines (DSM) dense-medium
cyclone and water-only cyclone designs, and more recently the reflux
classifier albeit with some mineral processing influence.
Milestones in Fine Coal Cleaning Development 5
In a previous chapter on this subject (Osborne 1986), it appeared that the

major technological thrust was in the direction of fines recovery and beneficia-
tion. This was interesting at that time because reference to similar publications
30 years beforehand emphasized an urgent need for improving fines treatment
(Arrowsmith 1957) and suggested the following reasons:
• Increased generation of fines resulting from
ǷǷ Greater mechanization and higher productivity in mining, and
ǷǷ Resultant loss in selectivity creating greater contamination
• Increased cost of fine coal cleaning
• Handling and blending problems associated with fine coal
In the 1950s, the following solutions were suggested:
• Use of selective froth flotation for “metallurgical coals”
• Use of oil agglomeration plus centrifugal dewatering
• Use of pressure filters for dewatering ultrafine coal
• Development of higher “g” centrifugal dewatering machines
• Closure of coal preparation plant water circuits
• Greater measurement and balancing of plant circuits
Most of this still applies, and in the 1980s, the trend was more directed toward
“surface chemistry” dependent separation and away from “gravity” dependent.
But in the past decade, elutriation has been more prevalent with teetered-bed
variants together with column flotation, occurring simultaneously with the
development of more efficient dewatering equipment via higher “g” forces
and higher pressures. The other continuing trend has been toward improv-
ing ancillaries such as wear resistance and handling via introduction of better
materials and design, increasing component interchangeability, and maximiz-
ing operational life and utilization. This pathway has also been accompanied
by improved plant design capability, process control, and management of data
(Lein 2011).
When IBM produced its first personal computer (PC) in the early 1980s,
a major transition occurred in terms of plant design. However, the PC could
not handle complicated drawings, and mini-computers evolved as computer-
aided design became a greater part of the engineering process. As the decade
progressed and the XT was introduced, greater use of the computer was made
in engineering and plant control.
In the mid-to-late 1980s, another revolution occurred. The process engi-
neering business suffered a major transition and the workload diminished
New Materials, etc. Intensified

Polyurethane/
Fibreglass Spirals
Mechanised
Dewatering Mining
Cast Basalt
Linings
Galvanised Ceramic
Settling Cyclones
Basket Structures Lining/Insert
Cones and Chutes
Centrifuges Membrane
Micro-wave
Press
Cyclones Dryer
Plate &
Birth of Modern
Frame Press Laminar Thickener Deep Cone Hyperbaric
Geology (Agricola) Centribaric
Thickeners Filter Centrifuge
1800 1850
1500 1890 1900 1910 1920 1930 1940 1950 1960 1970 1980 1990 2000 2010 2020
First Jigs Earliest Early Trough Baum Water Only

Sluices Coal Jigs and Barrel Jigs Cyclones
and Troughs (ex Minerals) Washers
Heavy Medium Reflux
Cyclones Classifier
Magnetic
Coal
Diester Separators Total
Shaking
Table Remote
Tables FGX Air
PLC Plant Control
Table
Spirals Control
Reichart Nuclear Pit-to-Port
Cone Density Infn Mgt
SCADA
Pneumatic (Air) Gauges System
Tacub and
Cleaning Jigs and Tables
Batac Jigs Computer-Aided
On-line Pit-to-Port
Moisture Infn Mgt
Design (CAD)
Other—control, etc. Analyser System
Figure 1 Development pathway for fine coal beneficiation
significantly with the result that coal preparation engineering work became
scarce and many experts migrated to other industries. However, with the intro-
duction of the Pentium chip, the engineering activity and storage of knowledge
and experience was transformed. By the mid-1990s, these mini-computer-
based systems became too expensive against the advancement of PC operating
systems, and with it the number of people needed to do engineering work of all
kinds was drastically reduced.
Finally, into the late 1990s and into this century, the transformation has
virtually become the norm and has translated into all facets (i.e., plant design,
construction, scheduling, supervisory control and data acquisition [SCADA]),
and perhaps the greatest benefit from all of this has been seen in the improved
treatment and cost of fine coal beneficiation.
CONTRIBUTING TECHNOLOGIES
Figure 1 is a time chart illustrating the pathway for the development of fine
coal beneficiation, which includes key milestones that have occurred during
this period.
The intent of this chapter is to set the scene for the comprehensive treatise
on fine coal beneficiation that follows in this book. The sections covered in
this chapter provide details of the various types of beneficiation and associated
equipment, including related aspects; and, for convenience, these are subdi-
vided into the following:
• Water-based separation
• Chemical separation
• Dry separation
• Magnetic and electrical separation
• Dewatering
• Computers and process control
• Materials science and engineering
Water-Based Separation
The earliest forms of separation employed flowing-film, sedimentation, or jig-
ging mechanisms to effect the separation.
Coal Jig
The use of jigging to create stratification and thereby clean coal is probably
among the oldest of coal cleaning techniques. Jigs in their simplest form have
been around from as early as the 15th century, when hand-operated versions
consisted of a framed sieve manually pulsed in a water tub. The first mecha-
nized coal jigs were recorded from as early as the 1840s, after which evolution
over the next 50 years focused on the jigging action, which changed from
moving a sieve frame to using a piston, then a diaphragm to pulse the medium
through a static sieve. In all cases, the raw coal was fed to the jig either after slur-
rying in water or after a step to “deslime” the feed, but in most cases there was
no further attempt to recover or beneficiate the fines.
By 1891, the father of the modern-day jig, the Baum jig, was developed
in Germany by Fritz Baum. This jig used air to displace water from one sealed
chamber, forcing the water through the jig bed in an adjacent chamber. The
use of air allowed for a more versatile control on the jig cycle over the older
diaphragm- or piston-driven Jeffrey and Harz jigs. With the introduction of
the Baum jig to Great Britain’s coal industry in 1903 and the Americas in 1928,
both the capacity and utilization of the jig rapidly grew until the 1950s when it
was felt that the capacity of the Baum jig had been reached with a 2.5-m-wide
unit. Wider units later appeared following the inclusion of the rotary air valve
that allowed the bed width of the Baum jig to be increased to 5 m (Lyman
1994).
In 1958, Takakuwa of Japan invented the Tacub jig (later replicated as the
Batac) by placing the air header directly below the bed, allowing bed widths of
up to 6 m. It was also in the 1950s that the innovation of using a mineral feld-
spar as “ragging” allowed the processing of finer sized coal by means of both of
these jig types. The modern-day versions of these jigs are capable of processing
material down to ~0.1 mm; however, there is a rapid deterioration in separation
efficiency below 0.5 mm, and this approach is not often employed when the
cleaned fines have little or no commercial value (Osborne 1988a). The modern
Batac jig is capable of processing up to 600 t/h of –10 mm + 0.5 mm coal.
More recent developments more specifically applied to fines treatment have
been the so-called in-line pressure jig and another variant, the Kelsey jig, the
latter of which incorporates centrifugal motion (Falconer 2003). Both employ
a ragging material and, although developed for mineral applications, so-called
“slurry jigs” have been seriously considered for fine coal cleaning applications
with various degrees of success reported (Bhattacharya 2009).
Troughs, Spirals, and Cones

Trough washers were the forerunners of a family of flowing-stream separators
that now include spirals and tables. All rely on various forms of hindered set-
tling and stratification. Early designs quite commonly used in Europe in the
late 1930s include the Hoyois trough (see Figure 2) developed by a Belgian,
L. Hoyois (1935), and the Rheolaveur trough washer. Most treated “fines” sized
below ~6 mm, but clean coal required dewatering via settling cones and fine
screens or later by early versions of basket centrifuges. With the introduction
of mechanical flotation cells, the ultimate fines (–0.5 mm) were recovered and
also cleaned. This was the forerunner of the intermediate cleaning step now
occupied by spirals and teetered-bed separators.
Spirals separators have been employed in the minerals processing industry
since their invention by Ira Boyd Humphreys in 1943. They were originally
used in the concentration of mineral sands where many other variants emerged,
including the Reichert cone. The earliest spirals were formed by cast-iron or
molded concrete sections and were considered excessively heavy in regard
to their limited unit capacity. In the 1960s, this changed dramatically with
the adoption of polyurethane-coated fiberglass as the preferred construction
materials together with “nesting” or stacking of two or three spirals on one
supporting pole. However, it was not until the early 1980s that this technology
was introduced to the coal industry. Since this introduction, their application
has grown as the preferred form of separation for the “intermediate” size range
of ~2.5 to 0.25 mm. Further evolution, driven by the coal application, has
Settling Cone
Feed Water to Hoyois Troughs
Slurry Screens
Raw Slack
Elevator
Hoyois Slack Screens

Raw Slack Troughs
Feed and Barrels Clean Slack
Re-wash
Shale
Raw Slack Shale
Sump Sump
Overflow
Overflow
Main Pump
Sump
Figure 2 Hoyois trough washer, circa 1935 (Colliery Eng. Ltd., later Head Wrightson)
included many changes to the traditional mineral spirals, specifically improv-

ing their performance in coal applications. These changes include a decrease in
pitch, to decrease velocity and increase residence time, and enlargement of the
spiral diameter to increase processing capacity. Also, the flowsheet configura-
tions offered have resulted in compounded improvement in both recovery and
efficiency (Luttrell et al. 2007).
They are probably still the preferred technology in modern coal prepara-
tion plants due to their ease of operation, low capital and operational costs, and
inherently low coal losses (Bethell 2007; see Figure 3). The modern-day spiral
has up to four starts, producing a separation with a density of above 1.8 for par-
ticle sizes below 3.2 mm (Palowitch et al. 1991b). However, particle sizes less
than 0.25 mm still suffer from poor performance (Dennis 1998).
Shaking Tables
The modern wet shaking tables were originally derived from the continuous
belt concentrators used in the mid-1800s. In these early separators, the belt
concentrator used a flowing film to wash the light fraction (coal) down a belt,
leaving behind the minerals and shales. The first step toward current shaking
Single Stage Primary-Secondary Primary-Secondary Primary-Secondary

(Midds Retreat) (Clean/Midds Retreat) (Clean/Midds Retreat -
Secondary Midds Recycle)
Feed Feed Feed Feed
Refuse Clean Refuse Clean Refuse Clean Refuse Clean
Source: After Bethell 2011.
Figure 3 Alternative spiral configurations
tables was taken by Hartwig in 1860 with the development of a vibrating end-
less belt (based on the mineral vanner). Further development by Linkenback,
with the circular stationary table—Anaconda-Evans with multiple round table
and Campbell with the bumping table—brought the dawn of water-based shak-
ing tables to the coal preparation industry, with the first bumping table instal-
lation in 1890. It was between 1896 and 1898 that Arthur Wilfley developed
the first differential motion table, which is still used for cleaning fine coal. The
Massco table was the name given to the original Wilfley table that was specifi-
cally designed for coal. Since then, many brands of shaking tables, all originally
designed for concentrating ores, have been marketed for coal cleaning. In 1918,
the most renowned coal cleaning table was introduced by Deister-Overstrom.
Since then, many hundreds of installations have been built, and probably the
most significant changes to the Deister table have been the incorporation of
multiple decks to a single unit, significantly increasing the capacity per floor
area, and also the improved material used for the table surface. By 1976, there
was an estimated 2,690 Deister tables operating in the United States. The
modern coal Deister tables are capable of processing feed between 0.15 mm
and 25 mm (although 15 mm is the norm) with capacities of up to 13.6 metric

tons per hour per deck (Palowitch et al. 1991a), but new installations are rare.
Elutriation “Teeter-Bed” Separators

Sometimes called upward current washers, these separators date back to the
mid-19th century, an early version being the spitzkasten, originally designed
to work in conjunction with feldspar jigs in Germany. They comprised a series
of boxes of increasing width within which water remained more or less station-
ary and the feed slurry flowed across each box. Coal particles of decreasing
density and size then settled out from box to box, and a valve enabled removal
from each spigot. Another upward current separator was the Draper washer,
which emerged in the early 20th century (Know 1918). Several individual
units were assembled in a battery so that capacities of up to ~50 metric tons/h
were possible. This led to numerous other similar devices in the United King-
dom, France, and the United States. Feed sized up to ~10 mm was cleaned in
these separators, which were the forerunner for a number of U.K. Coal Board’s
Bretby separators, including the Hirst “fine coal cone” and the “hybrid cone”
that ultimately led to the adoption of hindered-settling classifiers typified by
the (Stokes) hydrosizer and the Linatex classifier. Both used a chamber fitted
with a teeter plate to ensure an evenly distributed upward current creating what
was eventually to become known as a “teetered bed.” The raw coal feed is usu-
ally 2 mm upper size at a concentration of up to ~50% solids by mass (Hillman
et al. 2003).
These units became quite widely used in U.K. plants during the 1980s,
particularly in coal recovery plants used for cleaning up old waste piles, and
eventually migrated to Australia, the United States, and South Africa. The most
recent advancement in upward current separation is the reflux classifier (Galvin
et al. 2010), which combines three separation technologies (i.e., a lamella set-
tler, an autogenous [natural] dense medium, and a fluidized bed). It comprises
inclined lamellar channels that deliver better hydraulic conditions compared to
conventional hydrosizer types.
Most recent work suggests it should be capable of efficient separation at
coarser sizes of up to 5 mm by altering the spacing of the lamellar plates, but
normally it is used to treat 2 mm to >0.03 mm. The reflux classifier is also being
regarded as an attractive alternative to spiral separation because of its small
footprint and ease of control. Perhaps there is merit in having another look at
the separation mechanism of units like the Draper washer and its contempo-
raries; with the benefit of new material and a control system, these could also
emerge again.
Courtesy of FLSmidth.
Figure 4 Krebs water-only cyclone
Water-Only Cyclones
The earliest cyclones were used to reclaim and thicken solids out of a slurry
material. Just prior to World War II, operators at the Maurits mine in The
Netherlands observed that when cyclones became plugged, the overflow
predominantly contained clean coal. From this observation, beneficiation
cyclones were developed by technologists at DSM. Since then, there has been
a significant amount of research into cyclones, resulting in increased capacity,
wear life, and optimization of performance. The basic design of the water-only
cyclone has not radically changed since the inception of the technology and is
typified by the wider cone angle (≥60°) compared with the classifying cyclone.
This smaller slope appears to encourage stratification at its surface, thereby con-
centrating the denser material. Several variants have emerged since the DSM
version, one of which is shown in Figure 4.
Water-only cyclones have typically been used to process relatively fine mate-
rial, usually sized between 0.1 mm and 4 mm. The most common applications
have been twofold: reduction of pyrite from high-sulfur raw coal and effective
cleaning of fine coal with below 10% of near-gravity material. Because of the
relatively low single-unit efficiency, water-only cyclones have often been success-
fully used in two stages with a variety of configurations depending on coal type.
Dense-Medium Cyclones
Dense- (heavy-) medium cyclones were developed from the same beginnings as
the water-only cyclone, though more development was required to control the
dense medium before it emerged as a useful technology. This probably explains
why the first DSM dense-medium pilot plant was not built until 1945 (Sokaski
et al. 1968). Figure 5 shows the original flowsheet developed by researchers
at DSM. The first commercial-scale dense-medium cyclone plant specifically
intended for fine coal cleaning was built in Belgium in 1957 by DSM-Stami-
carbon and Evence Coppee (Mengelers and Dogge 1979). It was another 9
years before the second dense-medium plant was built, also in Belgium, but it
was not until 1961 that the normal dense-medium cyclone design was first used
in the United States, perhaps due to the availability of low-cost, readily easily
cleaned coal that was available in the United States at that time. Since then,
the use of dense-medium cyclones for coarser coal has grown considerably, but
the application of fine dense-medium cyclones has not enjoyed the same level
of adoption despite some confident attempts in the United States as well as in
Europe. Once again, there has been significant work into the cyclone with the
most progress for the dense-medium cyclones involving optimizing the separa-
tion, medium grades, spigot design, and retention of the dense medium.
All forms of fine coal cleaning have advanced in performance as a result of
the introduction of improved screening (or sieving), but perhaps none more
so than dense-medium cleaning where the introduction of the sieve bend,
shown in Figure 6, was perhaps a major breakthrough in enabling accurate siz-
ing and improved medium recovery in DSM circuits. Its application was more
widely utilized in other fine coal circuits, and innovations such as rapping or
induced vibration were introduced in attempts to overcome distribution and
blinding problems. Some circuits introduced tandem arrangements and others
used sieve bends to pre-scalp fines before multi-slope (banana screens) ahead
of dense-medium cyclone circuits (Leeder et al. 1986). Other innovative fine
coal screens have included the Stack Sizer by Derrick Company in the United
States, which introduced a high-frequency fines screen with a unique vibrator
in the 1960s from which has evolved several types for various applications. A
Raw Coal
Reject Clean Coal
Effluent
Source: Driessen 1945, reprinted with permission from DSM.
Figure 5 Flowsheet for the DSM cyclone using magnetite medium, circa 1945
variety of other dual-purpose, high-frequency screens have also emerged via

other manufacturers, an example of which is the Fordertechnik screen (now
Thyssen-Krupp), now using electrically synchronized gear box reducers, which
are mostly used for solid–liquid separation.
Feed Inlet
Screen Deck
Underflow
Overflow
Reprinted with permission from DSM.
Figure 6 Original DSM sieve-bend design
Despite the fact that much attention has been paid to the potential of the
dense-medium cyclone for fine coal cleaning, after almost 50 years of deter-
mined effort with at least a dozen commercially operated plants being built, the
process has yet to become the commercial proposition many thought would be
inevitable. The various attempts have been described elsewhere (Lathioor and
Osborne 1984) and are summarized in Table 1.
Since this publication, new plants have been built in South Africa with
radically improved medium recoveries, one of the major reasons why the earlier
plants failed, which has raised hopes that the application will meet with com-
mercial success and restore confidence. The numbers in Table 2 provide a clear
indication as to why this process is so appealing. Several others researchers have
fresh ideas, and a new generation of dense-medium cyclone circuits for cleaning
fine coal could well emerge within the next decade.
Chemical Separation
Flotation
Flotation of fine coal, like most other fine and ultrafine coal cleaning processes,
had its origins in mineral processing, and therefore, early flotation equipment
was adapted for coal applications with almost immediate success, primarily
because coal is a naturally hydrophobic material (Bury et al. 1921). Early tech-
niques were froth flotation, now referred to as conventional flotation, involving
cells agitated by a variety of mechanisms and differing tank designs offered by
Table 1 Published performance data for fine coal dense-medium cyclone plants
16
U.S.A. Belgium Holland South Africa

1. Location/Test U.S. Bur. Mine Childress Homer R & S. R & S. Marrow- Tertre Winterslag Stamicarbon F.R.I. (R.A. Greenside
Facility City (A) (B) bone Lathioor)
2. Cyclone Type H&P H&P Krebs H&P DSM DSM Krebs DSM DSM DSM DSM Velmet DSM
3. Cyclone Diameter 102 203 610 203 610 610 254 350 350 250 250 250 150 250 150
(mm)
4. Feed Inlet Pressure 70 70 122 62 140 140 152 140 min 140 min 150 150 150 150 156 150
(kPa)
5. Probable Error (Ep) 0.055 0.028 0.065 0.053 0.149 0.188 0.024– 0.059 0.1 — 0.06 0.075 0.01 0.025 0.045 0.035/?
0.081
6. Imperfection (I) 0.108 0.08 0.071 0.156 0.169 0.028 0.041– 0.094 0.12 — 0.115 0.095 0.1 0.052 0.064 0.07/?
0.103
7. Cut-Point Density 1.51 1.35 1.92 1.34 1.88 1.73 1.59– 1.63 1.83 — 1.52 1.79 2 1.48 1.69 1.5/1.8
(d50) 1.79
8. Feed Rate (t/h) 3.3 6.5 12.0(100) 8 15.0(200) — 10–12 — — — — 5 — 36
9. Clean Coal Ash (%) — 1.6 7.9 5 6.6 10.3 — 5.3 6.8 8.2 13.3 8.3 12 13.7 7.4 9.9 7.5/15.0
10. Organic Efficiency — 96.7 96.8 77.9 97.4 — — — — — — — — 93.2 97.68 —

(%)
11. Grain Size Range 0.30/0.15 0.34/0.044 0.6/0.15 1.0/0.15 0.6/0.15 0.6/0.15 0.6/0.1 0.3/0.75 0.3/0.15 1.0/0.3 0.3/0.15 1.0/0.1 1.0/0.1 1.0/0.1 0.5/0.075 0.5/0.075 0.5/0.15
(mm) (considered)
12. Feed Size Range 0.30/0.15 0.34/0.044 9.5/0 1.0/0.15 9.5/0 9.5/0 0.6/0.1 0.75/0 0.75/0 — — — — — 0.5/0.075 0.5/0.075 0.5/0.15
(mm) (to cyclones)
Perspective on Coal Fines
13. Magnetite Losses — 2.7 — — — — 1.5 — — — — 2.5 2.5 2.5 1.3 2.5 Max 1.5
Total kg/t feed
14. Type of Facility Test Plant Modular Test Plant Plant Plant Plant Plant Test Pilot Plant Test Pilot Test Plant Plant
Plant Plant Plant
15. Test Data 1974 1978 1981 1978 1979 1980 1958 1966 1966 1978 1978 1980
16. Comment ZnCl2 Magnetite Clean to 0 — Clean to 0 — Operated Until Rewash
Medium Medium Concept Concept 1983 Low Den-
sity/High
Density
Table 2 Comparison of the probable error (Epm), specific gravity of separation

(SG50), and normalized Epm for different separation options
Medium-Based Water-Based
W-O
Cyclone Cyclone Cyclone Cyclone Cyclone Cyclone Spirals Reflux TBS
800 mm 660 mm 500 mm 500 mm 200 mm Two- Two-

Stage Stage
Size, mm 3 × 0.5 2 × 0.5 3×1 1×6 3.35 × 0.2 1 × 0.1 1 × 0.1 3 × 0.5 3 × 0.5
Epm 0.043 0.079 0.021 0.031 0.026 0.054 0.15 0.101 0.104
SG50 1.56 1.75 1.31 1.38 1.35 1.74 1.80 1.63 1.67
Epm/SG50 0.028 0.045 0.016 0.022 0.019 0.031 0.083 0.062 0.06
the United Kingdom (Mineral Separation), Holland (DSM), Germany (Ekof

and Humboldt), and the United States (Callow). All recognized the need for
an accompanying filter; and vacuum filters, including the Oliver vacuum drum
filter, were widely utilized for dewatering the froth concentrate. To ensure
optimized recovery, overcoming slime coating (via clays) and dealing with feed
rate fluctuations, froth flotation was treated in either multiple stages of trough-
type cells whereby progressive recovery was achievable. Multiple cells were also
employed to enable staged addition of reagents.
Another innovation was the Elmore vacuum (Coppee) flotation cell intro-
duced in the late 1930s and used in European plants until the 1950s when
more sophisticated and higher-capacity froth flotation cell banks became more
capable and productive. Unifloc cells became popular in the United King-
dom following the incorporation of a total slurry treatment approach using
flocculants for improving concentrate filtration and tailings thickening. The
introduction of the Denver subaeration and subsequent variants cell drove a
rapid growth in froth flotation through the 1950s into the early 1960s when
the Wemco Fagergren froth flotation cell emerged, with its rotor-disperser unit
that provided pulp recirculation and aeration as well as significant scale-up
potential in cell volume.
By this time, the rotary vacuum drum filter had become an almost standard
dewatering partner to froth flotation. However, the Dorr-Oliver disc filter
with an agitated bowl emerged as an appealing alternative being demonstrated
as highly suited to fine coal applications. In the modern era, cell volumes and
hence treatment capacities continued to grow reflecting success in mineral
applications, but these did not enjoy the same success in coal applications,
and large separated tanks emerged typified by the Eimco “Smart Cell” in the
United States, which squeezed the froth inside an inverted cone (so-called
froth crowder).
Column flotation for coal applications arrived in the late 1980s (Miller
1988). Since their introduction, applications have spread to all metallurgical
coal-producing countries due to the inherent ability to recover ultrafine coal
down into the submicrometer sizes. Each manufacturer’s design differs, with
the bubble generator perhaps being the most significant identifying feature.
One of the earliest columns was developed in Canada, acting on the need for
processing mineral ores and utilizing the bubble column concept that origi-
nated early in the 20th century in chemical industries. The development path
saw companies like Deister introduce the Floataire column, and universities
develop various cells, including the Ken-Flote column from University of
Kentucky Center for Applied Energy Research, the Microcell from Virginia
Polytechnic Institute and State University, and the Heyl and Patterson Miller
Cyclo-Cell in the United States; the Jameson cell from the University of New-
castle in Australia; and the Wemco/Leeds column from the University of Leeds
in the United Kingdom.
More recent developments have included introducing different forms of
microbubble design or improved variants of those currently used (i.e., porous
venturi tube, in-line mixer, turbo-generator, and various forms of spargers).
Most need fresh water to function effectively, but the Microcell uses slurry
recirculated from the bottom of the column, thereby creating a form of retreat-
ment of coal microparticles that might otherwise have been lost to the tail-
ings. Features like this will be important as applications in the preparation of
micronized coal for ultraclean coal applications such as coal-water slurry fuels
are commercialized. Combined with fine grinding mills such as high-pressure
rolls or stirred-ball mills, microflotation cells will probably be the separator of
choice to produce ultra-low ash coal (<0.5%) in such circuits. Directly injected
diesel engines (DICEs) and gas turbines are the likely users of these fuel types,
and interest appears to be growing, especially in China.
Agglomeration and Briquetting

Agglomeration and briquetting techniques also appear to have a future in fines
recovery and treatment. With the growing need to maximize recovery of high-
value metallurgical coal and the trend toward greater mechanization in all min-
ing operations, compositing of the fines will help solve handling and moisture
problems that have created the need for compromizing on plant yield to avoid
high moisture or sticky coal handling problems. Neither technique is new;
processes like the Trent agglomeration process, introduced in 1918, and the
Convertol process, which followed about 30 years later, were the pioneers, and
then progressive work in Canada at the National Research Council (NRC),
SPS and Olifloc processes in Europe, Central Fuel Research Institute’s process
in India, and IPTACCS (integrated pipeline, transportation and coal cleaning
system) by BHP in Australia (Mehrotra et al. 1983) all had varying degrees of
success. However, all were impacted by the fluctuating oil price and improved
flotation techniques to the extent that there are still no commercial plants in
operation. The Otisca T-Process was developed for producing ultra-clean coals
with the ultimate objective of producing coal-water fuels. Most of the other
processes also anticipated this application, but some (BHP’s pipeline process
and NRC’s process) also intended the process to beneficiate valuable metal-
lurgical coal types for coke-making applications.
Briquetting, however, has been commercialized for many years and is
regarded as a mature technology. In 1848, a patent was granted to William
Easby for a briquetting process, but it took more than 50 years to emerge as
a commercial process. The coal briquetting process, as it ultimately evolved in
the United States, consisted of first drying the coal, then crushing and screen-
ing it, mixing the dry coal with about 6% molten asphalt binder, briquetting
this mixture in roll-type briquette machines, and finally cooling the briquettes
on a conveyor before loading them into railcars or diverting them to stockpile.
More than 6 Mt of coal briquettes were produced annually in the United States
before the process was hit by cheap oil and gas just after World War II.
The briquettes made by this process were used primarily for domestic heat-
ing, and many attempts were made to eliminate the asphalt binder, as the smoke
from the binder was the major user objection to the product. Coal briquetting
today has more definite applications. Coal is briquetted as an initial step in the
production of activated carbon.
There is undoubtedly growing interest in briquetting coal recovered from
tailings ponds or stockpiles of abandoned fine coal piles. Excessive amounts of
fine coal that have resulted from mining or crushing for greater liberation cannot
readily be shipped without some form of aggregation to a larger size to ensure that
a moisture specification is achieved. Briquetting is also used in the production of
form coke and has advantages in the production of metallurgical-grade coke as
well as in the coke-making process itself. Current machines, similar to the one
shown in Figure 7, can operate on coal feeds up to ~30 metric tons/h. So coal,
which launched a major briquetting industry in the first half of last century, may
well come full circle in the first half of this century (Komarek 1991).
Dry Separation
Mechanical dry separators for cleaning coal have been around since the early
1900s, with the first pneumatic oscillating machine for cleaning bituminous
Courtesy of K.R. Komarek, Inc.
Figure 7 Commercial briquetting and compacting roll press
coal introduced in 1916 and the first air dense-medium separator in 1923. Just
as with water-based cleaning, the pneumatic (dry) separators can be split into
three technologies: jigs, tables, and dense-medium separators.
All of the dry technologies appear to have emerged around the same time
(first half of the 20th century) and have undergone numerous developments at
different times since then, although no great breakthrough in their overall use,
let alone in fine coal treatment, has yet materialized.
The air jig is reportedly capable of separating coal down to 0.5 mm, whereas
the dense-medium vessels and pneumatic tables are only capable of separating
down to 0.8 mm and 3 mm, respectively. In the period between the 1930s and
the 1960s, as mining became increasingly more mechanized, dry separators
were more widely accepted, and several were incorporated into the design
of coal preparation plants, many as an initial step to “de-stone” run-of-mine
(ROM) coal.
The culmination of several factors challenged more widespread use of
mechanical dry separators between the late 1960s and early 1990s. These
included
• Inherent inefficiencies of the various technologies,
• Mechanized mining resulting in greater mining dilution in the feed
and a finer mean size of the ROM coal,
• General deterioration in the quality of available resources,
• Pre-dominant dust issues and resultant increased government legisla-

tion, and
• Resultant cost of removing these environmental factors.
Since the introduction of high-productivity mechanized underground and
open-pit mining methods, together with greater use of sprays for dust control at
the mine face, the portion of fines below 0.5 mm has increased fivefold from 5%
to 25% and ROM moistures have also risen. These two factors, together with
the unavoidable dust generation of the separators, probably drove the demise of
dry cleaning in the latter part of the 20th century.
However, since the 1990s, a clear resurgence of interest in dry technology
has occurred, with the introduction of low capital cost dry cleaning units in
China and India, but this has so far not included fine coal cleaning. This resur-
gence has been driven by a combination of increased demand for coal recovery
and environmental impacts from using water, which has encouraged develop-
ments in arid and arctic areas where water technologies are not feasible.
An opportunity exists for some form of agglomeration to capture poten-
tially lost fines. The introduction of this new generation of dry cleaners has
been accompanied by radically improved dust extraction systems that are very
efficient and also economical. Drawing from applications in fly ash recovery
and other similar industrial applications, this could well be the largest single
factor as to why dry cleaning methods are now likely to become increasingly
applied.
Magnetic and Electrical Separation

Magnetic separation applicable to fine coal recovery was introduced in the early
1970s. High-gravity magnetic separator units have been used with acceptable
performance for removing mineral matter from fine coal slurries, but applica-
tions have proved limited mainly because of limited unit capacities.
Tribo-electric separation is a dry beneficiation method applicable for
ultrafine coal particles of <75 µm (i.e., typical power station pulverized fuel).
Thermal coal with a significantly lower ash and sulfur content, plus a higher and
more consistent heat value, can potentially be obtained by this process. Also,
many potential pollutants can be controlled before the coal is fired. Mineral
impurities and coal macerals are tribo-charged differently and can be separated
when an external electrostatic field is applied.
As a dry process, tribo-electric separation has many advantages over tra-
ditional coal cleaning processes, notably the ability to operate in areas where
water resources are scarce, thereby effectively eliminating the need for dewater-
ing. Tribo-electric separation is readily adaptable to the captive power station,
especially in arid areas where coal resources are abundant. Hence, the opportu-
nity to incorporate optimal coal cleaning across the supply chain exists, and the
coal supplier could actually contemplate supplying pulverized fuel to the power
plant at an attractive cost.
Additionally, tribo-electric beneficiation has other possibilities that include
the production of ultra-low ash coal (<1% ash) for injection metallurgy via the
generation of coal-based carbon for use as high-quality active carbon, or for
micronized refined coal for coal-water slurries for use in DICEs or coal-based
fuel cells.
Dewatering
In the beginning, fine coal was not recovered; and where the total raw coal
was washed, the raw fines were invariably rejected as tailings via settling cones
and thickeners. Hence, coal operations worldwide created coal-rich tailing
ponds and lagoons, the recovery of which would ultimately become of interest
when their value started to increase in the last two decades of the 20th century.
Another driver was the desire to close coal preparation plant water circuits, and
this would be most effectively achieved by introducing dewatering equipment
into the plants.
Mechanized mining and perhaps a greater awareness of the potential value
of coal fines, via enhanced petrography and research and development (R&D)
in utilization (particularly in the metallurgical coal market), were probably the
other major drivers in seeking to recover this fraction. New metallurgical coal
deposits with unprecedented fines contents, such as those in the Rocky Moun-
tain areas of Western Canada in the 1970s, probably added impetus to the need
for improved dewatering (Butcher et al. 1979). Because of the greater surface
area involved, the moisture retention properties of this fraction added signifi-
cantly to the moisture content and difficulty in handling the combined prod-
uct. This in turn drove the development of classifying cyclones, vacuum and
pressure filters, fine coal centrifuges and fine screens, together with improved
thickening and settling and filtration chemicals and even thermal drying.
So-called screen-scroll basket centrifuges, a relatively recent development,
are now widely used for dewatering small coal (–2 + 0.1 mm) from spirals or
teetered-bed separators. They have a conical filtering basket for drainage, while
an internal scroll with a different rotation speed assists in material transport
and determines the residence time of the coal. Higher centrifugal acceleration
is required for the finer material, so these centrifuges have higher speeds (maxi-
mum 700 rpm) and treat up to 100 metric tons/h of –2 + 0.1 mm coal produc-
ing about 12%–18% moisture. The onward trend is toward a larger basket for
the horizontal axis machines as per the European design, which reduces the
footprint and eases basket change and maintenance. Vertical axis machines,
which were preferred in North America, have the advantage of the coal being
presented to the scroll top, which provides full acceleration of the feed slurry.
The horizontal axis design feeds against the basket wall and is reported to pro-
duce marginally lower moisture contents at equivalent g-force.
Vacuum filtration, screen-bowl (including decanter-type) centrifuges, and,
to a lesser extent, hyperbaric and pressure filters are used to dewater fine coal
concentrate material. Perhaps the earliest entrants were the rotary vacuum disc
and drum filters. Early rotary disc filters, such as the Eimco Agidisc, became
the popular choice for flotation concentrates; and rotary drum filters, perhaps
more forgiving of coal type, were often used for more-difficult-to-filter fine
coal. Vacuum drum filters were popular in the 1990s, and some are still used
with string, belt, or snap-blow cake discharge options. A new breed of large-
diameter, high-capacity disc filters developed for the alumina industry in the
1970s were adapted and introduced into the coal industry for flotation con-
centrate dewatering.
The horizontal belt types have emerged as a popular choice in recent years
as they provide ease in operation, a high level of reliability, and are the most
flexible vacuum filters to adsorb feed material changes. Their top-fed filtration
principal uses the gravity effect and, therefore, if fed correctly, a layer of coarser
coal settles onto the filter cloth, forming a drainage bed, easing cake release, and
improving filtrate clarity.
The higher driving force of pressure filters compared to vacuum types
provides faster filtration kinetics and lower final product moisture, but it also
incurs higher installation and operating costs. Thus, pressure filtration is nor-
mally applied for difficult-to-treat ultrafine coal tailings or for the recovery of
“lost fines” from screen-bowl centrifuges. Pressure filtration is also sometimes
applied to ultrafine coal slurries for briquetting and agglomeration.
Variants to those previously mentioned include the hyperbaric filter, which
is an enclosed disc filter, installed inside a pressure vessel so that the dewater-
ing driving force is compressed air pressure. They provide up to twofold solids
throughput capacity compared to vacuum filters and up to 40% lower product
moisture for coal concentrates. A heating component can also be applied to
further enhance dewatering potential.
Another interesting variant is the Centribaric filter, which combines cen-
trifugation and pressure filtration within one process to substantially reduce
moistures over what can be achieved using conventional dewatering systems.
Ultrafine coal can be dewatered to as low as 20% by mass by using this type
of unit, which has already shown great promise in commercial applications

in the United States, particularly in recovering ultrafines lost in screen-bowl
centrifuges, and may well become an integral part of future circuits (Keles et
al. 2010).
A variety of batch operating filter presses have evolved over a 50-year
period as a means of treating ultrafine coal or tailings slurries. These originated
as chamber-type filter presses, sometimes referred to as plate-and-frame filter
presses, with either steel or polypropylene plates. A later development, the
membrane chamber filter press, has rubber membranes or diaphragms for
squeezing the cake by mechanical pressure in between the feeding/filling and
air blowing cycle.
The belt press filter or twin wire presses are used for coal tailings to dewater
thickener underflow with the usage of sufficient flocculant so that the slurry
drains by gravity and is after the wedge zone viscous enough not to ooze out
between the filter belts. Belt presses utilize pressure filtration with the assistance
of shear through the differential traveling velocity of the two belts wrapping
around rollers. High flocculant dosage is usually required in particular for
ultrafine tailings containing clays.
Screen-bowl centrifuges are a combination of cylindrical and conical sec-
tions, producing effluent (overflow from the solid bowl), a screened centrate,
and the dewatered coal. Centrate generated via a screen section (~0.25 mm
aperture) is recycled, while the ultrafines (–0.045 mm) in the effluent are
discarded, or recovered by filtration or via the Centribaric unit. Screen-bowl
decanter centrifuges have an internal scroll that transports the coal. They
have evolved to sizes of 1.1 m in diameter and 3.3 m long, and units can treat
up to 100 metric tons/h. Screen-bowl units can treat coal down to nominal
–0.125 mm. Their main application is similar to that of vacuum filters, which
provide very high solids recovery compared to solid-bowl units that tend to
deslime at ~50 µm, recovering only about 50% of this ultrafine size fraction.
Solid-bowl (or decanter types) have no screen section and thus can only
separate the solids by centrifugal sedimentation. The product is compacted
because of the bed depth while transported by the scroll up the cone to the
solids discharge. Hence, they tend to behave more like thickeners operating in
the centrifugal field. There are a number of variants to the more conventional
screen-bowl and solid-bowl units, including the Centribaric unit previously
mentioned.
Mechanical dewatering is more economical than thermal drying and is
mostly regarded as adequate, especially when the most recent developments are
utilized. Therefore, most coal preparation plants currently dewater their fines
products by a combination of screens, centrifuges, and filters. Thermal drying is

only utilized in cold climates to avoid product freezing during transportation.
A number of novel thermal driers have been used for higher-value metallurgical
coals with extremely high fines contents. Holoflite (oil-filled heated scroll) and
Bearce (hot air rotary drum) units with small unit capacities were tried in the
1980s, and FMC and McNally offered fluidized-bed driers with higher unit
capacities (up to more than 1,000 metric tons/h). Very few new installations
currently exist, which tends to suggest that centrifugal/mechanical units have
improved to the extent that acceptable transportable moistures are now achiev-
able in all but the harshest cold climates.
Computers and Process Control

Since the late 1990s, the transformation in computer technology and potential
applications to the minerals industries has translated into all facets (i.e., plant
design, construction, scheduling, SCADA), and as stated earlier, perhaps the
greatest beneficiary from all of this has been seen in the improved treatment
and cost of fine coal beneficiation. A fundamental need, especially for large coal
mining operations, is to reliably measure and/or control important parameters
on a real-time, on-line basis, both quantitatively and qualitatively (Firth 2008).
With the increase in fines content of the salable product, the need for this level
of control has grown significantly, but this remains an industry weakness.
Computer technology was also applied to the prediction of performance
and the design of plant flowsheets, and numerous computer models were intro-
duced and commercialized. During the early 1970s these models, into which
empirical data were incorporated, were used to simulate process performance
toward the development and optimization of flowsheet design (Osborne
1988b). Later versions incorporated other flowsheet components such as
screens, hydrocyclones, and dewatering units. Eventually, development has led
to dynamic modeling of the entire plant and its associated handling equipment.
With the emergence of programmable logic control (PLC), an era of
sophisticated process control became possible and with it the opportunity for
centralizing data via the so-called SCADA system. Early plant control systems
incorporated simple on/off controllers located in a central control room with a
plant control mimic similar to that shown in Figure 8. From these have evolved
systems capable of sophisticated onstream analysis and optimized control using
integrated computer-based systems.
In the early plants, most controllers were separate, discrete pneumatic or
electronic units, but the advent of cheaper computers enabled a single com-
puter to control a wide range of control loops. The use of a single computer was
Figure 8 Plant control center with integrated mimic (circa 1980s)
limited in terms of system reliability, and with the introduction of distributed

control systems (DCSs) with discrete electronic loops for each controller net-
worked into a single data system, SCADA became a reality. With the advent
of DCSs and PLC/SCADA systems and the introduction of modern high-
resolution operator screens, plant operators are able to handle huge amounts
of information and tailor displays to provide the most important information
in the format that best suits them. These control systems incorporate advanced
control strategies that create a highly interactive control system and allow
operators to optimize the plant operation.
With such a control capability comes the absolute need for reliable “fit-
for-purpose” condition sensors, and a whole array of these are now available,
many of them having only emerged relatively recently as commercially reliable
devices; the following is a detailed though not exhaustive reference list:
• Conveyor operational devices, including belt line and misalignment
switches, pull cord switches, belt motion detectors and speed indica-
tors, position switches, plug chute switches, etc.
• Weigh or volumetric feeders to control and measure flow rates for
granular bulk solids
• Sampling and blending devices, including belt scales (weightometers)

and controllers, metal detectors, tramp metal detection systems, sam-
pling systems, and controllers
• Chute, bin, and silo devices, including tilt switches, capacitance level
probes, rotary level controls, flow detectors, etc.
• Power quality monitoring instruments, including instruments for
detecting and monitoring radiation exposure
• A whole family of on-line analyzers ranging from natural gamma and
dual-gamma to prompt-gamma neutron activation analysis (PGNAA)
and the very sophisticated neutron inelastic-scattering and thermal-
shattering analysis (NITA) covering elemental and also separate
devices capable of accurate moisture and particle size analysis
• Laboratory Information Management Systems (LIMSs) for CQ param-
eters including energy, ash, sulfur, and moisture; environmental moni-
toring for ground, air, and water control; other laboratory tests, e.g.,
pH, temperature, oxides, clay, limestone, conductivity, etc.
• In-plant control elements, including nuclear density gauges; nuclear,
magnetic, ultrasonic, and other types of flow meters; particulate (dust)
monitors; smoke and gas detectors, etc.
• SCADA-related data collection providing historical graphing, data
exporting to spreadsheet applications, report generation, network-
ready setup for data transfer to and from other locations
R&D is ongoing for most of these devices and control systems, and a lot of
this work is focused on fines treatment areas (Firth and O’Brien 2010). Process
and plant modeling is another very active area of research applied to both pro-
cess control and process simulation and design.
One area yet to emerge in the coal preparation arena is the integrated
application of automation and robotics. The minerals industries have already
successfully implemented robotics in the sampling and analysis area (Osborne
2004), but the real challenge is to extend the application to other areas in min-
ing and processing.
The Australian Centre for Field Robotics at the University of Sydney,
which has established the Rio Tinto Centre for Mine Automation (RTCMA),
is moving ahead with a number of automation innovations in a bid to improve
mine productivity and safety (CSIRO 2009). The RTCMA has programs
under way in areas such as sensing, machine learning, data fusion, and systems
engineering. This work has resulted in many research advancements in fun-
damental and applied areas. The team has also been working closely with the
mining company Rio Tinto to facilitate the transition of this new technology
to mine operations, which is the current Rio Tinto iron-ore focus.
In the processing area, treatment plants already featuring many control
features are an obvious target area, where the other major emphasis is safety.
The simple fact that those are areas of the plant which are almost always messy
and/or less accessible demonstrates that reaction times needed to avoid this
outcome are inadequate, and, hence, in the fine coal areas of the process circuit,
many opportunities exist for introducing automation and robotics solutions.
Materials Science and Engineering

Although wear and corrosion was an important topic in the past, the materials
of construction now in use in coal preparation plants—for example, hard-
ened alloys, ceramics, protective coatings, rubber lining, polyurethane, and so
forth—have greatly extended equipment life, even when compared to the end
of the 20th century. This has resulted in a significant reduction in maintenance
time required for repair, resurfacing, and replacement, thereby also reducing
operation costs. However, little appears to be known about other potentially
useful materials that could be cost-effectively used right now in preparation
plants, and there is still a distinct tendency for coal preparation engineers to
rely on other industries to do the research, test the applications, and thereby
take the lead.
Global mining companies that have wider interests are probably most
likely to recognize the benefits of cross-fertilization and transference of knowl-
edge and experience. Global suppliers (e.g., original equipment manufactur-
ers and chemical manufacturers) regard this as an opportunity for working
together in some form of collaborative partnership. Their engineers, scientists,
and technologists are encouraged to fully understand the needs and challenges
for coal producers and, via life-cycle partnerships or co-development, strive to
produce the required solutions. Figure 9 provides an indication of how these
relationships can develop to build value and solve problems.
FUTURE DIRECTION
When the milestones are reviewed, clearly there have been very few new ben-
eficiation technologies emerging in the 21st century—just better designs of
existing technology. This is enhanced in many cases by the potential introduc-
tion of new materials, leading to improved maintenance and longer component
life, and the implementation of better control and design, in turn leading to
improved operation and greater efficiency. Further improvements in both areas
High
Life-Cycle
Partnerships
Integrated
Solutions R&D Co-development
Cooperation
Service/Maintenance Consulting Operating
Spare Parts Support Models
Growth in
Importance Mechanical
Automation
Components
Systems
Electrical
Components
Value Cone
Low
Product Supply Process Know-How

Business Model
Courtesy of SMS Demag.
Figure 9 Development path for equipment improvement
remain a major R&D challenge (Osborne 2010), and in this regard there is still
“lots to do.”
Some examples to whet the appetite are listed here:
• Efficient, economical coal dewatering of “microfine” coal (–0.015 mm)
• Effective cleaning of poorly floating ultrafine coal (0.150 × 0.045 mm)
• Accurate size classification at 0.045 mm or lower
• On-line coal characterization and species recognition in fine coal
streams
• Commercialization of high-capacity, high-resolution dry sorting
technologies
• “Whole of supply-chain” beneficiation solutions (e.g., pit to pulver-
ized fuel)
• Power and water conservation activated by cost of carbon and legal
liability
• Robotic sampling and on-line analysis of slurries and pastes
• Handling and possible utilization of ultrafine reject material (clays,
sands, sulfides, etc.)
The chapters that follow will no doubt address some of these challenges.
ACKNOWLEDGMENTS
The author thanks colleagues Kelly Walton and Mark Jackson, coal prepara-
tion engineers with Xstrata Coal in Queensland, for their assistance in the
preparation of this chapter and other colleagues for proofreading and valuable
comments for improving the final version.
Opinions expressed in the chapter are the author’s own views and are not
necessarily shared by his colleagues or others at Xstrata Coal.
REFERENCES
Arrowsmith, G.H. 1957. Possible changes in coal preparation. In 2nd Symposium on
Coal Preparation. Paper 1. University of Leeds. October. pp. 11–30.
Bethell, P.J. 2007. Coal preparation: Current status and the way ahead. Presentation
to the Report Committee of the National Commission on Energy Policy—Coal
Study, Denver, CO, September 17.
Bethell, P.J. 2011. Personal communication.
Bhattacharya, S. 2009. Thermal coal preparation in India: Drivers, barriers, prospects
and challenges. South African Coal Processing Society. IV Biennial Coal Prepara-
tion Conference. Secunda, RSA.
Bury, E., Broadbridge, W., and Hutchinson, A. 1921. Froth flotation as applied to
washing of industrial coal. Trans Inst. Min. Eng. 110:243.
Butcher, S.G., Osborne, D.G., and Walters, A.D. 1979. Applications of computer
technology and special analytical methods developed for the design, commission-
ing and operation of plants treating highly friable Canadian coals. In IX Interna-
tional Coal Preparation Congress, Proceedings. Donetsk, Ukraine. Paper G3. pp.
145–160.
CSIRO. 2009. Mine automation. Earthmatters 19(March/April). Available: www
.csiro.au/files/files/pp19.pdf.
Dennis, P.I. 1998. Optimum processing of 1 mm by zero coal. Ph.D. dissertation. Vir-
ginia Polytechnic Institute and State University, Blacksburg, VA. pp. 1–3.
Driessen, G. 1945. The use of centrifugal force for cleaning fine coal in heavy liq-
uids and suspensions, with special reference to the cyclone washer. J. Inst. Fuel.
XIX(105):33–45.
Easby, W. 1848. Method for Converting Fine Coal into Solid Lumps. U.S. patent,
Aug. 29.
Falconer, A. 2003. Gravity separation: Old technique/new methods. Phys. Sep. Sci.
Eng. 12(1):31–48.
Firth, B.A. 2008. The Intelligent Plant—Measurement Requirements in Fine Coal
Cleaning and Dewatering Circuits. ACARP Project. Report C11069. p. 93.
Firth, B., and O’Brien, M. 2010. Fine coal measurement needs for improved control. In
XVI International Coal Preparation Congress, Proceedings. Edited by R. Honaker.
Littleton, CO: SME. pp. 679–688.
Galvin, K.P., Callen, A., Spear, S., Walton, K., and Zhou, J. 2010. Gravity separation of
coal in the reflux classifier: New mechanism for suppressing the effects of particle
size. In XVI International Coal Preparation Congress, Proceedings. Edited by R.
Honaker. Littleton, CO: SME. pp. 345–351.
Hillman, J. 2003. Upward current washers. In A History of British Coal Preparation.
Edited by J. Hillman. United Kingdom: The Mineral Engineering Society. pp.
141–161.
Hoyois, L. 1935. GB456770 patent. Chaussee de Ransart, Gilly, Belgium, May 16.
Keles, S., Luttrell, G., Yoon, R.H., Estes, T., Schultz, W., and Bethell, P. 2010. Devel-
opment of the CentribaricTM dewatering technology. In XVI International Coal
Preparation Congress, Proceedings. Edited by R. Honaker. Littleton, CO: SME.
pp. 488–495.
Know, G. 1918. The Draper washer. Proc. Soc. Wales Inst Eng. 34:291.
Komarek, K. 1991. Binderless briquetting of peat, lignite, sub-bituminous and bitumi-
nous coals in a rolls press. In 22nd Biennial Conference of the Institute for Briquet-
ting and Agglomeration, Proceedings. San Antonio, TX. 22:233–242.
Lathioor, R.A., and Osborne, D.G. 1984. Dense medium cyclone cleaning of fine coal.
In 2nd International Conference on Hydrocyclones. Cranfield, UK: BHRA.
Leeder, W.R., Hogg, J.W., Jacobs, E.M., and Osborne, D.G. 1986. Application of high
capacity multi-slope screens for coal de-sliming applications in heavy media plants.
In X International Coal Preparation Congress, Proceedings. Edmonton, Canada.
1:300.
Lein, L. 2011. Mining’s new future: How the industry will change in the next decade.
Min. Eng. 63(February):41–46.
Luttrell, G.H., Honaker, R.Q., Bethell, P.J., and Stanley, F.L. 2007. Design of high effi-
ciency spiral circuits for coal preparation plants. In Designing the Coal Preparation
Plant of the Future. Edited by B.J. Arnold, M.S. Klima, and P. Bethell. Littleton,
CO: SME. pp. 73–87.
Lyman, G.J., and Davis, J.J. 1994. Cleaning of Coarse and Small Coal. Coal Beneficia-
tion—Part 7. Monograph Series. Australian Coal Preparation Society. 3:47–59.
Mehrotra, V.P., Sastry, K.V.S., and Morey, B.W. 1983. Review of oil agglomeration tech-
niques for processing of fine coals. Int. J. Miner. Process. 11(3):175–201.
Mengelers, J., and Dogge, C. 1979. A new technique for the treatment of 0.1mm fine
coal by means of heavy media cyclones. In VIII International Coal Preparation
Congress, Proceedings. Donetsk, USSR. Paper B1. pp. 104–111.
Miller, K.J. 1988. Novel flotation technology—A survey of equipment and processes.
In Industrial Practice of Fine Coal Processing. Edited by R.R. Klimpel and P.T.
Luckie. Littleton, CO: SME. pp. 347–363.
Osborne, D.G. 1986. Fine coal cleaning by gravity methods: A review of current prac-
tice. Coal Prep. 2:207–242.
Osborne, D.G. 1988a. Coal Preparation Technology. Vol. 2. London: Graham and
Trotman. pp. 276–285.
Osborne, D.G. 1988b. Coal Preparation Technology. Vol. 2. London: Graham and Trot-
man. pp. 961–999.
Osborne, D.G. 2004. Coking coal and iron ore—Comparisons and contrasts. Arthur
LePage Lecture to the Australian Coal Preparation Society Conference, Pokolbin,
October.
Osborne, D.G. 2010. Value of R&D in coal preparation development. In XVI Inter-
national Coal Preparation Congress, Proceedings. Edited by R. Honaker. Littleton,
CO: SME. pp. 845–857.
Palowitch, E.R., Deurbrouck, A.W., and Parsons, T.H. 1991a. Hydraulic Concentra-
tion. In Coal Preparation. 5th ed. Edited by J.W. Leonard. Littleton, CO: SME.
pp. 414–416.
Palowitch, E.R., Deurbrouck, A.W., and Parsons, T.H. 1991b. Hydraulic Concentra-
tion. In Coal Preparation. 5th ed. Edited by J.W. Leonard. Littleton, CO: SME.
pp. 437–438.
Sokaski, M., Geer, M.R., and Yancey, H.F. 1968. Chapter 10: Dense medium separa-
tion. In Coal Preparation. 3rd ed. Edited by J.W. Leonard and D.R. Mitchell. New
York: AIME. p. 10-3.
Dealing with the Challenges
Facing Global Fine Coal
Processing
Peter J. Bethell
INTRODUCTION
This chapter covers some of the most important challenges facing fine coal
preparation internationally as we move forward in a dynamic, ever-changing
global energy scene. The challenges covered include
• Processing ever more poorly liberated coals for quality-constrained
markets,
• Plant water shortages,
• High reject coals, and
• Pyrite-rich coals.
Processing, dewatering, and waste disposal strategies to deal with these chal-
lenges will be tabled.
THE CHALLENGES
Poorly Liberated Coals
Most well-liberated coals in the coal-producing regions of the world have
already been mined. This leaves major challenges in providing high-quality
products from coal having more near-gravity material and very little low-
density, low-ash “cream.” Coals in India, Mozambique, and South Africa for the
most part will require substantial crushing. This will provide material liberated
enough to generate a marketable product for coking coal or high-heat steam
coals at economical yields. Crushing down to a top size of 12 mm, and maybe
even finer, will probably become commonplace. The impact of changing top
size from 50 mm to 12 mm can be seen from the washability curves generated
for an Indian coking coal (see Figure 1). Massive yield increases occur, which
33
50
45 50 mm Topsize
12 mm Topsize
40
35
30
% Ash DB
25
20
15
10
0
0 10 20 30 40 50 60 70 80 90 100
% Yield
Figure 1 Liberation of an Indian Gondwana coal at different top sizes
make processing economics attractive at marketable quality by reducing plant

feed top size. Crushing technology will have to apply the “sizer philosophy” of
only crushing oversize material to an ever smaller top size to ensure good libera-
tion without the generation of massive volumes of ultrafines.
With a very fine plant feed size distribution, the efficiency of fines (nomi-
nally –1 mm) cleaning becomes extremely important, as does the effectiveness
of product dewatering. Dense-medium cyclone (DMC) circuits will almost
certainly be used to treat the coarser material, although the bottom size feed-
ing these DMC units may return to the traditional 0.5 mm compared to the
coarser cuts seen more recently. This coupled with the use of smaller-diameter
DMC units will allow the efficient treatment of +0.5 mm particles to be maxi-
mized. The trend to ever-larger-diameter cyclones may be reversed, or, more
likely, multiple DMC circuits will evolve to treat intermediate-sized particles
and fines separately, similar to the approach used in South Africa. Here, coarser
material is treated in large-diameter cyclones (±1.1 m diameter) and finer coal
is treated in smaller units (Cresswell 2010).
As far as the –0.5 mm (or –1 mm) is concerned, the new reflux classifier,
capable of high efficiencies over a fairly narrow size range (4:1 ideal) and with
a capability to operate in a fairly wide density of separation range, has emerged
for coking coal plants, as well as for steaming coal plants where appropriate
(Galvin 2010). The ability to cut in the density range below 1.5 is vital for many
middlings-rich coals for all sizes to be processed, including the 1 × 0.2 mm
material. Current spirals, particularly the compound variety, are highly efficient
Challenges Facing Global Fine CoaL Processing 35
Feed Inlet
Lamella Chamber
Overflow
Feed Inlet Chute (2)

Feed Entry Zone
Mixing Chamber
Pressure Probes (2)
Access Door
Fuidization Nozzles Fluidization Chamber
Underflow Valve
Figure 2 Basic reflux classifier
Figure 3 Basic reflux classifier
(Luttrell 2003). Spirals, however, are incapable in their current configuration

of producing plant cuts much lower than 1.65 specific gravity on a continuous
and efficient basis. This will restrict their usage, without modification to pro-
duce a lower cut going forward.
A schematic drawing of the reflux classifier is shown in Figure 2. The rein-
troduction of fine dense-medium cycloning, which is being pioneered in South
Africa (Dekorte 2002), may also provide critical and highly efficient cleaning
in the 0.5 × 0.15 mm range. The circuitry has been running at the Leeuwpan
plant for some time (Figure 3). The bottom size feeding these units will be
determined by the froth flotation characteristics of the ultrafines.
Good “floating” coals will probably be treated in flotation circuits hav-
ing a top size of 0.25 mm. Column flotation provides superior selectivity in
Figure 4 Vacuum disc filter
comparison to conventional flotation (Bethell 2007), particularly for coals rich

in clays. Conventional cells carry entrained slimes into the product, which are
minimized by the deep froth depth and wash water used in column flotation.
Consequently, it is highly likely for clay-rich coals that column flotation will be
practiced, most likely in the rougher/scavenger mode pioneered at Middlefork
(Kennedy 2006). Most Australian and Mozambiquan plants built or upgraded
recently either employ Jameson cells, or Microcell or SlamJet column cells. For
poor floating coals, fine coal DMCs, spirals, or reflux classifiers will probably
be fed material down to 0.15 mm with discard of the ultrafines, until ultrafine
density separation devices come into their own.
Screen-scroll dryers will most likely be used on the DMC product
and screen-bowl centrifuges on the spiral/reflux/fine DMC product. Either
vacuum filtration or pressure filtration (depending on required product mois-
ture content and market moisture constraints) will be used to dewater the
froth concentrates. New developments such as the Bokela filter have greatly
improved vacuum filtration plant performance (Hahn 2011). Similar advance-
ments in the field of pressure filtration have also occurred, with plate-and-frame
filters being widely used in South African plants (Creswell 2010). Vacuum and
pressure filtration units are shown in Figures 4 and 5 along with the hyperbaric
filters widely used in China (Figure 6). Recent developments of efficient dewa-
tering of ultrafines on high-speed centrifuges by Decanter Machine (Shultz
2010) could also contribute to minimizing total plant product moisture in a
finer coal environment.
With increasing fines levels, high-efficiency screening at fine sizes will also
be critical. Efficient fine screens such as the Derrick Stack Sizer (Figure 7),
successfully applied at various coal cleaning facilities (Brodzik 2007), will no
doubt play a role in optimizing fine circuit efficiency.
Figure 5 Ishataki Lasta filter press
Figure 6 Chinese hyperbaric filter
The use of on-line nuclear analyzers will become more important as the
quality of mined coals deteriorates and exacting quality specifications must be
met from inferior coals. Major changes in product ash levels for minor density
changes will be a standard problem. Plant density control circuits will need to
be maintained at a very high level in dealing with coals possessing high per-
centages of near-gravity material. Low tolerances for density variation will be
essential to maintain coal product quality. Briquetting or agglomeration may
be necessary in view of the large percentage of fines in the finished product.
Water-Constrained Plants
Most of the coal-producing areas of the world have water supply issues. Coal
processing plant design will need to be innovative and resourceful to conserve
water supplies to ensure project viability.
Figure 7 Derrick Stack Sizer screening system
Alternative strategies are present to deal with lack of water. These include
closed water circuits for conventional plant circuit design, “coarse wash-only
plants,” and dry cleaning circuitry. Desalination plants are also being intro-
duced to provide plant water sources, significantly pushing up treatment costs.
Closed Water Conventional Circuitry

Whenever possible, current processing plants dispose of fine waste material in
slurry form. This is done by pumping the plant fine refuse, typically thickener
underflow, to dams, impoundments, slurry cells, underground, and so forth.
Where fresh water supplies are abundant and wet disposal areas permitted,
these methods typically provide the most economical disposal methods.
With scarcity of water supply and environmental permitting constraints
on wet fines disposal in most coalfields, this is no longer possible. Alternatives
include the use of belt presses (higher moisture content) and plate-and-frame
filters (lower moisture content). These allow fine refuse to be disposed of
“dry” and provide for minimal water usage. Many plants in the United States,
Australia, China, and South Africa have already moved into this mode of fine
Figure 8 Lone Mountain deep cone thickener
refuse disposal, due to water availability and environmental permitting con-

straints. An alternative to minimize water consumption and provide a more
handleable product is the deep cone thickener (DCT). This device is widely
used in the diamond industry to thicken tailings into a “stackable paste” for
disposal. This technology has been applied in the coal industry at the Lone
Mountain plant in the United States (Figure 8; Gupta 2010). This plant now
pumps conventional thickener underflow to a DCT, which replaced four belt
presses. The DCT product runs between 45% and 50% solids with more than
50% of the conventional thickener underflow water being returned to the plant
in the form of DCT clarified water. This installation has allowed substantial
flocculant savings to occur and maximizes the return of clarified water to the
processing plant.
Coarse Wash-Only Plants

Washing of coarse coal with back blending of raw fines with the washed coal
in some circumstances can provide a market-acceptable product quality. This
type of plant is designed for minimum water consumption and little fine refuse
disposal. Typically, these plants will be used when the quality of the raw coal
material, which traditionally served thermal coal markets unwashed, has dete-
riorated to the point that its quality raw is now unacceptable to the customer.
Examples of this phenomenon include the Western United States where coal
seams mined in Colorado and Utah have become progressively thinner and
higher in ash, delivering run-of-mine (ROM) products no longer salable on a
raw basis.
In these cases and in similar circumstances in some South African mines,
washing of the coarse coal will only provide sufficient beneficiation to render
an unsalable ROM coal salable. Typically in these plants, ROM coal is screened
at between 6 and 9 mm. The oversize material passes to a dense-medium bath
with the fines bypassed to product.
Conventional screening techniques will not provide efficient separation
dry at fine sizes. Consequently, innovative screen technology such as roller
screens and “flip flop” screens have been utilized to achieve high levels of
screening efficiency.
Recent plants incorporating this circuit design concept include Mid-
delburg in South Africa (Bivitec screen), which uses a Larcodems to clean
the coarse material. The Castle Valley (Roxon roller screen) and the Taggart
Global–built West Elk (Bivitec screen) plants of Arch Coal in the United
States use Peters vessels for coarse coal cleaning.
In both of the latter plants, dry screening is achieved on the novel screen
technology with the oversize (±6–8 mm) treated in a dense-medium bath cir-
cuit at high-efficiency levels. The Castle Valley circuit is shown in Figure 9 (Kel-
ley 2008). Figure 10 shows the West Elk plant. The flow in the plant from the
Bivitec screens to the pre-wet screens and after that the dense-medium bath fol-
lowed by appropriate screening, crushing, and drying can be seen in Figure10.
Wet fines generation and fine refuse disposal needs are kept to a minimum,
and, therefore, water requirements are also minimized, which is essential in the
dry, environmentally constrained Western United States.
The economic benefits of the coarse wash-only plants versus conventional
washing on the appropriate coals are enormous, both in terms of reduced
plant capital cost and increased plant yield. For the Western U.S. coals treated
in coarse wash-only plants, an advantageous feature of the coals has proved
to be the much higher ash level in the coarse material than in the fines. This
coupled with typically extremely well-liberated coarse material has enabled
these plants to produce a very salable combined product of washed coarse coal
and bypassed fines.
Roller
FEED
Screen
Magnetic Separator
Reversible Dense Medium Bath

Conveyor
Wetting
TO Screen Flume
Effluent Screen
CLEAN
Cyclones
COAL
BYPASS D&R D&R
FINES Crusher Screen Screen
Thickener Dewater
Screen
Centrifuge
CLEAN Pump
Conveyor COAL Conveyor
Pump COARSE
FINE Pump REFUSE
REFUSE
Figure 9 Castle Valley plant
Figure 10 Arch Coal’s West Elk preparation plant
Dry Cleaning
The ability to have an efficient dry cleaning process for coals, particularly in
arid regions or in those lacking suitable water supplies or wet waste disposal
sites, is obvious. Air jigs and air tables, many resurrected from old designs,
have been developed to fulfill this need. Although the levels of processing
efficiency reached by these dry cleaning devices are considerably less than for
dense-medium processes, the ability to clean coal with no water requirements
has major benefits. To date, cleaning in the 50 mm × 6 mm size range has
proved successful, particularly where the devices have been used as a de-stoner
(de-shaler) reducing the ash level of previously unsalable coal to produce a
marketable product.
The FGX device (Figure 11) developed in China has found widespread
worldwide use in de-shaling (Orhan 2010). Work continues to drive the
efficient cleaning size for these units finer to increase their usage and enable
dirtier coals to be handled (Honaker 2010). For good dry cleaning to occur, it
is important that the feed material be reasonably dry. The FGX units operate
similarly to the wet shaking tables previously so common in plants of old. The
shaking action is supplemented by fluidizing air.
High-Reject Coals
Not only are we continuing to see coals with more near-density particles, we are
also encountering higher rejects in the coals feeding our plants. The reduction
in plant clean coal yield has an adverse impact on revenues. De-shaling ahead of
the conventional plant has been successfully applied in Germany and Australia
to maximize salable coal production by the use of Romjigs (Ziaja 2007).
Pyrite-Rich Coals
Dealing with pyrite-rich coals poses yet another challenge, particularly where
these coals have good coking characteristics. Density separating devices can
control sulfur levels within the constraints of the coal’s washability and the
density cut range of the device. Flotation, on the other hand, poses significant
additional challenges. Coal-associated pyrite on most occasions is very hydro-
phobic. Consequently, even if ash levels can be well controlled by the latest flo-
tation techniques, sulfur tends to be concentrated in the product giving a froth
concentrate much higher in sulfur than is acceptable in coking coal products.
This poses a new set of challenges, which will probably be addressed by complex
sizing/density separation/flotation circuitry.
Figure 11 Tangshan FGX plant
T H E W AY A H E A D
Coal preparation in the future will face major challenges. The need to produce
a high-quality product will remain, but it will have to be done from low-yield,
middlings-rich coals. Water supplies will be harder to find and stricter environ-
mental regulations will be present. The positive side will undoubtedly be excel-
lent global demand for high-quality coking and steam products, particularly
from emerging markets such as India, China, and so forth.
Improving fine wet cleaning techniques such as spiraling and hindered-
bed separation as well as successfully driving down the bottom size treated by
DMCs will be needed. The ability to reduce spiral circuit cut points to the 1.5
specific gravity range would be extremely advantageous. Continual improve-
ment in techniques to effectively dewater the fines coal products from the new
plants will be required. Depending on handleability and dustiness, briquetting
and pelletizing processes may well be required to produce a similar product.
Improvements in dry screening efficiency and processing will be extremely
helpful in arid areas.
The area of ultrafine wet coal cleaning currently reserved for flotation (col-
umn and conventional) will need to be expanded to include efficient density
separation to process oxidized coals or those not efficiently treated by flotation.
Applications of devices such as the Falcon concentrator and Kelsey jig, modi-
fied to achieve efficient separation in coal cleaning, may well prove viable.
Through innovation and a “can do” attitude, the coal processing engineers
of the future will rise to the challenges facing them as coal preparation engi-
neers always have. Adaptability will be the key to survival and success.
REFERENCES
Bethell, P., and Barbee, C. 2007. Today’s coal preparation plant a global perspective.
In Designing the Coal Preparation Plant of the Future. Edited by B.J. Arnold, M.S.
Klima, and P.J. Bethell. Littleton, CO: SME. pp. 9–20.
Bethell, P., and Luttrell, G. 2005. Effects of ultrafine desliming on coal flotation cir-
cuits. In Proceedings, Century of Flotation Symposium. Brisbane. p. 43.
Brodzik, P. 2007. Application of Derrick Corporation’s Stack Sizer in clean coal spiral
product circuits. In Designing the Coal Preparation Plant of the Future. Edited by
B.J. Arnold, M.S. Klima, and P.J. Bethell. Littleton, CO: SME. pp. 89–96.
Cresswell, G. 2010. Process design of the Phola Coal Preparation Plant. In XVI Inter-
national Coal Preparation Congress, Proceedings. Edited by R. Honaker. Littleton,
CO: SME. pp. 66–75.
Dekorte, D.J. 2002. Dense media beneficiation of fire coal revisited. In Proceedings of
the International Coal Preparation Congress. Johannesburg. p. 43.
Galvin, K., Walton, K., and Zhou, J. 2010. Gravity separation and classification of fine
coal using the hydrodynamics of inclined channels. In Thirteenth Australian Coal
Preparation Conference. p. 224.
Gupta, B.K., and Bethell, P. 2010. Deep cone thickener at the Lone Mountain plant.
In Proceedings of the XVI International Coal Preparation Congress. Edited by R.
Honaker. Littleton, CO: SME. pp. 674–678.
Hahn, J. 2011. Performance, operation and maintenance experience of coal ultrafines,
filtration with modern high speed disc filters. In Proceedings of the South African
Coal Processing Society Meeting, Sekunda.
Honaker, R., Luttrell, G., and Mohanty, M. 2010. Coal preparation research in the
USA. In Proceedings of the XVI International Coal Preparation Congress. Edited by
R. Honaker. Littleton, CO: SME. pp. 864–874.
Keles, S., Luttrell, G., Yoon, R.H., Estes, T., Shultz, W., and Bethell, P. 2010. Devel-
opment of the CentribaricTM dewatering technology. In Proceedings of the XVI
International Coal Preparation Congress. Edited by R. Honaker. Littleton, CO:
SME. pp. 488–495.
Kelley, M., and Bethell, P. 2008. The design commissioning and operation of the Castle
Valley plant. Presented at 2008 Annual SME Meeting, Salt Lake City, UT.
Kennedy, D., et al. 2006. Improvements in flotation column recovery using cell-to-cell
circuitry. In Proceedings of the 23rd Annual International Coal, Aggregate and Min-
eral Processing Exhibition and Conference, Lexington, KY. pp. 89–98.
Luttrell, G., Stanley, F., Honaker, R., and Bethell, P. 2003. Operating guidelines for coal
spirals. In Proceedings of the 20th Annual International Coal Preparation Exhibi-
tion and Conference, Lexington, KY. p. 69.
Orhan, E.C., Ergun, L., and Altiparmak, B. 2010. Application of the FGX separator
in the enrichment of Catalagzi coal: A simulation study. In Proceedings of the XVI
SME. pp. 562–570.
Ziaja, D., and Yannoulis, G.F. 2007. Is there anything new in coarse or intermediate
coal cleaning? In Designing the Coal Preparation Plant of the Future. Edited by B.J.
Arnold, M.S. Klima, and P.J. Bethell. Littleton, CO: SME. pp. 43–59.
Coal Properties and Challenges
for Fines Processing
Barbara J. Arnold
ABSTRACT
Coal is used as a fuel as well as a source of carbon for metallurgical processes. It can
also be a source of carbon-based chemicals. For each of these markets, specifications
must be met for various coal properties. Generally, these are related to either the
organic (macerals and elemental analysis of carbon, hydrogen, etc.) or inorganic
(e.g., minerals or moisture) components of a coal, and coal rank, the increase in
coalification of a deposit of organic material overlays changes in composition—
higher carbon content and lower moisture, for example.
The organic and inorganic components also create opportunities and challenges
for the selection of fine coal processing technologies. This chapter reviews these prop-
erties and gives information related to fine coal processing options.
INTRODUCTION
According to the International Organization for Standardization (ISO 2005),
organic deposits of peat become “coal” when the moisture content is below
75%. In addition, the highest ash content recognized as coal is 50%. Of course,
some refuse piles containing greater than 50% ash are being reprocessed to
recover valuable coal products. The ash value and moisture are certainly impor-
tant coal properties associated with producing a salable product. But other
properties are also important—both physical and those related to the petrology
and mineralogy of the coal.
Physical properties like the size distribution of a coal are included in many
contracts. Some have a top size limit—perhaps 50 mm (2 in.) for coal used in
pulverized coal combustion systems or 6 mm (0.25 in.) for fluid bed systems.
Many contracts will limit the amount of fines that can be in a coal to limit
coal handling issues (Arnold 1995b). Moisture constraints, especially those
for surface moisture, can be considered a physical property. Limits on surface
47
moisture content are related to coal handling issues; and total moisture (surface
plus inherent) will impact the evaporative load on power plant boilers.
The petrologic and mineralogic properties are related to the organic and
inorganic components in a coal. Properties like carbon or hydrogen content—
rank-related phenomena—determine whether a coal can be used for metallur-
gical purposes. The composition of the minerals also influences the slagging or
fouling behavior of the resultant ash in a boiler.
These properties are also important when considering selection of coal pro-
cessing equipment. Of course, size distribution and washability will be impor-
tant for the selection of all coal cleaning equipment, including those for the
coarse coal fractions. However, as coal particles become finer, surface properties
become more important and associated minerals, such as clay minerals, play an
increasingly important role in equipment selection and chemical additive selec-
tion for processing and dewatering. Slurry properties will become increasingly
important as well. pH and the concentration of other elements like Ca2+ or
Mg2+ will influence the behavior of coal and mineral particles.
The froth flotation process has been described by many investigators, and Han-
sen and Klimpel (1985) noted that three components—chemistry, operational,
and equipment parameters—influence results. Mineralogy, which for coal includes
the petrologic macerals as well, was included under operational parameters. Other
operational parameters included feed rate, particle size, pulp density, and tempera-
ture. These same parameters will be important for any other fines processing device,
whether for beneficiation, classification, dewatering, or thickening devices.
And these operational parameters will influence the “chemistry” required
for efficient processing. For flotation, this can be collectors, frothers, activa-
tors, depressants, and pH control. For dewatering or thickening, chemistry will
include dewatering aids or flocculants, and pH control can also come into play.
And perhaps, attention to chemistry can play a role in classification with the
use of dispersants or pH control for dispersion. And one can ask the question
as to whether “chemistry” is required for processing coal in gravity separation
devices like water-only cyclones, spirals, reflux classifiers, and so forth. But the
“chemistry” that is required for a particular coal is based on its organic and
inorganic composition and the resulting slurry properties.
This chapter will delve into these coal properties—these operational
parameters—and the challenges they bring to fine coal processing.
COAL
No two coals are alike. Even within a given coal seam, coal properties vary. A
coal started out as a deposit of plant matter that underwent decomposition and
Coal Properties and Challenges for Fines Processing 49
burial. At different times, inorganic minerals or elements infiltrated the swamp

to be included in the coal resource today. The original types of plant material
influence the coal’s properties as do its history of pressure and temperature from
that burial. In some cases, the organic deposit was subjected to heat from igne-
ous intrusions; and in other cases, the deposit was subjected to increased pres-
sure from tectonic activity—these actions cause an increase in the rank of coal.
Different types of plants or differences in decomposition result in different coal
macerals—distinct microscopic components. So, again, no two coals are alike.
Organic Components—Macerals and Coal Rank

Coal macerals are grouped into three general categories based on either the
original plant material or its decomposition, as shown in Table 1. They have
different properties and react differently when subjected to processes for coke-
making or liquefaction. Vitrinite macerals are relatively reactive, for example,
while inertinite macerals remain inert.
Macerals also have different properties that can influence their behav-
ior during processing. Fusinite, for example, is typically very friable and will
degrade into finer size fractions. The exinite macerals are typically very resinous
and are resistant to grinding. Macerals also have different hydrophobicities, as
indicated by contact angle measurement (captive bubble or sessile drop). As
shown in Tables 2 and 3, fusinite is less hydrophobic than vitrinite from the
same coal. If exinite macerals are present, they are generally the most hydropho-
bic. This is likely related to their higher aliphatic content, as noted in Table 1.
Also noted in Tables 2 and 3 are levels of coal rank. The hydrophobicity of
fusinite and vitrinite generally increases with increasing coal rank, as indicated
with an increase in the sessile drop contact angle. But also in Table 2, note that
the sessile drop soaks into the fusinite surface of the mvb coal from Illinois.
This could indicate a difference in the actual structure of the fusinite and could
indicate that some types of fusinite could be hydrophilic even in high rank coal
samples.
Of course, coal rank has a specific definition, and the names for coal rank
change based on country. In the United States, coal rank is defined as in Table 4
and is based on fixed carbon and volatile matter (dry, mineral matter-free basis)
for higher rank coals and calorific value on a moist, mineral matter-free basis
for lower rank coals. Coal rank can also be defined by bed moisture for low rank
coals or vitrinite reflectance for higher rank coals as in the ISO classification
given in Table 5. Furthermore, vitrinite reflectance can be related to hydropho-
bicity, as shown in Figure 1. Hydrophobicity peaks with a vitrinite reflectance
of 1%–2%. This is at the mvb to lvb range.
Table 1 Coal macerals—classification, origin, and properties
50
Origin Appearance Technological Behavior

Maceral Color in Thin Reflected Carboniza-
Group Maceral Source Process Section Light Chemistry tion Liquefaction
Vitrinite Telinite (cell Woody tis- Mummification Red-brown Intermediate Intermediate Principal reac- Susceptible to
walls) sues, bark, gray hydrogen tive constitu- liquefaction
leaves, etc. content and ent in coking
Collinite
volatiles coals
(cell fillings,
structureless
vitrinite)
Exinite (or Resinite Resins Resistant Yellow Dark gray Higher hydro- Reactive Susceptible to
Liptinite) remains gen content during liquefaction
Sporinite Spore exines
and volatiles; carbonization
Cutinite Cuticles more aliphatic
Alginite Algae
Inertinite Micrinite Plant materials Degradation Opaque White, yellow- Lower hydro- Inert during Resistant to
(massive and products ish, light gray gen content carbonization liquefaction
granular) and volatiles;
more aromatic
Fusinite Woody tis- Fire; biochemi-
sues, etc. cal oxidation
Semi-fusinite Woody tis- Intermedi-

sues, etc. ate between
vitrinite and
fusinite
Selerotinite Fungal sclero- Resistant
tia, hyphae remains
Source: Davis et al. 1976.
Table 2 Contact angles on fusinite as compared to vitrinite from selected samples

Rank (Location) Maceral ӨCB ӨSD
mvb (British Columbia) Vitrinite 61 61
Fusinite Very slight cling 58
mvb (Illinois) Vitrinite 60 66
Fusinite No contact Soaks into sample
hvAb (West Virginia) Vitrinite 60 62
Fusinite No contact 40
hvBb (Indiana) Vitrinite 56 57
subA (Utah) Vitrinite No contact 35
Source: Arnold 1985; Arnold and Aplan 1989.
Note: CB is captive bubble; 5-minute induction time allowed for fusinite samples. SD is sessile drop.
Table 3 Contact angles on exinite samples

Rank (Location) Maceral ӨCB ӨSD
NA (Utah) Resinite concentrate 114 120
hvBb (Utah) Vitrinite 44 58
Resinite (pressed pellet) 120 132
Source: Arnold 1985; Arnold and Aplan 1989.
Note: CB is captive bubble; SD is sessile drop.
Many other properties of coal also change with coal rank, as given in
Table 6. A number of these will impact processing options.
For example, the Hardgrove grindability index (HGI) increases up to about
the low volatile bituminous rank, with anthracite becoming more difficult to
grind. Knowing the HGI of a coal will give an indication of the amount of fines
that can be generated through handling and cleaning. In some cases, for even
coarse coal circuitry, very high HGI coals have been handled very carefully,
including gravity-fed dense-medium cyclone circuits to eliminate pumping the
coal to the cyclone. Fines circuit capacity should also be increased to accom-
modate these increases in fines. One source has indicated that projected fines
should be increased by at least 40% for very high HGI coals (Stanley 2007).
The oxygen content of coal decreases with increasing coal rank, as does the
amount of oxygen present as OH and COOH groups in and on the surface
of coal. The higher oxygen content is related to low hydrophobicity and poor
floatability (Sun 1954; Gutierrez and Aplan 1984). It is thought that the pres-
ence of reactive oxygen groups that form upon oxidation—like those of humic
Table 4 ASTM classification of coals by rank
52
Fixed Carbon Limits, % Volatile Matter Limits, % Calorific Value Limits, Btu/lb / MJ/kg
(dry mineral matter-free basis) (dry mineral matter-free basis) (moist* mineral matter-free basis)
Agglomerating
Class/Group = or > < > = or < = or > < Character
Anthracite
Meta-anthracite 98 — — 2 — —
Anthracite 92 98 2 8 — — Nonagglomerating
Semiantracite 86 92 8 14 — —
Bituminous
Low volatile 78 86 14 22 — —
Commonly agglomerating
Medium volatile 69 78 22 31 — —
(there may be nonag-
High volatile A — 69 31 — 14,000†/32.6 — glomerating varieties with
†
notable exception in the
High volatile B — — — — 13,000 /30.2 14,000/32.6
high volatile C group)
High volatile C — — — — 11,500/26.7 13,000/30.2
10,500/24.4 11,500/26.7 Agglomerating
Subbituminous
Subbituminous A — — — — 10,500/24.4 11,500/26.7
Subbituminous B — — — — 9,500/22.1 10,500/24.4

Subbituminous C — — — — 8,300/19.3 9,500/22.1
Nonagglomerating
Lignitic
Lignite A — — — — 6,300/14.7 8,300/19.3
Lignite B — — — — — 6,300/14.7
Source: ASTM 2005.
Note: This classification does not include a few coals, principally nonbanded varieties, that have unusual physical and chemical properties and which come within
the limits of fixed carbon or calorific value of the high volatile bituminous and sub-bituminous ranks. All these coals either contain less than 48% dry mineral
matter-free fixed carbon or have more than 15,500 moist mineral matter-free Btu per pound (Btu/lb × 2,326 = kJ/kg).
* “Moist” refers to coal containing its natural inherent moisture but not including visible water on the surface of the coal.
† Coals having 69% or more fixed carbon on the dry mineral matter-free basis shall be classified according to the fixed carbon, regardless of calorific value.
Table 5 Summary of ISO coal classification

Rank Category Mean Random Vitrinite Reflectance Range, %
Low (lignite and sub-bituminous)
Low rank C (lignite C) <0.4 and bed moisture >35% and <75%, ash free
Low rank B (lignite B) < 0.4 and bed moisture ≤35%, ash free
Low rank A (sub-bituminous) 0.4 ≤ R < 0.5
Medium (bituminous)
Medium rank D (bituminous D) 0.5 ≤ R < 0.6
Medium rank C (bituminous C) 0.6 ≤ R < 1.0
Medium rank B (bituminous B) 1.0 ≤ R < 1.4
Medium rank A (bituminous A) 1.4 ≤ R < 2.0
High (anthracites)
High rank C (anthracite C) 2.0 ≤ R < 3.0
High rank B (anthracite B) 3.0 ≤ R < 4.0
High rank A (anthracite A) 3.0 ≤ R < 6.0 or mean maximum <8.0
Petrographic categories
Vitrinite Content, % by volume,
mineral free Vitrinite Class Category
<40 Low
≥40 and <60 Medium
≥60 and <80 Moderately high
≥80 High
Ash yield categories
Ash Yield, % by mass dry basis Ash Class Category
<5 Very low
≥5 and <10 Low
≥10 and <20 Medium
≥20 and <30 Moderately high
≥30 and <50 High
Source: ISO 2005.
acids—cause this recovery loss. Highly oxidized coal and humic acid extracts
from an oxidized bituminous coal exhibited a large negative zeta potential
with essentially no point-of-zero charge (PZC) (Wen 1977). Firth and Nicol
(1981) studied coal flotation behavior in the presence of clay and humic acids.
They hypothesized that clay absorbed more oil in the presence of humic acids so
that there was less collector available to float the coal. Table 7 gives results from
Arnold and Aplan (1986b) showing that humic acids depress coal flotation in
70
50
Water Sessile Drop
30
Contact Angle, θ°
Captive Bubble
60
40
60
20
50
0.5 1.0
0
0 1.0 2.0 3.0 4.0 5.0
Mean Maximum % VRO
Eastern Coals Interior Coals

British Columbia mvB Western Coals
Selected for Range of Vitrinite Reflectance “Pseudovitrinite” (higher reflectance
Half-filled Symbols from Purcell 1982. vitrinite from the same sample)
Source: Arnold and Aplan 1989.
Figure 1 Contact angle as a function of vitrinite reflectance for vitrinite in coal
both the presence and absence of clay. pH also has an effect in this humic acid/
clay/coal system.
The PZC—the pH at which the surface charge is zero—also varies with
coal rank. Below the PZC, the coal surface is positively charged, and above the
PZC, the coal surface is negatively charged (see example in Figure 2). Most

Table 6 Approximate values of some coal properties in different rank ranges (dry, mineral matter-free basis unless otherwise noted)
High Volatile Bituminous Bituminous
Sub-
Property Lignite bituminous C B A Medium Volatile Low Volatile Anthracite
Equilibrium moisture, % 35 16–29 12 7 4 3 3 5.5
Volatile matter, % 33–62 60 34–52 36–50 31–50 20–31 10–20 <10
Calorific value, moist mineral 7000 / 16.3 10 000 / 23.3 12 000 / 27.9 13 500 / 31.4 14 500 / 33.7 15 000 / 34.9 15 800 / 36.7 15 200 / 35.3
matter free, Btu/lb / MJ/kg
%C, average 73 76 80 82 85 89 91 95.5
%C, range 66–76 71–81 76–84 79–85 83–89 86–92 87–92 92–97
%H 4.5 5 5.5 5.5 5.5 4.5 3.5 2.5
%O 25-16 25-11 18-8 13-7 13-4 4-3 3 2
%O as OH 11-7 9-5 9-4 6-3 6-2 3-1 3-1 1-0
%O as COOH 10-6 9-4 low Tr-? 0 0 0 0
Aromatic C atoms, % of total C 50 65 ? ? 75 80–85 85–90 90–95
Average number of benzene 1–2 ? 2–3 2–3 2–3 2–3 5? >25?

rings/layer
Reflectance, %, vitrinite 0.2–0.3 0.3–0.4 0.5 0.6 0.6–1.0 1.4 1.8 4
Plasticity and coke formation No No Yes Yes Yes Yes Yes No
Hydrophobicity ←Decreases ←Decreases ←Decreases ←Decreases ←← Maximum →→ Decreases
Point-of-zero charge, pH 2 2 ←Decreases ←Decreases ←Decreases 7-8 7-8 Variable
Density (He), vitrinite ←Increases ←Increases ←Increases ←Increases Minimum Greatly increases→
250 100–300 100–220 100–220 100–220 200–250 200–250 200–400
Surface area (CO2), m2/g
Hardgrove grindability index 35–50 40–55 45–80 45–80 45–80 70–115 70–115 25–50
Friability, average, % 12 20–30 43 43 43 70 70 33

Coal Properties and Challenges for Fines Processing
Source: Aplan 1993 modified and expanded from the original tabulation of Given 1984.
55
Table 7 Effect of pH on Pittsburgh coal flotation in the presence of humic acid and
clay
Coal Recovery (0.1 kg/t MIBC), %
pH 3.5 pH 6.5
Clean coal alone
Tap water 84.7 87.7
Tap water +0.08 kg/t humic acid 68.9 56.4
Distilled water 82.5 80.6
Distilled water +0.08 kg/t humic acid 50.9 58.2
20% Kaolinite
Tap water 62.4 87.9
Tap water +0.08 kg/t humic acid 46.4 73.5
Distilled water 30.3 40.2
Distilled water +0.08 kg/t humic acid — 56.8
Source: Arnold and Aplan 1986b.
PZC results in the literature are reported as the surface charges in distilled
water—the absence of cations and anions. The presence of these ions, as in tap
water (or preparation plant water), will reduce the magnitude—either negative
or positive—of the surface charge.
As alluded to in Table 7 and in the discussion related to clay later in this
chapter, it will be shown that the presence of the ions affects the behavior of
the clay and coal in a flotation system. It will almost certainly have an effect on
fine coal dewatering and thickening and the reagents applied in these systems.
This coal/mineral/water system needs to be dewatered or thickened. In a
review of dewatering algorithms in the 1990s for the development of a flow-
sheet simulator sponsored by the U.S. Department of Energy and the Electric
Power Research Institute (Arnold 1995a, 1996, 1997, 1999), it was found
that important properties for predicting final moisture content included the
following:
• The average particle size using a mean diameter based on the volume.
That is, it uses both size and specific gravity distributions in the calcula-
tion. The amount of fines and ultrafines will be a big determinant of
the final moisture content.
• Bed or cake thickness.
• Dewatering force (pressure, gravity, vacuum, etc.) and dewatering time.
But also of importance were properties related to the slurry—surface
tension and liquid density for the screening algorithms; and surface tension,
+40
+30
+20
+10
Zeta Potential, ζ, mv
–10
–20
–30
–40
–50
2 3 4 5 6 7 8 9 10 11 12
pH
Distilled Water Tap Water

Pittsburgh Seam Coal
Upper Freeport Seam Coal
Lower Kittanning Seam Coal
Figure 2 Zeta potential of coal
viscosity of filtrate, and temperature for centrifuge and filter algorithms. A coal
property—contact angle (hydrophobicity)—was included in the screening
algorithm. The oxidation of coal or pyrite, ions in solution, and so forth, will all
play a role in the moisture content from dewatering devices as they effect these
properties. These were developed from work by Wakeman and Rushton (1977)
and Wakeman (1979).
Inorganic Components
Inorganic components in coal vary greatly. As-mined coal includes the mineral
impurities within the coal seam, those in partings, and those from the roof and
floor. These all become part of the feed to a coal preparation plant, and certain
of these minerals are found more prominently in the tailings from preparation
Table 8 Average mineralogical composition of mineral constituents in major U.S.

coal seams and in preparation plant tailings
Mineral Content of
Mineral Content
Coal Preparation Plant Tailings†
of Major U.S.
Coal Seams* Eastern‡ Western‡
Big Dirty Lower
Mean, Range, Mean, Range, (Washington), Sunnyside
Mineral % % % % % (Utah), %
Kaolinite 34.8 0–85 11 6–22 24
Illite 7.8 0–35 53 47-65
Mixed layer 3.2 0–20

Illite-
Montmorillonite
Montmorillonite 0.7 0–10 70 31
Chlorite 1.5 1–10 4 0–7
Quartz 10.1 0–40 15 8–22 12
Gypsum 11.9 0–60 §
Rutile 2.3 0–10
Calcite § 12 0–22 17
Pyrite § 4 1–11 2
Feldspar § § 30 <1
Dolomite § § 14
* O’Gorman and Walker 1972.

† Bradley et al. 1980a, 1980b.
‡ On average, 83% of the mineral matter particles were found to be finer than 0.044 mm.
§ Others: Pyrite, siderite, dolomite, calcite, aragonite, ankerite, muskovite, plagioclase,
hematite, jarosite, thernardite.
plants (O’Gorman and Walker 1972; Bradley et al. 1980a, 1980b). Table 8
summarizes the minerals present in U.S. coals and preparation plant tailings.
Inorganic ions are then present from the dissolution of these minerals or the
oxidation products of these minerals.
Minerals
The minerals in coal can be present as large, discrete, easily removed particles
or as finely disseminated particles that would require significant particle size
reduction to liberate the mineral from the coal. But mineral particles are
also present throughout the range of sizes of coal and in varying degrees of
liberation or locking. As given in Table 9, these coal–mineral particles have
Table 9 Coal–mineral associations or carbominerites

Carbargilite Coal + 20%–60% clay
Carbopyrite Coal + 5%–20% pyrite
Carbankerite Coal + 20%–60% carbonate
Carbosilicite Coal + 20%–60% quartz
Carbopolyminerite Coal + 20%–60% various minerals
Source: Stach et al. 1975.
Table 10 Difficulty of separation based on near-gravity material

Quantity Within ±0.10
Specific Gravity Range, % Separation Problem
0–7 Simple
7–10 Moderately difficult
10–15 Difficult
15–20 Very difficult
20–25 Exceedingly difficult
Above 25 Formidable
been named by Stach et al. (1975). The degree of coal–mineral association in

a particle will influence its recovery in gravity separation processes—particles
with higher mineral contents will have higher specific gravities. The amount of
these “locked” particles in any specific gravity range will give an indication of
the amount of near-gravity material at a separating gravity. Near-gravity mate-
rial is defined generally as the amount of material present in the ±0.10 gravity
fraction, though in some instances it has been re-defined as the ±0.05 gravity
fraction. Higher amounts of near-gravity material result in more difficult sepa-
rations, as indicated in Table 10. Knowledge of the specific gravity distribution
of a coal will be important for any fine coal gravity-based separation.
For fine coal processing, the clay minerals (illite, kaolinite, and montmoril-
lonite) and for some coals, pyrite, are of special interest.
Clays. Clay slime coatings on coal have been studied by many investiga-
tors (Brown and Smith 1954; Jowett et al. 1956; Burdon et al. 1976; Mishra
1978; Yancey and Taylor 1935; Szczypa et al. 1973; Neczaj-Hruzewicz et al.
1974; Firth and Nicol 1981). However, as noted by Arnold and Aplan (1986a,
1986b), the type of coal, the type of clay, and the concentration of ions in solu-
tion (including pH) effect the formation of clay slimes and the inhibition of
bubble contact and subsequent flotation. These results will also be important
in investigations of flocculants as clay slimes may inhibit their contact with coal
or mineral surfaces. As mentioned previously, clays can also absorb reagents,

like oils, and reduce the amount available for flotation (Firth and Nicol 1981).
Fine, dispersed clay mineral particles present in a fine slurry will split
generally as the water splits. For example, it has been shown that in froth flo-
tation, the “free” clay reports basically in proportion to the water—through
water carryover (Arnold and Aplan 1986). This carryover has been noted by
other investigators and is indicated as the major problem with clay in froth
flotation (Yancey and Taylor 1935; Neczaj-Hruzewicz et al. 1974; Lynch et
al. 1981). The use of countercurrent rinse water as in column flotation cells or
as described in froth washing or froth sprinkling studies of conventional cell
banks (Luttrell et al. 2000) limits the amount of carryover, though some could
still be collected through the presence of clay slime coatings.
Fine, dispersed clays and other ultrafines will follow the water split in
classifying cyclones. Clays will therefore be present in the devices treating, for
example, 1 × 0.1 mm coal—spirals, reflux classifiers, teeter-bed separators, and
water-only cyclones even if the feed is classified prior to these units. If condi-
tions are present that allow clay slime coatings to form, they will add to this
clay content in these circuits. This misplaced material will need to be addressed
during these processes.
But what are the conditions that allow clay slime coatings to form? Clay
minerals are plate-like. The flat surface has a negative charge, while the edge
exhibits a PZC at about neutral pH (van Olphen 1963). As shown in Figure 3,
the zeta potential of clay changes with pH and with the introduction of ions in
solution. Moreover, different clay minerals exhibit different behavior. Keeping
in mind the data from Figure 2 for coal surface charge, it could be expected that
slime coatings could form at low pH values where the flat surfaces of the clay are
negatively charged and coal surfaces are positively charged—opposites attract.
As shown in Table 7, at lower pH with humic acid in distilled water (no
ions present) and in the presence of kaolinite, coal flotation recovery is reduced.
In Table 11, additional data are presented to show the flotation recovery
behavior with different clays and coals. In tap water and at pH 6.5, essentially
no reduction in coal recovery was found in the presence of kaolinite and illite.
However, bentonite clay caused decreased recovery except for the very highly
hydrophobic low volatile bituminous Lower Kittanning coal. During this test
work (Arnold 1985), filtration of any samples with bentonite took considerably
longer than filtration with the other clays present. Also, a limit of 5% bentonite
addition to the samples was required. Any higher concentration resulted in
samples that would not readily filter and flotation was completely depressed.
Certain clay minerals may wreak havoc with thickening and dewatering circuits.
+40
+30
+20
+10
–10
–20
–30
-40
2 3 4 5 6 7 8 9 10 11 12
pH
Kaolinite: Distilled Water Tap Water

40 ppm Ca2+/20 ppm Mg2+ 40 ppm Ca2+
+40
+30
+20
+10
–10
–20
–30
-40
2 3 4 5 6 7 8 9 10 11 12
pH
Illite: Distilled Water Tap Water

Bentonite: Distilled Water Tap Water
Figure 3 Zeta potential of kaolinite, illite, and bentonite in tap and distilled waters or
with cations present
Table 11 Effect of clays on the flotation of coals

Coal Recovery, % (0.1 kg/t MIBC, pH 6.5, tap water)
Coal Coal Alone 20% Kaolinite 20% Illite 5% Bentonite
Pittsburgh hvAb 87.7 87.9 84.3 54.6
Upper Freeport hvAb 88.2 — 86.8 59.3
Lower Kittanning lvb 95.4 96.1 96.6 91.2
Source: Arnold and Aplan 1986a.
It is very important to characterize the coal–clay system along with the

pH and presence of other ions in the slurry to determine if clay slime coatings
may be an issue. Although most research has been conducted in regard to froth
flotation, these observations will also have implications for the use of reagents
for dewatering aids and thickening.
Pyrite. Just as no two coals are alike, no two coal pyrites are alike. Work by
Esposito et al. (1987) compared the properties of various pyrites, as shown in
Table 12. Only the specific gravity (SG) of anthracite pyrite approaches that
of ore pyrite. Pyrite is present in coal in many morphologies, including some
forms that are finely disseminated in the associated coal particles like framboids
or as cell fillings. Pyrite has been observed to have some natural hydrophobicity
and sometimes reports to the clean coal in froth flotation. As noted in Table 12,
there was some inherent floatability for some of the coal pyrites tested.
SUMMARY
In summary, there are important coal properties to be considered when process-
ing coal, especially fine coal.
1. No two coals are alike.
2. Consider the macerals present in a coal. Vitrinite macerals are gener-
ally the predominant maceral found in a coal, but higher concentra-
tions of inertinite macerals like fusinite could result in lower flotation
yields than expected, as these macerals are less hydrophobic than the
accompanying vitrinite. This could also play a role in the dewatering
behavior of coal fines. And it could also explain differences noted in
the behavior of dewatering aids that seemingly work for some but not
all coals. Fusinite is also a more friable maceral and could generate
more fines than expected.
Table 12 Various properties of pyrites

Surface Areas, Inherent
Apparent Semi- m2/g* Flotation,
Pyrite SG Berman Optical % conductor %
Sample Balance Anisotropic Type Blaine BET Recovery Morphology†
Anthracite 4.86 ± 35 p 0.0260 0.101 0 SS, PG
(wall) 0.10
Anthracite 4.74 ± 50 p 0.0272 0.266 0 SS, PG
(seam) 0.19
Lower 3.64 ± 15 p 0.0323 0.362 7.3 SS, PG,
Kittanning 0.21 GS,CF
Lower/ 3.53 ± 15 p 0.0243 0.854 2.7 SS, PG,
Middle 0.06 GS, CF, FR
Kittanning
Pittsburgh 3.78 ± 10 p 0.0339 1.502 2.2 GS
0.08
Upper 4.12 ± 25 p 0.0310 0.951 3.1 PG, GS,
Freeport 0.10 CF, FR
Kentucky 4.33 ± 5 p 0.0308 0.597 0 SS, PG,
#11 0.21 GS, CF
Ore 5.00 ± 50 n 0.0295 0.045 0 SS
0.07
Source: Esposito et al. 1987; Aplan 1993.
* 0.147 × 0.037 mm (100 × 400 mesh) particles.
† Morphology: SS = smooth surfaces, PG = porous grains, GS = granular or spherical masses,
CF = cell filings and imprints, FR = framboids.
3. Consider coal rank. Higher rank coals are more hydrophobic. Lower
rank coals have higher oxygen contents and humic acids may be pres-
ent in solution in greater concentration. Higher rank coals are more
friable and more fines will be present.
4. Consider the minerals present in a coal. Pyrite is easily oxidized and
can contribute to low pH. Clay minerals should be characterized and
have differing behavior with pH, the presence of other ions in solution,
and the presence of humic acids.
Although liberated mineral particles may be easily rejected, their
dissolution will affect the ions in solution in preparation plant waters.
These ions will affect pH, the charge on coal particle surfaces, the
charge on the mineral particles (especially noting the two differently
charged surfaces for clay minerals), and certainly the utilization of
reagents for flotation, dewatering, and thickening. The final moisture
content in dewatering may also be affected by the water or slurry
chemistry related to surface tension. Dispersants for clay minerals, for

example, may be suggested when trying to deslime prior to flotation.
However, there could be an effect on flotation reagents or flocculent
behavior in other processes. Some physical techniques that have been
suggested for clay dispersion, such as ultrasound, may be of some
benefit.
5. Consider slurry temperature. Although not directly related to coal
properties, it will affect dewatering and, perhaps, the use of some
reagents.
REFERENCES
Aplan, F.F. 1993. Coal properties dictate coal flotation strategies. SME Trans. 294.
Arnold, B.J. 1985. Coal froth flotation: The effects of petrological and mineralogical
constituents. M.S. thesis, University Park, PA: Pennsylvania State University.
Arnold, B.J. 1995a. A Guide to Coal Handling. TR-105110. Palo Alto, CA: Electric
Power Research Institute.
Arnold, B.J. 1995b. Predicting the moisture content of coal dewatered by vacuum fil-
ters. In Proceedings of the Twelfth Annual International Pittsburgh Coal Conference.
pp. 455–460.
Arnold, B.J. 1996. Predicting the moisture content of coal dewatered by centrifuges.
SME Preprint 96-139. Littleton, CO: SME.
Arnold, B.J. 1997. Predicting the moisture content of coals dewatered by screens. SME
Preprint 97-128. Littleton, CO: SME.
Arnold, B.J. 1999. Simulation of dewatering devices for predicting the moisture con-
tent of coals. Coal Prep. 20:35–54.
Arnold, B.J., and Aplan, F.F. 1986a. The effect of clay slimes on coal flotation, Part I:
The nature of the clay. Int. J. Miner. Process. 17:225–242.
Arnold, B.J., and Aplan, F.F. 1986b. The effect of clay slimes on coal flotation, Part II:
The role of water quality. Int. J. Miner. Process. 17:243–260.
Arnold, B.J., and Aplan, F.F. 1989. The hydrophobicity of coal macerals. Fuel
68:651–658.
ASTM D388. 2005. Standard Classification of Coals by Rank. West Conshohocken,
PA: ASTM International. Available from www.astm.org/Standard/index.shtml.
Bradley, P.B., Hogg, R., and Aplan, F.F. 1980a. Mineralogical characterization of black-
water solids. Trans. AIME 268:1831–1836.
Bradley, P.B., Hogg, R., and Aplan, F.F. 1980b. Particle size consist of black-water sol-
ids. Trans. AIME 268:1836–1841.
Brown, D.J., and Smith, H.G. 1954. Continuous testing of frothers. Colliery Eng.
31:245–250.
Burdon, R.G., Booth, R.W., and Mishra, S.K. 1976. Factors influencing the selection
of processes for the beneficiation of fine coal. In Proceedings of the Seventh Inter-
national Coal Preparation Congress, Sydney, Australia.
Davis, A., Spackman, W., and Given, P.H. 1976. The influence of the properties of coals
on their conversion into clean fuels. Energy Sources 3(1):55–81.
Esposito, M.C., Chander, S., and Aplan, F.F. 1987. Characterization of pyrite from coal
sources. In Process Mineralogy VII. Edited by A.H. Vassiliou et al. Warrendale, PA:
TMS-AIME. pp. 475–493.
Firth, B.A., and Nicol, S.K. 1981. The influence of humic materials on the flotation of
coals. Int. J. Miner. Process. 8:239–248.
Given, P.H. 1984. An essay on the organic geochemistry of coal. In Coal Science, Vol.
3. Edited by M.L. Gorbaty, J.W. Larsen, and I. Wender. Orlando, FL: Academic
Press. pp. 63–252.
Gutierrez-Rodriguez, J.A., and Aplan, F.F. 1984. The effect of oxygen on the hydropho-
bicity and floatability of coal. Colloids Surf. 12:27–51.
Hansen, R.D., and Klimpel, R.R. 1985. The influence of frothers on particle size and
selectivity in coal/sulfide mineral flotation. SME Preprint 85-9. Littleton, CO:
SME.
ISO (International Organization for Standardization). ISO 11760:2005(E). 2005.
Classification of Coals. Geneva, Switzerland: ISO.
Jowett, A., El-Sinbawy, H., and Smith, H.G. 1956. Slime coatings of coal in flotation
pulps. Fuel 35:303–309.
Luttrell, G., McKeon, T., and Bethell, P. 2000. An in-plant evaluation of froth wash-
ing for conventional coal flotation circuits. SME Preprint 01-191. Littleton, CO:
SME.
Lynch, A.J., Johnson, N.W., Manlapig, E.V., and Thorne, C.G. 1981. Mineral and Coal
Flotation Circuits. New York: Elsevier. pp. 21–56.
Mishra, S.K. 1978. The slime problem in Australian coal flotation. In Australasian
I.M.M. Mill Operators Conference. Mt. Isa. pp. 159–168.
Neczaj-Hruzewicz, J., Szczypa, J., and Czarkowski, N. 1974. Influence of flota-
tion reagents on formation of gangue slime coatings on coal. Trans. I.M.M.
83:C261–263.
O’Gorman, J.V., and Walker, P.L. 1972. Mineral Matter and Trace Elements in U.S.
Coals. Report No. 61, Interim Report No. 2. U.S. Department of Interior, Office
of Coal Research.
Purcell, R.J. 1982. Unpublished data. University Park, PA: Pennsylvania State
University.
Stach, E., Mackowsky, M-TH., Teichmuller, M., Taylor, G.H., Chandra, D., and
Teichmuller, R. 1975. Stach’s Textbook of Coal Petrology. Berlin: Gebrüder
Borntraeger.
Stanley, F. 2007. Personal communication.
Sun, S.C. 1954. Effect of oxidation of coals on their flotation properties. Trans. AIME
199:396–401.
Szczypa, J., Neczaj-Hruzewicz, J., and Sablik, J. 1973. Some properties of slime coating
in coal-gangue systems. Trans. I.M.M. 82:C167–169.
van Olphen, H. 1963. An Introduction to Clay Colloid Chemistry. New York: Inter-
science Publishers.
Wakeman, R.J. 1979. The prediction and calculation of cake dewatering characteristics.
Filtr. Sep. (Nov.-Dec.):655–669.
Wakeman, R.J., and Rushton, A. 1977. Dewatering properties of particulate beds.
J. Powder Bulk Solids Technol. 1(2):64–69.
Wen, W.W. 1977. Electrokinetic behavior and flotation of oxidized coal. Ph.D. thesis,
University Park, PA: Pennsylvania State University.
Yancey, H.F., and Taylor, J.A. 1935. Froth Flotation of Coal; Sulfur and Ash Reduction.
RI 3263. U.S. Bureau of Mines.
Maximizing Fine Pyrite
Rejection at the Arch Coal
Leer Plant
Cory Chafin, Peter Bethell, Greg DeHart,
and Randy Corder
ABSTRACT
Optimized fine pyrite rejection has been used by Arch Coal, Inc., at their Leer
plant to maximize coking coal recovery. Original circuit design produced 187 tph
(tons per hour) of minus 1 mm material that would be sold on the steam market.
By the redesign and optimization of the fines circuit it is anticipated that the Leer
plant will produce an increase of 149 tph of high volatile A coking coal and a net
reduction of 141 tph of middlings generating substantial revenue increases. Pyrite
concentration can be achieved by the use of classification and density separation
devices. A series of simulation and test programs has shown the benefit of pyrite
rejection through cyclone and sieve classifiers, as well as compound spiral concentra-
tors and reflux classifiers. The findings of these studies were used in the redesign of
the Leer Mining plant flowsheet. Additional sulfur reduction benefits are expected
when complete circuit optimization has occurred.
BACKGROUND
Year 2011 brought many changes to the coal business. One of these changes was
the Arch Coal, Inc. (ACI), acquisition of International Coal Group that was
completed in June 2011. Part of that acquisition was the Tygart #1 property,
which was in the process of being developed and is located in Taylor County,
W.V., near the town of Grafton. At the time of the acquisition, a flowsheet
had been developed and was in the very early stages of construction. The flow-
sheet was designed to clean coal from the Tygart #1 mine, which is mining in
the Lower Kittanning seam. The original flowsheet design envisaged placing
all minus 1 mm clean coal into a low realization steam market along with
67
middlings generated from the plus 1 mm dense-media circuits. The original

plant design utilized primary/secondary coarse coal dense-medium vessels and
primary/secondary dense-medium cyclones to wash plus 1 mm size fractions.
Primary low-density dense-medium product was to serve the coking coal
market with a high volatile A coal. The original fines circuit incorporated com-
pound spirals along with conventional froth flotation. The entire minus 1 mm
product was to source a low-realization steam market.
An immediate review of the fines (minus 1 mm) raw coal washability was
conducted. Upon initial examination, the washability indicated that by making
a lower-density separation (i.e., <1.50 specific gravity [SG]) and by minimizing
pyrite in the flotation cell feed, a substantial portion of the minus 1 mm mate-
rial could be shipped as a coking coal at +$100/ton higher revenue.
A new flowsheet was rapidly designed to achieve the aforementioned lower
ash and sulfur fines product. This circuit was bid out and Powell Construction
of Johnson City, Tenn., was awarded the contract to build the plant. Com-
missioning is planned for the new plant in October 2012, with full capacity
of 1,400 tph being needed when the longwall production begins in October
2013. The property is now known as Leer Mining.
ORIGINAL FINE COAL FLOWSHEET

The minus 1 mm material passing the desliming screens was to be split into
nominally ±150 µm (100 mesh) material via 381 mm (15 in.) classifying
cyclones. The plus 150 µm material was to pass to a compound spiral circuit
while the minus 150 µm material was to be treated in conventional flotation
cells. The product from the compound spirals was to be deslimed on two-stage
fine wire sieves with the effluent passing to an effluent cyclone circuit. The
overflow of the effluent cyclones would be treated in the above-mentioned
conventional flotation cells, while the underflow would to be recirculated back
to the spiral clean coal sieves.
Both the deslimed spiral product and the flotation concentrate were to
be dewatered in screen-bowl centrifuges. The projected clean coal quality for
the screen-bowl centrifuges combined spiral/froth product was 9.5% ash and
1.60% sulfur. These qualities prevented this material from being blended back
with the coarse coking coal product from the dense-media processes and sold
into a coking coal market.
Maximizing Fine Pyrite Rejection 69
Table 1 1 mm × 150 μm raw coal washability data, which represents 14.42% of the
total sample
Direct Data Cumulative Recovery
Specific
Gravity Weight, % Ash, % Suflur, % Weight, % Ash, % Sulfur, %
1.35 Float 62.72 4.98 1.38 62.72 4.98 1.38
1.40 Float 5.21 13.64 1.71 67.93 5.64 1.41
1.45 Float 2.82 19.53 2.19 70.75 6.20 1.44
1.50 Float 1.86 27.22 2.73 72.61 6.74 1.47
1.60 Float 1.01 34.67 3.00 73.62 7.12 1.49
1.70 Float 1.37 42.25 3.50 74.99 7.76 1.53
1.80 Float 0.34 47.89 3.62 75.33 7.94 1.54
2.00 Float 0.78 54.25 3.82 76.11 8.42 1.56
2.20 Float 0.59 64.95 4.92 76.70 8.85 1.59
2.20 Sink 23.3 89.65 3.93 100.00 27.68 2.13
UPSIDE POTENTIAL TO THE ORIGINAL FLOWSHEET

1 mm × 150 μm Circuits
A review of the mine washability data (Table 1), particularly the 1 mm × 150
µm fraction, showed that by operating in the 1.45–1.55 SG range by using
a reflux classifier, sulfur levels in the product could be reduced by ±0.1% in
comparison to the simulated spiral product. In addition to the lower sulfur
levels, superior ash levels would also be expected. A secondary circuit could be
incorporated to recover a higher-ash, higher-sulfur middlings product.
150 μm × 0 Circuit
Previous testing on nominal minus 0.6 mm (28 mesh) samples showed poor
sulfur rejection in flotation (Table 2) with concentrate sulfur levels in the
1.65–1.85% range when maintaining combustible recoveries greater than 80%.
Flotation feed was highly pyritic, and, unfortunately, the hydrophobic nature of
the pyrite made it highly problematic. This sulfur level was obviously unaccept-
able in a coking coal product where the rest of the plant circuits would have had
to be operated at low separation densities (i.e., <1.50 SG) to make an acceptable
overall sulfur specification (nominally <1.2%).
Release analysis testing on a pit bottom coal sample of minus 150 µm mate-
rial also showed very high product sulfur levels of >2.5% (see Table 3). A more
detailed review of the raw coal washability data showed that the high sulfur
levels in the flotation concentrate were almost certainly due to pyritic sulfur. In
Table 2 Results for flotation of minus 28 mesh Lower Kittanning seam coal
Product Reagent Drops Weight, % Ash, % Sulfur, %
7% Solids Loading
Tails 80.02 18.29 3.23
Concentrate 2 19.98 7.4 1.15
Tails 44.83 26.97 4.52
Concentrate 4 55.17 6.98 1.41
Tails 15.41 56.61 8.23
Concentrate 6 84.39 8.93 1.66
Tails 11.82 65.4 9.83
Concentrate 8 88.18 9.63 1.77
Tails 10.63 68.78 9.92
Concentrate 10 89.37 10.36 1.85
10% Solids Loading
Tails 75.55 18.38 3.23
Concentrate 2 24.45 9.98 1.39
Tails 46.08 25.91 5.34
Concentrate 4 53.92 8.4 1.47
Tails 23.33 39.86 6.47
Concentrate 6 76.67 8.94 1.63
Tails 17.78 47.57 7.73
Concentrate 8 82.22 10.06 1.82
Tails 13.61 62.22 9.58
Concentrate 10 86.39 9.02 1.63
fact, the organic sulfur of this coal is typically 0.75% (see Table 4). A review of
the raw coal washability showed excellent liberation of the pyrite into the finer
fractions of the feed stock.
Simulation modeling of the classifying cyclones in the plant showed that
liberated pyrite with 5.0 SG would be classified at a much finer size than the
coal and rock with 50% and 95% passing sizes of 35 µm and 45 µm, respectively,
for the pyrite. If this cyclone overflow stream was fed alone to the flotation cir-
cuit, feed and concentrate sulfur levels would be much less. In-plant testing has
also been conducted to confirm this simulation. Testing around a bank of clas-
sifying cyclones with a coal feed similar to that of the Leer washability showed
a major reduction in cyclone overflow sulfur levels (Table 5). Sulfur levels were
reduced from 2.28% in the feed to 1.11% in the overflow, with 5.60% sulfur in
the underflow of the minus 150 µm fractions.
Table 3 Flotation release analysis of minus 100 mesh Lower Kittanning seam coal,
representing 8.14% of the sample*
Direct Values† Cumulative Values†
Sulfur, Sulfur,
Increment Wt, % Ash, % % CR‡ Wt, % Ash, % % CR‡
Concentrate 1 37.29 8.59 2.66 67.70 37.29 8.59 2.66 67.70
Concentrate 2 7.38 10.79 2.55 13.08 44.67 8.95 2.64 80.78
Concentrate 3 2.01 15.26 3.16 3.38 46.68 9.23 2.66 84.16
Concentrate 4 3.49 24.01 3.31 5.27 50.17 10.25 2.71 89.43
Concentrate 5 1.48 31.82 3.00 2.00 51.65 10.87 2.72 91.43
Tailings 2 1.61 76.28 4.71 0.75 53.26 12.85 2.78 92.18
Tailings 1 46.74 91.58 0.59 7.82 100.00 49.65 1.76 100.00
* Collector: diesel; frother: MIBC.
† Dry basis.
‡ CR = combustible recovery.
Table 4 Clean coal statistical summary

Standard Relative Standard
Analysis Mean Variance Deviation Deviation
Total sulfur, % 1.309 0.007 0.085 0.065
Pyritic sulfur, % 0.558 0.005 0.073 0.130
Organic sulfur, % 0.752 0.005 0.072 0.096
Heating value, Btu/lb 13957 29672 172.260 0.012
Ash value, % 9.367 0.068 0.780 0.083
Total moisture, % 7.180 4.672 2.161 0.301
Emission rate, lb 1.875 0.017 0.131 0.070
SO2/MM Btu
The minus 150 µm material in the cyclone underflow in the plant would
pass through a cleaning device, a desliming device, and then would typically
report to froth flotation. A review of the ultrafines sulfur levels from the
aforementioned plant shows some sulfur removal in the spiral circuit for the
ultrafines (i.e., 5.65% in the feed down to 2.63% in the product). However,
this 2.63% sulfur material passes through the desliming sieve (see Table 6) and
would ultimately end up in the flotation circuit, resulting in elevated sulfur
levels.
Testing performed on minus 150 µm material on compound spirals has
shown great promise in reducing pyritic sulfur levels. Figures 1 and 2 represent
an Illinois plant in which spiral clean coal sieve effluent samples were treated on
compound spirals and then with flotation or treated in spirals or with flotation
Table 5 508 mm (20 in.) raw coal classifying cyclone samples

Yield*
Feed (6.78% solids) Underflow (65.97% solids) Overflow (1.99% solids) (72.85%)
Wt, Ash, Sulfur, Wt, Ash, Sulfur, Wt, Ash, Sulfur, Partition
Size Fraction % % % % % % % % % Coefficient
+1.5 mm 0.99 35.41 1.75 1.47 35.27 1.90 0.00 0.00 0.00 100.00
1.5 × 1 mm 5.73 21.69 2.03 9.27 25.38 2.63 0.00 0.00 0.00 100.00
+1 mm 6.72 23.71 1.99 10.74 26.73 2.53 0.00 0 0 100.00
1 × 0.25 mm 46.89 18.75 2.21 64.81 18.84 2.37 0.60 9.68 1.19 99.66
0.25 × 0.15 mm 10.05 18.33 2.73 12.41 20.26 3.33 2.79 3.95 0.85 92.27
+0.15 mm 63.66 19.21 2.27 87.96 20.00 2.52 3.39 4.96 0.91 98.58
0.15 mm × 11.52 18.79 2.76 8.43 31.94 4.54 17.71 5.01 0.96 56.08
75 μm
75 × 45 μm 5.38 21.30 3.29 2.03 52.24 7.64 13.56 8.13 1.08 28.65
45 μm × 0 19.44 28.74 1.71 1.58 54.09 9.09 65.34 30.56 1.15 6.09
Composite 100 21.13 2.27 100.00 22.20 2.90 100 22.13 1.10 72.84
0.15 mm × 0 36.35 24.49 2.28 12.04 38.27 5.66 96.62 22.73 1.11 25.05
*Based on % solids.
Table 6 Fine wire sieves data

Yield*
Feed (15.41% solids) Product (57.55% solids) Effluent (4.18% solids) (72.85%)
Wt, Ash, Sulfur, Wt, Ash, Sulfur, Wt, Ash, Sulfur, Partition
Size Fraction % % % % % % % % % Coefficient
+1 mm 9.47 9.49 1.31 13.43 9.69 1.52 0.00 0.00 0.00 100.00
1 × 0.25 mm 68.76 7.92 1.34 77.14 7.74 1.31 18.72 8.10 1.24 91.94
0.25 × 0.15 mm 8.39 7.71 1.27 6.53 7.77 1.35 39.17 10.71 1.59 31.58
+0.15 mm 86.62 8.07 1.33 97.10 8.01 1.34 57.89 9.87 1.48 82.28
0.15 mm × 0 13.38 24.06 2.63 2.90 15.29 2.05 42.11 23.74 2.79 16.01
Composite 100 10.21 1.50 100 8.22 1.36 100 15.71 2.03 73.46
*Based on ash.
alone. The immediate reduction in froth product sulfur levels is apparent by

pre-treating flotation feed with spirals. Similar improvements were seen by
feeding the same coal seam being mined at Leer (Kittanning) through spirals
prior to flotation (Figure 3).
From Figure 3 it can be seen that flotation sulfur concentrations were
reduced by 0.2% sulfur by pre-spiraling the flotation feed. These huge, antici-
pated benefits in the fines product sulfur reduction achieved by removing as
much pyrite as possible ahead of the flotation circuit have been incorporated
into the new flowsheet to maximize coking coal recoveries.
100
90
80
70
60
Yield, %
50
40
30
20
Flotation Only
10 Spiral Only
Spiral + Flotation
0
0 10 20 30 40 50
Ash, %
Figure 1 Illinois plant ash removal results
THE NEW FLOWSHEET

Based on the aforementioned test work and the dramatic financial impact of
moving material from the steam market into the coking coal market, the origi-
nal flowsheet was abandoned and the circuits shown in Figures 4 and 5 were
constructed.
Coarse coal cleaning is achieved by combining the 76 mm (3 in.) × 1 mm
material and washing it through a 1,150 mm primary/900 mm secondary
(high/low) dense-medium cyclone circuit. The initial cut made by the primary
cyclones will be at a high gravity, removing the coarse rock from the circuit. The
secondary cyclones will be operated at a lower gravity, making the metallurgical
coal/middlings separation (Bethell 2010).
The minus 1 mm material will report to a bank of 381 mm (15 in.) raw coal
classifying cyclones where the pyrite concentration will begin. These classifying
100
90
80
70
60
Yield, %
50
40
30
20
Flotation Only
10 Spiral Only
Spiral + Flotation
0
0 2 4 6 8
Sulfur, %
Figure 3 Illinois plant sulfur removal results
100
90
Cumulative Yield, wt %
80
70
60 Pre-Spiral Release Sulfur
50
Post-Spiral Release Sulfur
40
Pre-Spiral Rate Sulfur
30
20 Post-Spiral Rate Sulfur
10
0
1 1.2 1.4 1.6 1.8 2 2.2 2.4
Cumulative Sulfur, wt %
Figure 2 Lower Kittanning seam flotation performance pre/post-pyrite rejection
Figure 4 Coarse coal flowsheet (plus 1 mm circuit)—ACI Leer plant
cyclones were chosen because of the superior pyrite rejection they provide.
The nominal 1 mm × 150 µm material along with a large portion of the minus
150 µm pyrite-rich material will be treated in primary/secondary reflux clas-
sifiers. The primary cleaning devices in this circuit will make the coking coal
separation operating at approximately 1.50 SG. The reject from the primary
reflux classifiers will be retreated by secondary devices where the middlings
material will be recovered. The primary and secondary reflux classifier products
Figure 5 Fine coal flowsheet (minus 1 mm circuit)—ACI Leer plant
will then be deslimed on independent metallurgical coal and middlings fine

wire sieves. The material carrying over the sieves will be combined with the
later-mentioned primary and secondary column flotation concentrates and
dewatered in independent screen-bowl centrifuges producing a coking and
steam coal component (Galvin et al. 2008).
The low-sulfur product from the 381 mm (15 in.) classifying cyclone
overflow will be gravity fed to a bank of 152 mm (6 in.) classifying cyclones.
These cyclones serve to deslime at nominally 45 µm (325 mesh), discarding

high-ash clays to the thickener. Desliming at 45 µm ahead of column flotation
was selected for several reasons. These reasons include the fact that the Leer coal
has a high Hardgrove grindability index of 80, which leads to a higher percent-
age of fines and difficulty making customer moisture specifications without a
desliming circuit to eliminate the high-moisture ultrafines. Also, the expected
nominal minus 45 µm material will be high in ash and have low coal recoveries.
The underflow of the 152 mm (6 in.) desliming cyclones will be treated in 4.6 m
(15 ft) primary column flotation cells. These SlamJet columns were selected in
view of their great success in other ACI plants (Cardinal, Lone Mountain, and
Pardee). SlamJet columns provide a great deal of selectivity coupled with high
efficiencies and ease of operation (Bethell and DeHart 2006). These attributes
have made this type of flotation unit very successful in this particular flotation
application. This 152 mm (6 in.) cyclone underflow feed stream was viewed as
being the lowest sulfur stream in the plant and was therefore kept as the sole
feed stream to the coking coal flotation circuit. Concentrate from the primary
flotation cells will be combined with the primary reflux classifier clean coal
sieves product and be dewatered in the coking coal screen-bowl centrifuges.
The overflow of the 152 mm (6 in.) deslime cyclones will be discarded to the
thickener (Baumharth et al. 2005; Bethell and Luttrell 2005).
Effluent from the primary and secondary fine wire sieves will be combined
and pumped to another bank of 152 mm (6 in.) cyclones. The purpose of these
cyclones is to aid in solids concentration so that this effluent stream can be
treated through compound spiral concentrators. From the above-mentioned
test work and as seen in Figure 3, sulfur levels can be reduced by 0.2% in this
stream by this technique. The refuse of these spirals, rich in pyrite, will be dis-
carded on a high-frequency screen and later to the coarse refuse disposal site.
The product of the desulfurization spirals as well as the overflow of the 152
mm (6 in.) concentrator cyclones and the primary column flotation cell tailings
will be treated in 4.6 m (15 ft) secondary column flotation cells. Because of the
anticipated higher sulfur levels, the concentrate of the secondary flotation cells
will be mixed with the middlings fine wire sieve product, which comes from the
middlings reflux classifiers, and dewatered in the middlings screen-bowl centri-
fuge. It is hoped after full circuit optimization that the desulfurization effect of
the spirals will allow some (if not all) of the secondary flotation concentrate to
be directed to the coking coal product. The tailings of the secondary flotation
cell will be discarded to the thickener. The secondary screen-bowl centrifuge
screen drain (rich in pyrite) will be recirculated back to the raw coal screens for
further desulfurization (Honaker et al. 2007; Yoon and Luttrell 1994).
CONCLUSION
Because of the concentration and isolation of liberated pyritic sulfur in the
minus 1 mm material at Leer Mining, a large percentage of fine coal will be
moved from a low-realization steam market for which it was initially destined
to a higher-realization coking coal market. The anticipated results of the rede-
signed fines circuit will take 187 tph originally destined for the steam market
due to high ash and sulfur levels, and put 149 tph and 46 tph on the coking
coal and steam markets, respectively, generating substantial revenue increases.
This has been achieved by selecting equipment for the redesigned circuits that
will allow for maximum concentration and removal of ultrafine pyritic sulfur
from the coking coal circuitry. The concentrated pyrite is then treated through
compound spirals where the differential density of pyrite and coal is exploited
prior to secondary column flotation. This provides further potential enhance-
ments to the coking coal yield. The ability to run lower densities and therefore
lower sulfur levels has also been exploited by the incorporation of reflux classi-
fiers. Additional levels of sulfur rejection are expected when full plant process
is achieved and circuits can be fully optimized.
REFERENCES
Baumharth, T., Bethell, P.J., and Gupta, B. 2005. Recovering an additional 20 tph
of coal through a deslime column flotation circuit addition at Lone Mountain
Processing–Virginia. In Proceedings, Coal Prep AGG 2005. Stamford, CT: Penton
Media. pp. 39–50.
Bethell, P.J. 2010. Arch Coal processing philosophy, east and west. In Proceedings, XVI
SME. pp. 1–8.
Bethell, P.J., and DeHart, G. 2006. The design, construction, and commissioning of
the new 2000 tph Arch Coal Cardinal preparation plant. In Proceedings of the XV
International Coal Preparation Congress and Exhibition. Beijing: China National
Coal Association. pp. 79–88.
Bethell, P.J., and Luttrell, G. 2005. Proceedings, Centenary of Flotation Symposium. Bris-
bane, Australia, June 5–9. The Australasian Institute of Mining and Metallurgy.
Galvin, K.P., Callen, A.M., and Spear, S. 2008. Extending the size range of the reflux
classifier. ACARP Project C16040.
Honaker, R.Q., Boaten, F., and Luttrell, G.H. 2007. Ultrafine coal classification using
150mm gMax cyclone circuit. Miner. Eng. 20:1218–1226.
Yoon, R.H., and Luttrell, G.H. 1994. Microcel column flotation scale-up and plant
practice. CMP Proceedings of the 26th Annual Meeting. Paper 12.
Technology
Developments and
Plant Installations 2
StackCell Flotation—A New
Technology for Fine Coal
Recovery
Michael Kiser, Robert Bratton, Gerald Luttrell,
Jaisen Kohmuench, Eric Yan, Lance Christodoulou,
Van Davis, and Fred Stanley
A bstract
During the past decade, column flotation cells have become widely accepted for
the upgrading of fine coal streams. This popularity can be largely attributed to the
ability of columns to remove high-ash clays from the froth product via the addition
of wash water to a relatively deep froth. Although there are numerous successful
column installations, discussions with both end-users and engineering firms have
identified certain design criteria that can make these installations challenging.
The greatest of these challenges is the overall size of the cells and the associated
foundation loads. To address this problem, a new high-intensity flotation system
known as the StackCell has been developed. This technology makes use of pre-
aeration coupled with a high-shear feed canister. This arrangement provides effi-
cient bubble–particle contacting, thereby substantially shortening the residence
time required for coal collection and virtually eliminating most of the column
height. This article reviews the design features of this innovative technology and
presents recent data obtained from full-scale installations.
I ntroduction
Column flotation has become the dominant method of recovering fine frac-
tions in the coal industry. The use of columns has led to increased metallurgi-
cal performance when compared to that of mechanical flotation cells. This
improvement in product quality has been proven by comparing plant flotation
data to a release analysis curve (Dell et al. 1972). Studies have also been per-
formed that show how the use of column flotation affects the bottom line of
81
82 Technology Developments and Plant Installations
a plant (Luttrell et al. 1999; Kohmuench et al. 2004; Baumgarth et al. 2005).
These studies report that the plants benefit from an increase in overall yield
because of the improvement in product grade from the flotation circuit. This
increase in product grade is linked to the application of wash water used in
column flotation. This countercurrent flow of water is applied to the froth and
minimizes the non-selective recovery of high-ash ultrafine material that is nor-
mally hydraulically entrained in the froth of conventional flotation machines.
Column flotation does have its own set of design challenges though. The
first of these challenges is simply the size of the column. The cell must be tall
to achieve the desired residence time and minimize internal mixing, which
can be detrimental to cell performance. This design minimizes the plant floor
space required for the cell but increases the foundation loads. The large size of
the column also leads to difficulties with fabrication and installation of column
cells. The economics associated with plant design typically lean toward fewer
large-diameter cells. The largest diameter cell that can be shipped in the United
States as a single piece is 4.5 m (15 ft). Larger cells can be installed, but these
cells must be shipped in multiple sections and require more on-site assembly.
Additionally, larger-diameter cells must be taller to maintain the proper aspect
ratio, at least 2:1, which then adds to the overall foundation load.
The design challenges previously mentioned show that there is a need for
a new generation of flotation machine. A machine is needed that is capable of
delivering column-like performance, while also improving some of the design
and operational challenges associated with column flotation. Based on experi-
ence gained over the last decade with the design, engineering, and operation
of coal flotation circuits, Eriez has developed a new flotation cell that offers
high capacity, reduction in both size and horsepower, and superior metallurgi-
cal performance. Although column flotation will be a requirement for some
applications, this new approach offers an alternative that provides column-like
performance with reduced capital, installation, and operating costs.
technology D E S C R I P T I O N
Figure 1 illustrates the working features of the StackCell technology. During
operation, feed slurry is introduced to the cell through a side (or bottom) feed
port. At this point, low-pressure air is added to the feed slurry. The aerated
feed slurry then travels into the aeration chamber where significant shear is
imparted to the system. The shear forces imparted to the system are used to cre-
ate bubbles for bubble–particle collisions. In fact, all of these bubble–particle
collisions occur in the aeration chamber prior to discharge into the outer tank.
StackCell Flotation—A New Technology 83
Aeration/
Contacting
Chamber Wash-Water
Manifold
Froth
Launder
Tails
Feed Inlet
Outlet
Air Manifold
Figure 1 Eriez StackCell
When the slurry enters the outer tank, phase separation occurs between the
froth and pulp. A pulp level is maintained in the outer tank to provide a deep
froth that can be washed to minimize the entrainment of ultrafine high-ash clay
material. The froth overflows into a froth collection launder, while the tailings
are discharged using either a control valve or mechanical weir system. The sys-
tem is specifically designed to have both a small footprint and a gravity-driven
feed system. This allows multiple units to be “stacked” in series on subsequent
levels in the plant or placed ahead of existing column or conventional flotation
circuits.
Why Multistage?
The enhanced performance made possible by the stacked arrangement can be
mathematically quantified using the standard tanks-in-series flotation model
(Lynch et al. 1981). According to this model, the recovery (R) of a given species
from a single well-mixed flotation tank can be estimated by
kτ
R= (EQ 1)
1 + kτ
where k is the flotation rate constant and τ is the residence time. The rate con-
stant represents how quickly particles float and is normally reported in units of
min–1 (i.e., mass floated per unit mass in the cell per unit time). This parameter
is largely dependent on the coal properties, chemical types/dosages, aeration

rate, and the design and operation of the bubble–particle contacting system.
The residence (or retention) time, which is usually reported in minutes, repre-
sents the average amount of time that particles stay within the flotation pulp.
As a general rule, a mean residence time of about 4 minutes is typically required
in conventional flotation machines to achieve good recoveries of bituminous
coals. The required residence time may be longer for difficult-to-float coals that
are very fine or oxidized.
According to Equation 1, a 90% recovery would require a relatively large
kτ value of 9 (i.e., 9/(1 + 9) = 90%). One effective method of reducing the kτ
requirement is to arrange the flotation cells in series to reduce potential losses of
floatable particles to the reject stream. In this case, the total recovery (RN) for a
bank of N tanks in series can be determined from the arithmetic series given by
RN = Ri + Ri (1 − Ri ) + Ri (1 − Ri ) + Ri (1 − Ri )
2 3
(EQ 2)
+  + Ri (1 − Ri ) = 1 − (1 − Ri )
N N
where Ri is the fraction recovery defined at τi = τ/N. Combining Equations 1

and 2 gives
N
 N 
RN = 1 −  (EQ 3)
 N + kτ 
This expression is plotted in Figure 2 as a function of residence time for differ-

ent numbers of cells (N) in series and an assumed rate constant of 0.8 min–1.
Based on these estimates, a single cell would achieve a recovery of only 76.2%
after 4 minutes of residence time. For the same total 4 minutes of residence
time, the recovery would increase to 85.2% after two cells, 88.7% after three
cells, 90.5% after four cells, and 91.6% after five cells. As such, this analysis sug-
gests that three to four cells in series provides a good balance, since additional
cells provide little incremental improvement in recovery compared to the
increased cost of purchasing more cells.
Previous studies by Stanley et al. (2006) demonstrated the advantages
of cell-to-cell circuitry for full-scale column flotation plants. Unfortunately,
cell-to-cell circuitry is difficult to apply for columns because of their tall aspect
ratio and large volumetric footprint. On the other hand, the modular design of
the StackCell easily accomodates the in-series configuration to take advantage
of improved mixing conditions. Therefore, as shown in Figure 3, the preferred
100
N=∞
54
3
90 2
N=1
80
Recovery, %
70
60
N
N
R=1–
N + kτ
50
0 1 2 3 4 5 6
kτ
Figure 2 Effect of residence time and number of cells in series on the recovery of
floatable material (assumes k = 0.8 min–1)
arrangement of the StackCell technology is three to four sequential stages for

new installations. The technology can also be employed as a retrofit scalping
unit placed ahead (or behind) existing column cells or mechanical flotation
machines for additional capacity in the flotation circuit.
Why Intense Agitation?

Another unique feature of the StackCell technology is the use of a high-shear,
bubble–particle contactor in place of the conventional rotor-stator mechanism
historically used by mechanical flotation cells. Instead of operating with a large-
volume tank, the StackCell forces the bubbles and particles to contact within a
very small confined area within an aeration chamber. Under this highly turbu-
lent environment, the flotation rate constant (k) can be expressed as
Feed
Combined Tails
Product
Figure 3 Preferred layout of StackCell flotation modules
k ∝ CbC p E (EQ 4)
where Cb is the concentration of bubbles, Cp is the concentration of particles,

and E is the specific energy imparted to the system (Williams and Crane 1983).
The high-shear environment within the aeration chamber provides an energy
dissipation level that is substantially higher than that produced by conven-
tional flotation machines, thereby enhancing the recovery of difficult-to-float
particles. The contactor is specially designed to efficiently impart energy for
bubble–particle contacting and to avoid unnecessary pumping or unwanted
recirculation of the feed slurry (Kohmuench et al. 2008). This allows the input
energy to be used for gas dispersion and contacting and not for particle sus-
pension. Moreover, the intense mixing shears the low-pressure air blown into
the machine into extremely small bubbles, which substantially increases the
concentration of bubbles present in the contacting chamber. This approach
ensures that the maximum concentration of floatable particles and gas bubbles
are present during the high-shear contacting.
According to Equation 1, the very high rate constant (k) created by the
high-shear environment within the aeration chamber allows the StackCell to
operate at a correspondingly lower residence time (τ) without adversely impact-
ing the recovery. Field studies conducted with a pilot-scale unit showed that a
residence time of less than 10 seconds was often adequate for good contacting
when using StackCell technology. Consequently, the required cell volume for
a StackCell installation is significantly less, thereby reducing both equipment
and installation costs. Structural steel requirements are considerably less due to
the reduction in tank weight and live load. For a typical installation, the overall
space requirement for the stacked-cell design is half the volume of an equivalent
column circuit. Shipping and installation are also simplified, because the units
can be shipped fully assembled and lifted into place, complete and without field
welding. Moreover, the energy input per unit ton processed is typically lower
for the StackCell, since energy is only expended for the purpose of creating
bubbles and for bubble–particle contacting, and not for particle suspension like
conventional flotation cells. In addition, the aeration chamber operates under a
near-atmospheric pressure in a manner that removes the need for a compressor
to overcome the hydrostatic or dynamic head. As a result, a low-pressure and
maintenance-friendly blower can be used as opposed to a compressor.
Why Froth Washing?

Much like column flotation, StackCell technology makes use of a froth wash-
ing system to avoid the hydraulic carryover of ultrafine high-ash slimes into the
froth product (Kohmuench et al. 2004). In the case of the StackCell, an over-
head drip pan wash-water distributor is utilized to reduce plugging problems
that are often associated with submerged wash-water distributors (see Figure 4).
The drip pan design also provides excellent coverage of the entire sufrace area
of the cell and is easier for plant workers to maintain. For best performance,
sufficient wash water must be added to fully displace the water carried by the
froth into the clean coal launder. This requirement is normally reported as the
number of dilution washes (i.e., wash-water flow rate divided by the froth-water
flow rate). Normally, 1.1 to 1.3 dilution washes are required for good perfor-
mance in coal applications. Despite the low profile, the StackCell is designed
so that a relatively deep froth (45–75 cm [18–30 in.]) can be maintained to
maximize the froth washing action. Also, limitations associated with froth
overloading that frequently occurs with column-type cells are greatly reduced
with the StackCell system because of the greater surface area resulting from the
use of multiple cells.
Figure 4 Drip pan used for wash-water distribution
I ndustrial E VA L U AT I O N
To demonstrate the performance capabilities of StackCell technology, a full-scale
unit was installed and commissioned at an industrial coal preparation plant. The
plant processed run-of-mine coals from several seams supplied by both under-
ground and surface mines. The StackCell unit consisted of a single 3.7-m- (12-
ft-) diameter cell equipped with a 76-cm- (30-in.-) diameter aeration chamber.
The single StackCell unit was installed as a scalping system ahead of two existing
flotation columns. Historical data suggested that the two column cells were often
overloaded because of plant production demands. The tailings stream from the
StackCell was equally split and fed to the two existing columns.
Figure 5 shows the impact of the StackCell installation on the combustible
recovery and refuse ash for the entire flotation circuit. For the first 149 samples
taken prior to the installation, the two-column cells provided an average recov-
ery of 74.4% and a combined refuse ash of 72.5%. After the installation, the
combined recovery for the StackCell and two-column cells improved to 83.7%
and the refuse ash increased to 80.7%. The increased recovery is significant
considering that less than 10% more cell volume was added to the circuit via
the installation of StackCell technology. In fact, the aeration chamber provided
an additional residence time of only about 5–10 seconds to the total flotation
circuit. More recently, the average monthly plant recoveries have increased to
more than 90% (i.e., 90.88%), while the average monthly tailings ash values
have increased to nearly 86% (i.e., 85.9%).
Close inspection of the test data indicates a gradual improvement in overall
performance since the StackCell was installed. The continued improvement
can be largely attributed to the optimization of operating variables such as
reagent dosage, froth depth, aeration rate, and wash-water addition rate that
occurred over time as a result of fine tuning by the plant operators. For example,
100
80
Recovery, %
60
Avg = 74.4% Avg = 83.7%
40
20
0
100
80
Reject Ash, %
60
Avg = 72.5% Avg = 80.7%

40
20
0
0 50 100 150 200 250 300
Sample Number
Figure 5 Change in flotation circuit performance due to the installation of StackCell

technology (the dashed line represents the sample where the changeover occurred)
Figures 6 and 7 show the impact of optimization on the clean coal quality and
recovery for the +0.30 mm (+48 mesh), 0.30 × 0.15 mm (48 × 100 mesh),
0.15 × 0.045 mm (100 × 325 mesh), and –325 mesh size fractions. The high
ash content in the –325 mesh fraction was substantially reduced from about
43.4% to less than 13.3% once the froth washing system was optimized. This is
due to the elimination of entrained ultrafine, non-floatable high-ash material
in the –325 mesh fraction. Before the opitmization of the wash-water addition,
roughly 13% of the concentrate was made up of this entrained material. After
the wash water was optimized, roughly 2.84% of the concentrate was made
up of this high-ash material. The plant data continue to show that the quality
of the froth product is sensitive to froth depth and wash-water addition rate.
Therefore, it is important that these values be properly monitored and con-
trolled to optimum settings.
50
Before Optimization
45
After Optimization
40
35
30
Clean Coal Ash, %
25
20
15
10
0
+48 48 × 100 100 × 325 –325
Size Class, mesh
Figure 6 Effect of wash-water optimization on StackCell size-by-size clean coal ash
In light of the importance of froth depth and wash-water addition rate,

several series of parametric tests were also conducted to demonstrate how the
StackCell would react to changes in these important variables. The plant’s nor-
mal operating point was used as the baseline for the parameter sweep. The cell
was swept through a total of four operating points for each variable, while the
other variables were held constant at their normal operating condition. The test
matrix is summarized in Table 1.
Figure 8 shows how the cell performed with respect to a change in froth
depth. Increasing the froth depth led to an improved concentrate ash and
reduced the amount of non-floatable material present in the concentrate. The
minimum froth depth tested, 30 cm (12 in.), produced the highest concentrate
ash of 8.37% and contained the largest amount of –325 mesh hydrophilic
material (6.61% of the total concentrate weight). As froth depth increased, the
14
Before Optimization
After Optimization
Weight of Non-Floatable Material Present in Concentrate, % of Conc.
12
10
0
+48 48 × 100 100 × 325 –325
Size Class, mesh
Figure 7 Effect of wash water on the size-by-size amount of non-floatable material

present in the concentrate
Table 1 Test matrix used for StackCell testing

Test ID Froth Level, cm (in.) Wash-water Rate, m3/h (gpm)
A 76 (30) 82 (360)
B 61 (24) 82 (360)
C 46 (18) 82 (360)
D 30 (12) 82 (360)
E 61 (24) 82 (360)
F 61 (24) 0
G 61 (24) 59 (260)
H 61 (24) 78 (345)
I 61 (24) 91 (400)
9 9
8 8
7 7
6 6
Dilution Washes
5 5
Percent
Total Conc. % Ash

4 4
Non-Float –325% Weight of Total
3 Dilution Washes 3
2 2
1 1
0 0
0 5 10 15 20 25 30 35
Froth Depth, in.
Figure 8 Concentrate quality and dilution wash data for various froth depths
flotation process became more selective. At a more acceptable froth depth of

60 cm (24 in.), a better concentrate ash of 7.12% was produced, whereas the
–325 mesh hydrophilic material in this sample only made up 4.88% of the total
concentrate weight. As expected, the deepest froth depth tested, 76 cm (30 in.),
produced the best results with respect to concentrate ash and hydrophilic mate-
rial present in the concentrate. This deep froth resulted in a concentrate ash of
5.42% and the –325 mesh hydrophilic material comprised only 3.53% of the
total concentrate weight. The spike seen in the dilution-washes data is likely
linked to the drop in total weight recovery caused by the improved ash rejec-
tion. When a smaller amount of high percent solids product is being produced,
less water reports to the concentrate, and the effective number of dilution
washes increases.
Figure 9 shows how the StackCell performance changes with varying
wash-water rates. As with the previous set of data, the cell follows the expected
trend with respect to total concentrate ash; that is, ash content decreased as the
amount of wash water increased. Overall, the concentrate ash was a maximum
of 11.0% when no wash water was added to the cell. The weight of undesirable
–325 mesh hydrophilic material present in the concentrate was also highest at
8.7% when no wash water was added. The minimum ash value occurred with
the addition of 78 m3/h (345 gpm) of wash water, resulting in a 6.7% con-
centrate ash. The 59 m3/h (260 gpm) test also produced a very similar, albeit
slightly higher, concentrate ash of 6.9%.
12 12
Total Conc. % Ash
10 Non-Float –325% Weight of Total 10
Dilution Washes
8 8
Dilution Washes
Percent
6 6
4 4
2 2
0 0
0 100 200 300 400 500
Wash-Water Rate, gpm
Figure 9 Concentrate quality and dilution wash data for various wash-water
addition rates
The concentrate samples from the three non-zero rates tested were made
up of similar amounts of –325 mesh hydrophilic material ranging from 4.7%
to 5.6% of the total concentrate weight at the 78 m3/h (345 gpm) and the
91 m3/h (400 gpm) tests, respectively. Interestingly, the ash content of the
concentrate increased slightly at the highest wash-water rate. One possibility
for the higher value was that pressure variations due to pump cycling/surg-
ing in the preparation plant created fluctuations in the wash-water flow rate.
Although this could be a contributing factor to the high ash value, it is unlikely
that it is solely responsible since the other data points were subjected to the
same testing conditions. A more likely explanation for the unexpected increase
in ash is short-circuiting of the wash water into the concentrate. This possibil-
ity is supported by the fact that the dilution washes also did not increase as the
water rate was increased to the highest rate. Thus, more of the wash water must
have reported to the concentrate, which reduced the dilution washes for that
test. Further testing is suggested to determine whether this phenomenon is site
specific or an inherent characteristic of this particular flotation machine.
SUMMARY
A new high-capacity flotation technology, called the StackCell, has been devel-
oped as an alternative to both conventional and column flotation machines. This
technology makes use of pre-aeration and a high-shear aeration chamber that
provides efficient bubble–particle contacting, thereby substantially shortening

the residence time required for coal flotation. Other potential advantages of the
process include low air pressure requirements, low capital and installation costs,
and increased flexibility in plant retrofit applications. Recent full-scale plant
trials suggest that the technology can provide product qualities comparable to
column flotation systems using a low-profile design. Further testing shows that
the StackCell follows the expected trends with respect to concentrate ash when
it experiences changes in froth depth and wash-water addition rate. Although
this new technology is not expected to replace the need for column flotation, it
does provide an alternate means to efficiently achieve column-like performance
when plant space and/or capital is limited. In particular, the small size and low
weight of this new technology makes it amenable to low-cost plant upgrades
where a single unit can be placed into a currently overloaded flotation circuit
with minimal retrofit costs.
R eferences
Baumgarth, T., Bethell, P., and Gupta, B. 2005. Recovering an additional 20 tph coal
through a deslime column flotation circuit addition at Lone Mountain Process-
ing–Virginia. In Proceedings, 22nd Annual International Coal Preparation and
Aggregate Processing Exhibition and Conference, Lexington, KY, May 2–5. pp.
41–50.
Dell, C.C., Bunyard, M.J., Rickelton W.A., and Young, P.A. 1972. Release analysis: A
comparison of techniques. Trans. IMM (Sect C), 81:787.
Kohmuench, J.N., Davy, M.S., Ingram, W.S., Brake, I.R., and Luttrell, G.H. 2004.
Benefits of column flotation using the Eriez Microcel. Tenth Australian Coal
Preparation Conference, Proceedings, Polkolbin, NSW, Australia, October 17–21.
pp. 272–284.
Kohmuench, J.N., Mankosa, M.J., and Yan, E.S. 2008. An alternative for fine coal Flota-
tion. Coal Prep. Soc. Am. J. 7(1):29–38.
Luttrell, G.H., Kohmuench, J.N., Stanley, F.L., and Davis, V.L. 1999. Technical and
economic considerations in the design of column flotation circuits for the coal
industry. SME Preprint No. 99-166. Littleton, CO: SME.
Lynch, A.J., Johnson, N.W., Manlapig, E.V., and Thorne, C.G. 1981. Mineral and Coal
Flotation Circuits. New York: Elsevier Scientific. pp. 44–55.
Stanley, F., King, P., Horton, S., Kennedy, D., McGough, K., and Luttrell, G. 2006.
Improvements in flotation column recovery using cell-to-cell circuitry. In 23rd
International Coal Preparation Exhibition and Conference, Proceedings, Lexington,
KY, May 1–4. pp. 87–98.
Williams, J.J.E., and Crane, R.I. 1983. Particle collision rate in turbulent flow. Int. J.
Multiphase Flow 9(4):421–435.
Fine Coal Processing
Developments in Anglo
American Thermal Coal
South Africa
Chris Swanepoel
ABSTRACT
This chapter details the approaches adopted in fine coal processing within Anglo
American Thermal Coal South Africa, outlining the journey followed since 1995.
This historical review of fine coal processing focuses on beneficiation, dewatering,
and agglomeration with particular reference to (1) flotation, detailing the devel-
opment of Multicell technology and the subsequent evolution to Dual cell; (2)
successful dewatering of ultrafine thermal coal by achieving specified low moistures
and ensuring product handleability; and (3) briquetting of ultrafine material
resulting in the creation of a salable product, which adds value and minimizes
environmental liability.
COAL INDUSTRY OVERVIEW

Coal is the most abundant source of fossil fuel energy in the world, consider-
ably exceeding known reserves of oil and gas. The bulk of all coal produced
worldwide is thermal coal, which is used as a fuel for power generation and
other industries, notably the cement sector. In 2011, seaborne thermal coal
global demand accounted for approximately 790 Mt and was supplied from
numerous countries, with coal producers operating in a highly competitive
global marketplace.
Thermal coal usage is driven by the demand for electricity and influenced
by the price of competing fuels, such as oil and gas and, increasingly, the cost
of carbon. Global thermal coal demand is also affected by the availability of
alternative generating technologies, including gas, nuclear, hydro-electricity,
and renewables.
95
Figure 1 Thermal Coal South African operations
Anglo American Thermal Coal Overview

Anglo American Thermal Coal (AATC) operates in South Africa and is a joint
partner in Cerrejon, Colombia. In South Africa, Thermal Coal (see Figure 1)
wholly owns and operates nine mines and has a 50% interest in the Mafube col-
liery and Phola washing plant. Six of the mines collectively supply 22 Mtpa of
thermal coal to both export and local markets. New Vaal, New Denmark, and
Kriel collieries are domestic product operations supplying 30 Mtpa of thermal
coal to Eskom, the state-owned power utility. The Isibonelo colliery produces
5 Mtpa of thermal coal for Sasol Synthetic Fuels, the coal-to-liquids producer,
under a 20-year supply contract.
Anglo American Inyosi Coal is a broad-based black economic empower-
ment (BEE) company with 73% held by Anglo American and the remaining
27% is held by Inyosi, a BEE consortium. Anglo American Inyosi Coal, in turn,
owns the Kriel colliery, the new Zibulo multi-product colliery, and the green-
fields projects of Elders, New Largo, and Heidelberg.
Thermal Coal’s South African operations currently route all export ther-
mal coal through the Richards Bay Coal Terminal (RBCT), in which it has
a 24.17% shareholding, to customers throughout the Med-Atlantic and Asia-
Pacific regions. Within South Africa, 62% of total sales metric tons are made
to the Eskom power utility, of which the majority are on long-term (i.e., life-
of-mine) cost-plus contracts. A further 8% is sold to Sasol, and 2% is supplied
to the industrial sector consumers. The remaining 28% is exported through
RBCT.
Fine Coal Processing Developments 97
INTRODUCTION TO FINE COAL PROCESSING

DEVELOPMENTS
Up to the early 1990s, fine coal processing consisted of spiral classifiers for
the –500+150 µm material, with the –150 µm material being thickened and
disposed of in slimes compartments. In some cases the thickener underflow,
if quality allowed, was dewatered by making use of horizontal belt filters and
added to the final product to increase salable production. Mineral residue
deposit facilities consisting of coarse material outer walls and inner slimes com-
partments were, and still are, the order of the day. In particular, the slimes are
of high quality and have economic potential, due to the fragmentation of the
higher-quality friable coal into the slimes fraction. The slimes deposits, how-
ever, remain mostly dormant due to their handleability and salable contractual
size limitations. This is obviously a resource for the future, which may require
further beneficiation, certainly further dewatering and generally agglomeration
in the form of briquetting to increase its economic marketability.
Witbank coal was not considered to be amenable to froth flotation to
upgrade the ultrafines (slimes) fraction using technology of the day, and was
also not deemed economically feasible as a result of the poor yields obtained,
and the energy adjustment made for moisture content being a thermal coal. Ini-
tial test work on flotation dating back to the 1980s showed that Witbank fine
coal was very difficult to float due to the nature of the surface properties, which
require a substantial amount of reagents to enhance the hydrophobic nature
of the coal. In addition, the capability to dewater ultrafine coal sufficiently was
very limited from a technological perspective and was seen to be costly.
With new developments in flotation technology and more specifically in
flotation reagents, AATC metallurgists were re-energized in their quest to ben-
eficiate ultrafines. This resulted in fresh, innovative thinking to find solutions
to the challenges at hand.
The broad timeline depicts the journey that Thermal Coal has taken since
1995, with the development of flotation technology, a paradigm shift with
regard to dewatering, further evolution along the way, and the introduction of
agglomeration. This journey will be expanded on and described in detail in the
sections that follow.
Thermal Coal’s flotation, dewatering, and briquetting journey timeline is
outlined here:
• Initial flotation test work (1995)
• Small-scale test work at Goedehoop—Wemco and Jameson cells (1996)
• Kleinkopje fines plant commissioned (1998)
• Multicell concept developed at Goedehoop (1998)

• Multicell patented (2000)
• Pilot unit constructed and commissioned at Goedehoop (2000)
• Greenside fines feasibility project approved (2003)
• Greenside fines plant commissioned (2004)
• Goedehoop fines feasibility project approved (2005)
• Goedehoop fines plant construction completed and commissioned
(2007)
• Briquetting project approved (2010)
• Goedehoop B stream conversion from Multicell to Dual Cell
(2010/2011)
• Briquetting plant commissioning and first coal (2011/2012)
• Greenside Multicell to Dual Cell conversion (2012)
• Greenside solid-bowl replacement with Jingjin press (2012)
A gap in development occurred during the 2008/2009 period, which coin-
cided with the credit crunch and downturn in the world economy.
BACKGROUND
Flotation
The process of froth flotation, which separates minerals according to the surface
properties, has been in existence since 1860, and flotation is used in the treat-
ment of fine ground ores. The extent to which minerals are separated is known as
recovery, and the selectivity of the process is defined by grade, which means that
the recovery is defined as the mass ratio of the valuable mineral in the concentrate
to the valuable mineral in the feed. Grade is defined as the ratio of the mass of the
valuable mineral in a processing stream to the total mass of that stream.
Water is used as a medium within the cell in which more dense particles
would sink. As the solid particles are suspended in solution (i.e., by means of
mechanical agitation), a second process, based on the surface properties (i.e.,
the particles are either hydrophobic or hydrophilic), takes place at the same
time. This process requires air to be either injected into the system or intro-
duced by agitation of the slurry, which produces bubbles within the slurry.
Hydrophobic particles attach to the bubbles and are floated to the surface
of the slurry as a result of the air being less dense than water. For a particle to
stay attached to a bubble, the gravitational force experienced by the particle
must not overcome the adhesion to the bubble.
Coal Flotation
To distinguish between coal flotation and other mineral flotation processes, it
is necessary to consider that in other mineral processes, the material is specially
ground to be selectively processed with flotation, but in the case of coal, the
process is designed to generate the least amount of fines possible. This means
that the fines arise from various processing stages, which could have been
exposed to different environments, and, as such, the properties may also differ
throughout the processing stream.
When considering coal, the combustible mass is considered to be the valu-
able mineral and ash constitutes the gangue mineral. The flotation process is
utilized in industry, because flotation is the only process for effectively and
economically beneficiating –150 µm coal.
Flotation Reagents
Even though coal is naturally hydrophobic, the use of collecting reagents when
processing coal increases the floatability of the coal by altering the behavior of
coal within the process. Bubbles have a natural tendency to break the air–water
interface as it reaches the water surface, which is not a desired effect in the
flotation process, because the valuable minerals need to be removed from the
cell before the interface is destroyed. Therefore, a substance called a frother is
used to facilitate air dispersion into fine bubbles and increase froth stability.
This keeps the material floating above the pulp surface long enough for froth
removal.
Dewatering/Filtration
Slurry (i.e., a mixture of solids and water) is typically dewatered/deslimed using
sieve bends, static screens, vibrating screens, hydrocyclones, and centrifuges.
The slimes fraction (i.e., typically <150 µm) poses a great challenge to the
industry. Generally this fraction is disposed of by making use of slurry ponds,
which allows settling of the material and minimum reclamation of water back
to the operation. This, however, creates various environmental liabilities for the
industry and therefore a major global drive to maximize water recovery effec-
tively minimizing raw water consumption and slimes disposal. At this point in
time, filtration is the most common method of dewatering, and by which the
typical types of equipment used are horizontal belt filters, disc filters, drum
filters, and plate-and-frame filter presses.
Other than direct filtration, froth flotation can also be used as an upstream
process, in which case a salable high-value product can be produced. This
product is then filtered and blended with the typical export product. However,
given that the product contains moisture, the lowest possible product moisture
is required. Additionally, flotation tailings are produced, which can then be
filtered in the same manner to maximize overall water reclamation.
Filtration is the removal of solid particles from a fluid, by passing the fluid
through a filtering medium on which the solids build up. Many factors can be
important in selecting a filtration process, but because mineral processing oper-
ations are concerned primarily with recovering the solids at large throughputs,
the selection of equipment is considerably narrowed. Filtration equipment size
is specified by the surface area necessary to produce the required product. As
with sedimentation, particle properties cannot be adequately measured, and
small-scale filtration tests must be carried out to obtain basic data. Filters can be
operated in two basic modes. Constant pressure filtration maintains a constant
pressure so that the flow rate falls slowly from a maximum at the start of the
cycle. Most continuous filters can be considered to operate on this principle,
using a vacuum to provide the pressure difference. Constant rate filtration
requires gradually increasing pressure as the cake builds up and increases the
resistance to flow. A common approach is to use the constant flow rate until the
pressure builds up to a certain level and then to use constant pressure filtration
for the remainder of the time. The cycle can conveniently be achieved by using
centrifugal pumping.
Briquetting
Coal briquetting is one of the oldest applications in the forming process by
agglomeration. Originally, piston presses were used to produce coal bricks
of about 0.5 to 10 kg, made of coal fines that were mixed with binders such
as pitch, bitumen, or tar. These presses had reduced capacities (1,000 bricks
per hour maximum). The development of the double roll press (see Figure 2)
took place during the second half of the 19th century, and it offered higher
production capacities under acceptable economic conditions. Thanks to this
technique, the annual production has reached up to 10 million metric tons of
briquettes. Coal briquetting technology by means of the double roll press was
mainly developed for the upgrading of coal fines coming from coal screening
and washing, and used as fuel for domestic and/or industrial heating in the
same way as coal.
The new coal briquetting units are located in the larger coal-producing
countries (China and India), the developing countries (Turkey), or used for
coke production in Japan, India, and China by upgrading of lower-quality cok-
ing coals.
Figure 2 Double roll briquette press
Raw Materials Used in the Briquetting Process

The raw materials used are coal fines and a binder.
Coal fines. The coal fines should possess the following characteristics to
obtain good-quality briquettes:
• Good combustion characteristics, including high calorific value, low
ash content, medium % volatile matter, and low sulfur content
• Amenable to compaction
Size range of fines. The coal fines usually have a size range between 0 and
3 mm; a small quantity (a few percent) of grains larger than 3 mm and smaller
than 5 mm is acceptable. To obtain good-quality briquettes, it is essential to
have the widest grain size distribution possible, with very fine grains among
larger grains. During briquetting, the fine grains will interlock with the larger
grains and will ensure the highest possible compaction. This parameter is
directly connected with the quality of briquettes, in particular their crushing
and handling strength. If the quantity of fine grains is insufficient, a crushing of
the large grains will occur while the material is running through the press rolls;
these crushed and cracked grains will give more fragile briquettes.
Binder. The coal fines are not self-agglomerating (i.e., they require the addi-
tion of a liquid or solid binder to obtain briquettes of a sufficient quality). This
has an essential effect on the investment and operating costs of coal briquetting.
Various Types of Binder Used for Coal Briquetting

Binders used in the coal briquetting industry are
• Coal pitch (residue of coking) under solid or liquid form,
• Petroleum bitumen (oil refining) under solid or liquid form,
• Tar,
• Molasses (beet or sugarcane) with possible addition of lime or possibly
phosphoric acid,
• Lignosulfonate (residue from paper mill),
• Starch (maize, corn, potatoes), and
• Polyvinyl alcohol.
S U M M A R Y O F A AT C F L O TAT I O N I N T H E W I T B A N K
C O A L F I E L D A N D D E V E L O P M E N T O F T H E M U LT I C E L L
Until the early 1990s, it was thought that Witbank coal was not amenable to
froth flotation. The first flotation plant to treat Witbank ultrafine coal was only
installed in 1995. Since that first plant, there have been more installations, par-
ticularly the 30-t/h Kleinkopje (KK) fines plant, utilizing a Multotec column
flotation cell and a Latham plate-and-frame filter press, which was constructed
and commissioned in 1998. The KK flotation plant was relatively successful
but was adversely impacted by weathered and burnt (spontaneous combustion)
feed material. In addition, the performance in the finer fraction (<100 µm) was
not acceptable, primarily as a function of bubble size. The Latham filter press
had many technical challenges but performed relatively well. Ownership of the
fines plant by plant personnel was a major challenge, as they viewed it as non-
core and this impacted negatively on the performance of this plant.
Generally, the Witbank flotation test work and practice steered away from
the traditional flotation technology routes as they were not considered suc-
cessful. It was highly accepted that high shear mixing was necessary to achieve
success. In addition, it was believed at the time that there was no justification
to float and produce steam coal. Steam coal generally attracts a low price, which
is made worse by the adjustment made for moisture. This caused metallurgists
to investigate other forms of dewatering inclusive of thermal drying, which is
often considered economically unjustifiable.
With new developments in flotation technology and flotation reagents,
the decision was made to conduct further flotation tests at Goedehoop col-
liery. Initial test work was conducted on a conventional Wemco mechanical
test cell using new, improved reagents, and promising yields were obtained.
Subsequently, the decision was made to test column flotation ( Jameson)

because of the envisioned smaller footprint, lower operating costs, and ease of
operation, but the performance was poor. While running the Jameson, metal-
lurgists developed a mechanism to produce microbubbles that greatly improved
flotation performance at the colliery. Throughout the test work at Goedehoop,
high mixing energy was required to achieve acceptable recoveries and optimal
use of reagents. The high-intensity mixing and generation of very small bubbles
increases the probability of bubble–solid and solid–reagent contact, resulting
in lower dosing rates being required to achieve acceptable recoveries.
A cell was developed combining the features of the mechanical Wemco
(high energy) and column Jameson (selectivity) cells while incorporating the
small bubble generation stumbled on as referred to previously. The design
combined the high-intensity mixing of the mechanical cell with the quiescent
nature of column cells. This cell became known as the Multicell and was devel-
oped in three stages:
1. A 1-m3/h cell, which produced good results and gave confidence to
proceed with the next scale-up.
2. The 1-m3/h unit was scaled up to 100 m3/h and also performed well.
3. The decision was made to scale-up to a production size unit, which
could treat 250 m3/h, and this also proved to be successful.
The performance of the Multicell was superior to the Wemco and Jameson
cells. The Multicell maintained a high-quality product at high yields without
using excessive amounts of reagents (Opperman et al. 2002).
The Multicell
The Multicell incorporates a primary and secondary flotation cell together with
a single Multicell flotation pump. The pump consists of a modified spindle
pump (i.e., the barrel) and the shaft and intake are modified to achieve an
efficient and high-dispersion energy flotation cell. The material is fed into the
barrel together with air and reagents. The material is then dispersed into the cell
through a grated “window” at which point the bubbles start rising to the top
of the cell. The froth (i.e., the product) overflows the “lip” of the tank into a
launder, whereas the primary cell discard is pumped to the secondary cell. A
distribution manifold disperses the material into the tank volume for addi-
tional retention time to increase the process yield. As there is no pump, a much
calmer flotation process is experienced (Opperman et al. 2002). The Multicell
schematic is shown in Figure 3.
Air
Feed
Slurry Reagent
Air
Multi-Cell Not to Scale
Figure 3 Multicell schematic diagram
D E W AT E R I N G C O N U N D R U M
With all the excitement around the Multicell, there was still a large hurdle to
overcome with regard to dewatering of this very fine flotation concentrate. This
flotation concentrate consisted of 150 by 0 µm particles and at times as much
as 80% <45 µm. Steam or thermal coal is sold on an energy basis, and thus
moisture must be kept to a minimum. Initial test work centered on Humboldt
screen-bowl centrifuges by feeding a combined flotation concentrate and spirals
product in a predetermined ratio. Test work indicated that a 75% spiral/25%
flotation ratio achieved the best results. The coarser spiral fraction is required
to recover the ultrafine fraction because it acts as a filter bed. The recovery at
this optimum was in the order of 90% but dropped off dramatically as the flota-
tion material (ratio) was increased and the surface moisture of the final product
also increased significantly. As most of the loss of recovery occurred through
the screen section of the unit, it was therefore decided to test a solid-bowl cen-
trifuge by converting the existing units from screen-bowl to solid-bowl. This
increased the recovery of the ultrafine material but at higher surface moisture,
which also impacted on the throughput of the unit.
Greenside Colliery Fines Plant

Based on the development of the Multicell and the test work on dewatering car-
ried out at Goedehoop, a decision was made to install the first Multicell flota-
tion plant at Greenside colliery making use of solid-bowl centrifuges to dewater
the resultant concentrate together with spiral product in a predetermined ratio.
The Greenside fines plant consisted of four primary and secondary cells. The
plant was constructed and commissioned, and first coal was produced in 2004.
The Multicell performed satisfactorily but was constrained by the solid-bowl
performance with concentrate recoveries in the order of 50%. The unrecovered
portion was lost to the effluent and ultimately to tailings as this recirculating
load of ultrafines could not be accommodated.
Dewatering Developments
During this Multicell installation period at Greenside, Goedehoop continued
to conduct dewatering test work. In an effort to glean industry best practice,
a team set out to visit several operations within South Africa comprised of
various commodities. This was primarily to establish what was being used suc-
cessfully elsewhere. What came out of this exercise was that plate-and-frame
presses were the route to pursue. Historically, there had been a perception that
plate-and-frame presses were expensive, difficult to operate, complex, and being
a batch process. This initiated an in-depth investigation into plate-and-frame
presses. What transpired from this work was that there were more advantages to
plate-and-frame press filtration than disadvantages. The main advantages were
• Lower cake moistures,
• Dewatering that did not require the addition of coarser material as a
facilitator,
• Product that was easier to handle, and
• Virtually 100% recovery and no circulating ultrafines load.
Equipment selection is not the only factor that determines the amenabil-
ity to filtration and final product moisture when considering different process
streams and ore bodies. Important particle characteristics include
• Particle size distribution, and
• Clay content.
Test work investigating the effect of various parameters and ore character-
istics on the final product moisture as well as the flux (the rate of dewatering,
i.e., l/h/m2) is critical. Additionally, when considering plate-and-frame filter
press units, the filter press functionality (i.e., feeding, membrane squeeze, and
air blowing) can have various effects on the final product moisture.
A trade-off study was conducted, and the Lasta filter press (see Figure 4)
from Ishigaki in Japan was chosen for the envisaged Goedehoop fines plant
based on superior technical performance. In addition, the Lasta unit made use
of technology considered to be essential to ensure total cake discharge, namely,
Figure 4 Lasta filter press
traveling cloths and also automated cloth washing to ensure that no cloth blind-
ing takes place.
At this stage of the journey, considerable work was done on thermal dry-
ing, as it was believed that the moisture content in the filter cake may adversely
impact the overall gross as-received quality of the final product, expressed in
kcal/kg. Test work was successfully conducted on flash driers with acceptable
results. Any moisture could be achieved based on retention time in the unit.
It was generally believed that going below 8% was counterproductive as there
would be re-adsorption from the atmosphere and surrounding coal as the cake
would be added to the main product stream (the carrier) prior to stockpiling
and train loading.
This route was discontinued, however, as it was believed that the safety risk
in terms of fire and explosion was too great; the lack of expertise and the capital
requirement impacted on the feasibility of the project. In addition the calcu-
lated impact on the overall energy content of the total AATC product grade
complement was found to be small (approximately 10 kcal/kg).
G O E D E H O O P F I N E S P L A N T I N S TA L L AT I O N
Toward the end of 2005, Goedehoop submitted a capital application to con-
struct a fines beneficiation plant. The application was approved in October
2005, and commissioning of the new plant commenced by mid-2007. The
project was aimed at implementing a fine coal beneficiation plant making use
of flotation and filtration to produce a salable and acceptable product to satisfy

the market. The plant consisted of
• Nine Multicell flotation units,
• Two plate-and-frame filter presses, and
• A tailings thickener.
The flotation plant was designed to process approximately 96 t/h of feed
slurry at 5% solids. The flotation circuit is fed from three thickeners. Each
thickener provides feed to three flotation units comprised of a primary and a
secondary cell. The concentrate is pumped to the Lasta presses. The dewatered
cake is discharged onto a collection belt below, which in turn discharges onto
a conveyor belt routing the material to the main plant product streams. The
tails from all nine flotation units are combined in a launder and gravitate to
a tailings thickener where the thickened tails are pumped to a slime compart-
ment at the mineral deposits tailings facility. The capital motivation stated that
beneficiating the ultrafine material would result in a total yield increase of 3%.
Goedehoop Fines Plant Evaluation

The final technical project analysis yielded the following.
Positives
The average product filter cake surface moisture of 18.9% was achieved com-
pared to the target value of 20%. The filter cloth life exceeded the guaranteed
5,000 cycles.
Negatives
The flotation plant has not achieved the yield increase of 3%. This is due partly to
head grade issues but also inherent constraints with the Multicell design, namely:
• A throughput constraint as a result of the primary pump, which dou-
bles up as both a transfer mechanism and high-energy shearer, which
was further impacted by the conscious decision to lift the secondary
cells up for the tailings to gravitate to the thickener. This resulted in
lower throughput and higher feed solids.
• The low energy and recoveries in the secondary cells due to poor cell
dynamics.
The Goedehoop fines plant produced in excess of 15,000 tons per month, but
this did require a large amount of attention, technical input, and ownership by
plant management and operational teams.
A G G L O M E R AT I O N
Anglo Thermal and Enprotec embarked on an ultrafine coal agglomeration
feasibility study in 2009. A decision was made to conduct the test work at a
specific operation where the ultrafine coal was being stockpiled after dewater-
ing the material using Technicas Hydraulicas plate-and-frame filter presses. This
material presented constraints as a result of the particle size with the power
utility refusing to accept this material.
Briquetting and granulation are two of the most widely used agglomera-
tion processes. Technology and affordable binders have made it possible for the
South African market to seriously consider briquetting to solve the ultrafine
coal challenge and, in the process, clean up the environment. Both briquetting
and granulation tests were conducted.
The aim of the project was to evaluate the influence of variables on the
final briquette and granule. This included the influence of varying binder addi-
tions, the effect of a sorbent, the optimum feed moisture, green strength, cured
strength, water proofing properties, and so forth.
Experimental Procedure Employed

Only polyvinyl alcohol was used as a possible binder due to its availability,
history, and affordability. The binder was added in various dosage rates to the
sample at various feed moistures. A catalyst was added to each test run, which
constitutes to 10% of the wet binder mass. This is done to enhance the effec-
tiveness of the binder. In some instances, some bark (sorbent) was added to test
its ability to improve handling and reduce apparent moisture for the wet, as
received, sample.
The filter cake/binder mixture was pressed with a lab press at 5 tons of
pressure. The briquette was then tested for green strength, cured strength, and
waterproofing properties in the following manner:
1. Green strength. Drop the freshly produced briquette from 2 m high
onto a concrete floor until it breaks.
2. Cured strength. Let the freshly formed briquette stand for a period
of one day to cure. Tests include determining the amount of moisture
that was lost as a result of drying. Also drop the briquette from 2 m
high onto a concrete floor until it breaks.
3. Waterproofing. Submerge the freshly produced briquette into water
for a period of 1 hour. Measure the amount of water absorbed. Drop
the briquette from 2 m high until it breaks. Submerge the 1-day-cured
briquette under water for a period of 3 hours. Measure the amount of
water reabsorbed. Drop the briquette from 2 m high onto a concrete

floor until it breaks.
Briquetting Results
On completion of all the tests, it was evident that the trend with regard to bri-
quette strength is the same (i.e., as the binder dosage increases, the briquettes
can withstand more 2-m drops compared to the same briquettes at a lower
binder dosage). Another observation was that the waterproof characteristics of
the briquettes improve as the binder dosage increases.
Granules Results
Notably, there exists a feed moisture range where granulation is possible. When
the feed moisture drops below the range, no granulation will occur, as the feed
is too dry to ensure sufficient blending of the binder. Similarly, when the feed
moisture exceeds the range, a pulp will form rather than granules. With higher
binder addition and also higher feed moisture, the granules can withstand long
periods of time within water without decomposing. Importantly, the freshly
produced granule had no strength or waterproof properties. It also sticks to
other granules and forms large agglomerates. Fresh granules are not a final
product and need to be thermally dried before stockpiling. As a result of the
test work conducted, the decision was made to follow the briquetting route.
AUDIT CONDUCTED BY ANGLO RESEARCH

In line with the timeline given in this chapter's introduction, the focus shifted
to the Goedehoop flotation performance challenges. A flotation circuit audit
was conducted in 2009. The purpose of the hydrodynamic study was to char-
acterize the flotation cells and assess their performance. Three parameters were
used to analyze cell performance: superficial gas velocity ( Jg), which is a mea-
sure of the flow-rate of air in the pulp phase of the flotation cell; froth depth to
assess possible changes to froth characteristics; and bubble size representing the
size distribution of bubbles generated.
Based on the superficial gas velocity, values were taken on the day the
primary cells were within an acceptable range (2.30 cm/s) while the velocity
for the secondary cells was out of range (0.54 cm/s), low to be specific. The Jg
value for the secondary cell can be improved by adding air as residual air from
the primary cells, but this has not proven to be effective for favorable flotation
conditions. This confirmed the poor secondary cell dynamics that had been
identified previously.
Some cells had a shallow froth depth of 30 cm and there was a potential of
recovering a lot of gangue material. This was easily rectified.
The bubble size was found to be fine enough to float such fine material, as
shown by the Sauter mean bubble diameter of <0.5 mm. This confirmed the
advantage that the Multicell had over traditional flotation technology. This
indicated that there were challenges specifically around the throughput and sec-
ondary cell performance that appeared to be inherent to the Multicell design.
As time went by, the ownership deteriorated and skills were also lost as person-
nel were transferred, which resulted in declining performance. This initiated an
evaluation study utilizing a third party, namely Enprotec Mineral Processing.
E N P R O T E C P E R F O R M A N C E E VA L U AT I O N O F T H E
G O E D E H O O P F L O TAT I O N C I R C U I T A N D C O N V E R S I O N
TO DUAL CELL
Outcomes from this investigation confirmed previous findings. Challenges with
the Multicell system were due to the fact that the mounted pump is utilized for
both pumping as well as agitation. If, for whatever reason, the pump efficiency
and thus the pumping capacity is lost, the plant throughput is reduced and the
retention time of material within the cell is increased. An increase in retention
time increases the amount of discard material floated, effectively reducing the
efficiency of the process. A reduction in plant throughput causes an increase in
feed density due to an increasing recirculation load. The increase in feed density
overloads the froth phase (i.e., the froth carrying capacity is exceeded), which
reduces the yield.
The feed density was regulated by the dilution water addition and the
thickener underflow variable-speed drive (VSD) pump. Therefore, as soon as
the feed density to the plant is above the set point, the thickener underflow
VSD pump slows down, thus reducing the flow rate to the flotation feed tank.
The problem, however, is that the feed density was therefore approaching the
thickener underflow density (see Figure 5). This could be explained by the
fact that the minimum speed of the VSD pump is set at 50%. In addition, one
feed tank feeds one bank consisting of three process streams; therefore, to run
at optimal conditions, all three streams should be operational. It can thus be
shown that the feed density tends to the thickener underflow density when the
material is not moving through the cells fast enough as a result of the primary
pump transfer capability challenge as explained previously.
Approaches Thickener
Underflow Density
Figure 5 Trends illustrating how the flotation feed density approaches the thickener
underflow density
The Dual Cell

The Dual Cell was developed by Enprotec under license to Anglo American
and uses the same principles as the Multicell (i.e., high energy dispersion);
however, a second agitation pump is introduced into the second cell, as can
be seen schematically in Figure 6. The pump delivery is directed back into the
modified barrel such that the probability of floating a specific particle is effec-
tively increased. A major benefit in the Dual Cell design is the use of gravity
to feed the secondary cell, because the pump is not required to pump material.
This removes flow constraints as well as increases pump life. The mass yield is
increased with additional internal launders on the primary cell. Other major
advantages within the Dual Cell arises with respect to a unique air injection
system, paddle (shear plate) design, as well as the unique fluid dynamics consid-
ered in the method of feeding into the barrel (Kruger et al. 2011).
The Dual Cell has the following characteristics:
• Four stages of flotation in two separate tanks (recirculation of the tails
in each of the two tanks by pumping the tails to the feed section of
each tank)
• Energy transfer in both tanks (each tank consists of its own patented
spindle pump system)
Feed
Slurry Reagent - 75%
Slurry Reagent - 25%
Air
Air
Tailings
Dual-Cell Not to Scale
Figure 6 Dual cell
• Each tank can be controlled individually with respect to reagent addi-

tion and air input
• Gravity flow to the first cell and subsequently second cell from a
header box main feed (the total system is gravity controlled)
• No restriction of air inputs to the cell (both tanks can be fed
independently)
Conversion from Multicell to Dual Cell (B Stream)

From the study conducted by Enprotec, the Anglo Research hydrodynamic
study and internal test work, a decision was made to upgrade one of the streams,
namely, B stream. This would entail the conversion from the Multicell to a Dual
Cell. The required modifications to convert the Multicell circuit to a Dual Cell
circuit are shown in Figure 7.
Main Objectives
The main objectives were as follows:
• Increase the throughput. It must not be constrained by the primary
pump’s inability to pump slurry and air.
• Improve performance of the flotation plant by addressing dynamics in
the secondary cells.
The following changes were made such that the feed is gravitated through the
process system:
Figure 7 Modifications to convert B stream cells to Dual Cell configuration
• The header box was lifted.

• The stream was rerouted such that the current primary cells become
the secondary cells.
• A tails tank and pump should be incorporated in the process stream to
transfer the tails to the tailings thickener.
• The Multicell should be converted to the Dual Cell.
Some mechanical changes were also made:
• The current paddles incorporated in the Multicell are an austenitic
stainless steel, which are prone to wear and are very malleable. A tem-
pered stainless steel is utilized, which is more brittle and more resistant
to wear.
• The spindle pump shaft needs to be extended from 1.2 m to 1.8 m
from the top of the shaft, effectively increasing the area for the flota-
tion process to take place.
• The Dual Cell makes use of a specially designed pump in each tank for
recirculation purposes as compared to the Multicell, which only makes
Table 1 Direct comparison of the Multicell to the Dual Cell

Multicell Dual Cell
Feed ash, % 26.4 28.1
Product ash, % 14.9 15.0
Feed CV,* MJ/kg 22.14 22.38
Product CV, MJ/kg 27.65 27.67
Ash balance yield to product, % 34.8 66.2
Measured tonnage mass yield, % 32.3 64.4
d50, µm 60 60
% of Theoretical yield achieved 49 97
Source: Kruger et al. 2011.
*CV = calorific value.
use of a single larger pump in the primary tank to transfer pulp to the
secondary tank.
These changes increased the throughput of the plant by removing the flow
constraint as well as increased the yield to the product stream at the specified
product quality by duplicating the primary in the secondary cell.
Pre- (Multicell) and postconversion (Dual Cell) performance measure-
ment tests were conducted on module B, and the results are summarized in
Table 1 (Kruger et al. 2011).
Results and Benefits

The following results and benefits were obtained:
• Yield improvement
• Lower reagent consumption
• Lower power consumption
• Lower complexity and less operator intervention
• Higher product carrying capacity
• Feed density control
GREENSIDE FINES PLANT CHALLENGES

The Multicell flotation plant was commissioned in 2004 with the primary aim
of increasing the plant yield by 3% as well as reducing the volume of slimes
reporting to the co-disposal facility, which would reduce the environmental
Figure 8 Solid-bowl centrifuge
liability. Since commissioning, these yields were not achieved as a result of the
same Multicell inherent constraints experienced at Goedehoop.
Greenside colliery was thickening their flotation concentrate, blending it
with spiral product, and dewatering it using solid-bowl centrifuges (see Fig-
ure 8). This process had the following major drawbacks:
• High product moisture content
• Significant loss of product in the effluent
• Product that is difficult to handle
The high maintenance cost, low availability, low recovery, and high product
moistures of the three solid-bowl centrifuges resulted in a total yield loss of at
least 1.4% of the forecasted 3%.
Goedehoop colliery was making use of two fully automatic filter presses
to dewater their flotation concentrate. The problems that Greenside was facing
had been eliminated at Goedehoop due to the installation of filter presses. The
filters in use at Goedehoop had unfortunately become very expensive due to the
weakening of the rand. This fact was a key driver in Greenside deciding to test
other filter presses on the market that could achieve the following objectives:
• Fully automatic
• Low in capital and operational cost
• Produce acceptable final moisture content
• Produce a product that is easier to handle
• Operate at a high availability/utilization
Enprotec was at the time using and promoting a filter press from Sepro-
tech to dry their flotation concentrate, which had the characteristics required
by Greenside. The filter is known as the rapid filter press and is manufactured
locally in South Africa. Enprotec together with Anglo Thermal Greenside col-
liery embarked on a project to test the rapid filter press at Greenside colliery
in 2008 using an arising flotation concentrate as feed to the filter. The project
was divided into two phases with thickened flotation concentrate and flotation
concentrate used as feed in each respective phase.
Rapid Filter Press

The rapid filter press has various features worth mentioning: except for normal
cake fill and filtration, it has a membrane squeeze and cake blow, both of which
contribute to achieving the lowest possible moistures. The period taken for each
of these features in the cycle can be adjusted to optimize operating conditions
and performance.
Filter cake moistures remained fairly constant with the average moisture
for phase 1 being 30.7%, which was much higher than expected. During test
work, the filter cake from phase 1 was incompressible. In addition, high vol-
umes of flocculant were required to be added to the flotation concentrate to
thicken the flotation product properly. The flocculant was the major contribu-
tor to the higher-than-expected final cake moisture content. After phase 1 was
completed, some tests were conducted at 16 bar membrane squeeze pressure
and 16 bar air blow pressure to determine whether the cake moisture could
be brought down further. This did not have any effect on the final product
moistures, proving the incompressibility of the thickened flotation product
filter cake.
The average moisture for phase 2 was 23.0%, which was in line with expec-
tations. The main reason for the lower moistures was the absence of flocculant
in the feed.
Jingjin Filter Press

In 2009/2010, the credit crunch was in full swing and the downturn in the
worldwide economy necessitated further evaluation of less expensive equip-
ment. Although the rapid press had effectively halved the cost of the filter unit
as compared to the Lasta unit, it was believed that further cost reductions were
necessary. The outcome was a focus shift to the east, specifically China. A desk-
top study was carried out and an AATC group was sent to China to investigate
claims made firsthand. The outcome of the technical visit was the exciting
Figure 9 Jingjin filter press
discovery of Jingjin and the presses they provided. The presses were custom
designed and cost effective, a further reduction of ±50% when compared to the
Rapid press, and they were of acceptable quality.
The major risk was that the Jingjin press (see Figure 9) would be a first in
South Africa. This risk was mitigated as follows. A technical visit was made to
the Jingjin factory in October 2010. The visit also included viewing the filters
operating in the field in coal applications. The Jingjin factory visit revealed a
professional, quality-controlled manufacturing process using computerized
machinery at all stages of manufacture. All critical components of the machine
are manufactured in-house. Quality control was of an acceptable standard and
the stock holding of all components was extensive.
F I LT E R P R E S S C O M PA R I S O N
A comparison between the Ishigaki, Seprotech, and Jingjin filter presses was
conducted and the Jingjin was chosen based on the following:
• Fit for purpose
• Cost effective
• User-friendly maintenance
• Operational understanding
The new fines plant at Greenside, which consists of four primary and
secondary cells converted to Dual Cell and two 1.5-m plate Jingjin presses, is
currently in the commissioning phase (see Figure 10).
Figure 10 Greenside filtration plant
I M P L E M E N TAT I O N O F B R I Q U E T T I N G AT A AT C
Dewatered fines are transported to a fines stockpile area for storage. Market-
ability of the fines in its current form is very limited. Because of environmental
concerns and the availability of technology, the decision was made to build
a 56-tph briquetting plant. The aim of the project was to take the stockpiled
fines and produce a sized salable product (refer to the conceptual layout of the
briquetting plant in Figure 11). The briquette plant’s main process streams are
briefly described as follows.
• Stream 1: This assumes optimum feed (surface) moisture of 15%–20%
to be loaded at 66 tph from a stockpile prepared by a front-end loader
(FEL). From the feed bin, material is fed onto a conveyor discharging
into the Eirich mixer.
• Stream 2: The binder addition takes place continuously in a range of
0.3%–0.5% by mass. The addition percentage is reassessed from time
to time. The dissolution of binder and the operation of the binder
plant are controlled by a programmable logic controller.
• Stream 3: After thorough mixing, the Eirich mixer discharges onto
a conveyor that feeds the briquette machine. The briquette machine
product is then transferred to the stacker conveyor for final stacking.
• Stream 4: Fines screening allow fines generated by the briquetting
process to be screened out and discharged onto a fines stockpile via
a fines stockpile conveyor. These fines are then circulated to the main
prepared feed stockpile via FEL ready to be loaded back into the
mixers.
• Stream 5: An Osborn KPI Series 33-30130 Super Stacker with luffing
and slewing capability places the briquettes on the stockpile area for
air drying to take place. Five days of rolling stock will allow the bri-
quette moisture to reduce and the binder to cure.
• Stream 6: Water is drawn from the clear water dam and is used for the
binder solution.
• Stream 7: Potable water is required for steam generation within the
boiler section of the binder plant.
• Stream 8: Binder is delivered by bulk truck and off-loaded into the
binder storage bin. Binder at the required percentage is added into the
mixing tanks and prepared for addition to the mixers.
The briquetting plant was commissioned and first briquettes produced in
January 2012. There have been challenges, most notably the following:
• High % fines creation post the briquetting machine—green strength
• Impact of the weather has been significant on briquette curing strength
and time taken to cure
• Inconsistent briquette strength
In essence, the cause of the challenges referred to can be summarized as follows:
• Inadequate and homogeneous binder mixing
• Briquette machine speed setting to ensure efficient pocket filling
• Transfer points and stacker luffing contributing to the fines creation
FUTURE PLANS
Flotation
Plans have been made to convert the remaining flotation streams at Goedehoop
from Multicell to Dual Cell during 2013 as a result of the improved recovery
and quality control experienced since conversion of the B Module was initiated.
The findings of the project concluded that the Dual Cell delivered improved
recoveries (yields) and improved quality control due to the fundamentally
improved flotation cell dynamics. An improvement of 0.4% yield (percentage
of total feed to plant) was experienced with a concurrent 0.34 MJ/kg improve-
ment in product quality. In addition, because of higher recoveries and increase
in froth solids content to filtration, optimum dewatering can be facilitated.
Figure 11 Conceptual three-dimensional layout of a briquette plant
Ultrafine Coal Dewatering

Historically, AATC has disposed of tailings in slimes compartments, which
is an environmental liability and complicates mineral residue deposit facility
design. In other cases, it is the major source of acid generation, necessitating
expensive water treatment plants and enormous operating costs. Air space con-
straints on discard facilities are also at a premium and therefore are minimized
through the dewatering of thickener underflow, especially if the dewatered
product can be sold into the marketplace. It is the intention to convert these
slurry disposal systems to “dry disposal systems” by dewatering the ultrafines or
tailings.
Fine Coal Briquetting

Briquetting is still to be optimized but presents a huge opportunity to gener-
ate profit, clean up the environment, and also minimize the closure liability.
Large quantities of potential feedstock exist for briquette making, either for the
domestic or export market.
CONCLUSION
South African coals, specifically the Witbank Coalfield, are classified as gener-
ally difficult to beneficiate. This holds true for fine coal, particularly the ultrafine
fraction that was considered not amenable to flotation until the mid-1990s and
also presents major difficulties in terms of dewatering. This has resulted in large
quantities of unbeneficiated ultrafine coal being stored in slimes compartments.
The journey of fine coal processing in AATC outlined in the chapter has
confirmed that Witbank ultrafine coal can be floated employing Multicell
flotation technology, which has evolved further in the Dual Cell. The ultrafine
flotation concentrate and thickener underflow can successfully be dewatered by
utilizing plate-and-frame filter presses, as proven by several installations within
AATC increasing salable thermal product and reducing the environmental
liability.
The briquetting of ultrafine material can be done successfully and econom-
ically, which will continue to add value, eliminate environmental liability, and
comply with South African regulations. The optimization journey in AATC
continues and will improve and evolve fine coal processing into the future.
ACKNOWLEDGMENTS
The author thanks his many colleagues at Anglo American Thermal Coal and
Enprotec Environmental and Process Technologies.
REFERENCES
Kruger, M., Vermaak, M.K.G., Buitendag, M.J., and Nxele, J. 2011. Coal flotation:
A case study of Dual Cell performance. South African Coal Processing Society
Conference.
Opperman, S.N., Nebbe, D., and Power, D. 2002. Flotation at Goedehoop Colliery.
South African Institute of Mining and Metallurgy. SA ISSN 0038-223X/3.00.
pp. 405–409.
Fine Coal Processing—
Technical Developments in
Australia
Bruce Firth, Mike O’Brien, and Graham O’Brien
ABSTRACT
Coal preparation plants incorporate a wide array of solid–solid and solid–liquid
separation equipment. The types of processes employed and configuration of these
unit operations vary considerably from plant to plant and across different geo-
graphic regions depending on the coal type and end use. Location-specific factors
influence plant design protocols and operating practices that need to be recognized.
This chapter highlights the differences between the Northern Hemisphere and
Australian requirements and includes some of the recent developments to overcome
problems arising in fine coal processing.
D I F F E R E N T N AT U R E O F R E G I O N A L C O A L S
There are both similarities and distinct differences in the design and operation
of Australian coal preparation plants with respect to other regions. This has
led to robust discussions and questioning of the respective approaches by these
parties. It does need to be recognized that for coal preparation, and many other
process industries, conservatism and fashion can dictate many decisions. But
there are some significant differences in the objectives of the respective regional
industries.
The Feed to the Plant

The first major difference that needs to be considered is the feed materials to
the majority of the plants in each country. If coals such as Powder River Basin
coals are discounted given that they are of a low rank with very little prepara-
tion required (crushing and screening), the Northern Hemisphere coals are
dominated by coal originating from the deposition of organic material dur-
ing the Carboniferous period (van Krevelen 1993). The main Australian coal
123
measures originated during the Permian period when the Gondwanaland

supercontinent was in existence (Australia, India, South Africa, South America,
and Antarctica [van Krevelen 1993]).
Sixty million years separated the genesis of the Northern Hemisphere
and Australian coal basins with significant differences in the starting plant
material, the climate, and the environment. The most abundant plant forms in
the Carboniferous period were cryptogams (ferns), and they were deposited
in subtropical lagoons or deltaic swamps (this semi-marine environment led
to relatively high sulfur levels). The younger Permian coals were laid down in
much cooler conditions in continental basins with pines and firs as the vegeta-
tion base (van Krevelen 1993; Cook 1975).
Volcanoes in the vicinity of the Gondwanaland coal seams during forma-
tion had a major influence on the coal’s characteristics (Creech 2002). This
resulted in a characteristic feature of these coal seams where the volcanic tuffs
were transformed into various clay minerals, which are evident in all the coal
seams from this period. The clay is either
• Intimately dispersed in the organic material,
• Thin clay bands inter-dispersed within the coal seams, or
• Clay/shale bands closely associated with the seams.
This results in these coals being relatively more difficult to clean with respect
to final product ash value and higher levels of near-gravity material (Whitmore
1979).
The end result is coals that have very different coking and thermal poten-
tialities but, importantly, washability characteristics, which are different to the
Northern Hemisphere coals.
The major impact of this difference is the amount of near-gravity mate-
rial present and locked minerals within the coal, which need to be dealt with
to realize the potential worth of the coal. The majority of U.S. Carboniferous
seams do not have a problem with respect to significant amounts of clay inclu-
sion. A different problem in the United States is the occurrence of sulfur (par-
ticularly pyrite), which requires differing cleaning approaches to be considered.
Markets
The markets for coal from the Northern Hemisphere and Australia are very
different. For example, more than 90% of U.S. production is for internal use in
thermal electricity production. There is a strict commercial association between
specific energy of the delivered coal and its value. Hence, both the ash value and
the moisture content of the coal are of importance. The Australian context is
Fine Coal Processing—Some technical developments 125
strongly influenced by the international market for hard and soft coking coals
and pulverized fuel injection coals, with more than 80% of the coal production
exported. The ash value is of strict importance but the emphasis on moisture
is decreased with the recovery of material with prime coking characteristics
becoming an additional objective.
As a result, the majority of Australian plants operate at lower relative den-
sity of separations in their dense-medium cyclone (DMC) circuits compared
with the U.S. plants. The introduction of the pneumatic flotation cells ( Jame-
son and Microcel) with wash water applied in significant quantity to the thick
beds of froth has resulted in lower fine coal product ash values with a commen-
surate increase in the DMC separation density. This effectively has allowed the
plants to operate closer to the preferred situation having all cleaning circuits
operating at the same incremental ash value (Abbott 1981; Cierpisz and Gott-
fried 1977). If an intermediate cleaning step such as spirals is included, then it
would have difficulty in attaining a separation density of about 1.5, which is a
common value for the coking coals. This needs to be considered in conjunction
with potential problems of poor probable errors for the particles below 2 mm
in a DMC and flotation having difficulty with recovering plus 0.5-mm wedge
wire coal particles that may be present as the desliming screen wears.
In the United States, the density of separation is higher on average because
of favorable washabilities. The need to reduce the sulfur level to a reasonable
value is usually satisfied at these higher densities. The separation densities
among the dense-medium bath, DMCs, spirals, and flotation are much closer,
hence satisfying the optimization criteria of separating at the same incremental
ash value (Abbott 1981; Cierpisz and Gottfried 1977).
With regard to dewatering, the typical moisture content of Australian
export coal is in the order of 10% by mass for coking coal, whereas thermal coal
is mainly sold on a specific energy basis. The moisture level for the coking coal
is a result of a long-term compromise of recovering as much of the fine coal due
to its good coking properties at an acceptable moisture level. Vacuum filtration
has a long history of providing this need in Australia with horizontal belt filters
becoming the preferred unit due to ease of feed presentation and control.
Though there are a few installations of screen-bowl centrifuges in Australia,
this unit is dominant in the United States. It provides consistent low-moisture
values and performs a secondary size classification step by discarding signifi-
cant amounts of –0.04 mm material. Considerations have indicated that the
amount of water which this size fraction retains in the final product results
in it having a negative impact on the overall specific energy of a thermal coal
product (Hart et al. 2005; Luttrell et al. 2004). Another factor also needs to be
considered in the different dewatering approaches; the topology of the land in

the Appalachian coalfields is characterized by mountainous land in which flat
areas for placing a coal preparation plant is difficult to obtain. Hence the need
for compact plants, which use height and high-capacity units to reduce the
floor plan area. In Australia there is ample flat land available to accommodate
the footprint of large horizontal belt filters.
O V E R C O M I N G C L AY
Clay in run-of-mine coal is a major issue in the Australian coal preparation
industry. This is particularly evident in the Hunter Valley region in New South
Wales and the rapidly developing Surat Basin in Queensland. The majority of
the mines in these areas produce thermal coal products due to their lower rank
and/or the high ash value even after cleaning due to significant amounts of fine
clay embedded in the carbonaceous material.
The main fine coal processing issues arising from this type of raw coal are
• Removal of ultrafine material,
• The high amount of near-gravity material, and
• Continual breakdown of clay bands.
These issues give rise to a major problem with respect to the design of a coal
preparation plant for a new mine. The sample preparation approach to be used
for estimating the size distribution and washability of the raw coal has to be
carefully controlled, otherwise significant problems may arise with respect to
capacity in the fines circuit of a plant. There will be breakage due to handling
and continual breakdown of the clay/shale bands in the coal. This is critical to
the selection of process unit operations and their expected capacity require-
ments, particularly with regard to the fines circuit. Although there are some
variations on the approach in Australia, the work of Swanson et al. (1993) still
provides the basis.
Size Classification
Improvement in size classification at separation sizes below 0.25 mm has been
the subject of much investigation over a long period of time, and the classifi-
cation technology chosen for this separation is a compromise between many
factors such as
• Efficiency of the separation,
• Capacity of the unit operations,
• Level of operator attention and maintenance,
• Ultrafine particles following the water split, and

• Impact of particle density on separation size.
Hydrocyclone
A hydrocylone is the preferred unit because of its low cost, high capacity, and
simplicity in operation. The size of separation can be varied over a wide range
depending on the cyclone diameter and operating conditions. Unfortunately,
some significant deficiencies arise from this unit. These are clearly demon-
strated by data obtained by a detailed audit at a Hunter Valley coal preparation
plant. The fine coal circuit had two-stage desliming cyclones to remove the
ultrafine clay before the flotation circuit. There were two banks of 12 primary
cyclones and 12 secondary cyclones (all 250 mm in diameter).
Figures 1 and 2 show the size partition curves as well as those for selected
relative density fractions. Both cyclones had well-developed and stable spray
discharges. These results are typical of partition curves obtained from the other
cyclones.
The first finding is that for the individual particle density partition; there is
a significant difference in the separation sizes. The clays have a separation size of
about 0.02 mm, whereas the clean coal (relative density 1.28) is slightly above
0.10 mm and there is a major loss of coarse coal particles.
The ultrafine particles tend to follow the water split, and the fraction of
water reporting with the cyclone underflow (Rf ) is 0.07 and 0.20 for the pri-
mary and secondary cyclones, respectively. This is mainly due to the solids con-
tent of the respective feed flow streams being 10% and 18%, respectively. In the
Australian context, there are large amounts of clay in the ultrafine size fraction.
When the performance of the cyclones is considered with respect to par-
ticle size, it appears that a poor separation has been achieved. In Figure 2, the
size partition curve has a “bump” at about 0.05 mm size. This was observed
with some of the other cyclones and identified in some previous plant audits. It
was thought to be the result of poor laboratory analysis, but this investigation
has shown that it results from the density effect where the clay is separated at a
lower separation size than the coal and the washability of the feed coal.
Large-Diameter Hydrocyclone
The potential use of a 1-m-diameter classifying cyclone to separate fines in the
region of 0.2 to 0.3 mm would have some interesting advantages. There would
be reduction in capital and maintenance costs and a simpler plant layout. The
elimination of the slurry subdivision step would also lead to improved process
efficiencies. A detailed investigation of the size separation has been carried
0.9
0.8
0.7
Partition Coefficient
0.6
0.5
0.4
1.28
0.3
1.35
1.5
0.2 1.9
2.2
0.1 Size
Rf = 0.07
0
0 0.05 0.1 0.15 0.2 0.25 0.3
Size, mm
Figure 1 Individual partition curves for float/sink density fractions for one of the
primary cyclones
0.9
0.8
0.7
0.6
0.5
0.4
1.28
0.3
1.35
1.5
0.2 1.9
2.2
0.1 Size
Rf = 0.20
0
0 0.1 0.2 0.3 0.4
Size, mm
Figure 2 Individual partition curves for float/sink density fractions for one of the
secondary cyclones
out at the Stratford coal preparation plant north of Newcastle in New South
Wales (O’Brien et al. 2000). A number of operating conditions, which could
be encountered in normal plant operation, were considered, including feed
solids content and vortex finder and spigot diameters. The separation efficiency
obtained for the cyclone was comparable to much smaller cyclones over a wide
range of conditions, and the separation size varied from about 0.15 to 0.35 mm
(see Figure 3).
Size by density float/sink analysis of the solids in the various flow streams
has shown that the shale has a separation size of about 0.08 mm, but the separa-
tion size of clean coal (float 1.3) can be varied from about 0.25 mm to nearly
0.5 mm (see Figure 4). This indicates that the solids content of the feed, its
washability, and subsequent cleaning unit operations need to be considered
carefully before a large unit of this type is installed in a plant.
The performance of the existing cyclone bank/rapped sieve bend circuit
was also determined for comparison (see Figure 5).
Sieve Bends
There have been many attempts to use sieve bends (rapped or unrapped) to
overcome some of the deficiencies of the hydrocyclone. Unfortunately, these
units tend to be low capacity, higher in cost, and require more operator and
maintenance attention. In fact, because of wear on the screen surface, it has to
be rotated at regular intervals. That is, the efficiency and capacity fall and rise
with the maintenance cycles.
The feed rate is the important variable with respect to the performance of
a sieve bend. There tends to be a feed rate at which a sieve bend will operate at
maximum efficiency. The efficiency declines at lower or higher values, and the
tolerance to variation increases for screens with larger apertures (Schreckengost
1989). The work of Slechta and Firth (1983) showed that this effect is due to
the efficiency of the undersize monotonically decreasing with increase in feed
rate and the efficiency of the oversize approaching a plateau. The capacity value
for the highest combined efficiency is the value at which the plateau is reached.
Figure 6 shows the impact of the mass flow rate over the sieve bend used by
Slechta and Firth (1983) on the performance of the sieve bend as described
by the partition curves. The partition curves start to degenerate for solids
flow rates above about 7 tph, which closely matches the 8.3 Lps/m value. The
capacity for maximum efficient separation has been estimated for a number of
apertures, and these are shown in Table 1. The solids content of the feeds were
all about 20% by mass.
0.9
0.8
0.7
0.6
0.5
0.4
0.3
0.2
10% solids
0.1
20% solids
0
0 0.2 0.4 0.6 0.8 1
Size, mm
Figure 3 Comparison of the partition curves obtained by the 1-m-diameter cyclone

with varying solids content
0.9
0.8
0.7
0.6
0.5
0.4
1.28
0.3
1.35
1.5
0.2 1.9
2.2
0.1 Size
Rf = 0.06
0
0 0.2 0.4 0.6
Size, mm
Figure 4 One-meter-diameter cyclone size/density float/sink partition curves for

10% solids feed
0.9
0.8
0.7
0.6
0.5
0.4
0.3
0.2 Bank of Cyclones

Cyclone Bank and Sieve Bend
Large-Diameter Cyclone
0.1
0
0 0.2 0.4 0.6
Size, mm
Figure 5 Comparison between the size partition curves for the large-diameter
cyclone, the bank of conventional cyclones, and the complete cyclone bank/rapped
sieve bend circuit
Case Study
A recent paper (Bain 2012) provides an interesting case study of the changes
that a Hunter Valley mine had to make to overcome major clay problems, which
impacted on raw coal handling through to product quality. The plant was ini-
tially a DMC and spiral plant with no flotation. The mine has 19 seams with
different raw coal properties and washabilities. Nearly all of the potential plant
feed coals have high bentonite clay content, which has forced the plant circuit
to evolve over 20 years in the following manner:
• Low friction materials were required to facilitate flow of material
through the plant.
• Close monitoring of the amount of water being added at various parts
of the plant to prevent material “sticking” is required.
• The use of classifying cyclones only led to spiral products with high
clay contents, and hence high ash values and moistures. Initially, a sec-
ond bank of cyclones was added in series, which was aimed at remov-
ing all of the –0.08 mm material.
• The next step was to thicken the spiral product in a third bank of
cyclones prior to dewatering with a fine coal centrifuge.
0.9
0.8
0.7
0.6
0.5
0.4
0.3 1.3 tph

2.0 tph
2.6 tph
0.2
6.0 tph
7.5 tph
0.1
10.9 tph
0
0 0.05 0.1 0.15 0.2 0.25 0.3
Size, mm
Source: Slechta and Firth 1983.
Figure 6 Efficiency curves for the DSM rapped sieve bend
Table 1 Estimated capacities for maximum separation efficiency for sieve bends
Screen Aperture, mm Capacity for Maximum Efficiency, Lps/m
0.120* 8.3
0.25† 16
0.3‡ 20
0.5 ‡
34
1.8‡ 55
3.5 ‡
103
* Rapped DSM sieve bend (Slechta and Firth 1983).
† Variesieve (Firth et al. 1995).
‡ Schreckengost 1989.
• After investigations with rising current classifiers and water injection

cyclones, it was decided that sieve bends would be included in the cir-
cuit. With a preferred separation size of 0.30 mm, a complete revalua-
tion of the fine coal circuit was undertaken.
• This led to a circuit with the bottom size of the feed to the spirals being
0.30 mm and flotation being used to clean the –0.30 mm material. The
concept of only having sieve bends for the classification was not viable
due to the amount of screen area required.
• Because the 0.125 mm separation size was not required, a 1-m-diam-
eter classifying cyclone was installed to replace the bank of smaller
cyclones.
• Jameson cells with the capability of significant addition of froth wash
water were employed for treating the ultrafine size fraction.
This circuit has resulted in product quality targets being met with major
increases in yield.
U S E O F C O A L G R A I N A N A LY S I S
The optimum design of a fine coal circuit requires a good understanding of the
associations of the minerals and macerals. This controls the washability of the
fines and the potential response to flotation. Coal grain analysis (CGA) char-
acterizes the individual fine coal grains in terms of liberated maceral and com-
posite particles (O’Brien et al. 2003), and hence enables an estimation of the
washability of the fines and the flotation response of specific grain types to be
determined. This new level of detail provides a potentially very useful approach
to process optimization and hence increased resource recovery.
CGA processes the individual coal grains separately and provides quantita-
tive data on particle area, which can then be used to estimate a volume composi-
tion (vitrinite, inertinite, dark mineral, and bright mineral). The density of the
grains can then be estimated from the densities of the individual components.
A summary of the information generated for a particular coal is shown in
Figure 7.
Figure 8 shows the washabilities of a feed coal to a spiral and the resulting
spiral product using CGA data. For comparison, conventional float/sink analy-
sis was also carried out and a reasonable match has been achieved. From these
data, partition curves for fine coal processing unit operations can be obtained
relatively quickly without the problems associated with organic liquid float/
sink analysis.
In a similar manner, the feed, concentrate, and tailings from a flotation
process can also be examined with respect to the recovery of the various
components in the feed. In this case, the components have been classified as
vitrite (greater than 95% of the grain is vitrinite), inertite, mineral, vitrinite
rich (vitrinite is the dominant maceral), inertinite rich, and mineral rich. The
results are shown in Figure 9 for the various size fractions. The vitrite is floating
relatively easily except for the coarse vitrinite-rich grains. Inertite is not floating
Pixel Measurements Volume Percent Grain Information
Dark Bright Dark Bright

Area Vitrinite Inertinite mins mins Vitrinite Inertinite mins mins Density Ash %
249,345 63,972 169,375 15,998 0 25.66 67.93 6.42 0.00 1.368 10.82
3,586 3,586 0 0 0 100.00 0.00 0.00 0.00 1.232 0.29
73,132 46,122 24,530 2,480 0 63.07 33.54 3.39 0.00 1.302 5.96
61,341 48,654 7,260 5,427 0 79.32 11.84 8.85 0.00 1.366 14.59
53,507 41,350 5,063 7,094 0 77.28 9.46 13.26 0.00 1.428 21.13
19,558 0 0 19,558 0 0.00 0.00 100.00 0.00 2.662 69.88
94,148 0 0 94,148 0 0.00 0.00 100.00 0.00 2.662 69.88
245,597 137,115 36,001 72,443 38 55.83 14.66 29.50 0.02 1.664 41.84
251,677 23,415 205,710 22,538 14 9.30 81.74 8.96 0.01 1.413 14.76
5,909 2,852 688 2,362 7 48.27 11.64 39.97 0.12 1.816 52.45
411 0 411 0 0 0.00 100.00 0.00 0.00 1.297 0.29
251,544 2,328 238,036 11,164 16 0.93 94.63 4.44 0.01 1.357 7.67
68,988 68,071 354 563 0 98.67 0.51 0.82 0.00 1.244 1.67
97,794 69,158 19,126 9,510 0 70.72 19.56 9.72 0.00 1.384 15.92
2,847 2,847 0 0 0 100.00 0.00 0.00 0.00 1.232 0.29
251,544 2,328 238,036 11,164 16 0.93 94.63 4.44 0.01 1.357 7.67
68,516 48,401 6,071 14,035 9 70.64 8.86 20.48 0.01 1.531 31.01
3,574 3,534 25 15 0 98.88 0.70 0.42 0.00 1.238 1.00
98,935 27,401 58,045 13,458 31 27.70 58.67 13.60 0.03 1.466 21.67
256,763 0 0 256,763 0 0.00 0.00 100.00 0.00 2.662 69.88
61,422 50,872 9,268 1,282 0 82.82 15.09 2.09 0.00 1.272 3.81
228,103 197,311 28,908 1,884 0 86.50 12.67 0.83 0.00 1.252 1.69
228,754 83,708 102,465 42,581 0 36.59 44.79 18.61 0.00 1.527 28.54
231 223 8 0 0 96.54 3.46 0.00 0.00 1.234 0.29
Figure 7 Output from coal grain analysis
that well above 0.3 mm. Some of the locked mineral matter is reporting with
the product along with some ultrafine clay material. This technique allows coal
flotation to be considered in the same manner as used in mineral processing.
CONCLUSIONS
Development of improved fine coal processing circuits is progressing in Aus-
tralia within the constraints of the time and finance, and the coal quality issues.
New techniques to assist this development are becoming available and should
improve the recovery of more salable coal.
100
90
80
70
Cumulative Recovery, %
60
50
40
30
20 CGA Feed
CGA Conc
Lab Assay Feed
10
Lab Assay Conc
0
1 1.5 2.0 2.5 3
Grain Density, g/cc
Figure 8 Comparison of washabilities estimated from CGA with conventional float/

sink analysis
100.0
90.0
80.0 Mass Split

Vit
70.0 Inert
Component Recovery, %
Min
Vit Rich C
60.0
In Rich C
Min Rich C
50.0
40.0
30.0
20.0
10.0
0.0
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8
Characteristic Size, mm
Figure 9 Component recovery by flotation as a function of particle size
ACKNOWLEDGMENTS
The authors thank CSIRO for permission to publish this information, ACARP
for jointly funding the research, and the many mines who have allowed access
to their plants so that the projects could be carried out.
REFERENCES
Abbott, J. 1981. The optimization of process parameters to maximize the profitability
from a three component blend. In Proceedings of the First Australian Coal Prepara-
tion Conference. Edited by A.R. Swanson. pp. 87–105.
Bain, G. 2012. Conquering the clay at Rix’s Creek CHPP. In Proceedings of the 14th
Australian Coal Preparation Conference. Edited by D. Mathewson. Paper D3.
Cierpisz, S., and Gottfried, B.S. 1977. Theoretical aspects of coal washer performance.
Int. J. Miner. Process. 4:261–278.
Cook, A.C. 1975. Australian Black Coal—Its Occurrence, Mining, Preparation and
Use. Symposium organized by the Illawarra Branch of the Australasian Institute of
Mining and Metallurgy. pp. 63–84.
Creech, M. 2002. Tuffaceous deposition in the Newcastle Coal Measures. Int. J. Coal
Geol. 51:185–214.
Firth, B., Edward, D., Clarkson, C., and O’Brien, M. 1995. The impact of fine classifi-
cation on coal preparation performance. In Proceedings of the 7th Australian Coal
Preparation Conference. Edited by J. Smitham. Paper E2.
Hart, G., Townsend, P., Morgan, G., and Firth, B. 2005. Improving fine coal centrifug-
ing—Stage 3. Australian Coal Association Research Program. Project C9047.
Luttrell, G., Honaker, R., and Yoon, R-H. 2004. Optimisation of the Coal Fuel Supply
Chain: A Coal Preparation Prospective. 29th International Technical Conference
on Coal Utilisation and Fuel Systems, Clearwater, FL. Preprint 140.
O’Brien, M., Firth, B., Clarkson, C., and Edward, D. 2000. Operational performance
of a large diameter (one metre) classifying cyclone. Proc. 8th Aust. Coal Prep. Conf.
C1: 98–111.
O’Brien, G., Jenkins, B., and Beath, H. 2003. Coal grain analysis for improved predic-
tion of utilisation performance. In Proceedings of the 12th International Conference
of Coal Science.
Schreckengost, D. 1989. Sieve bends handle high volumes. Coal 42–50.
Slechta, J., and Firth, B. 1983. Operation of a rapped sieve bend. Proc. Australas. Inst.
Min. Metall. 288:7–12.
Swanson, A., Fletcher, I., and Partridge, A. 1993. Improved Prediction of Size Distribu-
tions and Their Effects in Materials Handling and Coal Preparation Systems. Final
Report for NERDDC Project 1290.
van Krevelen, D.W. 1993. Coal: Typology, Physics, Chemistry, Constitution. New York:
Elsevier.
Whitmore, R.L. 1979. The washability of Australian coals. Proc. Australas. Inst. Min.
Metall. 270:47–53.
Beneficiation
Technologies 3
Fine Coal Processing with
Dense-Medium Cyclones
G.J. de Korte
ABSTRACT
Dense-medium cyclones have been used for many years to beneficiate fine coal in
a number of countries around the world. The use of cyclones in this application is
not widespread, and at present, the process is employed only in South Africa and
China. The chapter provides a brief overview of past and current application of
dense-medium cyclones in the processing of fine coal and reviews some of the impor-
tant considerations for successful application of the technique.
INTRODUCTION
The dense-medium cyclone has, since its development by Dutch State Mines
(DSM) almost 70 years ago, become the main processing unit in the coal indus-
try. It is capable of efficient separations, even on coals containing high amounts
of near-density material, and accurate control of product quality is possible.
Although cyclones are capable of processing coal down to fine sizes, they are
generally applied only to coal coarser in size than 0.5 mm.
Beneficiation of fine coal, nominally below 0.5 mm, with dense-medium
cyclones was implemented in 1957, and cyclones are still being used to process
fine coal today, but history shows that the application of cyclones in fine coal
processing has not been nearly as successful as in the processing of coarser coal.
There is a perception in the coal processing industry that dense-medium
cyclone processing of fine coal is expensive, that it consumes much magnetite,
and that the separation efficiency is not good. There are, however, some who
realize that cyclones are still the most effective means to process fine coals
containing elevated amounts of near-density material at low cut-point densi-
ties, and, as such, the technology is presently in limited use in South Africa and
widely employed in China.
139
140 Beneficiation Technologies
Table 1 FRI results on fine coal

Test No. 31 A 31 B 31 C 31 D 31 F
RD of separation 1.35 1.37 1.37 1.37 1.32
EPM* 0.0387 0.0275 0.0283 0.0345 0.0336
Feed pressure, psi 25 17.5 10 7.5 4
Feed pressure, kPa 172 120 70 52 27.6
*Ecart probable moyen or probable error.
BRIEF OVERVIEW
South Africa—1949
Some of the earliest work on beneficiation of fine coal with a dense-medium
cyclone was carried out in South Africa by the Fuel Research Institute (FRI)
of South Africa under the guidance of P.J. van der Walt in 1949. Using a
240-mm-diameter cyclone and barites as the dense medium, the results shown
in Table 1 were obtained when processing minus 1 mm by 0.25 mm coal. The
results showed that fine coal could be processed at a low relative density (RD)
and with very good separation efficiency.
Belgium—1957
In 1957, a flowsheet was developed by Stamicarbon/DSM and Evence Cop-
pée for a new dense-medium cyclone plant at Tertre in Belgium to process
10 × 0 mm raw coal. The plant was unique in that the feed to the plant was
not deslimed before processing. The raw coal was mixed with magnetite and
gravity-fed to two 500-mm-diameter (D) cyclones at a feed pressure of 9D.
The minus 0.75 mm coal in the feed drained through the product and discard
screens into the circulating medium tank. A part of the circulating medium
was pumped, using a separate pump, to two 350-mm-diameter cyclones at a
feed pressure of approximately 30D to affect a separation on the minus 0.75
mm coal. The overflow and underflow medium from the 350-mm-diameter
cyclones were directed to separate product and discard magnetic separators
to recover a fine coal product and discard in the underflows of the magnetic
separators. The product and discard were dewatered using conventional dewa-
tering equipment. A simplified flow diagram of the circuit employed at Tertre
is shown in Figure 1.
The separation efficiency of the plant at Tertre was good, and the results
were reported by Mengelers and Absil (1976) (see Table 2). The magnetite
consumption for the operation at Tertre was approximately 1 kg per feed ton.
In 1965, a similar plant was constructed at Winterslag in Belgium. This plant
Fine Coal Processing with Dense-Medium Cyclones 141
10 × 0 mm
Raw Coal
10 × 0.75 mm
Discard
10 × 0.75 mm
Product
0.75 × 0 mm
0.75 × 0 mm Product
Discard
Figure 1 Simplified flow diagram of circuit used at Tertre
Table 2 Results obtained at Tertre plant

Coal Size Fraction, mm RD of Separation EPM
10 × 0.75 1.57 0.035
0.75 × 0.3 1.63 0.050
0.3 × 0.15 1.83 0.010
remained in operation for 17 years and was also reported to perform satisfacto-
rily (Lathioor and Osborne 1984).
United States—1970s
During the 1970s, a number of “wash-to-zero” plants, based on the Tertre
flowsheet, were built in the United States by Stamicarbon’s licensee, Roberts
and Schaefer. In these plants, efficient separation of the minus 0.5 mm size
fraction was not specifically targeted; therefore, the fine coal was processed,
together with the coarser coal, in a large-diameter cyclone. Recovery of the fine
coal from the drained medium was carried out in a similar manner as in the
Belgian plants. A simplified flowsheet for a typical wash-to-zero plant is shown
in Figure 2.
The results obtained from a wash-to-zero plant in West Kentucky, as
reported by Taylor and Chen (1980), are shown in Table 3. As can be seen, the
separation efficiency obtained on the fine coal was worse than that obtained
50 × 0 mm
Raw Coal
50 × 1 mm 50 × 1 mm
1 × 0 mm Discard Product
Discard
1 × 0 mm
Product
Figure 2 Wash-to-zero flow diagram
Table 3 Results obtained in wash-to-zero plant

Coal Size RD of
Cyclone Module Fraction, mm Separation EPM
Module 1: 28-in. (711-mm) cyclone 25 × 0.6 1.54 0.027
operating at 9D 0.6 × 0.15 1.95 0.149
Module 2: 24-in. (610-mm) cyclone 32 × 0.6 1.65 0.035
operating at 20D 0.6 × 0.15 2.03 0.159
in the Belgian plants. The finer coal was also separated at a much higher rela-
tive density than the coarser coal. Magnetite consumption for the plant was
reported to be around 0.5 kg/feed ton.
Similar results were reported by Island Creek Company, which con-
structed several wash-to-zero plants during the late 1970s and early 1980s.
Table 4 summarizes the results obtained at the Providence plant (Burch 1987).
Table 5 shows the names and commissioning dates for some of the wash-to-zero
plants built by Island Creek Company.
The main advantage offered by the wash-to-zero concept is simple plant
layout, which translates to low capital cost. The desliming screens and the fine
coal beneficiation section can be eliminated while still providing an efficient
separation on the coarse coal. The separation obtained on the fine coal, without
Table 4 Results obtained in Providence wash-to-zero plant

Test No. Coal Size Fraction RD of Separation EPM
1 32 mm × 600 μm 1.455 0.020
600 × 150 μm 1.61 0.168
2 32 mm × 600 μm 1.48 0.032
600 × 150 μm 1.71 0.205
Table 5 Wash-to-zero plants built by Island Creek Company

No. of Feed Size, Year
Plant Contractor Cyclones inches Commissioned
Pond Fork Childress 1 ³⁄8 × 0 1976
Coal Mountain #12 Childress 1 ³⁄8 × 0 1977
Providence J.O. Lively 4 1¼ × 0 1979
Fies #9 J.O. Lively 4 1¼ × 0 1979
Hamilton #1 J.O. Lively 6 1¼ × 0 1979
Hamilton #2 J.O. Lively 4 1×0 1979
Holden #22 J.O. Lively 4 1×0 1980
No. 25 Mine F.M.C. 3 1×0 1981
North Branch Envirotech 2 1¼ × 0 1981
North Branch Island Creek 2 1¼ × 0 1983
any additional equipment, is still better than that which can be had with most
water-based fine coal processing equipment.
Two plants designed to process 1 × 0.1 mm coal were also built in the
United States during the late 1970s. The plants were at Marrowbone in West
Virginia and Homer City in Pennsylvania. The coal preparation plant at Homer
City was built to prepare “deep cleaned” coal for burning in the power station.
The main purpose of the plant was to reduce the ash level and sulfur content
of the coal fed to the power station as an alternative to flue gas desulfurization.
The washability of the coal at Homer City is such that a very low cut-point
density of about 1.30 had to be employed to achieve coal of the required qual-
ity. Given that the minus 1 mm size fraction is beneficiated at a relatively higher
cut-point density than the coarser coal in a wash-to-zero operation and further
because this density cannot be controlled, it was necessary at Homer City to
employ a separate fine coal circuit to beneficiate the fine coal. The fine coal
plant was commissioned in July 1978. The performance of the plant during the
first year was reported to have been poor.
In February 1979, the utility owners initiated an in-depth investigation to

trace the problem areas in the circuit and to rectify such problems. The follow-
ing major “flaws” were identified as a result of the investigation (Sehgal et al.
1982):
• The circulating medium was contaminated with fine coal. This was
a result of entrapment of fine coal by the magnetic separators. The
entrapment of coal by the magnetic separators in turn was caused by
too high a percentage of solids in the magnetic separator feed.
• The high percentage of fines in the circulating medium increased the
viscosity of the medium, which in turn led to reduced efficiency of
separation in the 14-in. (350-mm) dense-medium cyclones.
• The high percentage of fine coal in the medium made it very difficult
to control the density of the medium.
• The magnetite in the circuit degraded as a result of high losses of, par-
ticularly, the ultrafine fraction.
Several modifications were made to the circuit following this investigation.
The most significant of these was to modify the size split of coal fed to the fine
coal beneficiation plant. Where a minus 3 mm feed was previously fed to the
circuit, the screening arrangement was modified so that the 3 mm × 0.5 mm raw
coal was removed and fed to a separate conventional dense-medium cyclone cir-
cuit employing conventional drain-and-rinse screens. Only the minus 0.5 mm
coal was sent to the fine coal dense-medium circuit, which resulted in a signifi-
cant reduction in feed tonnage. With the smaller top size and reduced tonnage
being fed to the circuit, it was possible to change the dense-medium cyclones
from 350 mm to more efficient 200 mm cyclones.
Additional Derrick screens were installed for desliming the feed. Rapped
sieve bends were installed to drain medium from the product and discard fol-
lowing the separating cyclones. This allowed a substantial part of the medium
to be recirculated directly back to the medium tank. The sieve bend overflow
material was diluted with sufficient water before being fed to the magnetic
separators to minimize entrapment of fine coal in the recovered medium. The
medium density control was improved by utilizing a nuclear density gauge in
conjunction with a Ramsey coil. This allowed contamination of the medium
to be “calculated,” and the actual density of the medium in the circuit could
thus be controlled (Esposito and Higgens 1982). The plant seemed to operate
satisfactorily after these modifications.
South Africa—1970s
In the early 1970s, South African coal companies concluded a contract with the
Japanese steel industry for the supply of a low ash blend coking coal. This coal
was to have an ash value of 7% and would be produced by processing the raw
coal from the Witbank no. 2 seam. The coal had to be processed at low rela-
tive density in the presence of very high amounts of near-density material, and
dense-medium separation was found to be the only process capable of affecting
the required separation. At the time, the fine coal could not be washed to 7%
ash and was therefore added raw to the lower-value middling coal.
Extensive research conducted in South Africa led to the conclusion that
dense-medium cyclones would be the only process capable of beneficiating the
fine coal to 7% ash (Horsfall 1976) and a 5-t/h pilot plant was subsequently
built at the FRI test facility in Pretoria in 1976. The plant was equipped
with a single 150-mm-diameter dense-medium cyclone supported by all the
additional equipment needed to feed the cyclone and recover magnetite. The
cyclone was fed at feed pressures typically ranging between 80 and 150 kPa
(40D to 70D). The plant was employed to test the beneficiation of fine coal
from several collieries in the country and proved that the production of low ash
coal from the fine fraction was possible. A simplified flow diagram of the FRI
fine coal dense-medium plant is shown in Figure 3.
The results obtained during a 10-day continuous operating period of the
pilot plant are shown in Table 6 (Fourie et al. 1980). Based on the success of
the pilot-plant work, a full-scale fine coal dense-medium plant was built at
Greenside colliery and commissioned in 1980. This plant produced low ash
coal from the fines fraction and remained in operation for 18 years. Three
additional plants, based on the FRI design, were constructed in South Africa
at Newcastle-Platberg colliery, High Carbon Products in Natal, and Rooiberg
tin mine. The 150-mm-diameter dense-medium cyclone used in the FRI pilot
plant is shown in Figure 4 (CSIR 1980).
Australia—1990s
A fine coal dense-medium plant was constructed at ARCO’s Curragh mine
in Central Queensland, Australia, in the early 1990s. As in South Africa, the
motivation for the plant at Curragh was to produce a product containing 7%
ash from the minus 0.5 mm size fraction.
Much of what had previously been learned about dense-medium cleaning
of fine coal was incorporated into the design of the Curragh plant, which was
carried out in cooperation with CLI Corporation of Pittsburgh, Pa. The lessons
–100 micron
Product
Raw –1 mm
Discard
Figure 3 Simplified flow diagram of FRI dense-medium fine coal pilot plant
Table 6 Results obtained from FRI dense-medium fine coal pilot plant
Test No.
1 2 3 4
Product ash, % 6.5 7.7 7.4 5.9
Product yield, % 53.1 64.2 58.8 43.3
RD of separation 1.46 1.50 1.48 1.43
EPM 0.022 0.031 0.025 0.020
Total magnetite losses, kg/t 0.81 1.11 1.32 0.9
learned from the Homer City plant (United States) in particular were taken
into account (Kempnich et al. 1993).
The plant was equipped with two 500-mm-diameter dense-medium
cyclones. Much attention was devoted to the design of the magnetite-recovery
circuits, as well as to the desliming of the feed coal. Vibrating sieve bends were
used to assist in both desliming of the feed and recovery of medium. The plant
started up “smoothly” during commissioning. Magnetite consumption seemed
low and there was no noticeable impact on the magnetite consumption of the
overall Curragh operation. However, problems were encountered with the effi-
ciency of the cyclones, and it was not possible to produce the required 7% ash
level in the product. Test work carried out onsite led to the conclusion that the
initial cyclone feed pressure of 9D, or approximately 45 kPa, was too low. The
Figure 4 150-mm cyclone in FRI dense-medium fine coal pilot plant
feed pressure was therefore increased to 120 kPa after which a 7% ash product
was consistently achieved (Kempnich et al. 1993).
The fine coal plant at Curragh closed down toward the end of 1996 for
reasons unknown.
South Africa—2001
In 2001, a 25-t/h fine coal dense-medium pilot plant was built in South Africa
by the Coaltech research program. The objective of the plant was to demon-
strate that coal from the Witbank no. 4 seam, which is difficult to process, can
be upgraded to export quality, something that could not be done using spirals.
The plant employed two dense-medium cyclones in series to improve separa-
tion efficiency and was run on standard “medium” grade magnetite—one of the
objectives for the research. A simplified flow diagram of the Coaltech circuit is
shown in Figure 5.
Extensive test work was conducted on the raw fine coal from several col-
lieries over a period of 5 years and much valuable information was gained from
the plant. As a result of the work done, the spiral plant at Leeuwpan colliery
was replaced with a dense-medium cyclone plant. This plant is presently in
operation at Leeuwpan. The dense-medium cyclones in the Leeuwpan plant
are shown in Figure 6, a photograph taken during the construction of the plant.
China—Recently
In recent years, many plants were built in China using the wash-to-zero
approach to beneficiate coarse and fine coal simultaneously without the use of
additional fine coal processing equipment.
–100 micron
Product
Raw –1 mm
Discard
Figure 5 Flow diagram of Coaltech pilot plant
Figure 6 Dense-medium cyclones in Leeuwpan fine coal plant
In the new Chinese plants, un-deslimed raw coal is fed to large-diameter

(up to 1,500 mm) three-product dense-medium cyclones and the plus 0.25 mm
product (coking coal), middling, and discard is recovered using conventional
drain-and-rinse screens. Part of the medium draining through the product
(coking coal) screen is diverted and pumped to small (350 mm) conventional
dense-medium cyclones where the minus 0.25 mm fine coal in the medium is
Table 7 Separation efficiency obtained on fine coal

Coal Size Range, mm RD of Separation EPM Comment
0.5 × 0.25 1.65 0.099 Processed in
1,500-mm-diameter
three-product cyclone
0.25 x 0.15 1.52 0.080 Processed in
350-mm-diameter
fine coal cyclone
beneficiated to yield a coking and a middling coal. The coal is recovered from
the small cyclone overflow and underflow with magnetic separators. The circuit
employed is similar to the one implemented at Tertre (Belgium) 55 years ago,
but by using the overflow medium from the three-product cyclones, fine mag-
netite is automatically ensured. The separation of the fine coal is reported to be
effective (Zhao and Yu 2012), and some typical results are shown in Table 7.
L E S S O N S L E A R N E D F R O M T H E PA S T
From the preceding overview on the subject of dense-medium processing of
fine coal, one can conclude that the use of dense medium to clean fine coal has
not always been easy or completely successful. Mixed results were obtained, but
some useful learning does emerge. Some of the more pertinent aspects of the
process are briefly discussed in this section.
Cyclone Diameter and Cyclone Geometry

There seems to be general consensus that smaller-diameter cyclones result in
better separation efficiency. This is principally due to the higher centrifugal
forces generated in smaller cyclones, which aids in the separation between fine
coal and fine shale particles. Very small cyclones, however, are prone to block-
ages when oversized particles are present in the feed and a compromise is there-
fore necessary in practice between separation efficiency and practical operation.
It appears that the optimum cyclone diameter for fine coal beneficiation is thus
between 250 mm and 350 mm.
Cyclones with longer body sections and lower cone angles appear to be bet-
ter suited for fine coal processing. Van der Walt (1949) found that a 25° cone
angle cyclone gave better results than a 38° cyclone.
Magnetite Sizing
Magnetite size consistency is perhaps the most confusing aspect of fine coal
processing. It is well known that finer magnetite improves the separation effi-
ciency in cyclones for the plus 0.5 mm size fractions (DSM 1970; de Korte
Table 8 Separation results obtained with brine medium at RD 1.20

Test No. Conditions RD of Separation EPM
1 Coal 320 g/L 1.41 0.0790
2 Coal 185 g/L deslimed 1.32 0.0543
3 Coal 235 g/L deslimed 1.33 0.0457
2007). In general, it is assumed that for finer sized coal, even finer magnetite is
required, and the FRI (Fourie et al. 1980) insisted that magnetite which is 50%
finer than 10 μm is necessary for efficient separation in the 150-mm-diameter
cyclone. Sokaski and Geer (1963) found that the best separation could be
obtained using grade “B” magnetite. They further found that using finer magne-
tite improved the cyclone performance for coal particles coarser than 350 µm,
but no improvement was noted for the finer sizes.
Tests conducted in a 250-mm-diameter cyclone, operating at a feed pres-
sure of 117 kPa and using saturated brine as a medium (Mengelers 1982), indi-
cated that results which were very similar to those obtained using magnetite
were obtained. Table 8 shows a summary of the results obtained by Mengelers.
Notably, there is a differential of between 0.12 and 0.21 relating the
medium relative density of 1.20 and the actual relative density of separation.
The EPM values obtained are similar to those obtained elsewhere with magne-
tite. Figure 7 shows the normalized partition data for the brine tests compared
to the normalized results obtained at Homer City on a 200-mm-diameter
cyclone operating at a feed pressure of 70 kPa and using grade B magnetite
(Esposito and Higgens 1982). One can see that the results compare well.
The results obtained by Deurbrouck (1974) when processing fine coal
sized between 350 µm and 150 µm, using zinc chloride as the medium, also
indicated EPM values in the range of 0.04 to 0.065. These values compare well
to the results obtained when processing fine coal using magnetite as a medium.
It is difficult to conclude from these results what grade of magnetite would
be best in practice, but it would most probably be safe to assume that the fin-
est magnetite available should be used. Cyclone overflow medium from coarse
coal cyclones can be effectively used to provide fine magnetite to a fine coal
processing circuit. This is the practice at the Leeuwpan mine and also in the
Chinese plants.
Cyclone Feed Pressure

In general, an increase in cyclone feed pressure results in an improved separa-
tion. The most convincing proof is the experience at Curragh in Australia. The
plant at Curragh was initially designed to operate at a cyclone feed pressure of
100
90 Homer City Tests
Brine Tests
80
70
Partition No.
60
50
40
30
20
10
0
0.85 0.95 1.05 1.15 1.25 1.35 1.45 1.55
RD/d50
Figure 7 Comparison of normalized partition data (d50 = relative density of

separation)
45 kPa. When the plant started operation, the required 7% ash product could
not be produced. When the cyclone feed pressure was increased, initially to
65 kPa and then to 120 kPa, the required low ash product could be obtained
(Kempnich et al. 1993). The influence of the cyclone feed pressure on the per-
formance of the cyclones at Curragh is summarized in Table 9.
Tests conducted at the Coaltech pilot plant in South Africa indicated that
the normalized EPM value obtained when processing 1 × 0.1 mm coal could
be improved from 0.066 at 18D to 0.028 at 24D (McGonigal and de Korte
2005). Tests conducted at Homer City, however, found that better separation
efficiency was obtained at low cyclone feed pressures when processing fine coal
at very low relative density (Chedgy et al. 1986). It therefore seems that specific
conditions can dictate the required cyclone feed pressure.
Magnetite Consumption
The magnetite consumption reported from the majority of dense-medium
fine coal plants, both stand-alone and wash-to-zero plants, mostly fall within
normal limits. In general, these plants have magnetite consumption of between
1.0 and 1.5 kg/feed ton, which compares favorably with that of conventional
dense-medium plants processing deslimed feed. In the modern Chinese wash-
to-zero plants, magnetite consumption of around 0.55 kg/t is reported (Zhao
and Yu 2012).
In recent years, the performance of magnetic separators has improved
dramatically through the use of more sophisticated magnetic materials. The
Table 9 Effect of feed pressure on cyclone performance at Curragh

Cyclone Feed Pressure
45 kPa 65 kPa 120 kPa 45 kPa 65 kPa 120 kPa
Parameter +0.25 mm size fraction –0.25 mm +0.125 mm size fraction
RD of separation 1.596 1.752 1.480 1.841 2.078 1.663
EPM 0.1513 0.0704 0.0367 0.2663 0.1369 0.1492
Imperfection* 0.0948 0.0402 0.0248 0.1446 0.0659 0.0897
*EPM/RD of separation.
counter-rotation magnetic separators, installed in the Coaltech pilot plant in

South Africa, are shown in Figure 8 and the performance of these units, mea-
sured over a period, is summarized in Table 10. The separators were manufac-
tured by two companies and were found to be equally effective.
Older dense-medium fine coal plants used rapped sieve bends and multiple
stages of magnetic separation to recover magnetite. With the new generation
of magnetic separators, very good magnetite recovery is possible with only
magnetic separators—this can simplify plant configuration significantly. For
optimum recovery of magnetite with these magnetic separators, the feed to the
separators must be diluted to ensure that the amount of magnetite as well as
nonmagnetic material in the feed falls within the design loading for the units.
This also results in minimum entrapment of coal in the over-dense medium
recovered by the magnetic separator.
Even with modern, efficient magnetic separators, magnetite lost in the
magnetic separator underflow is the finer size fraction. Tests conducted on
magnetic separators in the Coaltech pilot plant confirmed that the magnetite
lost in the magnetic separators effluent is mostly below 10 µm in size (de Korte
2002). The size distribution of the magnetite lost via the product and discard
magnetic separators is shown in Figures 9 and 10. A photograph of the magne-
tite in the magnetic separator tailing is shown in Figure 11.
Particle Size of Coal Processed

From the work done on dense-medium processing of fine coal to date, it is clear
that the process is not effective for the very fine sizes, especially below about
100 µm. Evidence of this can be seen from the work done by Deurbrouck
(1974) using a 200-mm-diameter cyclone operating at a feed pressure of 70 kPa
and using a magnetite medium at a feed relative density of 1.30. A summary of
the results is shown in Table 11 and in Figures 12 and 13. The density of separa-
tion is shown to increase with decreasing particle size, and the EPM value also
increases as the particles become smaller.
Figure 8 Magnetic separators in Coaltech pilot plant
Table 10 Performance of magnetic separators

Date of Determination
Nov. 1, May 31, May 15, June 8, June 15,
Product 2001 2002 2006 2006 2006 Average
Feed, g/L 252.90 54.49 172.04 218.56 170.20 173.64
magnetite
Concentrate, 1022.31 1138.93 513.35 743.59 602.22 804.08
g/L magnetite
Tailing, g/L 0.16 0.19 0.18 0.51 0.19 0.24
magnetite
Efficiency, % 99.954 99.675 99.929 99.836 99.921 99.86
Discard
Feed, g/L 113.17 122.17 527.84 519.90 508.48 358.31
magnetite
Concentrate, 1862.47 1542.24 1316.04 1061.40 1657.65 1487.96
g/L magnetite
Tailing, g/L 0.003 0.22 0.13 0.15 0.04 0.11
magnetite
Efficiency, % 99.998 99.837 99.986 99.985 99.994 99.96
Processing of coal finer than 100 µm results in poor separation efficiency;

therefore, if possible, this size fraction should be excluded from the feed to
fine coal dense-medium plants. In the case where conventional desliming of
the plant feed is employed and the fine coal routed to a stand-alone fine coal
circuit, sieve bends or Derrick screens are capable of removing the bulk of the
minus 100 µm coal from the feed. In wash-to-zero plants, the minus 100 µm
900
800
700
600
500
Counts
400
300
200
100
0
1 4 7 10 13 16 19 22 25 28 31 34 37 40 43 46 49 52 55 58
Size, micron
Figure 9 Size distribution of magnetite lost via product magnetic separator
700
600
500
400
Counts
300
200
100
0
1 4 7 10 13 16 19 22 25 28 31 34 37 40 43 46 49 52 55 58
Size, micron
Figure 10 Size distribution of magnetite lost via discard magnetic separator
material can be removed from the product after beneficiation. The fine coal
below 100 µm, being poorly separated and normally having an ash value higher
than that of the required final product, can degrade the quality of the final
product. Removal of the ultrafines from the product will therefore improve its
quality. This also applies in the case of other fine coal processes such as spirals
or teetered-bed separators.
2 μm Disc Tails EHT = 5.00 kV WD = 6 mm Mag = 5.00 KX
Figure 11 Magnetite in magnetic separator effluent
Table 11 Results obtained with 200-mm-diameter cyclone

Particle Size Fraction, mm RD of Separation EPM
3 × 1.5 1.33 0.024
1.5 × 0.64 1.36 0.024
0.64 × 0.24 1.41 0.034
0.24 × 0.150 1.44 0.038
0.15 × 0.100 1.53 0.064
0.100 × 0.075 1.57 0.103
Source: Deurbrouck 1974.
The presence of ultrafine coal in the feed to dense-medium cyclones had

very little influence on the separation of the coarser particles in the feed (King
and Juckes 1986). On the other hand, Deurbrouck (1974) suggests that the
presence of coarse particles in the feed to a dense-medium cyclone could nega-
tively influence the separation of smaller particles due to crowding—especially
in the apex area of a cyclone. This seems to be confirmed by the fact that better
separation is obtained in fines-only processing than in wash-to-zero circuits
where the coarse and fine coal are processed together.
One other consideration in favor of dense-medium processing of fine coal
is that any “stray” oversized particles in the feed will be correctly separated in a
dense-medium cyclone. In water-only processes, oversized particles end up in
the reject stream, irrespective of quality or density.
1.60
1.55
1.50
RD
1.45
1.40
1.35
1.30
0.000 0.500 1.000 1.500 2.000 2.500
Particle Size, mm
Figure 12 Increase in separation density vs. particle size
0.12
0.10
0.08
EPM
0.06
0.04
0.02
0.00
0.000 0.500 1.000 1.500 2.000 2.500
Particle Size, mm
Figure 13 Increase in EPM vs. particle size
CONCLUSION
Dense-medium processing of fine coal is still the most efficient method of fine
coal cleaning available. It has been successfully implemented at several plants
around the world in the past and is presently in use in South Africa and China.
Provided that the appropriate combination of cyclone geometry, magnetite

medium, and cyclone feed pressure is applied, very good separation efficiency
and accurate control over final product quality can be obtained when process-
ing fine coal in dense-medium cyclones.
REFERENCES
Burch, E. 1987. Island Creek’s feeding-to-zero concept simplifies coal prep circuit at
Providence plant. Min. Eng. 39:775–777.
Chedgy, D.G., Watters, L.A., and Higgens, S.T. 1986. Heavy medium cyclone separa-
tions at ultralow specific gravity. In 10th International Coal Preparation Congress,
Proceedings. Edmonton, Canada. pp. 60–79.
CSIR. 1980. Fuel Research of South Africa 1930–1980. CSIR Brochure B65.
de Korte, G.J. 2002. Dense-Medium Beneficiation of Fine Coal: Coaltech 2020 Pilot
Plant. CSIR Report no. 2002-0189. March.
de Korte, G.J. 2007. Dense medium cyclone tests conducted with coarse magnetite.
Presented at South Africa Coal Processing Society International Coal Conference:
Coal Processing in a Changing World, Johannesburg, South Africa, September.
Session 4, Paper 2.
Deurbrouck, A.W. 1974. Washing Fine-Size Coal in a Dense-Medium Cyclone. USBM
Report of Investigations 7982.
DSM (Dutch State Mines). 1970. The Heavy Medium Cyclone Washery for Minerals
and Coal. DSM.
Esposito, N.T., and Higgens, S.T. 1982. Modification of the fine coal circuit at Homer
City coal preparation plant. Paper presented at the SME-AIME Annual Meeting,
Dallas, TX, February 14–18.
Fourie, P.J.F., van der Walt, P.J., and Falcon, L.M. 1980. The beneficiation of minus
0.5 mm coal by dense medium cyclone. J. S. Afr. Inst. Min. Metall. (October):
357–361.
Horsfall, D.W. 1976. The treatment of fine coal: Upgrading –0.5 mm coal to obtain a
low-ash product. ChemSA ( July): 124–129.
Kempnich, R.J., van Barneveld, S., and Lusan, A. 1993. Dense medium cyclones on
fine coal—the Australian experience. In 6th Australian Coal Preparation Congress,
Proceedings. Paper E1.
King, R.P., and Juckes, A.H. 1986. Beneficiation of Fine Coal in a Dense Medium
Cyclone: The Effect of Slimes in the Feed. Report CSP Coal-6. University of the
Witwatersrand, Johannesburg, South Africa, January.
Lathioor, R.A., and Osborne, D.G. 1984. Dense medium cyclone cleaning of fine coal.
In 2nd International Conference on Hydrocyclones, Proceedings. BHRA The Fluid
Engineering Centre, Bath, England. Paper G1.
McGonigal, S., and de Korte, G.J. 2005. Dense-Medium Beneficiation of Fine Coal:
Coaltech 2020 Pilot Plant at Navigation, Landau Colliery. CSIR Report
2005-0566.
Mengelers, J. 1982. Report on Tests Carried Out at Stamicarbon with Coal 0–1.0 mm,
Treated in a Heavy Medium Cyclone of Diameter 250 mm. Stamicarbon BV Report
4950/Stac/95-1. March.
Mengelers, J., and Absil, J.H. 1976. Cleaning coal to zero in heavy medium cyclones.
Coal Min. Process. (May): 62–64.
Sehgal, R., Matoney, J.P., and Esposito, N.T. 1982. Innovative heavy media fine coal
cleaning: Homer City coal cleaning plant. In Ninth International Coal Preparation
Congress, Proceedings. New Delhi, India. Paper 143. C1–C15.
Sokaski, M., and Geer, M.R. 1963. Cleaning Unsized Fine Coal in a Dense-Medium
Cyclone plant. USBM Report of Investigations 6274.
Taylor, B.S., and Chen, W.L. 1980. Heavy medium cyclone cleaning of 1½" × 0 raw
coal. Paper presented at the 1980 American Mining Congress International Coal
Show, Chicago, IL, May 5–8.
Van der Walt, P.J. 1949. Operating characteristics of the cyclone washer. In Fuel
Research of South Africa Report No. 17.
Zhao, S., and Yu, J. 2012. Novel efficient and simplified coal preparation process. Inter-
national Coal Prep 2012, Lexington, KY, April 30–May 3. Paper 9.
Development of the
Reflux Classifier
K.P. Galvin
ABSTRACT
The reflux classifier is a new gravity separation technology for processing fine coal
and other dense minerals, developed through collaboration between the University
of Newcastle and Ludowici Australia. This chapter provides an account of that
development, covering a number of the key findings that have led to the final
design, and indeed the separation performance achieved. The chapter also outlines
the theoretical basis used to quantify the capacity of the separator. The technology
embodies the benefits of a high shear rate similar to that observed for spirals, the
benefits of an autogenous dense medium similar to that observed in a teetered-bed
separator, and an increased throughput advantage similar to that produced by
a lamellae thickener, with a strong synergy achieved among all three. Moreover,
the laminar shear mechanism that arises within closely spaced inclined channels
permits a powerful separation on the basis of density, delivering a significant per-
formance advantage at the low cut points required for the recovery of metallurgical
coal.
INTRODUCTION
The reflux classifier, shown in Figure 1, consists of a fluidized bed, with a sys-
tem of parallel inclined channels above. Feed enters the system and distributes
into an overflow and underflow stream in accordance with a mechanism that
amplifies the role of particle density by suppressing the effects of particle size.
Fluidization through the base is used to establish a suspension of higher-density
particles that acts as a dense medium. Lower-density particles thus transport
up through the system of inclined channels. Fine, dense particles entrained
into the inclined channels segregate from the flow and onto the inclined sur-
faces, joining the lower bed. When the density of the lower bed exceeds the set
point value, an underflow valve opens, discharging the high-density particles.
159
Feed
Overflow
Fluidization
Underflow
Figure 1 Schematic representation of the reflux classifier showing the lower

fluidized bed housing and system of parallel inclined channels above
Meanwhile. the lower-density particles emerge with the overflow. The physics
that underpins this separation is outlined in this chapter.
The reflux classifier, which followed a period of more than 10 years of col-
laborative research and development by the University of Newcastle (Australia)
and Ludowici Australia, is delivering significant benefits to the coal industry,
especially in the beneficiation of fine metallurgical coal. Similar performance
has been evident in the beneficiation of dense minerals (e.g., chromate, iron
ore, and mineral sands). This chapter provides a unique historical perspec-
tive behind the development of the technology, the rationale for the physical
arrangement, and the fundamental theoretical basis that governs the particle
separation. This chapter also outlines for the first time the theoretical basis used
to assess potential applications and in turn to predict solids throughput.
Consider the batch settling of fine particles suspended in a long tube. The
upper interface is observed to move downward at a steady rate. A slight tilt of
the tube, however, leads to a dramatic increase in the rate of collection of par-
ticles at the base of the vessel, with particles depositing first onto the inclined
Development of the Reflux Classifier 161
surface, leading to the rapid sliding of the sediment toward the base, coupled
with the upward movement of the return fluid. Boycott (1920) was the first to
report about the effect of inclining a suspension, whereas Ponder (1925) and
Nakamura and Kuroda (1937) provided the first theoretical explanation, the
so-called PNK kinematic model. Other work led to the development of the
first industrial application, the lamellae thickener (see for example, Yao 1973).
Others (Carty 1957; Acrivos and Herbolzheimer 1979; Leighton and Acrivos
1986; MacTaggart et al. 1988; Davis et al. 1989; Zhang and Davis 1990; Smart
et al. 1993; Galvin et al. 2001) exploited the physical arrangement to explore
additional applications and some truly fundamental questions concerned with
the way particles interact with solid surfaces.
My first observation of the remarkable effects of inclined settling was
made during a visit to a laboratory in Houston, Texas, in 1992, concerned with
directional drilling in the oil industry. Later, a simple experimental system was
established consisting of a 25-mm-diameter tube, 1 m long inclined at 45° to
the horizontal, with feed entering near the base, underflow withdrawn from
the sediment near the base, and the balance of the flow reporting through the
top of the device as overflow (Thompson and Galvin 1997). The initial work
was concerned with the classification of fine coal on the basis of particle size,
whereas later work examined the potential to separate on density. The scale-up
of this single tube arrangement to meet the requirements of an industrial sepa-
rator were unclear.
At this time, teetered-bed separators (TBSs) were being introduced to the
Australian coal industry. The introduction of this technology was motivated
by the increasing need to overcome the limitations of spirals, that is, the need
to lower the separation density, and in turn recover the more valuable metal-
lurgical coal. I subsequently became involved in a study on TBSs, led by Stuart
Nicol. This work (Galvin et al. 1999) was funded by the Australian Coal
Association Research Program (ACARP), with direct support from Rio Tinto.
The benefit of an autogenous dense medium in promoting relatively large low-
density particles to segregate upward from the bed was in evidence, while the
considerable entrainment of fine dense particles into the overflow when the
feed pulp density was reduced was also apparent.
Following the completion of these two projects, there was a need to con-
ceive of a method to exploit the benefits of inclined settling. The problem
identified in scale-up of the inclined tube separator was solved by combining
the TBS arrangement with a system of parallel inclined channels. Though simi-
lar to a lamellae separator, this system incorporated the delivery of fluidization
water through the base of the vessel, allowing a fluidized bed to be established,
with a mixing zone permitted to form in the zone between the top of the bed
and below the inclined plates. At this time, the arrangement was generalized to
cover multiple sets of inclined channels separated by mixing zones, as described
in the patent, a reflux classifier (Galvin 2001).
Intuitively, this basic arrangement made a great deal of sense, with a sig-
nificant increase in the effective sedimentation area generated by the presence
of the inclined plates. There should also be significant interaction between the
sediment sliding down the inclined surfaces and the upward fluidization of the
suspension from below.
RESEARCH PROGRAM
A Research and Development Agreement was established between the Uni-
versity of Newcastle and Ludowici Australia to develop this new technology,
referred to as a reflux classifier. This collaboration first developed through
Maurie Munro of Ludowici, and later through Taavi Orupold of Ludowici. The
research consisted of two main parts, a pilot-and-full-scale study at the nearby
Bloomfield coal preparation plant and work of a more fundamental nature at
the University of Newcastle. Funding was secured from ACARP and the Aus-
tralian Research Council.
Pilot-Scale Research
There was confidence that the pilot-scale study would lead to early success,
given the simplicity of the concept, coupled with our past experience and
hence understanding of the TBS. However, the commencement of the pilot-
scale research proved to be premature, with a series of design flaws evident
early on, requiring the reengineering of the whole system. Many problems were
encountered during the first 6 months of the study with the control of the feed
delivery, feed entry, and PID (proportional–integral–derivative) control of the
underflow discharge. The feed was also highly variable with high clay content
during the early period.
The selection of an inclined channel angle of θ = 60°, spacing of z = 60 mm,
and length of L = 1,000 mm was based on the objective of delivering a signifi-
cant solids throughput advantage over the TBS, while ensuring that the system
internal design was free-flowing and economic. The PNK model predicted a
throughput advantage of order Lcosθ/z~10. It was evident from the research,
however, that the throughput advantage was in the range of three- to fourfold.
The final work was published in Minerals Engineering (Galvin et al. 2002),
with partition curves presented for narrow particle size fractions over the range
0.25 to 2.0 mm. The composite probable error (Ep) reported for this overall
size range was 0.14. With a solids throughput of 47 t/m2h, the work, in the
end, proved to be sufficiently successful for a larger-scale trial to be undertaken.
Bed Expansion in a Fluidized Bed

The homogeneous expansion of suspensions in fluidized beds is well known
and understood via the Richardson and Zaki (1954) empirical relationship
given by
U = U t (1 − φ )
n
(EQ 1)
where U is the superficial fluidization velocity, Ut is the particle terminal veloc-

ity for a monocomponent system, φ is the volume fraction of solids, and n is
an exponent. Figure 2 shows the monotonic reduction in the volume fraction
of the solids versus the fluidization velocity, approaching a value of zero when
U = Ut.
Galvin and Nguyentranlam (2002) provided the first account of the
effects of inclined surfaces on particle segregation in a fluidized bed, demon-
strating the result in a series of photographs, with the elevation of the system
of plates gradually lowered down through the fluidized bed. The fluidized
suspension appeared to “compress” when the plates were lowered. Doroodchi
et al. (2004), in a more precise set of experiments, quantified the effects of an
inclined channel on the bed expansion, using a channel spacing equal to that
used in the earlier pilot-scale study. Figure 2 shows a sharp departure from the
conventional curve for different bed inventories when the bed expanded into a
single inclined channel. It is evident that a substantial bed concentration can be
maintained even when the fluidization velocity, U, exceeds the terminal veloc-
ity of the particles, Ut. The inclined channel breaks the standard one-to-one
relationship between the suspension concentration and the superficial velocity,
providing an extra degree of freedom.
Full-Scale Trial of the Reflux Classifier

Following the success of the pilot-scale study, and with the knowledge gener-
ated from that work, a full-scale reflux classifier, the RC1800, was installed at
the Bloomfield plant, as shown in Figure 3. There were some initial problems
experienced with the design, requiring some further reengineering. In particu-
lar, a significant scale-up problem emerged. The particles reporting to the over-
flow not only rise vertically, they must also convey in the horizontal direction.
With a full-scale device, there is a significant increase in the horizontal distance
to the overflow weir; hence, particles were observed to segregate from the
0.4
6.0 kg
Solids Volume Fraction
0.3 Expansion Into

Inclined Channels
4.0 kg
0.2
2.0 kg
0.1
Conventional Fluidization
Expansion
0.0
0.0 0.01 0.02 ut 0.03 0.04
Fluidization Velocity, m/s
Figure 2 Homogeneous bed expansion showing conventional behavior and three

cases of expansion into inclined channels. Each case involved a different inventory
of silica particles (in kilograms).
suspension, producing sediment over portions of the vessel. Laboratory experi-

ments also showed that there was some flow communication between adjoining
channels, which can lead to elevated flow in one channel, compensated by lower
or downward flow in the next.
A new overflow launder design was therefore developed, the aim being
to (1) minimize the horizontal distance the particles needed to travel and
(2) ensure that the flow through each of the inclined channels was virtually
the same, thus removing the flow communication between the channels. This
work led to a second key patent on the reflux classifier, the overflow launder
(Galvin and Munro 2004). A series of internal launders were introduced to
intersect with the inclined plates. Improvement in the system hydrodynamics
Source: © University of Newcastle, Australia. Used with permission.
Figure 3 Original installation of the first RC1800 at the Bloomfield plant in 2003
and transport of the overflow coal was immediate, with uniform flow achieved
through each of the channels.
With the introduction of the new launder design, the channel spacing was
in fact increased to 120 mm by removing every second plate. This decision was
based on the discrepancy observed in the earlier pilot-scale work where the
throughput advantage of 3.5 was much lower than the kinematic theoretical
value of 10. This result suggested that half the inclined plates were redundant.
The available feed, at this time, was reduced to only 60 t/h; hence, it was not
possible to properly assess the effect of this change. However, several years later,
the additional plates were returned to the unit.
Other major hurdles were encountered during the full-scale trial work,
notably the effects of wear on the fluidization nozzles and significant effects
due to wear on the underflow valve. There were also problems encountered due
to the presence of grit in the fluidization water, which on occasions caused the
blockage of some of the nozzles, most notably those just below the pressure
transducers. The bed would lose fluidization in this location, which would
undermine the control of the whole system. A novel device for removing this
grit solved the problem.
Finally, in late 2004, all of these problems were solved, and the plant pur-
chased the RC1800 and incorporated the device into their circuit on a perma-
nent basis. The composite Ep reported for the overall size range from 0.25 to
2.0 mm was 0.15 (Galvin et al. 2005). The unit continues to operate to this day.
Semi-Batch Elutriation Experiments

From as early as 2002, the objective was to induce an inertial lift force to
selectively convey low-density particles to the overflow. A new laboratory-scale
reflux classifier, with dimensions 60 mm × 100 mm internal cross section, was
established in 2004. Several inclined sections, with different angles of inclina-
tion, were used, leading in the end to a preferred inclined channel angle of 70°.
This angle was adopted primarily because of the interest in promoting inertial
lift. The laboratory-scale design also allowed the inclined channel spacing to
be reduced appreciably to promote inertial lift. King and Leighton (1997)
had previously reported on the inertial lift of a moving sphere in a shear field,
developing a theoretical model and validating the work using a Couette device.
Their inertial lift criterion can be expressed as
Re 2s
> 32 (EQ 2)
Retn
where Res is the shear Reynolds number and Retn is the sedimentation Reynolds
number in the normal direction of the plane. This result has been expressed
using Reynolds numbers based on the particle diameter and, hence, has been
altered accordingly from the original form. Early work, which typically involved
inclined channels with a spacing of z = 60 mm, demonstrated no evidence or
even remote prospect for inertial lift. In fact, it was concluded that the inclined
channel spacing would need to be in the vicinity of a few millimeters and,
hence, concluded that the lift force criterion had little relevance to this work.
A comprehensive investigation of semi-batch elutriation from the new
laboratory system commenced, focused on the effect of the inclined channel
spacing. The PNK model had failed to explain the throughput advantage in the
earlier pilot-scale study. In this new work, a given system of particles was elutri-
ated at a specific superficial velocity. The number of equally spaced plates was
increased and the separation size determined. The separation size was found to
decrease. The number of equally spaced plates was increased again, producing
a further reduction in separation size. Eventually the separation size reached
a minimum before eventually increasing. A mechanism of shear-induced par-

ticle resuspension was thought to be responsible, favoring the lower-density
particles. Dimensional analysis was applied to the data set leading to a reliable
empirical relationship for describing the efficiency of the segregation given by
(Laskovski et al. 2006)
1
η= (EQ 3)
1 + 0.133cos θ Re0.33
t ( L / z)
where θ is the angle of inclination with the horizontal, Ret is the particle Reyn-
olds number, and L/z is the channel aspect ratio. The term segregation efficiency
should not be aligned with the efficiency of the separation. Low segregation
efficiency implies that the particle has a strong tendency to convey, whereas,
conversely, higher segregation efficiency is consistent with a tendency to seg-
regate onto the inclined surface and slide downward. In fact, lower-density
particles tend to have lower segregation efficiency, which in turn promotes their
separation from the higher-density particles.
Using the approach of Ponder (1925) and Nakumara and Kuroda (1937),
the effective vessel area increases by the factor
 L
F = 1 +   cos θ sin θ (EQ 4)
 z
Laskowski et al. (2006) combined Equations 3 and 4 to obtain the actual

throughput advantage given by
U / U t = ηF (EQ 5)
Zhou et al. (2006) showed that the maximum throughput advantage is

reached at θ = 70°. Equation 5 is applied in the following section to establish
the throughput achievable for a given application in an RC2020.
The laboratory system influenced the design of the commercial systems,
leading to a preferred channel spacing of 25 mm. Pilot-scale systems were also
constructed with a channel spacing of 12 mm, and longer channels with L =
2 m were also introduced. The design was clearly evolving toward more closely
spaced inclined channels.
In 2008, a semi-batch elutriation experiment was conducted using a feed
of silica sand with a channel spacing of z = 1.77 mm. At the time, it was antici-
pated that the results would either conform to the theory of Laskovski et al.
(2006) or lead to the inertial lift that had been sought since 2002. The results
were extraordinary, with a constant partition number obtained for particles
of different size ranging from about 0.1 to 0.4 mm. The partition curve, based
on particle size, was horizontal. Moreover, the result was repeated for particles
of different density, with a very different superficial velocity required in each
case (Galvin et al. 2009). Continuous steady-state experiments were then con-
ducted on metallurgical coal feeds (Galvin et al. 2010) using a wider channel of
z = 4.2 mm, with the composite Ep = 0.06 over the particle size range 0.25 to
2.0 mm. It is emphasized that this work was conducted at a relatively low solids
throughput of about 12 t/m2·h, with satisfactory performance achieved down
to a lower particle size of 0.075 mm. Data were also presented at significantly
higher solids throughput, with satisfactory performance maintained down to a
particle size of 0.25 mm. Figure 4 shows the results obtained at two cut points
and a comparison with the earlier work (Galvin et al. 2004). Figure 4 shows
the variation in the relative density of separation (d50) with particle size was
significantly lower than previously possible. Hence, there is a clear step change
in separation performance. Figure 5 shows that Ep values remained remarkably
low across a broad particle size range. The Ep values are much lower across the
size range than previously reported (Galvin et al. 2004). There was a clear step
change of improvement in separation performance.
When closely spaced inclined channels are used, the flow Reynolds number
decreases, resulting in the formation of a laminar flow profile, with a high shear
rate near the planar surfaces. The laminar flow profile produces a strong, and
very precise, variation in the local fluid velocity with the normal distance from
the wall. Thus, relatively fine particles, which migrate toward the upward fac-
ing planar surfaces during their transport through the channels, are ultimately
exposed to relatively low fluid velocities, while coarser particles are exposed to
proportionally larger fluid velocities. The velocity gradient at the wall delivers a
fluidization velocity to the particles that is proportional to the diameter of the
particle. This in turn reduces the effective dependence of the particle settling
velocity on the particle diameter by a full decade. Applied to particles that settle
within the intermediate settling regime, this mechanism effectively removes the
dependence of settling velocity on particle size. Fine, dense particles that reside
on the upward-facing inclined surfaces are exposed to low local fluid velocities
and therefore slide downward. Fine low-density particles are, of course, readily
conveyed, while much larger low-density particles experience a strong inertial
lift force that exposes such particles to even higher local fluid velocities, causing
them to lift and convey.
2.4
Galvin et al. 2010

Galvin et al. 2004
2.2
2.0
Relative Density, d50
1.8
1.6
1.4
1.2
0.1 1.0 10.0
Particle Size, mm
Figure 4 d50 vs. particle size at two cut points as reported by Galvin et al. (2010) and
compared to that previously reported (Galvin et al. 2004)
Galvin et al. 2010

Galvin et al. 2004
0.10
Ep
0.05
0.00
0.1 1.0 10.0
Particle Size, mm
Figure 5 Ep vs. particle size at two cut points as reported by Galvin et al. (2010) and
compared to that previously reported (Galvin et al. 2004)
The research findings were brought to the attention of Ludowici with a

view to developing an entirely new reflux classifier design, the RC2020. A deci-
sion was made to adopt a slightly wider channel spacing of 6 mm, while ensuring
laminar flow. When transporting particles through pipes or valves, it is common
practice to adopt a general rule that the maximum particle diameter be less than
one-third the pipe diameter. This rule is applied to prevent blockages. However,
within inclined channels, this rule does not apply, because the channel width
is many times larger than the channel gap. Indeed, particles with a diameter
close to the size of the channel gap do not block the channels. However, the
separation performance will tend to decline if particles of diameter larger than
about one-third the channel gap are used, given that the corresponding velocity
gradient will tend to decline. Particles with a minimum dimension that is larger
than the channel gap will fail to enter the incline and, if low in density, will tend
to accumulate, eventually discharging via the underflow. Oversize particles with
an unusual shape could in principle become wedged within the channels. So,
clearly, oversize protection via well-maintained screens is desirable.
There was significant interest in the new RC2020, with the first commer-
cial sale in June 2009, and sales into seven countries during the first year. The
Australian company Sedgman conducted a full-scale trial of the technology
in late 2009. Figure 6 shows the full-scale unit being installed at a plant in the
Bowen Basin in Queensland, Australia. The Ep obtained over the size range
from 0.25 to 2.0 mm was 0.07, while the variation in the D50 with particle size
was very similar to the laboratory data. In another Bowen Basin installation,
there was evidence of uneven overflow. This uneven flow was eliminated by
removing air entrainment with the feed.
Low cut points provide a means for recovering a metallurgical product.
However, it is generally accepted that a low Ep is also essential when target-
ing low cut points. Figure 7 provides a good illustration of the effect of Ep on
performance at low cut points. For illustrative purposes, the extreme case of an
even washability distribution of coal and mineral matter across the relative den-
sity range 1.25 to 2.20 is assumed. The particle size range is assumed to be 0.25
to 2.0 mm. The goal is a product with an average relative density of 1.4. With
an Ep = 0.14, the cut point is forced to drop to only 1.3, leading to low yield
values for the lowest density particles, with the overall product yield only 12%
to achieve the target product density. Figure 8 shows the actual yield relative to
the maximum possible yield (for a perfect separator) as a function of the relative
density of the product. At Ep = 0.14, it is impossible to generate any product
with an overall relative density less than 1.37.
Source: © Ludowici Australia. Used with permission.
Figure 6 Installation of the first RC2020 by Sedgman in the Bowen Basin in

Queensland, Australia
Conversely, Figure 7 shows that for an Ep of 0.06, the yield of the lowest
density particles is nearly 100%; hence, a higher cut point of 1.53 can be used.
The target product, with an average relative density of 1.4, is then achieved
at a much higher yield of 30%. Figure 8 shows the improved performance in
terms of the actual yield relative to the maximum possible yield (for a perfect
separator) at an Ep of 0.06. Evidently, a product relative density as low as 1.3
can be obtained. Although a feed washability with an even density distribution
is extreme, this distribution is relevant to more poorly liberated coals as found
in India. With more typical coals, the benefits remain significant with a much
higher yield achieved via a low Ep.
Galvin and Liu (2011) developed a theoretical elutriation model of the
laminar-shear mechanism incorporating the inertial lift force model of King
and Leighton (1997). Under laminar flow, the particles at the upward-facing
surface of the inclined channel experience a local fluid velocity much lower
than the superficial velocity. The throughput advantage, based on this mecha-
nism, is approximately
1.00
0.75
Partition Number
D50 1.53
Ep 0.06
0.50
RD product 1.40
Y 30%
0.25
D50 1.3
Ep 0.14
RD product 1.40
Y 12%
0.00
1.0 1.2 1.4 1.6 1.8 2.0 2.2
Relative Density
Figure 7 Representation of partition curves showing the benefit of a low Ep at low

cut points
100
Ep = 0.06
75
Ep = 0.14
Relative Yield, %
50
25
0
1.2 1.3 1.4 1.5 1.6 1.7 1.8
Relative Density of Product
Figure 8 Relative yield vs. relative density of the product for Ep values of 0.06 and 0.14
U z
= (EQ 6)
U t 3d
where d is the diameter of the target particle and Ut is the corresponding termi-
nal velocity. Equation 6 applies to relatively fine particles that fail to experience
inertial lift. For larger particles, the shear rate leads to the lift force as
ρ1.8 γ 2.8 d 5.6

L f = 0.0567 0.8
= 0.0567 Re0.8
s
ργ 2 d 4 (EQ 7)
µ
Figure 9 shows the theoretical elutriation achieved via the laminar shear
mechanism (Galvin and Liu 2011) for the original channel spacing of 1.77 mm,
in particular the strong potential to separate particles on the basis of density.
The original paper includes experimental data and hence reports the full valida-
tion of the theory, requiring no adjustable parameters. Figure 10 shows the lift
force fraction versus the channel spacing for a particle of density 2,600 kg/m3
and diameter of 0.310 mm in water subject to a superficial velocity of 0.05 and
0.1 m/s. The lift force fraction is the ratio of the buoyant weight of the particle
in the normal direction to the planar surface relative to the shear-induced lift
force. Clearly, the lift force fraction is negligible when the channel spacing is
large and does not reach a value of 1.0 until the channel spacing is relatively
small. It is evident that closely spaced channels are essential for producing suf-
ficient lift.
A P P L I C AT I O N O F T H E O R Y
This section is used to illustrate the application of the theory to the design of
the reflux classifier.
Design Calculations
The commercial reflux classifier is the RC2020, which has a diameter of 2.0 m
and, hence, cylindrical cross-sectional area of 3.14 m2. The solids feed flux is
always quoted relative to the value of this lower vessel area. The upper lamellae
section is constructed across a square section 2.0 m × 2.0 m = 4.0 m2, with an
overflow launder external to this section. The square system consists of rela-
tively large internal launders and 18 generic boxes of inclined plates, each of
internal horizontal dimensions of 0.653 m × 0.212 m = 0.138 m2. The inclined
plates are 0.212 m wide and 1.5 mm thick. In total, the horizontal cross-
sectional area of the boxes is 18 × 0.138 m2 = 2.49 m2; hence, the combined
0.5
1,400 kgm–3 2,600 kgm–3 4,450 kgm–3
0.4
Particle Diameter, mm
0.3
0.2
0.1
0.0
0.00 0.02 0.04 0.06 0.08 0.10 0.12
Superficial Channel Velocity, m/s
Figure 9 Theoretical prediction of elutriation achieved via the laminar shear

mechanism, showing clear separation of particles on the basis of density
1.0000
0.1000
0.1 m/s
Lift Force Fraction
0.0100 ρs = 2,600 kg/m3

d = 0.310 mm
0.0010
0.05 m/s
0.0001
0.1 1.0 10.0 100.0
Channel Spacing, mm
Figure 10 Lift force fraction vs. channel spacing for a particle of density 2,600 kg/m3
and diameter of 0.310 mm, subject to a superficial velocity of 0.05 and 0.1 m/s
area of the internal launders and internal walls is 4.0 – 2.49 = 1.51 m2. Thus,
the active box area relative to the area of the lower cylindrical cross section is
2.49/3.14 = 0.79.
The inclined channels are 1 m long, inclined at 70° to the horizontal, with
a 6-mm perpendicular spacing. For the generic application where the objective
is to achieve gravity separation over the particle diameter range 2 mm × 0.25
mm, the units are rated at 100 t/h of solids nominal and 120 t/h maximum.
The capacity of the unit, however, depends on the nature of the feed. For
example, a feed with very low density coal and little mid-range density mate-
rial can be processed at much higher rates than a feed with significant mid-
range density material. Thus, the so-called maximum rating may well be easily
achieved for a feed with little mid-range density material. But, of course, the
industry seeks nominal values, so in summary, a value of 100 t/h is stated for
this application.
Another key design factor is the yield, which dictates the relative solid
flows to overflow and underflow. At a low yield, relatively little material reports
to the overflow, and hence in principle the system is not constrained by the
capacity of the inclined channels. This suggests that the system could operate at
a higher feed solids throughput. However, this is a false assumption. The higher
solids rate to underflow results in a higher water split to underflow and, hence,
increasingly higher portions of particles become misplaced to the underflow.
Thus, a constraint is imposed requiring the water split to the underflow to be
10% or less. This constraint in turn limits the feed solids throughput to realistic
levels, and imposes the need to consider lower feed pulp density. In the follow-
ing section, this result is achieved by setting a lower value for the volume frac-
tion of solids in the overflow.
The design capacity is governed by the smallest dense particle to be
retained below the system of inclined channels. This particle is required to
ultimately report to the underflow. In the example, the target is a 0.25-mm-
diameter particle of density 1,900 kg/m3. The upper particle size requires more
judgment or direct knowledge of past performance, with a maximum diam-
eter ratio of eight times the lower target diameter considered appropriate. A
more conservative limit of fourfold could be equally applied. So the top size is
approximately 2 mm. The autogenous dense medium needs to prevent the low
density 2-mm-diameter particles from reporting to the underflow. By ensuring
that the right hydraulic flows into the inclined channels, there is also a strong
tendency for large low-density particles to convey to the overflow. The theoreti-
cal work associated with the laminar shear mechanism is very much concerned
with the mechanism that underpins this hydraulic conveying. However, when
assessing the capacity of the reflux classifier in a given application, the primary
consideration is the smallest dense particles to be retained below the system
of inclined channels. This assessment, as described in the following section, is
based on Equations 3, 4, and 5.
Example Calculations
Consider an upper section of parallel inclined channels covering an open box
area of 0.653 m × 0.212 m. Each plate has a thickness of 1.5 mm, width of
0.212 m, active length to the overflow weir of 1.0 m, and angle of inclination
to the horizontal of 70°. The target particle has a diameter of 0.25 mm and a
density of 1,900 kg/m3. The design is based around preventing this particle
from reaching the overflow. This condition is established using the theory of
Laskovski et al. (2006). The choice of the density of 1,900 kg/m3 is based on
the anticipated D50 of the partition curve applicable to the target particle size.
The terminal velocity of the target particle of size d = 0.25 mm and density
1,900 kg/m3 settling in water is obtained using the Zigrang and Sylvester equa-
tion (1981) given by
2
 d 1.5 
0.5

(
Ret  14.51 + g ( ρs − ρ) ρ )
0.5
1.83 − 3.81 (EQ 8)
 µ  
 
where Ret is the particle Reynolds number, d is the particle diameter, ρs is the
particle density, ρ = 1,000 kg/m3 is the fluid density, μ = 0.001 Ns/m2 is the
fluid viscosity, and g = 9.8 m/s2 is the acceleration due to gravity. The particle
Reynolds number is
2
 0.5

( (
Ret =  14.51 + 9.8 1,900 − 1,000 1,000 ) ) 1.83 × 0.000251.5 / 0.001
0.5
− 3.81 = 4.79
 
 
ρU t d
By definition, Ret = , and hence the terminal velocity is
µ
U t = ( 4.79 × 0.001) / (1,000 × 0.00025) = 0.0192 m/s
For 82 channels across the box width of 0.653 m, with plate thickness
1.5 mm, the perpendicular channel spacing, z = 6.0 mm. That is,
z = 0.653* sin ( 70° ) − (82 − 1) * 0.0015 / 82 = 0.0060 m = 6.0 mm
The channel aspect ratio is L/z = 1.0/0.006 = 166.19. Thus, the segregation
efficiency, η, is
1 1
η= = = 0.0731
0.33
1 + 0.133cos θ Ret L / z ( )
1 + 0.133cos 70° 4.790.33 166.19 ( ) ( )
The effective vessel area factor is
 L
F = 1 +   cos θ sin θ = 1 + 166.19cos ( 70° ) sin ( 70° ) = 54.41
 z
The throughput factor for a given inclined channel is
U / U t = ηF = 0.0731 × 54.41 = 3.98
Notably, this value is consistent with that achieved in practice (Galvin

et al. 2002) and is significantly lower in value to that of a 10-fold advantage
referred to earlier using the kinematic PNK theory. Laskovski et al. (2006)
showed that the throughput factor of an inclined channel asymptotes to a value
of 7.5Ret–0.33 in the limit as the channel aspect ratio, L/z, approaches infinity.
Corrected for the angle of inclination using sin(70°), the asymptotic value
becomes sin(70°) × 7.5(4.79)–0.33 = 4.18.
The throughput factor needs to be corrected for the space occupied by the
plates. The area correction is
α = 0.653 − (82 − 1) * 0.0015 / sin ( 70° )  / 0.653 = 0.80
The area occupied by the boxes relative to that of the lower cylindrical cross sec-
tion is f = 0.79. Thus, the open area available for the inclined channels is 0.79
× 0.80 × 3.14 m2 = 2.0 m2.
Therefore, the actual throughput factor = fαU/Ut = 0.79 × 0.80 × 3.98
= 2.53. Hence, this system should process at a solids feed rate 2.5 times that
of a TBS having the same lower cross-sectional area. Note that when finer
particles are targeted, this throughput advantage will increase markedly due
to the particle Reynolds number in Equation 3. For example, if the relevant
particle Reynolds number is 0.1, the asymptotic throughput factor is sin(70°)

× 7.5(0.1)–0.33 = 15.2.
Now the upward hydraulic velocity needs to be determined. This velocity
will be applicable to dilute conditions in the absence of hindered settling.
V = f α (U / U t ) × U t = 2.53 × 0.0192 = 0.0486 m/s
The hindered settling that exists within the inclined channels is a complex
question, outside the scope of this chapter and the simple design approach
outlined here. It is assumed that the volume fraction of the suspended par-
ticles within the inclined channels is the same as the level in the overflow that
emerges. This assumption is reasonable for a system operating close to maxi-
mum capacity.
For systems incorporating particles of different density, the level of hin-
dered settling is best determined using the equation first proposed by Asif
(1997) given by
n
 ρ − ρsusp 
h= p  (EQ 9)
 ρp − ρ f 
where n is 4.5. This exponent should arguably be smaller in value when these
larger particles are involved; however, in practice, a significant fraction of the
feed particles are finer than the size of the target particle, especially when slimes
are present. The effect of slimes is best assessed by introducing a modified fluid
density and viscosity into the calculation of the particle Reynolds number
and hence terminal velocity. Additional hindered settling is then applied via
Equation 9 based on the suspension density produced by the slimes and by the
nonslimes particles.
The hindered-settling factor will be lower for the mineral matter than for
the coal. This fact explains the autogenous dense-medium effect responsible for
preventing low-density coal from reporting to the underflow and also underpins
the inversion phenomenon of fluidized beds (Di Felice 1995). Here ρs is the
density of the target particle, 1,900 kg/m3, ρsusp is the suspension density of the
suspended solids in the inclined channels, and ρf is the density of water. In these
calculations the volume fraction of solids suspended in the inclined channels
is specified as φ = 0.2. Although a higher concentration is possible, this level
is regarded as boarding on the maximum. Assuming the solids transported up
through the inclined channels have an average solids density ρs = 1,400 kg/m3 for
a coal separation, then the suspension density is
ρsusp = φρs + (1 − φ ) ρ = 0.2 × 1,400 + 0.8 × 1,000 = 1,080 kg/m 3
Therefore, the hindered-settling factor applied to the inclined channels for the
target particle is
4.5
 1900 − 1080 
h= = 0.66
 1900 − 1000 
Conversely, the hindered-settling factor applied in the inclined channels

for the coal particles is
4.5
 1400 − 1080 
h= = 0.37
 1400 − 1000 
Thus, the hydraulic velocity applicable to the target particle, corrected for hin-
dered settling, is
V ' = 0.0486 × 0.66 = 0.0321 m/s
This is the highest velocity permitted up through the vertical section of the
reflux classifier, just ahead of the inclined zone. Allowing for the vessel geom-
etry, the volumetric overflow rate can now be determined. The reference vessel
area used here is A = 3.14 m2. Thus the volumetric overflow rate is
Q = V ' A = 0.0321 × 3.14 = 0.101 m 3/s = 363 m 3/h
It is assumed that the volume fraction of the solids in the overflow is equal
to that of the suspended particles within the inclined channels, φ = 0.2. This
is clearly not true for a system that is underloaded; however, it is a reasonable
assumption for a system operating at a throughput close to its maximum. Thus,
• The solids concentration of the overflow is C = φρs = 0.2 × 1,400 =
280 kg/m3
• The solids rate to the overflow is MO = QC = 363 × 280/1,000 =
101.5 t/h
• The water concentration of the overflow is C = (1 – φ)ρs = 0.8 × 1,000
= 800 kg/m3
• The water rate to the overflow is WO = 363 × 800/1,000 = 290.2 t/h
Assuming the proportion of the solids to the overflow (yield) is y = 0.85,

the feed rate of solids is MF = Mo/y = 101.5/0.85 = 119.5 t/h. Therefore the
rate of solids to the underflow is MU = MF – MO = 119.5 – 101.5 = 17.9 t/h.
The volume fraction of solids in the underflow is assumed to be φ = 0.45.
Assuming the average density of the solids equals the target density, 1 m3 of
underflow equals 0.45 m3 of solids or 1,900 × 0.45 = 855 kg solids. Similarly,
1,000 × 0.55 = 550 kg water. Therefore, the pulp density of the underflow is
855/(855 + 550) × 100 = 60.9%.
The water rate to the underflow is WU = 17.9 × (550/855) = 11.5 t/h.
The water fluidization velocity and hence the fluidization water rate needs
to be specified in t/h. The fluidization rate, WD, is estimated using the value
previously established under plant conditions, adjusted for the geometric mean
size, d*, and target density, ρs. The upper particle size of 2.0 mm combined with
the lower target size has a geometric mean size of 0.71 mm. Hence, a particle
size factor of (d*/0.71) is applied, where d* = 0.71 is the geometric mean for
this case. The density factor applied is (ρs – 1,000)/(1,900 – 1,000), where ρs =
1,900 kg/m3 is the target particle density in this case. The generalized relation-
ship (with d* in units of millimeters) is
WD = 72 * ( d * /0.71) * ( ρs − 1,000 ) / (1,900 − 1,000 ) = 72 m 3/h = 72 t/h
The basic principal is to ensure that the bed is “just” suspended, and that
the rate used gives good stability. Insufficient fluidization will lead to a failure
in the measurement of the suspension density and hence a failure in the process
control, as well as the trapping of lower-density particles in the bed, and in turn
poorer Ep values. It is crucial to ensure that the bed is not excessively fluidized,
as this lowers the suspension density and induces unwanted mixing near the
underflow exit, while preventing fine dense particles from reaching the under-
flow. Clearly, by mass balance, the water rate with the feed solids is
WF = Wo + WU − WD = 290.2 + 11.5 − 72 = 229.7 t/h
Therefore the feed pulp density is 119.5/(119.5 + 229.7) × 100 = 34.2%.

Finally, the water split to the underflow must be checked to ensure a value
of less than 10%. This water split is 11.5/(72 + 229.7) = 0.038, that is, 3.8%,
which is clearly less than 10%. At low yields this value can increase markedly,
requiring a lower value to be specified for the volume fraction of solids report-
ing to the overflow. In turn, this imposes a lower feed pulp density. It is empha-
sized, however, that the system is best designed around a feed pulp density of
Mo = 101.6
Wo = 290.2
MF = 119.5
WF = 229.7
RC
Fluidization Underflow
WD = 72.0 MU = 17.9
Wu = 11.5
Figure 11 Overall material balance of solids in t/h (M) and water (W) in the feed (F),
underflow (U), and overflow (O), with fluidization, D
about 30% solids for coal and mineral matter feeds, given this level allows the
system to self-adapt to variable solids loadings, assuming the water rate remains
reasonably steady.
The overall material balance is shown in Figure 11 in terms of the mass of
solids and water in each of the flow streams. Notably, if the target particle size
is 0.25 mm, a more conservative design would be based on a smaller size in the
calculation, perhaps 0.125 mm, if very high performance is sought. This will
drive the solids throughput down by a factor of about 2.0 to 60 t/h. Further,
in the presence of slimes, involving particles significantly finer than 0.25 mm,
hindered settling could be more significant. The level of hindered settling can
vary appreciably depending on the extent to which the clays are space filling.
Thus, care is required in estimating the effects of slimes.
It is appropriate to end with some comment on the application of the lami-
nar shear mechanism to the design calculations. This mechanism is primarily
concerned with providing a basis for the conveying of particles to the overflow.
Application of the theory of Galvin and Liu (2011) to a channel with z = 6.0
mm, and particles covering the range d = 0.25 to 2.0 mm, and density 1,900
kg/m3, produces a range of channel velocities, U '. An example calculation can
be found in the paper. Note, however, that the particle terminal velocity in
both the normal and tangential directions should be lowered by the relevant
hindered-settling factor, 0.66. This approach reduces the required inertial lift
force and local fluid velocity at the inclined surface. The value of U ' is then
reduced by the area correction factor of 0.79 × 0.80 to allow for the plate
thickness, and launders. A further factor of sin70° is applied to convert the
channel velocity to the vertical direction, based on continuity considerations.
Equation 6, when modified by these factors, gives the approximation U = [Ut z/
(3d)] × [0.66 × 0.79 × 0.80 × sin(70°)] = [0.0192 × 6/(3 × 0.25)] × 0.39 =
0.06 m/s. The value based on the exact velocity profile is 0.052 m/s. The more
complete theory, incorporating inertial lift, results in a velocity of 0.036 m/s for
0.5-mm-diameter particles, 0.032 m/s for 1.0-mm particles, 0.038 m/s for 1.5-
mm particles, and 0.045 m/s for 2.0-mm particles. This trend has similarities
to the curve shown in Figure 9, with an average value of 0.041 clearly higher
than the value of 0.0321 m/s obtained using Equations 3, 4, and 5. The aver-
age value obtained using a lower target density of 1,700 kg/m3 is 0.031 m/s,
which is lower than 0.0321 m/s, so there is reasonable consistency between
the methods. The laminar shear mechanism tends to favor the transport of the
lower-density particles, regardless of their size.
Moreover, the suspension velocity alone does not govern the actual sepa-
ration. Rather, it is a combination of the inclined channel velocity and level
of hindered settling that matters. These two values are controlled by the set
point density applied to the lower bed. When the set point density is exceeded,
underflow discharges, reducing the inclined channel velocity. Hence, in prac-
tice, the upper inclined channel zones self-adapt, building the volume fraction
of the solids in the inclined channels to a level sufficient for the transport of
the particles to the overflow. The lowest density particles will always tend to
transport ahead of the higher-density particles. An excessive velocity will lead
to a lower volume fraction of solids in the inclined channels, whereas a lower
velocity will lead to a higher volume fraction. Conversely, a relatively low solids
throughput will lead to the building of the volume fraction of solids within the
inclined channels, though the volume fraction in the actual overflow emerging
from the unit will be low. In general, the holdup, or volume fraction of solids,
required to convey the particles to the overflow will develop automatically, with
the lower-density particles always in the “front of the queue.”
CONCLUSIONS
This chapter outlined the development of the reflux classifier, from inception,
through the initial pilot-scale phase, to the development of the first prototype.
The importance of the ongoing research and subsequent advance achieved
when closely spaced inclined channels were used, were also described. This
work then led to a significant improvement in separation performance, with

a composite Ep over the size range of 0.25 to 2.0 mm of 0.06 at laboratory
scale. The technology thus underwent a new design, resulting in the RC2020.
The chapter also outlined the calculation of the throughput capacity using an
approach based on the research.
ACKNOWLEDGMENTS
The author acknowledges the financial support of the Australian Research
Council and the Australian Coal Association Research Program, the significant
contributions of Ludowici Australia, and the ongoing support of the Bloom-
field Group in hosting plant trials.
REFERENCES
Acrivos, A., and Herbolzheimer, E. 1979. Enhanced sedimentation in settling tanks
with inclined walls. J. Fluid Mech. 92(3):435–457.
Asif, M. 1997. Modeling of multi-solid liquid fluidized beds. Chem. Eng. Technol.
20:485–490.
Bird, B., Stewart, W., and Lightfoot, E. 1976. Transport Phenomena. New York: John
Wiley and Sons. p. 62.
Boycott, A.E. 1920. Sedimentation of blood corpuscles. Nature 104:532.
Carty, J.J. 1957. Resistance coefficients for spheres on a plane boundary. B.S. the-
sis, Department of Civil and Sanitary Engineering, Massachusetts Institute of
Technology.
Davis, R.H., Zhang, X., and Agarwala, J.P. 1989. Particle classification for dilute sus-
pensions using an inclined settler. Ind. Eng. Chem. Res. 28:785–793.
Di Felice, R. 1995. Hydrodynamics of liquid fluidization. Chem. Eng. Sci. 50:1213–1245.
Doroodchi, E., Fletcher, D.F., and Galvin, K.P. 2004. Influence of inclined plates on
the expansion behaviour of particulate suspensions in a liquid fluidised bed. Chem.
Eng. Sci. 59:3559–3567.
Doroodchi, E., Zhou, J., Fletcher, D., and Galvin, K.P. 2006. Particle size classification
in a fluidised bed containing parallel inclined plates. Miner. Eng. 19:162–171.
Galvin, K.P. 2004. A reflux classifier. University of Newcastle Research Associates Lim-
ited. U.S. Patent 6814241.
Galvin, K.P., and Liu, H. 2011. Role of inertial lift in elutriating particles according to
their density. Chem. Eng. Sci. 66:3686–3691.
Galvin, K.P., and Munro, M. 2008. Overflow launder. University of Newcastle Research
Associates Limited. U.S. Patent 7334689.
Galvin, K.P., and Nguyentranlam, G. 2002. Influence of parallel inclined plates in a
liquid fluidized bed system. Chem. Eng. Sci. 57:1231–1234.
Galvin, K.P., Pratten, S.J., and Nicol, S.K. 1999. Dense medium separation using a
teetered bed separator, Miner. Eng. 12(9):1059–1081.
Galvin, K.P., Zhao, Y., and Davis, R.H. 2001. Time-averaged hydrodynamic roughness
of a noncolloidal sphere in low Reynolds number motion down an inclined plane.
Phys. Fluids 13(11):3108–3119.
Galvin, K.P., Doroodchi, E., Callen, A.M., Lambert, N., and Pratten, S.J. 2002. Pilot
plant trial of the reflux classifier. Miner. Eng. 15:19–25.
Galvin, K.P., Callen, A., Zhou, J., and Doroodchi, E. 2004. Gravity separation using a
full-scale reflux classifier. In Proceedings of the Tenth Australian Coal Preparation
Conference. Edited by W.B. Membrey. Paper H21.
Galvin, K.P., Callen, A., Zhou, J., and Doroodchi, E. 2005. Performance of the reflux
classifier for gravity separation at full scale. Miner. Eng. 18:19–24.
Galvin, K.P., Walton, K., and Zhou, J. 2009. How to elutriate particles according to
their density. Chem. Eng. Sci. 64:2003–2010.
Galvin, K.P., Walton, K., Zhou, J. 2010. Application of closely spaced inclined channels
in gravity separation of fine particles. Miner. Eng. 23:326–338.
King, M.R., and Leighton, D.T. 1997. Measurement of the inertial lift on a moving
sphere in contact with a plane wall in shear flow. Phys. Fluids 9(5):1248–1255.
Laskovski, D., Duncan, P., Stevenson, P., Zhou, J., and Galvin, K.P. 2006. Segrega-
tion of hydraulically suspended particles in inclined channels. Chem. Eng. Sci.
61:7269–7278.
Leighton, D., and Acrivos, A. 1986. Viscous resuspension. Chem. Eng. Sci.
41(6):1377–1384.
MacTaggart, R.S., Him-Sum Law, D., Masliyah, J.H., and Nandakumar, K. 1988. Grav-
ity separations of concentrated bidisperse suspensions in inclined plate settlers.
Int. J. Multiphase Flow 14:519–532.
Nakamura, H., and Kuroda, K. 1937. La cause de l’acceleration de la vitesse de sedi-
mentation des suspensions dans les recipients inclines. Keijo J. Med. 8:256–296.
Ponder, P. 1925. On sedimentation and Rouleaux formation. Q. J. Exp. Physiol.
15:235–252.
Rampall, I., and Leighton, D.T. Jr. 1994. Influence of shear-induced migration on tur-
bulent resuspension, Int. J. Multiphase Flow 20(3):631–650.
Richardson, J.F., and Zaki, W.N. 1954. Sedimentation and fluidization: Part I. Trans.
Inst. Chem. Eng. 32:35–53.
Smart, J.R., Beimfohr, S., and Leighton, D.T. 1993. Measurement of the translational
and rotational velocities of a noncolloidal sphere rolling down a smooth inclined
plane at low Reynolds number. Phys. Fluids A 5:13.
Thompson, P.D., and Galvin, K.P. 1997. An empirical description for the classification
in an inclined counter-flow settler. Miner. Eng. 10(1):97–109.
Vance, W.H., and Moulton, R.W. 1965. A study of slip ratios for the flow of steam-
water mixtures at high void fraction. AIChE J. 11(6):1114–1124.
Wills, B.A., and Napier-Munn, T.J. 2006. Mineral Processing Technology: An Introduc-
tion to Practical Aspects of Ore Treatment and Mineral Recovery. 7th ed. London:
Butterworth-Heinemann.
Yao, K.M. 1973. Design of high-rate settlers. J. Environ. Eng. Div. 99(5):621–637.
Zhang, X., and Davis, R.H. 1990. Particle classification using inclined settlers in series
and with underflow recycle. Ind. Eng. Chem. Res. 29:1894–1900.
Zhou, J., Walton, K., Laskovski, D., Duncan, P., and Galvin, K.P. 2006. Enhanced
separation of mineral sands using the reflux classifier. Miner. Eng. 19:1573–1579.
Zigrang, D.J., and Sylvester, N.D. 1981. An explicit equation for particle settling veloci-
ties in solid-liquid systems. AIChE J 27(6):1043–1044.
Column and Nonconventional
Flotation for Coal Recovery:
Circuitry, Methods, and
Considerations
J.N. Kohmuench, E.S. Yan, and L. Christodoulou
ABSTRACT
Column flotation continues to be an efficient and economical means of recovering
fine coal from wash plant effluent streams. The use of deep froths and the applica-
tion of wash water ensure a high-quality float product. As well, the use of robust
sparging systems ensures high combustible recoveries through the generation of copi-
ous amounts of fine bubbles. With more than a dozen years of supplying flotation
systems for the coal industry, Eriez Manufacturing Company continues to gain
insight into the proper application of these devices and how the flotation circuit
interacts with the remainder of the plant circuitry. This understanding is essential
as a handful of specific parameters can have a large impact on several aspects of
the design of the flotation circuit including, but not limited to, unit equipment
selection and sizing, circuit arrangement, layout, up- and downstream ancillary
requirements, ease of operation, and ultimately capital cost. Incorporating these
findings and taking advantage of new technology and approaches can help end
users achieve the desired rate of return for a flotation project.
INTRODUCTION
Flotation columns derive their name from the geometric shape of the vessel.
Unlike conventional mechanically agitated flotation machines, which tend to
use relatively shallow rectangular tanks, column cells used in the coal industry
are tall vessels with heights typically ranging from 7 m to 16 m depending on
the application. Unlike conventional flotation machines, columns do not use
mechanical agitation. The absence of intense agitation promotes higher degrees
of selectivity and aids in the recovery of coarse particles. In general, feed slurry
187
enters the column at one or more feed points located in the upper third of the
column body and descends against a rising swarm of fine bubbles generated by
the air sparging system. Particles that collide with and attach to the bubbles rise
to the top of the column, eventually reaching the interface between the pulp
(collection zone) and the froth (cleaning zone).
For an equivalent volumetric capacity, the cross-sectional surface area of
the column cell is much smaller than a conventional cell. This reduced area
is beneficial for promoting froth stability and allowing deep froth beds to be
formed. This is an important aspect of column flotation, as a deep froth bed
facilitates froth washing to remove unwanted impurities from the float product.
Wash water, added at the top of the column, percolates through the froth zone
displacing dirty process water and nonselectively entrained particles trapped
between the bubbles. In addition, froth wash water serves to stabilize and add
mobility to the froth. Sufficient water must be added to ensure that all of the
feedwater that normally reports to the froth product has been replaced with
fresh or clarified water. Less than 1% of the feed pulp and associated clays will
report to the froth in a well-operated column (Luttrell et al. 1999).
In contrast, conventional mechanical cells do not operate with deep froths.
Therefore, these devices allow some portion of the ultrafine mineral slimes to be
recovered with the water that reports to the froth, consequently reducing prod-
uct quality. This particle entrainment is the nonselective hydraulic conveyance
of gangue into the product launder. In fact, fine particles (<0.045 mm) have a
tendency to report to the froth concentrate in direct proportion to the amount
of product water recovered. As such, the flotation operator is often forced to
make the decision to either “pull hard” on the cells to maintain yield or run the
cells less aggressively to maintain grade.
The primary advantage of utilizing wash water is the ability to provide a
superior separation performance compared to conventional flotation processes.
This capability is illustrated by the test data summarized in Figure 1, which
compares column flotation technology with an existing bank of conventional
cells. As shown, the separation data for the column cells utilizing wash water
are far superior to those obtained from the conventional flotation bank. In fact,
the data for the column cells tend to fall just below the separation curve pre-
dicted by release analysis (Dell et al. 1972). A release analysis is an indication
of the ultimate flotation performance and is often regarded as “washability” for
flotation. This figure suggests that columns provide a level of performance that
would be difficult to achieve even after multiple stages of cleaning by conven-
tional machines.
Column and Nonconventional Flotation for Coal Recovery 189
100
Release
Analysis
80
60
Recovery, %
40
20
Full-Scale Conventional
Laboratory Column
Full-Scale Column
0
0 5 10 15 20
Product Ash Content, %
Figure 1 Comparison of column and conventional flotation cells
The importance of minimizing entrainment is better understood when

optimizing plant circuitry with respect to coal value. This optimization exercise
is commonly accomplished by utilizing the incremental ash concept, which
states that clean coal yield for parallel operations will be highest when all cir-
cuits are operated at the same incremental quality (Luttrell et al. 2009). In other
words, all circuits in a coal preparation plant should be recovering the same
quality particles with no one circuit recovering a disproportionate amount of
unwanted material such as high-ash mineral matter or moisture. When entrain-
ment is not minimized, large amounts of high-ash material is recovered with
the product. This is particularly problematic in flotation circuits that contain
large amounts of ultrafine clay, which elevates the ash content of the flotation
product. To compensate, other plant circuits must be operated at lower specific
gravity cut points in order to achieve the desired global product specification.
Based on a simple comparison between the specific gravity of the high-ash clays
and the low-ash coal, approximately 3 tons of salable coal are lost for every ton
0.15 mm (Norm.)
Column Feed
Filtrate
Adjustable
Wash-Water
Ring
To Spirals
or Clean Clean Coal
Coal De-aeration
Dewatering Tank
Circuit
Feed
Column Tails Filter Clean Coal
Figure 2 Traditional by-zero circuit
of clay that is recovered into the flotation launder. As such, there is significant
economic incentive to minimize entrainment and improve plant performance.
T Y P I C A L F I N E C O A L C O L U M N F L O TAT I O N C I R C U I T S
There are two typical circuits utilized for fine coal flotation. These include
the traditional minus-0.150-mm “by-zero” circuit and the 0.150 × 0.045 mm
“deslime” circuit. As would be expected, the by-zero approach typically incor-
porates the use of large-diameter cyclones that treat minus-1-mm feed as seen
in Figure 2. Typically, these cyclones are configured to achieve a cut point of
approximately 0.150 mm with the finer fraction reporting directly to flotation.
In some markets, the cut point can be as coarse as 0.5 mm, though it is recom-
mended to not exceed 0.250 mm.
The feed to flotation in a by-zero circuit is relatively dilute and typically
ranges between 2.5% and 6.0% solids, by weight. As a result, volume flows in
these circuits can be quite large. Consequently, significant cell volume must be
available to achieve the desired retention time to ensure adequate combustible
recovery. To attain the needed retention time, a number of large-diameter cells
may be required, which can range in height from 8.5 m to 16.0 m. The ultimate
height is typically based on kinetic data generated from laboratory testing, but
it is generally constrained by practical limits as dictated by both economic and
site-specific engineering requirements. As a result, the height-to-diameter ratio
for an industrial cell ranges from 3:1 to as low as 2:1.
100
95
90
85
80
Recovery, %
75
70
65
60 Plug Flow
Two Cells in Series (3.2 Mixers)
Two Cells in Parallel (1.6 Mixers)
55
50
0 1 2 3 4 5 6 7 8 9 10
kτ
Figure 3 Theoretical recovery as a function of residence time and flotation rate

constant
Because of the high flow volumes found in by-zero circuits, bypassing of

feed can occur and is caused by the less-than-ideal mixing found within the
large-diameter and relatively short tanks commonly found in coal flotation
installations. The high degree of mixing allows some material within the cell to
report to tails without being influenced by the bubble swarm produced by the
sparging system. As the residence time is increased, the probability of bypass is
reduced. As a consequence of this relationship, the occurrence of bypass is more
often found in traditional by-zero circuits where the volumetric feed rates are
relatively high and retention times are typically low. However, recent studies
show that arranging column cells in series helps to minimize bypass (Stanley
et al. 2006). This study applied the concepts defined by Levenspiel (1972)
and showed that when cells are arranged in series, the theoretical maximum
combustible recovery is greater than when the cells are arranged in parallel.
This concept is illustrated in Figure 3 which shows the expected recovery for
an efficient plug-flow system compared to both parallel and in-series column
circuitry.
In Figure 3, combustible recovery is predicted with respect to a combina-

tion of both residence time (τ) and the flotation rate constant (k). Shown are
data from tracer studies, which indicate that when operated in parallel, a single
column cell achieves a recovery response curve similar to 1.6 mixers in series. A
mixer is defined as one perfectly mixed cell where an infinite number of mix-
ers in series reflects an ideal plug-flow reactor. Figure 3 also shows that when
arranged in series, a column circuit can achieve a response curve similar to 3.2
mixers in series. In other words, for the same residence time, float capacity, and
capital expense, the expected combustible recovery can be greatly increased
for any given residence time by simply rearranging the circuit flow scheme.
Conversely, the time at which a target combustible recovery can be achieved is
significantly reduced. For instance, to reach a combustible recovery of 90%, two
columns operating in parallel must exceed a kτ value of 5.0. However, two col-
umns operating in series will achieve this same target recovery at a kτ value of
3.5. Since the flotation rate constant is equal for a specific coal and/or system,
this reduction indicates that the same combustible recovery can be achieved in
30% less residence time by operating the cells in series.
The by-zero circuit works well in plants that serve the metallurgical coal
market where ash is the primary quality specification; however, studies have
shown that a by-zero approach is not the optimum circuitry for plants that
are producing coal for the thermal market where other quality specifications
such as moisture, heat content, and sulfur are also critical (Luttrell et al. 2009;
Bethell and Luttrell 2005). These studies show that the popular “deslime”
circuit is better suited to achieve a separation that maximizes profitability as
dictated by a typical utility coal sales agreement. Luttrell et al. (2009) show that
product moisture plays a key role in determining profitability of a plant produc-
ing coal for the thermal market. In fact, this study indicated that when taking
into account moisture and applying the incremental quality concept, it is neces-
sary to discard material that is finer than 44 μm. The deslime circuit (Figure 4)
achieves this rejection by incorporating a secondary bank of 150-mm cyclones,
which is used to further classify the flotation feed at approximately 0.045 mm.
The study by Bethell and Luttrell (2005) further indicates that this approach is
superior for circuits that contain clay-rich flotation feeds where the benefit of
wash water can be maximized.
Bethell and Luttrell (2005) indicate that there are several benefits to the
deslime flotation circuit which are realized due to the rejection of high sur-
face area particles. These include a lower ash product, reduced froth moisture,
an increase in flotation capacity, and an improvement in downstream froth
handling. The improved froth handling is a result of two factors, including a
Reject
Recycled
Screen-bowl
Effluent
Clean Coal
To Spirals 0.150 × 0.045 mm De-aeration
or Clean (Nominal) Tank
Coal
Screen Bowl
Dewatering
Circuit
Feed
Reject Clean Coal
Figure 4 Deslime circuit with additional classification
coarser float product and a significant reduction in total frother addition. The
coarser froth product tends to be less stable and will readily collapse. The lower
frother addition rate is a result of the volume split achieved in the smaller 150-
mm cyclones, which substantially reduces the total volume of slurry that must
be processed in flotation. Given that frother is added based on total flotation
volume flow, this allows high concentrations of surfactant to be realized in the
flotation circuit without causing upsets elsewhere in the plant. The obvious sec-
ondary benefit is the reduction in size and cost of the flotation and downstream
dewatering equipment. Overall, Bethell and Luttrell (2005) indicate that a
deslime circuit may cost less than half of an equivalent by-zero circuit.
CELL TECHNOLOGY
The ability to consistently maintain high combustible recoveries while pro-
ducing low ash flotation product is the result of operating with efficient and
robust sparging systems. Furthermore, the high grade float product is a result of
operating with a deep froth while incorporating the generous addition of wash
water. In order to run with deep froths, the sparging systems must be designed
to provide the maximum rate of bubble surface area through the column. As
such, the air sparging system is perhaps the most important component in a
column flotation cell. While details related to the specific design features of the
various sparging technologies have been presented in the literature (McKay et
al. 1988; Davis et al. 1995; Finch 1995), the most prevalent column sparging
systems are presented here.
Source: Courtesy of EFD.
Figure 5 SlamJet sparger with self-closing nozzle
SlamJet Technology
The first practical bubble generation system was developed by the U.S. Bureau
of Mines. This design essentially consisted of a series of small orifices (2–3 mm)
that were placed along the length of a pipe which was then inserted through
the column sidewall. This approach worked satisfactorily but was plagued with
maintenance problems with regard to plugging of the orifices. Canadian Pro-
cess Technologies (CPT) solved the plugging problems with the introduction
of their SparJet aeration system. This system uses a series of removable air lances
that include a single orifice located at the end of each sparger. High-velocity
air is injected into the column cell to create and disperse fine bubbles. This
technology was further refined when CPT, now the Eriez Flotation Division
(EFD), updated the SparJet spargers by incorporating a self-closing mechanism
that eliminates any backflow of slurry into the aeration system in the event of
an air supply failure. This updated version has been designated the SlamJet and
is presented in Figure 5.
SlamJet technology has best been applied to the coarser 0.150 × 0.045 mm
deslime flotation circuits where the coarser feed does not require intense colli-
sion energy to achieve particle collection. In addition, the lower volumetric feed
rate provided by this circuit ensures recovery by providing significant retention
time that can easily exceed 10–12 minutes in a typical deslime column that is
fed between 10% and 14% solids, by weight. The bubble size distribution pro-
vided by this sparging technique is well suited for this size class and provides
efficient bubble generation powered only by compressed air.
Cavitation-Tube Technology
One of the most popular sparging technologies used in the coal industry is
generally considered the dynamic sparging system. This technology was con-
ceived after several years of fundamental research which showed that the rate
of flotation could be enhanced through the use of smaller air bubbles. The small
(<0.8 mm) bubbles are generated by circulating slurry from the lower section
of the column through a set of parallel, in-line spargers into which compressed
air is injected (Figure 6). The spargers are typically mounted outside the col-
umn to simplify inspection and replacement. The dynamic sparging system has
been best applied on by-zero fine coal circuits. The primary reason is that the
dynamic sparging system has the ability to efficiently generate large amounts
of very small bubbles for a given airflow. As such, the probability of bubble–
particle collisions is greatly improved, which can increase capacity. This is an
added benefit when treating coal in a traditional by-zero circuit, which has a
high-volume flow, reduced residence time, and a finer feed size distribution.
More recently, CPT developed and implemented the Cavitation-Tube
sparger which can meet the rigorous performance demands of industrial instal-
lations. This sparger consists of an hourglass-shaped tube (Figure 6) constructed
of wear-resistant material. As part of a dynamic sparging system, a mixture of
flotation pulp and compressed gas is circulated at very high velocity through
the device using a centrifugal pump. The high velocity and throat geometry
creates cavitation in the slurry that is beneficial for the flotation of ultrafine
particles (Zhou et al. 1994, 2009). The gas nuclei that are created during cavita-
tion help to activate particles by promoting the attachment of larger bubbles.
In other words, the tiny bubbles act as a secondary collector, which can reduce
flotation collector dosage and enhance the probability of particle attachment.
Details and further descriptions of the advantages of cavitation in flotation
have been described elsewhere (Santos et al. 2012; Honaker et al. 2011).
StackCell
Over the last several years, EFD has developed a new flotation device called
the StackCell that offers high flotation capacity in a small footprint. This
approach provides column-like performance while offering the flexibiilty to
arrange by-zero flotation circuits in such a manner to optimize both metallur-
gical and economical efficiency. Although a detailed theory of operation and
Adjustable
Wash-Water
Distributor
Froth Launder
Feed Inlet
Froth Product
Slurry
Manifold
Slurry
Recirculation
Pump
Cav-Tube
Air
Manifold Tailing
Discharge
Valve
Figure 6 Cavitation-Tube sparging system with essential design features
cell description are provided elsewhere (Kohmuench et al. 2008), a depiction

is provided in Figure 7. In this device, feed slurry enters the separator through
either a bottom- or side-mounted feed nozzle at which point low-pressure air is
added. The slurry travels into an internal preaeration sparging device that pro-
vides significant shear and contacting prior to arrival into the separation cham-
ber. In fact, all of the necessary bubble–particle contacting is conducted in an
aeration chamber prior to injection into the primary tank, which is used only
for the phase separation between the pulp and the froth. A liquid slurry level
is maintained inside the tank so as to provide a deep froth that can be washed,
thereby providing a high-grade float product similar to column flotation cells.
This unit is designed specifically to overcome some of the challenges set
forth by both end users and engineering firms with regard to traditional column
cell installations. These challenges include cell size, foundation loads, as well as
fabrication and operating costs. The key design parameter of the StackCells is
that they have a small footprint and are gravity driven. This allows the cells to
be easily “stacked” in series or placed ahead of existing conventional or column
flotation cells. This approach was demonstrated by Davis et al. (2011) after
StackCells were installed preceding two parallel column flotation cells operat-
ing at the Alpha Natural Resources Roxana coal preparation plant. Historical
data suggested that the two cells were often overloaded because of production
demands and recent changes in feed quality.
Aeration/
Contacting
Chamber Wash-Water
Manifold
Froth
Launder
Feed Inlet
Tails
Outlet
Air Manifold
Figure 7 Three-dimensional illustration of a StackCell module
Figure 8 shows the impact of the StackCell installation on the combustible

recovery and refuse ash for the entire Roxana flotation circuit. For the first
149 samples taken prior to the installation, the two column cells provided
an average recovery of 74.4% and a combined refuse ash of 72.5%. After the
installation, the combined recovery for the StackCell and two column cells
improved to 83.7% and the refuse ash increased to 80.7%. Close inspection of
the StackCell test data indicates a gradual improvement in overall performance
since being installed. In fact, recent data received from the site indicate that
this circuit now regularly achieves a combustible recovery exceeding 90%. The
improvement can be largely attributed to the optimization of operating vari-
ables such as reagent dosage, froth depth, aeration rate, and wash-water addi-
tion rate. The increased recovery is significant considering that less than 10%
more cell volume was added to the circuit via the installation of the StackCell
technology. In fact, the aeration chamber provided an additional residence time
of only about 5–10 seconds to the total flotation circuit.
The decision of when to use StackCells and/or columns is based on both
economics and coal characterization data. Given that the retention time in the
StackCell is limited, it is best used for treating coal stocks that have relatively
high rate constants. For slower floating coals, it may be best to utilize a com-
bination of StackCells and columns. For instance, a recent in-house exercise
indicated that for a 70-tph by-zero flotation circuit, four StackCells placed
in-series are required to meet the carrying-capacity requirements for a fine
100
80
Recovery, %
60
Avg = 74.4% Avg = 83.7%
40
20
0
100
80
Reject Ash, %
60
Avg = 72.5% Avg = 80.7%

40
20
0
0 50 100 150 200 250 300
Sample Number
Figure 8 Change in flotation circuit performance after installation of a StackCell

(dashed line represents the sample where the changeover occurred)
metallurgical coal. In comparison, a hybrid circuit that utilized StackCells fol-

lowed by column flotation cells yielded a capital cost savings of 20%, as seen
in Table 1, when also considering the cost of the ancillary equipment such as
pumps, compressors, and blowers. Of course, this capital cost savings must be
weighed against the savings generated by the smaller footprint offered by the
StackCells. Regardless, in this exercise, a hybrid circuit incorporating both
StackCells and columns offered an optimized solution that provided both capi-
tal and operational cost savings while ensuring good metallurgical performance
through the in-series arrangement that combined the benefits of preaeration
and increased retention time.
Table 1 Various flotation options for 70-tph flotation by-zero circuit with 80% float
yield
StackCell Column Combination
Circuit Variable SC-3050 Cav-Tube SC-3050 Cav-Tube
Cell diameter, m 3.00 4.25 3.00 4.25
Cell, no. 4 4 2 2
Circulation pump — P — P
Blower P — P —
Compressor — P — P
Total connected power, hp 840 1,600 1,110
Live load, kt 404 1,140 671
Estimated equipment cost, US$ 1.5M 1.5M 1.2M
P I T FA L L S A N D B E S T P R A C T I C E S
Over the past decade, Eriez has continued its research and development efforts
with regard to circuit design and column sizing. Important design parameters
include product carrying capacity, retention time, aeration rate, wash-water
rate, and chemical dosage. Many of these parameters are determined from
either feed characterization studies and lab- or pilot-scale test data. However,
special consideration should be given to the following parameters to ensure that
the flotation circuit can be operated at maximum efficiency.
Particle Size Distribution and Carrying Capacity

The flotation feed size distribution can have a large effect on a flotation circuit,
including product moisture, but more importantly, product carrying capacity.
Carrying capacity is defined as the mass rate of flotation concentrate per unit
time per cell cross-sectional area. Maximum carrying capacity is achieved when
all available bubble surface area has been consumed. For mineral applications
where the float yields are typically low, this parameter is rarely attained. How-
ever, for coal applications, the float yield can be greater than 80% by weight,
and this parameter can be of great importance. Maximum carrying capacity has
been well defined by numerous researchers with extensive data available for coal
and mineral applications. These studies indicate that carrying capacity is lin-
early related to the size and density of particles in the froth (Sastri 1996). Stud-
ies and extensive test work conducted by Eriez also support this finding, as seen
in Figure 9. As seen in this figure, a direct correlation exists between capacity
and both the mean size (d50) and ultrafines content of the flotation feedstock.
Carrying Capacity
Carrying Capacity
Feed Particle Size Ultrafines Content
Figure 9 Correlation between capacity, mean size, and ultrafines content for coal
flotation
Inspection of Figure 9 shows that it is important to consider the vari-

ous particle size distributions of the feedstocks that may be applicable when
designing a coal preparation facility. After the maximum carrying capacity is
achieved, there is no bubble surface area available for any additional combus-
tible recovery. As a result, any additional coal within the system will be lost to
the tailings unless cell surface area is increased or feed rates are dropped. In the
design phase, the required cross-sectional area and number of cells required is
best determined by identifying the case with the finest flotation size distribu-
tion and the maximum floatable tons. By doing so, the worst-case scenario is
considered.
Froth Washing
The use of wash water in a column or StackCell is important and necessary to
provide the optimum separation efficiency as indicated by release analysis. In
addition, a froth depth of 0.6–1.2 m is typically required to ensure good distri-
bution of the wash water and to prevent short-circuiting. As with any column,
the flow of wash water must exceed the volumetric flow of water reporting to
the clean coal product to prevent entrainment of the high-ash slimes. In most
cases, less than about 1% of the feedwater will report to the froth product if the
wash water is properly controlled. The amount of water carried by the froth can
be calculated from:
water demand {m 3 /h/m 2 } =

 100  (EQ 1)
carrying capacity {t/h/m 2 } − 1
 froth % solids 
20
Number of Wash Water Rate

=
18 Dilutions Froth Water Rate
16
14
12
Froth Ash, %
10
Typical
2 Operation
0
0 1 2 3 4
Number of Dilution Washes
Figure 10 Effect of number of dilution washes on clean coal quality
Entrainment should theoretically be eliminated when the number of

dilution washes (defined as the wash-water addition rate divided by the froth
water demand) reaches a value of 1. However, as shown in Figure 10, froth
mixing usually requires that 1.25–1.50 dilution washes be used to fully suppress
hydraulic entrainment. Field data collected from columns operating in the coal
industry suggest that a wash-water flow rate of 7.3–12.2 m3/h/m2 (3.0–5.0
gpm/ft2) is normally adequate for most commercial installations. However,
higher gas and frother addition rates will typically increase the froth water
demand and, as a result, the amount of wash water required. Excessive wash-
water flows should be avoided since the extra wash water passing downward
through a column will create an undesirable reduction in the slurry retention
time and, hence, a potential reduction in recovery. Very high water additions
may also destabilize the froth by stripping surfactant (frother) from the bubble
surfaces. High water rates may also have a detrimental impact on product grade
by increasing axial froth mixing, thereby reducing the wash-water effectiveness
(Yianatos et al. 1988).
The design of the wash-water distributor can also significantly affect col-
umn performance. In some cases, the distribution piping is intentionally sub-
merged below the cell lip so that a drained froth can form above the distributor.
This arrangement allows for exacting control of the water split between the
clean coal and the refuse streams. Ultimately, changes to the vertical position of
the distributor can be used to control the product grade. In some cases, multi-
level concentric distribution rings can be used to overcome problems associated
with poor froth mobility. The inner rings are typically located above the outer
rings to reduce drainage and improve the fluidity of the froth in the center of
the column. More recently, one or more internal launders have been employed
in lieu of tiered wash-water rings. Unfortunately, it can be difficult to identify
and clean out plugged distributor piping, which can severely impact cell per-
formance. As a result, static wash pans have been employed that are located just
above the top of the froth. This arrangement does not provide froth mobility
control, but does allow for easy cleaning and reduced maintenance require-
ments given that they can be easily cleared using a plant water hose.
Aeration
The primary advantage of an efficient and commercially available sparging
system is the ability to generate large amounts of very small bubbles. This capa-
bility is commonly reported in terms of the superficial bubble surface area rate
(Sb), which is defined as the total bubble surface area per unit of time passing
through a given column cross-sectional area. This value can be calculated by
 superficial air rate {m 3 /h/m 2 }

Sb{1 / sec} = 1.67  (EQ 2)
 bubble diameter {mm} 
The impact of Sb on flotation recovery has been illustrated elsewhere

(Kohmuench et al. 2004). Ultimately, the data indicate that the relationship
between recovery and Sb is generally independent of sparger type and that a
superficial gas rate in the range of 70–95 m3/h/m2 would be suitable for most
coal applications. These values correspond to total aeration rates of approxi-
mately 990–1,350 m3/h (582–795 ft3/min) of air for a full-scale 4.25-m-
diameter column. The gas rates at the lower end of the range would generally
be used for spargers that generate smaller bubbles, whereas the higher gas rates
are typically needed for less efficient spargers. A proper combination of gas rate
Figure 11 Improper air instrumentation arrangement with no pressure regulator and
with air flowmeter incorrectly installed following the flow control valve
and bubble size will generally provide a gas holdup in the flotation pulp above
15%–18%. Values above 20% are not unusual for well-tuned circuits.
Caution should be used during the metering of gas flow rates. A properly
designed system should be equipped with a flowmeter that is calibrated to
read correctly at a specified operating pressure. The operating pressure should
be held constant by placing a pressure regulator ahead of the flowmeter and
followed by a control valve so that the meter will always operate at its design
pressure. If the flowmeter is placed after the control valve, then the operating
pressure is unknown and the true gas flow rate cannot be determined (see Fig-
ure 11). In addition, the air supply lines should be as large as reasonably possible
to minimize line losses. Supply lines with an excessive number of fittings, mul-
tiple changes in direction, and small pipe diameters can severely limit the total
volume and pressure of air that can be delivered to the flotation cell.
A great deal of confusion also exists regarding the specification of compres-
sors for column applications. Much of this confusion is related to improper
use of gas flow terminology (Sullair Corporation 1992). For example, column
manufacturers normally report gas flow rates as a “standard” volumetric flow
per time. This value is only valid at 1 atm of pressure and 20°C of dry air. The
“actual” flow rate specified by compressor manufacturers is typically reported
in terms of “inlet” conditions or “free air.” Although this amount of air enters
the compressor, it is not necessarily the amount of air delivered to the column
due to compressor seal leakage. As a result, the actual flow may be only 95% of
the inlet flow. Furthermore, corrections to the gas flow rate must be made to
account for differences in elevation (atmospheric pressure) and humidity. Air
temperature generally has little impact on the capacity of an oil-flooded screw
compressor, but it may affect the performance of an air-cooled compressor.
Table 2 Typical frother dosages based on open-circuit designs

Frother Type Blend, % Typical Dosage, ppm
Glycol 100 8-10
Glycol/alcohol 70/30 12–14
Glycol/alcohol 90/10 14–16
Alcohol 100 14–18
MIBC* 100 20–24
*MIBC = methyl-isobutyl-carbinol.
These complications generally require that professionals be consulted to ensure

that the compressor is properly sized for the specified air requirements.
Frother Rates and Froth Handling

The addition of a surfactant (i.e., frother) is required in most conventional
flotation applications. Column and StackCell flotation are no different and
also require the addition of a frother that will sufficiently lower the surface
tension of the pulp to allow for the creation of a deep and stable froth, which
can withstand the addition of wash water. Many types and blends of frothers
are available; however, the typical dosage of each type is dependent on many
factors that include frother strength, site water chemistry, flotation circuit con-
figuration, dewatering techniques, and whether the plant maintains an open or
closed water circuit. Regardless, sufficient frother must be added to allow for
maximum flotation yield. The ultimate chemical dosage may also have to be
lower if the reagent has the potential to cause other problems in the plant (i.e.,
froth buildup). As such, the choice and dosage of these chemicals should be
determined through consultation with professional chemical vendors. Table 2
provides general dosages for various types of frothers for open-water circuits.
The dosage rates are based on total cell supply volume flows including wash-
water addition.
One of the most important design considerations for column flotation is
the proper handling of the float product. Historically, this is one of the most
problematic issues encountered when utilizing column flotation cells due to the
large volumes of froth generated by this technology. In particular, concentrates
containing large amounts of ultrafine (–0.045 mm) coal can become excessively
stable, creating serious operating problems related to backup in launders and
other downstream unit operations. Bethell and Luttrell (2005) demonstrated
that coarser deslime froths readily collapsed, but finer froths had the tendency
to remain stable for an indefinite period of time. Attempts made to overcome
this problem by selecting weaker frothers or reducing frother dosage have not
been successful and have generally led to lower circuit recoveries.
Several circuit modifications have been developed to deal with the froth
stability problem. For example, column launders need to be considerably
oversized with steep slopes to reduce backup. Horizontal froth travel distances
must be kept as short as possible and adequate vertical head must be provided
between downstream operations and column launders. In addition, piping and
chute work must be designed such that the air can escape as the froth travels
from the flotation circuit to the next unit operation. Figure 12 shows how
small changes in piping arrangements can result in better process performance.
Figure 12a shows a column whose performance suffered because of the inability
to move the froth product from the column launder although a large discharge
nozzle (1 × 1 m) had been provided. In this example, the froth built up in the
launder and overflowed when the operators increased air rates. To prevent this,
the air rates were lowered, which resulted in less-than-optimum recoveries.
It was determined that the downstream discharge piping was air-locking and
preventing the launders from properly draining. The piping was replaced with
larger chute work that allowed the froth to flow freely and the air to escape. As a
result, higher air rates were possible and recoveries were significantly improved.
Some installations have resorted to using defoaming agents or high-
pressure launder sprays to deal with froth stability. However, newer column
installations avoid this problem by including large deaeration tanks to permit
time for the froth to collapse (Figure 13a). Special provisions may also be
required to ensure that downstream dewatering units can accept the large froth
volumes. For example, standard screen-bowl centrifuges equipped with 10-cm
(4-in.) inlets may need to be retrofitted with 20-cm (8-in.) or larger inlets to
minimize flow restrictions.
The use of screen-bowl centrifuges provides low product moistures; how-
ever, there are typically fine coal losses as a large portion of the –0.045 mm float
product is lost as main effluent. This material is highly hydrophobic and will
typically accumulate on top of the thickener as a very stable froth layer, which
increases the probability that the process water quality will become contami-
nated (i.e., black water). This phenomenon is more prevalent in by-zero circuits,
especially when the screen-bowl screen effluent is recycled back through the
flotation circuit, either directly or through convoluted plant circuitry. Reintro-
ducing material that has already been floated to the flotation circuit can result
in a circulating load of very fine and highly floatable material. As a result, the
capacity of the flotation equipment can be significantly reduced, which results
in losses of valuable coal. Most installations will combat this by ensuring that
A B
Figure 12 Difference in froth discharge before and after changes to downstream
piping
A B
Figure 13 (a) Column with deaeration tank and (b) thickener with floating boom and
sprays
the screen-bowl screen effluent is routed directly back to the screen bowl so
that it does not return to the flotation circuit. The accumulation of froth on the
thickener is also reduced by utilizing reverse weirs and taller center wells, as this
approach helps limit the amount of froth that can enter into the process water
supply. Froth that does form on top of the clarifier can be eliminated by employ-
ing a floating boom that is placed directly in the thickener (Figure 13b) and
used in conjunction with water sprays. The floating boom is constructed out of
inexpensive PVC (polyvinyl chloride) piping that can be attached to the rotat-
ing rakes. The boom floats on the water interface and drags any froth around to
the walkway that extends over the thickener. Water sprays are mounted on the
walkway and positioned to maximize the destruction of the froth.
For very coarse feeds, the froth product may be sufficiently handleable
to be pumped to classifying cyclones. In such cases, the cyclone underflow
(oversize) product is typically passed directly to the dewatering circuit (usually
filters), while the cyclone overflow (undersize) product is passed to a clean coal
thickener. The solids from the thickener underflow are pumped to the dewater-
ing circuit. The clarified overflow from the thickener is sent back to the column
circuit as wash water. This arrangement maintains a high frother concentration
in the column circuit and reduces the amount of fresh frother that is required,
thereby minimizing the impact of surfactant buildup on other plant circuits.
This approach can also be mimicked using vacuum belt and disk filters by rout-
ing the relatively clean filtrate back to the flotation circuit to be used as wash
water.
SUMMARY
As high-grade feed stocks continue to decline, the role of flotation will likely
increase in future plant designs. As such, it is important to understand how
both column and other nonconventional flotation can be applied to maximize
profitability. This includes ensuring that the circuit design and equipment selec-
tion compliment the product requirements and how they ultimately relate to
the coal sales contract. Experience has shown that the final market destination
(i.e., metallurgical vs. thermal) and contract specifications (i.e., moisture, tons,
and inerts) will be the primary factors when making these process decisions.
Although the installation of a column or StackCell circuit can provide a
beneficial financial return, the design and scale-up of this technology is chal-
lenging. Outside issues such as froth handling and dewatering techniques
must be fully considered to ensure a successful outcome. Careful engineering,
extensive testing, and attention to ancillary requirements will ensure that the
flotation performance agrees with that predicted by standard release analysis.
To ensure the success of the entire installation, the overall design of the flo-
tation circuitry must also be taken in account. This includes the determination
of the circuit type (i.e., by-zero or deslime) and whether or not the cells should
be placed in parallel or in series. By itself, the selection of a flotation column
or a StackCell requires the understanding of application-specific details, such
as coal quality and expected product mass yield. Through this understand-
ing, the cells can be properly engineered with regard to sparging technology,
product carrying capacity, retention time, wash-water rates, and aeration rates.
Adherence to the design criteria as outlined in this manuscript will provide the
basis for a successful flotation circuit.
REFERENCES
Bethell, P.J., and Luttrell, G.H. 2005. Effects of ultrafine desliming on coal flotation
circuits. In Proceedings, Centenary of Flotation Symposium, Brisbane, Australia. pp.
719–728.
Davis, V.L., Bethell P.J., Stanley F.L., and Luttrell G.H. 1995. Plant practices in fine coal
column flotation. High Efficiency Coal Preparation: An International Symposium.
Edited by S.K. Kawatra. Littleton, CO: SME. pp. 237–246.
Davis, V., Stanley, F., Kiser, M., Bratton, R., Luttrell, G.H., Yan, E.S., Kohmuench,
J.N., and Christodoulou, L. 2011. Industrial evaluation of the StackCell flotation
technology. Coal Prep. Soc. Am. (CPSA) J. 10(3):22–26.
Dell, C.C., Bunyard, M.J., Rickelton, W.A., and Young, P.A. 1972. Release analysis: A
comparison of techniques. Trans. IMM (Sect. C), 81(787):89–96.
Finch, J.A. 1995. Column flotation: A selected review—Part IV: Novel flotation
devices. Miner. Eng. 8(6):587–602.
Honaker, R.Q., Saracoglu, M., Kohmuench, J.N., and Mankosa, M.J. 2011. Cavitation
pretreatment of a flotation feedstock for enhanced coal recovery. 28th Interna-
tional Coal Preparation Conference and Exhibit, Lexington, KY, May 2–5.
Kohmuench, J.N., Davy, M.S., Ingram, W.S., Brake, I.R., and Luttrell, G.H. 2004.
Benefits of column flotation using the Eriez Microcel. In Tenth Australian Coal
Preparation Conference, Proceedings, Polkolbin, NSW, Australia, October 17–21.
pp. 272–284.
Kohmuench, J.N., Mankosa, M.J., and Yan, E.S. 2008. An alternative for fine coal flota-
tion. Coal Prep. Soc. Am. (CPSA) J. 7(1):29–39.
Levenspiel, O. 1972. Chemical Reaction Engineering. New York: Wiley.
Luttrell, G.H., Kohmuench, J.N., Stanley, F.L., and Davis, V.L. 1999. Technical and
economic considerations in the design of column flotation circuits for the coal
industry. In Proceedings, SME Annual Meeting and Exhibit, Symposium Honoring
M.C. Fuerstenau. SME Preprint 99-166. Denver, CO: SME.
Luttrell, G.H., Keles, S., and Honaker, R.Q. 2009. Implications of constant incremen-
tal quality on the design of fine coal dewatering circuitry. SME Preprint 09-093.
Denver, CO: SME.
McKay, J.D., Foot, D.G., and Shirts, M.D. 1988. Column flotation and bubble genera-
tion studies at the Bureau of Mines. In Column Flotation’ 88. Littleton, CO: SME.
pp. 173–186.
Santos, N.C., Duarte, F., Fan, M., Honaker, R.Q., Baldessin, A., and Pinheiro, C.P.
2012. Feed air jet—A cavitation sparger system for enhanced flotation recovery.
In Proceedings, 13th ABM Iron Ore Symposium, Rio de Janeiro, October 14–18.
Sastri, S.R.S. 1996. Technical note: Carrying capacity in flotation columns. Miner. Eng.
9(4):465–468.
Stanley, F., King, P., Horton, S., Kennedy, D., McGough, K., and Luttrell, G. 2006.
Improvements in flotation column recovery using cell-to-cell circuitry. In 23rd
International Coal Preparation Exhibition and Conference, Proceedings, Lexington,
KY, May 1–4.
Sullair Corporation. 1992. Sales bulletin. Industrial compressors—Compressor capac-
ity. In Bulletin E-78. April.
Yianatos, J.B., Finch, J.A., and LaPlante, A.R. 1988. Selectivity in column flotation
froths. Int. J. Miner. Process. 23:279–292.
Zhou, Z.A., Xu, Z., and Finch, J.A. 1994. On the role of cavitation in particle collection
during flotation—A critical review. Miner. Eng. 7(9):1073–1084.
Zhou, Z.A., Xu, Z., Finch, J.A., Masliyah, J.H., and Chow, R.S. 2009. On the role of
cavitation in particle collection during flotation—A critical review II. Miner. Eng.
22:419–433.
Mechanical Cells for Fine Coal
Flotation—An International
Perspective
Jeff Euston, Keith Wilkes, and Asa Weber
ABSTRACT
Froth flotation is currently the only practical and economic method for treatment of
the finest coal fractions. Flotation is a complex process calling on a range of chemi-
cal, hydrodynamic, and engineering disciplines. Fine coal usually comprises less
than 20% of the plant feed and often contains the highest ash. Dewatering is best
achieved by vacuum filtration, and even then, the final product may have relatively
high moisture content with the associated contractual and transport issues.
Thirty years ago, mechanical flotation circuits required large numbers of
small individual cells with high energy and operator demands. The introduction
of pneumatic and column flotation devices in the 1980s offered an apparently
simple solution to increasingly large banks of mechanical flotation cells. These new
technologies promised simplicity with reduced operator involvement in what was
essentially a single device. This simplicity has resulted in a compromise in terms of
combustibles recovery.
Mechanical flotation cells continue to play a major role in coal flotation around
the world, and there is evidence of a resurgence of interest in modern mechanical
flotation technologies. Increasing coal prices and concerns for the environment have
generated a new look at the opportunities for high recovery from the fines circuit.
This chapter summarizes the rationale behind the use of modern mechanical flota-
tion cells and describes their role in the modern coal preparation plant.
INTRODUCTION
Froth flotation has its origins at the beginning of the 20th century at Broken
Hill in Australia. In metalliferous applications, lower ore grades and larger
plants resulted in the need for increasingly fine grinding. Comminution
211
followed by froth flotation quickly became the only practical process to recover
metalliferous ores.
At the time, coal preparation consisted largely of dry beneficiation tech-
niques, such as spiral pickers, techniques that had themselves taken over from
manual sorting from the middle of the 19th century onward. Froth flotation
was first applied to coal in laboratory testing in 1915 with the first operating
plant in England in 1920 and the first U.S. plant in 1930.
Legislation, particularly in the United Kingdom, to reduce the health
effects of airborne dust in coal mines and dry beneficiation plants resulted in
the use of copious volumes of water for dust suppression in underground mines.
This resulted in a wet feed to the preparation plant, reducing the efficiency of
the traditional dry techniques.
Wet processing techniques were developed, and these led to the generation
of a large volume of slurry containing small but significant coal values in addi-
tion to the shale and clay waste. This waste stream was not only a loss of valuable
fine coal, but it generated a large volume of what was essentially dirty water,
often referred to as blackwater. The excellent volume on flotation, published by
the Australasian Institute of Mining and Metallurgy (Lynch 2010), provides an
excellent review of flotation development and includes an enlightening chapter
on coal flotation.
The fine coal fraction typically comprises less than 20% of the coal plant
feed and represents the most difficult to treat stream from both a cost benefit
and processing approach, often with low coal values and high clay content.
Some of the negatives associated with coal flotation are listed here:
• The clays often associated with coal report with the fines, resulting in
high ash values for the ultrafine fraction.
• There is potential to add moisture to the final product, which can in
some operations lead to freezing and an increase in the calorific value
of the final product.
• Certain coal types containing oxidized coal and high clay content do
not respond well to froth flotation.
• Coal specifications may place a limitation on fines content.
The balance of these factors coupled with the historically low value of coal
has resulted in mixed attitudes toward coal flotation. Not only the choice of
technology, but whether or not coal flotation is worth the effort is by no means
clear cut. Increasing coal prices with the mining boom, which began in 2007,
generated a renewed interest in improving coal recovery across the full size
range. Even though the boom has not continued, awareness has been created of
Mechanical Cells for Fine Coal Flotation 213
the coal values being lost to tailings. By the 1980s, coal producers who decided
to install froth flotation found themselves with large numbers of small cells
with individual mechanisms and manual control requirements and relatively
high energy requirements per metric ton of coal treated.
In 1996, an Australian ACARP report by Sanders (2006), summarized
by Heiser (1996), described mechanical flotation as suffering from a range of
issues:
• Highly varied design, with unclear advantages for each
• Poor air control, as observed from periodic extreme cell disturbance
• Unpredictable scale-up performance
• Continual operator/metallurgical attention
• Adversely affected as wear progresses
• No universally applicable control strategy
• Difficult to set up for optimization of recovery and grade
• Energy and reagent costs
At that time, innovative designs such as the Jameson cell and conventional
column devices promised to replace these large banks of mechanical cells with
simple devices with no moving parts, and having the potential for lower prod-
uct ash values and better theoretical grade/yield performance with a minimum
of operator intervention. In Australia, between 1982 and 2012, no complete
mechanical flotation circuits for coal were installed, although several expan-
sions did use mechanical cells. The convenience of these new processes was
seen as an acceptable compromise to produce a low-ash product. Elsewhere
around the world where coal preparation plants were generally smaller than in
Australia, mechanical cells have continued to play a predominant role in coal
flotation.
F L O TAT I O N F U N D A M E N TA L S
In the 1970s, flotation circuits typically comprised large numbers of cells, each
with their own motor and complex arrangements of paddles, and distribution
and froth collection systems. There have been significant developments in
flotation cell design in the intervening years. The introduction of round tank
designs in the 1980s heralded an almost exponential increase in individual tank
volumes. What may once have required literally hundreds of flotation cells can
now be achieved in a bank of four or five cells. Improved control systems and
the power reductions resulting from these economies of scale have eliminated
the perceived problems associated with mechanical flotation in the latter half
of the 20th century. The development of mechanical flotation cells has main-
tained the basic process needs of the flotation process, those of residence time
and flotation kinetics.
Residence Time
Mechanical flotation cells are primarily designed on the concept of residence
time in the flotation cell or bank of cells. Fundamental circuit design requires
the calculation of the total flotation volume required to give sufficient time and
opportunity for the physical and chemical processes of flotation such as activa-
tion and contact times to occur. Residence time is calculated from the flow into
the cell and the effective cell volume. The required residence time is usually
determined by test work from either bench-scale or pilot-plant test work. If
laboratory-scale test work has been performed, the operational residence time
is determined using a scale-up factor. Historically, a scale-up factor of 2 has
been used from laboratory work. In mineral applications, rougher/scavenger
circuits may require residence times from 20 to 30 minutes, whereas cleaning
applications will be lower at 5 to 15 minutes residence times. In coal flotation,
residence times of 5 minutes are typically used.
Hydrodynamics of Flotation
Essentially, froth flotation depends on the combination of several prob-
abilities. The probability of a successful froth recovery is the product of these
probabilities:
• Pc = probability of collision of particles and bubbles
• Pa = probability of adhesion of particle to bubble
• Pd = probability of detachment in the turbulent region of the flota-
tion zone
• Pl = probability of levitation and recovery of values in the concentrate
Collision
The probability of a particle colliding with a bubble depends on the relative
sizes of particles and bubbles, hydrodynamics of the flotation environment,
but not on hydrophobicity. The probability increases as particle size increases.
Adhesion
The probability of a particle attaching to one or more bubbles is largely a
function on particle hydrophobicity. The probability decreases as particle size
increases.
Detachment
The probability of the particle/bubble aggregate persisting depends on hydro-
phobicity and the hydrodynamics of the system. The probability increases as
particle size increases.
Levitation
The probability of the particle reaching the concentrate overflow is the final
stage. If a stable particle/bubble aggregate forms, the aggregate must pass
through the pulp/froth interface to the overflow launder. This part of the pro-
cess requires a quiescent zone but also sufficient energy for froth movement
toward the discharge. Tao (2004) has provided an excellent summary of these
processes with particular reference to coarse and fine particles.
Number of Cells
To prevent the effects of short-circuiting, it is necessary to divide the total
volume required into a number of individual cells. Depending on the mineral
and the duty, the minimum number of cells varies from one to six. For coal
applications, a minimum of four cells is recommended. In a bank of flotation
cells, it is reasonable to assume that the fast-floating material will float first.
Based on the probability of a particle floating and the competition for floating/
transfer across the pulp/froth interface, these fast-floating particles will be the
optimum-sized particles (neither too coarse nor too fine) and the strongly
hydrophobic particles. In subsequent cells, the competition is reduced and
slower floating particles are given an increased opportunity to float. From a
purely statistical point of view, the advantages of multiple cells is clear. It is the
practical equivalent of successive points on the release analysis/ultimate flota-
tion curve. Using the traditional hog trough design for illustration and from an
idealized point of view, it might be imagined that the first cell could recover
about 50% of the liberated coal. The second cell might recover 50% of the
remainder (or 25%), and so forth, to give an overall recovery that approaches
100%. Clearly, this is somewhat stylized but serves to illustrate the principle of
multiple stages of flotation.
The actual residence time, within the bubble/particle contact region or
flotation zone, is on the order of seconds, and in a mechanical cell, particles
will pass through the contact zone several times before it exits the vessel. In the
absence of recycle, a column or Jameson cell allows a single pass through the
flotation zone.
Table 1 Variation in flotation response with particle size

Feed Concentrate Tailings
% % % % % % Yield, Combustibles
mass ash mass ash mass ash % Recovery, %
–1000 + 500 4.7 9.5 4.7 5.0 12.8 36.8 85.8 90.1
–500 + 250 24.8 6.9 16.6 4.9 8.6 42.7 94.7 96.7
–250 + 125 18.0 8.5 22.1 5.9 3.2 67.8 95.8 98.5
–125 + 63 10.4 10.1 13.2 6.7 9.2 77.2 95.2 98.8
–63 + 45 9.4 10.6 11.0 6.9 9.0 72.6 94.4 98.3
–45 32.6 20.4 32.5 9.6 57.3 83.4 85.4 96.9
Total — 12.4 — 7.1 — 72.0 91.8 97.4
Particle Size and Particle Size Distribution

The relationship between flotation recovery and particle size has been repro-
duced many times. There is an accepted range of particle size over which flota-
tion is most efficient. At fine sizes, small bubbles and high shear are important
for recovery—conditions prevalent in Jameson cells, which lead to the excellent
performance of these cells at finer sizes. For coarse particles, reduced recovery
is due to the difficulties of maintaining the attachment of coarse particles to
bubbles in the turbulent environment of the flotation cell. For coal particles,
the accepted size range for conventional flotation is between 75 and 500 μm.
Column cells perform well at the lower end of this range and indeed toward
the finer sizes. Conventional cells are also capable of good recoveries at finer
sizes but are most impressive as particle size increases. Table 1 shows size-by-size
performance from a Queensland coal washery using mechanical flotation cells.
Flotation Circuits
It is well known that a single flotation device cannot be optimized for both
grade (ash) and recovery. Flotation cells are classified according to the duty. The
first stage is referred to as roughing. Here, the aim is to produce a concentrate
and tailings with maximum recovery at an acceptable grade. Scavenger cells aim
to recover valuable material displaced to tailings.
Scavenger cells will typically be used to refloat the rougher tailings and
“scavenge” any values from these tailings. A bank of flotation cells will often
perform roughing and scavenging in the same bank with the first cells effec-
tively roughing and the last two or three cells scavenging the tailings. In coal
flotation, it is usual for the easy floating coal to be recovered in the first cells
with the more difficult to float coal, slow floating, and coarse particles in the
last cells.
Table 2 Flotation circuits

Product Ash Tailings Ash
Option Yield, wt % (dry), wt % (dry), wt %
Single stage 59.6 11.6 27.0
Desliming 60.0 8.2 31.3
Separate conditioning of coarse and fine 51.0 13.0 23.5
size fractions prior to one stage of flotation
Two-stage conditioning prior to one stage 65.0 12.8 28.6
of flotation
Split feed (coarse and fine flotation) 75.2 11.2 39.5
Two-stage reagent addition 79.6 11.0 43.6
Re-flotation of classified tailings 81.0 9.9 58.0
Source: Firth et al. 1979.
The third basic category of flotation is referred to as cleaners. These are

used to clean the concentrate or, for coal, to lower the ash content. Cleaner
tails may report to final tailings or be recycled to feed. Cleaners are designed to
be highly selective. The simplest of multiple bank circuits are a rougher/cleaner
combination, which can produce a lower ash than a single stage alone and a
rougher/scavenger combination, which aims to recover the coal lost to tailings.
Complex Circuits in Coal Preparation

The traditional bank of mechanical coal flotation cells is best described as
being a single rougher stage. This simple circuit does offer the opportunity to
fine-tune the individual cells for optimum overall performance. This can be
achieved by adjusting the air rate along the bank or by the addition of extra
reagents in the second half of the bank. The fast-floating coal is recovered in
the first few cells, and the slow-floating and coarse particles are recovered in the
back of the bank.
The concept of multistage flotation in coal is not new, although the sensi-
tive economics of coal flotation have not historically encouraged their serious
consideration. Firth et al. (1979) reviewed flotation circuits with an emphasis
on poorly floating coals. They describe a range of options including the re-
flotation of classified tailings. Some of the results are summarized in Table 2.
The simplest of multiple-bank circuits are a rougher-cleaner combination,
which can produce a lower ash than a single stage alone and a rougher-scaven-
ger combination, which aims to recover the coal being lost to tailings. In coal
preparation, the justification for scavenging or cleaning will need to be based
on the extra capital and operating costs measured against the increase in coal
recovery of reduction in concentrate ash value.
Euston (2010a) provided a review of multiple circuits for coal and com-
ments that the concept has been revisited every 10 years since the original pub-
lication by Firth. Arnold (1999) describes the concept of a rougher-scavenger
circuit for coal as a “grab and run” concept. In the “grab” stage, gentle flota-
tion conditions are applied (low air rate, low impeller speed, low addition of
reagents) to recover the best coal. This stage focuses on grade or ash and might
be used to produce a coking coal product in a mixed plant. The second stage
of scavenging (“run”) is aimed at recovery with less stress on product grade or
ash. Flotation parameters such as air rate, froth depth, and so on, are optimized
for recovery rather than grade. The paper discusses the concepts of cleaning the
scavenger product and a classification stage prior to scavenging to discard the
high-ash ultrafines, both interesting concepts but a little beyond the scope of
the present paper. Continuing the 10-year tradition, more details can be found
in Euston (2010b).
To summarize:
• Effective flotation requires a consideration of the time required for
particles and bubbles to have sufficient multiple interactions for
attachment.
• The physiochemical processes of attachment to occur are based on the
product of several probabilities.
• Flotation conditions for coarser and finer particles are quite different.
• Optimum conditions for recovery and grade (ash) cannot be achieved
in the same unit operation.
I N T E R N AT I O N A L P E R S P E C T I V E S A N D D E V E L O P M E N T S
As discussed previously, the decision to install flotation for fine coal depends on
both economic and technological factors, with an element of corporate or even
personal preference about whether or not the value to be received is worth the
capital and ongoing operating costs. The decision on the choice of technology is
also based on corporate or individual preferences in addition to the coal quality
and the product specifications. In general, coals with high clay content in the
fines or if lower ash is required in the fines product may benefit from column-
type flotation, although this will normally be at the expense of recovery.
United States
In the Unites States, there is an increasing trend toward mechanical cells,
although, as mentioned previously, corporate preference plays a significant role.
Traditional square or “hog trough” designs dominate the market. These Wemco
Figure 1 Mechanical cells at Rosebud McVille, United States
machines are self-aspirated and as such are less operator intensive and produce
more consistent results without constant attention. Cell sizes are available up to
42.5 m3 (1,500 ft3) with a size range that allows for a bank of four to six cells in
a compact and efficient circuit. The majority of the cells in operation are 8.5–28
m3 (300–1,000 ft3). A typical flotation circuit will consist of one or more rows
of cells, with each row consisting of four to five cells in series. In general, a pulp
residence time within the separation vessel is in the range of 3–5 minutes. A
bank of cells can be seen in Figure 1.
For plants with a high ash or clay content, flotation columns are often used.
Recovery can often suffer, however, and both high recovery and low ash can be
achieved with a rougher-cleaner arrangement.
United Kingdom/Europe
The coal industry in Eastern Europe has declined in recent years. Coal flotation
plants in Eastern Europe tend to follow the U.S. philosophy of small, compact
banks of mechanical flotation cells with the focus on consistent operation with
minimum operator involvement. Figure 2 shows two typical banks of compact
14.2-m3 (500-ft3) Wemco flotation cells operating in South Wales.
Another popular concept in Europe is the self-contained flotation/filtra-
tion plant (see Figure 3). Dewatering of fine coal concentrate is best performed
with vacuum filtration. Horizontal belt filters are the industry preference, as
they provide consistent operation over a wide range of feed variability, both in
terms of quality and metric tons per hour.
Figure 2 Wemco 14.2-m3 (500-ft3) flotation cells installed in South Wales
Figure 3 A self-contained flotation and filtration plant
South Africa
In 2002, Opperman et al. (2002) reported that the Witbank coals of South
Africa were difficult to float with low yields essentially making the process
uneconomic. High energy and high reagent costs were the major sources of
problems. More recent private communications indicate that this attitude pre-
vails. Much South African coal is for power station use and the fines are often
used without processing.
Opperman et al. (2002) describe the Anglo Coal experience at Goede-
hoop colliery and their investigations of coal flotation options and results. The
authors compared the flotation response of Jameson and Wemco pilot test
units (see Table 3), concluding that although the mechanical cell gave the best
recovery and ash combination, both technologies had benefits. The Jameson
cell did not appear to provide sufficient energy in the downcomer to sufficiently
absorb the frother. The article describes in detail the development of the hybrid
Table 3 Comparison of flotation technologies at the Goedehoop colliery

Wemco Jameson Multicell
Test Cell Test Cell Production
Date 01/1997 to 12/1997 01/1998 to 12/1998 06/1999 to 12/2000
Feed solids, % 3.1 4.5 5.5
Product solids, % 9.4 11.0 14.2
Tailings solids, % 0.9 3.1 1.9
Feed ash, % 17.2 17.0 17.6
Product ash, % 9.8 8.0 8.5
Tailings ash, % 49.3 23.0 45.0
Product recovery, % 81.26 40.0 75.1
Combustibles recovery, % 88.53 44.34 83.4
multicell. The results are summarized in Table 3. The multicell has been further
refined as the dual cell, and a number of installations are currently being com-
missioned in South Africa. The dual-cell concept uses two cells per bank in a
rougher-scavenger combination.
Australia
Mechanical Flotation Stands the Test of Time
During the growth in the Australian coal industry in the 1980s, many of the
new, large coal preparation plants installed mechanical flotation technolo-
gies. These new generation plants with their high throughputs required large
numbers of mechanical cells. Figure 4 shows six banks of 14.2-m3 (500-ft3)
self-aspirated cells (then the largest available) installed at German Creek in
Queensland, Australia, in 1982. This installation has just been entirely refur-
bished with complete new mechanisms. Figure 4 shows the recently upgraded
flotation circuit at German Creek.
Rougher-Scavenger Flotation at Stratford Coal

Gloucester Coal operates the Stratford Coal Preparation Plant in New South
Wales, Australia. In common with many Jameson cell plants, limitations on
frother dose to 6–8 ppm resulted in lower recoveries than expected. Higher
frother doses resulted in many problems in the plant including buildup of frother
and fine coal in the circuit (Crisafulli and James 2008). In 2007, Stratford Coal
identified the potential to recover coal being lost to tailings, with the added
potential to remove excess frother from the circuit. A summary is presented in
Table 4.
Figure 4 Recently refurbished flotation cells at German Creek (Capcoal),

Queensland, Australia
Table 4 Test results from Stratford Coal

Jameson cell feed (air dried) ash including recycle 37.5%
Jameson cell product (air dried) ash 9.7%
Jameson cell tailings (air dried) ash 46.9%
SmartCell product (air dried) ash 16.0%
SmartCell tailings (air dried) ash 55.5%
Source: Crisafulli and James 2008.
A single 130-m3 (4,590-ft3) Wemco SmartCell was commissioned at

Stratford in October 2007. Wade (2008) reported an increased overall fine
coal yield of 2%, giving the mine an additional 60,000 metric tons per annum
of salable product. Installing the flotation cell made the whole processing and
disposal circuit much easier to manage and more profitable (Euston 2010a).
The results of a sampling program are presented in Figure 5. Jameson cell
yields were typically between 30% and 50% with the SmartCell contributing
an additional 15% to 55% depending on coal type. In addition to the direct
financial benefits of the extra coal, improved clarified water clarity was also
achieved. An important aspect of this concept is that the use of a rougher-
scavenger configuration allows the flotation circuit to be viewed holistically
rather than as separate unit operations. The ability of the scavengers to remove
excess frother from the plant circuit permits an increase in the frother dose to
the rougher cells (up to a point), a decreased load on the scavengers, and so on.
90.0
80.0 Smart Cell

J-Cell
70.0
60.0
50.0
Yield, %
40.0
30.0
20.0
10.0
0.0
1 2 3 BRN Coke + Duralie Duralie
Dirty Thermal Coke
Roseville
Coal Type
Source: Crisafulli and James 2008.
Figure 5 Summary of sampling program results
Figure 6 Tailings thickener at Stratford Coal without secondary flotation
Figure 6 shows the buildup of fine coal in the tailings thickener when not using
secondary flotation. Figure 7 shows the improved water clarity of the tailings
thickener when using secondary flotation.
Rougher-Cleaner at Rix’s Creek Coal

In a similar story to that of Stratford Coal, the Jameson cells installed at Rix’s
Creek coal suffered from the limitations common to this technology as a single
flotation stage, excess frother requirements, and poor combustibles recovery.
The approach here was somewhat different and has included several innovative
concepts. The installation comprises five 20-m3 (706-ft3) Wemco SmartCells,
each on a separate control level to give a very high level of process flexibility.
Figure 7 Tailings thickener at Stratford Coal with secondary flotation
The first three cells, shown in Figure 8 during construction, are used as rough-
ers. The other two cells can be configured as rougher or cleaner cells depending
on coal type. At the time of this writing, the cleaners are about to be commis-
sioned. The inherent flexibility of this circuit and the innovative approach of
the client will allow extensive on site research to be carried out.
China
China is currently the world’s largest coal producer with limited exports. A
search of flotation technologies in China reveals many local manufacturers, and
it is likely that these will dominate the local industry. Several types of flotation
cells are shown in Figure 9.
CONCLUSIONS
Froth flotation is currently the only economic and practical process for the
treatment of the finest coal fractions. In principle, froth flotation is a simple
concept, the attachment of bubbles to receptive particles and their subsequent
removal from the flotation device as a froth concentrate. In practice, however,
the process is technically complex, relying on a number of physicochemical and
hydrodynamic disciplines.
The need for residence time and the probabilities associated with the
recovery of values are best achieved in multiple reaction vessels and ideally
using separate circuits to achieve product grade and recovery. These require-
ments are well satisfied by the operation in traditional mechanical flotation
cells. The economics of treating the finest coal are often tenuous, and the use
of complex circuits is often not attractive, both financially and culturally. The
introduction of single-stage column and similar technologies to the coal indus-
try in the 1980s resulted in a rapid acceptance of these new technologies as they
promised simplicity and minimal operator involvement, albeit at the expense
of coal recovery.
Figure 8 Rougher flotation cells being installed at Rix’s Creek, Hunter Valley, New
South Wales, Australia
Figure 9 Selection of available Chinese flotation cells
However, mechanical flotation circuits continue to be recognized as the

technology most suited to the physical and chemical requirements of the flota-
tion process. Well-operated and optimized modern mechanical flotation cells
offer an efficient solution to the renewed demands for maximizing coal recov-
ery through their use of multiple cells in a bank and separate circuits to achieve
the maximum coal recovery at the required product ash.
REFERENCES
Arnold, B.J. 2000. The “grab and run” revisited—Improving selectivity between
organic and inorganic components in conventional coal flotation. Int. J. Miner.
Process. 58:119–128.
Crisafulli, P., and James, T. 2008. Secondary flotation at Stratford CHPP—A mechani-
cal cell processing Jameson cell tailings. In Proceedings of the 12th Australian Coal
Preparation Society Conference, Sydney. pp. 382–390.
Euston, J.A. 2010a. Two stage coal flotation using a mechanical cell. In XVI Interna-
tional Coal Preparation Congress, Proceedings. Edited by R. Honaker. Littleton,
CO: SME. pp. 382–390.
Euston, J.A. 2010b. Mechanical flotation cells in coal preparation—a technical review
and international perspective. In 13th International Coal Preparation Conference,
Mackay, Australia. pp. 339–351.
Firth, B.A., Swanson, A.R., and Nicol, S.K. 1979. Flotation circuits for poorly floating
coals. Int. J. Miner. Process. 5:321–334.
Heiser, N. 1996. Coal Flotation Technical Review—A Report on an ACARP Project.
Undertaken by Mike Williamson and Joe Sanders, The Australian Coal Review,
October.
Lynch, A.L., Harbort, G., and Nelson, M. 2010. Coal flotation. In History of Flotation.
Carlton, VIC: Australasian Institute of Mining and Metallurgy.
Opperman, S.N., Nebbe, D., and Power, D. 2002. Flotation at Goedehoop colliery. J. S.
Afr. Inst. Mining Metall. (October):405–410.
Sanders, G.J., and Williamson, G.J. 1996. Coal Flotation Technical Review C4047—An
ACARP Commissioned Study. February.
Tao, D. 2004. Role of bubble size in flotation of coarse and fine particles—A review.
Sep. Sci. Technol. 39(4):741–760.
Wade, W. 2008. Float cell breakthrough. Aust. Min. 10 March.
Gravity Separators for
Ultrafine Coal Cleaning
Rick Honaker
ABSTRACT
The application of density-based separators for upgrading ultrafine coal has been
studied and utilized in select applications. In some cases, conventional separators
such as the spiral concentrator have been commercially used to provide an effective
density-based separation for particles finer than 0.15 mm. Relative to the spiral
application, a higher level of efficiency and capacity may be realized by continuous
enhanced-gravity separators, which utilize fundamental density-based separation
principles to achieve ultrafine particle concentration in a centrifugal field. This
chapter reviews the processes available to achieve effective gravity separations for
ultrafine coal particles and their associated separation performances.
INTRODUCTION
The upgrading of coal below a particle size of 0.15 mm (100 mesh) is predomi-
nantly achieved using froth flotation in operating preparation plants world-
wide. The flotation process separates coal from the associated rock particles by
exploiting the natural surface hydrophobicity of the coal particles. However,
some coal sources are not effectively treated by froth flotation due to surface
oxidation. In these cases, collectors or other surface-modifying chemicals may
be applied to selectively generate a hydrophobic surface on the coal, which may
add significant operating cost. Another option is to utilize a density-based sepa-
rator with the capabilities to achieve an effective treatment of ultrafine coal.
Density-based separators may also be needed to separate pyrite from coal
prior to flotation. Several studies have reported that coal pyrite surfaces are
hydrophobic under certain flotation conditions, thereby reducing the effec-
tiveness of the flotation process in reducing the total sulfur content (Hurt
and Aplan 1991; Yoon et al. 1991). In fact, a few plant operators treating coal
containing moderate-to-high levels of pyritic sulfur choose to avoid flotation
227
because of unacceptable levels of sulfur in the flotation product. Other opera-

tors incur a significant coal loss as a result of the avoidance to recover the coal
that exists in the underflow stream of the sieve bends that are used to deslime
the spiral product. The sieve underflow stream is typically fed to the flotation
feed except when the pyrite content is high because of concentration in the clas-
sifying cyclone underflow. Figure 1 shows a fine coal cleaning circuit that could
be applied to a coal source with high pyrite concentrations. In this circuit, an
ultrafine density-based separator such as an enhanced-gravity separator (EGS)
is used to remove the pyrite and fine high-density rock particles prior to the
flotation process.
The application of conventional fine gravity separators for cleaning coal
particles smaller than 0.15 mm may be achievable by adjusting operating condi-
tions and parametric values. Spiral concentrators and fluidized-bed separators
have the potential to be used for this purpose. Alternatively, several continuous
EGSs have been developed over the past two decades that are commercially
used for heavy metals recovery and have been evaluated for the potential appli-
cation of upgrading ultrafine coal. Commercially available units include the
multi-gravity separator (MGS), Kelsey jig, Knelson concentrator, and Falcon
concentrator. Highly efficient rejections of both ash and sulfur from fine coal
have been reported for the Knelson concentrator (Paul et al. 1993) and the
Kelsey jig (Riley and Firth 1993). In comparison with single-stage treatment
using column flotation, Venkatraman et al. (1995) found significantly larger
reductions in both ash and total sulfur by combining column flotation with
the MGS unit. Exceptional separation performances by the MGS were also
reported by other research groups (Menéndez et al. 2007; Majunder et al.
2007). A few of the ultrafine gravity separation options will be discussed in the
following sections.
S P I R A L C O N C E N T R AT O R
The separator most commonly used for cleaning the 1 × 0.15 mm particle
size fraction is the spiral concentrator because of its operational simplicity
and relatively low capital and operating cost. Separation efficiency has been
significantly improved as a result of focused research efforts that led to the
development of the compound spiral. The typical spiral concentrator receives
feed from the underflow of a bank of raw coal classifying cyclones after being
diluted to a desired solids concentration of around 30% by weight. With mass
solids feed rates of around 2.5 tph per spiral start, highly effective specific grav-
ity cutpoints as low as 1.70 can be realized for the majority of the particles in
Gravity Separators for Ultrafine Coal Cleaning 229
–0.15 mm
–1 mm
Feed
Classifying
Cyclone
1 × 0.15 mm
Enhanced Flotation Thickener

Spiral Gravity Bank
Separator
Tailings
Sieve-bend Screen-bowl
Product
–0.15 mm
Thickener
Figure 1 Fine coal flowsheet utilizing an ultrafine gravity separation process to

concentrate pyrite and rock particles prior to froth flotation
the feed stream. However, performance deteriorates substantially for particle

sizes below 0.30 mm.
According to Richards et al. (2000), it is possible to adjust the feed charac-
teristics and operating parameters of the spiral concentrator to achieve an effec-
tive density-based separation for particles smaller than 0.1 mm. Honaker et al.
(2007) followed this observation with a detailed study to evaluate the potential
of cleaning ultrafine coal (<0.21 mm particle size) using spiral concentrators.
The study involved a parametric evaluation from which the data were used to
develop empirical models to describe the separation performance as a function
of the feed solids concentration, feed volumetric flow rate, and splitter posi-
tion. The set of parameter values needed to achieve optimum performance was
identified using empirical models.
Separation performance values achieved under conditions that maximize
yield over a range of product ash values are shown in Table 1. Notably, the
performances were achieved on material that was not deslimed and, thus,
contained a significant amount of material smaller than 0.044 mm. Removal
or even partial removal of the –0.044 mm material prior to treatment could
Table 1 Process performance and efficiency data achieved from cleaning 0.21 ×
0.044 mm Coalberg seam coal under optimum conditions in a spiral concentrator
Efficiency Parameters Test 1 Test 2 Test 3 Test 4 Test 5
Feed ash, % 31.85 36.82 35.26 36.04 38.45
Product ash, % 8.04 11.71 13.30 16.14 19.73
Tailings ash, % 69.10 74.92 85.61 90.51 91.10
Mass yield, % 50.19 60.28 69.63 73.24 83.02
Combustible recovery, % 74.99 84.23 93.25 96.03 97.78
Ash rejection, % 89.50 80.83 73.73 67.21 48.58
Separation efficiency, % 64.49 65.06 66.98 63.24 46.36
Separation density, d50 1.65 1.89 2.06 2.17 2.25
Probable error, Ep 0.22 0.23 0.22 0.13 0.13
High-density bypass, % 5.03 15.05 19.95 27.55 42.36
Low-density bypass, % 16.93 9.77 2.00 0 0
Organic efficiency, % 78.11 86.88 96.13 98.49 99.78
significantly improve performance. The samples taken from all process streams
were screened to remove the particles smaller than 0.044 mm. As such, the data
in Table 1 represent the assays obtained from the 0.21 × 0.044 mm (65 × 325
mesh) particle size fraction.
Product ash values ranging from 8.04% to 19.73% were achieved from 0.21
× 0.044 mm Coalberg seam coal. The ash value of the feed coal was greater than
30%. To achieve the lower product ash values, a substantial amount of low-
density material was bypassed to reject, which resulted in relatively low organic
efficiency values. Likewise, a large amount of high-density, high-ash value par-
ticles were bypassed to the product when targeting relatively high product ash
values. The most desirable performance corresponded to a product ash value of
13.30%, which provided an organic efficiency of 96%.
To achieve the desired performance, the mass throughput capacity was
required to be reduced to about 0.5 tph/start from a typical value of about
3 tph/start. This finding reflects (1) the impacts of increased particle popu-
lation with a reduction in particle size and (2) the need for greater particle
retention time. Similar findings were reported by Benusa and Klima (2008),
although the stated impacts of feed solids and volumetric flow were less signifi-
cant. The optimum volumetric feed rate was approximately 50 L/m (13 gpm)
at a feed solids concentration of 15% by weight.
An in-plant study was performed to evaluate the separation performance
of an SX7 Multotec spiral when applied to treat the underflow of a second-
ary 15-cm-diameter classifying cyclone at a preparation plant treating Illinois
Table 2 In-plant separation performance achieved by SX7 spiral when cleaning

nominally –0.15 mm Illinois No. 6 coal*
Feed Product Tailings
Particle Size
Fraction, mm Ash, % Sulfur, % Ash, % Sulfur, % Ash, % Sulfur, %
1.00 × 0.15 8.44 2.64 4.67 2.45 36.87 4.38
0.15 × 0.044 19.33 3.37 11.80 2.64 56.33 5.86
–0.044 53.74 5.77 45.66 4.44 68.68 9.91
% of Feed 100.00 65.08 13.03
*Middling stream assays and weights not provided.
No. 6 seam coal (USA). The nominal 0.15 × 0.044 mm spiral feed contained
about 12.0% solids by weight. The volumetric and solid mass feed flow rates
were 70 L/m (18 gpm) and 0.6 tph, respectively. As shown in Table 2, signifi-
cant ash and total sulfur reductions were achieved for all size fractions in the
feed. For the +0.044 mm coal, 60% of the ash-forming minerals were rejected
while 48.3% rejection of ash was obtained for the overall coal. Likewise, 47.6%
total sulfur rejection was achieved. To achieve an acceptable final product, the
0.044 mm fraction will need to be removed by classification or screening.
C E N T R I F U G A L F L O W I N G - F I L M C O N C E N T R AT O R S
Particles within a thin film of water moving across a solid surface segregate
according to density due to a differential streamline velocity profile. High-
density particles settle onto the solid surface, whereas light particles remain in
the high-velocity streamlines located a distance away from the surface. Spiral
concentrators and riffle tables are examples of flowing-film separators that are
commonly used for particles coarser than 0.15 mm. Continuous centrifugal
concentrators utilizing the flowing-film principle have been developed to allow
density-based separations for ultrafine particles. Commercial units include the
Falcon concentrator and the MGS.
Falcon Concentrator
The essential feature of a continuous Falcon concentrator is a vertically aligned,
open-topped cylindrical bowl that is mounted on a rotating shaft, as shown in
Figure 2. A centrifugal force up to 300 g’s can be produced to cause deposition
and stratification of the fine particles against the inside of a smooth centrifugal
bowl. The coal slurry is continuously introduced at the bottom of the spinning
bowl by means of a conduit extending downward along the axis of rotation. The
feed slurry is then impelled to the wall of the bowl by an impeller, which causes
Feed
Rinse Water
Heavies Heavies
Dilution Dilution
Water Water
Lights Outlet
Heavies Heavies
Outlet Outlet
Rotor Bowl
Cleanout
Outlet
Compressed Air Supply

(Introduced Through Rotating Union)
Courtesy of Sepro Mineral Systems.
Figure 2 Continuous Falcon concentrator
stratification along the inclined lower section of the bowl, called the migra-
tion zone, due to differential acceleration. In this zone, the enhanced-gravity
field is resolved into two force components. The strong component normal to
the wall is the concentrating gravity field that provides the strong g-forces for
the hindered-settling processes and density stratification of the particles. The
weak driving component parallel to the bowl wall pushes the stratified solids
up toward the top of the bowl. An overflow lip that has an internal diameter
less than the bowl diameter restricts the particle bed from reporting to the
overflow. The combination of the lack of a vertical force component and the
presence of an overflow lip causes the heavy pyrite and ash-bearing particles to
come to rest while the centrifugal force assists the heavy particles into a slot that
exists around the circumference of the bowl. Mass transport chutes and pinch
valve-nozzle assemblies placed at equal distances in the slot allow discharge of
the heavy particles into an underflow launder. At the same time, light particles
forming the particle bed furthest from the bowl wall move upward and over the
Table 3 Total sulfur reductions achieved on a particle size-by-size basis from the
treatment of –0.60 mm Pittsburgh No. 8 coal by C40 Falcon concentrator
Total Sulfur, %
Particle Size Combustible
Fraction, mm Feed Product Tails Recovery, %
+0.60 3.21 1.92 12.30 90.8
0.60 × 0.30 3.32 1.67 12.60 92.7
0.30 × 0.212 4.26 1.71 9.92 92.9
0.212 × 0.150 3.06 1.77 13.70 93.6
0.150 × 0.075 3.37 1.77 22.20 93.7
0.075 × 0.037 6.52 3.12 20.50 94.4
overflow lip of the bowl. These light particles report as final product with other
particles (i.e., colloidal clay) that are too fine to be affected by the enhanced
gravitational force.
Water-Only Separations
The ability of the Falcon concentrator to effectively clean coal to a particle
size as small as 0.037 mm has been extensively studied using a 25-cm-diameter
pilot-scale unit (Honaker et al. 1996) and a 1-meter-diameter full-scale unit
(Honaker 1999). Performance results from the treatment of a relatively large
number of coal samples indicate an exceptional ability to reduce sulfur content
in the ultrafine fractions while also achieving effective ash reductions. Using
the full-scale unit at a mass flow rate of 78 tph, nearly 75% of the ash-bearing
material and 60% of the sulfur from a Pittsburgh No. 8 seam coal were removed
while recovering 80% of the mass yield to the product. Table 3 shows the total
sulfur reductions realized on a particle size-by-size basis. Organic efficiency
(i.e., ratio of actual recovery over theoretical recovery) was reported to be about
96%.
Dense-Medium Separations
Research findings reported by Honaker and Patil (2001) indicate that the
separation performances provided by an enhanced-gravity concentrator can
be significantly enhanced using a dense medium. The medium was formed
using magnetite in which 90% of the material was smaller than 0.017 mm.
A test program was performed in which the operating parameter values were
significantly varied to obtain the data needed to develop the empirical models
used to optimize their values. The optimized performances shown in Figure 3
were obtained by varying the medium specific gravity from 1.35, to realize the
lowest product ash value, to 1.50 to produce the highest product ash value. The
100
80
60
Clean Coal Yield, %
40
20
Washability
Dense-Medium Falcon
Optimized Dense-Medium Falcon
0
0 5 10 15 20 25
Product Ash, %
Figure 3 Separation performances based on product ash value achieved using

magnetite-based dense medium in the Falcon concentrator on the 0.60 × 0.044 mm
particle size fraction of a Pittsburgh No. 8 seam coal sample (feed ash = 22.1%)
relatively small difference between the mass yield values from the optimized
tests and the theoretical values represented by the washability curve indicates a
very high level of efficiency over the entire range of product ash values.
C E N T R I F U G A L F L U I D I Z E D - B E D C O N C E N T R AT O R
The use of fluidized-bed units for achieving effective density-based separations
has historically been a common practice within the coal industry. The Chance
cone is an example in which an upward current of water is used to suspend sand
particles and create a medium through which the high-density, high-ash parti-
cles settle. Within the last few decades, fluidized-bed separators, which are also
known as teeter beds, have been used as an alternative to spiral concentrators
for upgrading coal within the particle size range of 1 × 0.15 mm. In this applica-
tion, the fluidization water is used to suspend high-density, high-ash particles
entering in the feed and, as a result, create an autogenous medium. As a result,
separation efficiencies achieved over a particle size range of around 4:1 are
exceptional. However, performances deteriorate significantly with a reduction
Feed
Overflow Overflow
Product Product
Underflow Underflow
Tailings Tailings
Elutriation Water
Figure 4 Knelson CVD concentrator
in particle size. The Knelson concentrator utilizes fluidized-bed principles in a

centrifugal field to effectively treat particles finer than 0.15 mm.
Knelson Concentrator
The Knelson continuous variable discharge (CVD) concentrator employs flu-
idized particle bed separation concepts in a mechanically applied centrifugal
field. As shown in Figure 4, the Knelson concentrator operates by introducing
water through a series of fluidization holes located in the rings that circle the
circumference of a bowl. The bowl, which has a truncated cone shape, is rotated
at speeds that provide a centrifugal field up to 200 times gravity. Feed slurry is
introduced through a tube that directs the material toward the bottom cen-
ter. Upon reaching the bottom, the slurry is driven outward and up the cone
wall toward the rings. The fluidization water entering in the rings provides an
inward velocity that allows the creation of a fluidized particle bed comprised of
heavy particles. The high-density particles that pass through the fluidized par-
ticle bed are extracted through a series of controlled pinch valves located along
the circumference in the center of a concentration ring. The discharge ports are
designed to handle mass flows equivalent to 1%–50% of the feed.
A pilot-scale unit of the Knelson CVD concentrator with a maximum
throughput capacity of 5 tph was evaluated in an operating coal preparation
plant treating both steam and metallurgical coal sources (Honaker and Das
2004; Honaker et al. 2005). The feed to the unit was provided from the under-
flow of a bank of 15-cm-diameter desliming classifying cyclones. Approxi-
mately 40% of the coal had a particle size between 0.15 mm and 0.044 mm and
an equal amount was finer than 0.044 mm. In an effort to improve recovery, air
100
90
80
70
60
Recovery, %
50
40
30 Coalberg Coal
20 Washability
Experimental
10 Release
0
0 4 8 12 16 20 24
Product Ash, %
Figure 5 Comparison of separation performance achieved by Knelson concentrator

on Coalberg coal with those obtained by washability and flotation release analysis
bubbles were injected into the feed stream to generate particle–bubble aggre-
gates that have a lower density than the individual coal particles.
The performance data shown in Figure 5 is an example of the ability to
enhance the separation performance of an ultrafine coal using EGS. The steam
coal (Coalberg seam, United States) was a poorly floating material and the
selectivity achieved by the flotation process was poor, as indicated by the release
data. The separation performances achieved by the Knelson CVD unit were
superior to the froth flotation performance represented by the release curve.
The optimum separation performances shown in Table 4 indicate that the
Knelson unit reduced the ash value from 17.8% to 10.1% while recovering 81%
of the combustibles. Injecting air bubbles into the feed increased the recovery
by 13 absolute percentage points. The relative separation specific gravity values
varied from 1.3 to 1.8 with corresponding probable error values around 0.20.
Significant bypass (low- and/or high-density particles) occurred in all tests.
CENTRIFUGAL JIG
Commercially available centrifugal jig units include the Kelsey and Altair jigs.
The Altair jig consists of a rotating bowl, which is placed inside a static casing
having separate launders for collecting the concentrate and tailing samples
Table 4 Comparison of optimum separation performance results achieved on

metallurgical coal coarser than 0.044 mm with and without air injection
Ash, %
Test Number Feed Product Tailings Recovery, %
No Air
1 16.90 5.28 20.31 25.86
2 17.65 6.88 35.70 70.81
3 17.84 10.07 39.64 80.68
4 17.07 12.05 43.72 89.25
Air Injection
1 20.59 4.93 46.14 74.24
2 20.84 6.95 57.64 85.33
3 21.11 9.76 74.30 94.27
4 20.47 11.82 73.82 95.40
(Figure 6). The rotating bowl contains a cylindrical screen with a lip, whose
height can be adjusted to vary the natural depth of the ragging bed, which
remains in a vertical position on the screen due to bowl rotation. The feed
slurry, which is introduced from the top at the center of the rotating bowl, is
distributed into the ragging bed on the screen by the diffuser plate placed under
the feed inlet. Pressurized water is periodically injected under the bed through
four pulse-blocks to cause alternating dilation and contraction of the ragging
and feed bed. This, coupled with the high centrifugal force generated by the
rotation of the bowl, provides a hindered-settling environment required for
the jigging to occur, which results in physical separation of particles of varying
density. The tailings material settles through the ragging bed and screen into
the hutch and reports to the tailings launder through the discharge ports. The
clean coal particles do not settle through screen and, thus, report to the con-
centrate launder.
Tests conducted on Illinois No. 5 coal containing particles finer than 1 mm
reduced the ash value from 28.00% to 6.84% while achieving a recovery of 82%
(Mohanty et al. 1999, 2002). Ash and total sulfur reduction averaged about
85% and 42%, respectively. The separation performances shown in Figure 7
indicate that tests performed with no ragging provided inferior results. When
granite ragging material was used, performance substantially improved. The
results indicate some loss of coal to the tailings stream. The specific gravity cut-
point ranged between 1.43 and 1.54 with probable error values varing between
0.08 and 0.17.
Figure 6 Altair centrifugal jig
100
80
Combustible Recovery, %
60
40
20
Feed Washability
Ragging Tests
No Ragging Tests
0
0 20 40 60 80 100
Ash Rejection, %
Figure 7 Separation performance achieved by the Altair centrifugal jig when

treating Illinois No. 5 seam coal finer than 1 mm
CONCLUSIONS
General applications of density-based separations for treating ultrafine coal
(i.e., particles finer than 0.15 mm) include coal sources that have difficult-
to-float characteristics and those that contain a significant amount of pyritic
sulfur. Spiral concentrators have been evaluated for these applications and
implemented into a few operating preparation plants. Available plant data
indicate the ability to significantly reduce the ash value and total sulfur content.
However, the capacity reporting to each spiral start is significantly reduced

from the level of 3.0 tph/start used to treat 1.0 × 0.15 mm coal to 0.5 tph/start
when treating material finer than 0.15 mm.
The use of enhanced-gravity separators may be needed to achieve high-
capacity, density-based separations for ultrafine coal. Several EGS units have
been developed and evaluated at a scale up from 5 to 75 tph. The separation
performances achieved on particles between 1 mm and 0.044 mm indicate the
ability to achieve relatively efficient separations (Ep values ranging from 0.08
to 0.17) over a range of specific gravity cutpoints from 1.50 to 2.00. Bypass
of both low- and high-density particles is a significant issue for all EGS units.
In addition, desliming of the product is needed to achieve the desired clean
coal specifications, and the availability of an efficient, cost-effective method to
achieve high-capacity removal of colloidal particles is limited.
REFERENCES
Benusa, M.D., and Klima, M.S. 2008. An evaluation of a two-stage spiral processing
ultrafine bituminous coal. Int. J. Coal Prep. Util. 28(4):237–260.
Hirt, W.C., and Aplan, F.F. 1991. The influence of operating factors on coal recovery
and pyritic sulfur rejection during coal flotation. In Processing and Utilization of
High Sulfur Coals. 4th ed. Edited by P.R. Dugan, D.R. Quigley, and Y.A. Attia.
Amsterdam: Elsevier Science Publishers. pp. 339–356.
Honaker, R.Q. 1998. High capacity fine coal cleaning using an enhanced gravity con-
centrator. Miner. Eng. 11(12):1191–1199.
Honaker, R.H., and Das, A. 2004. Ultrafine coal cleaning using a centrifugal fluidized-
bed separator. Coal Prep. 24(1-2):1–18.
Honaker, R.Q., and Patil, D.P. 2001. Parametric evaluation of a dense-medium process
using an enhanced gravity concentrator. Coal Prep. 22(1):1–17.
Honaker, R.Q., Wang, D., and Ho, K. 1996. Application of the Falcon concentrator for
fine coal cleaning. Miner. Eng. 9(11):1143–1156.
Honaker, R.Q., Das, A., and Nombe, M. 2005. Improving the separation efficiency of
the Knelson concentrator using air injection. Coal Prep. 25(2):99–116.
Honaker, R.Q., Jain, M., and Saracoglu, M. 2007. Ultrafine coal cleaning using spiral
concentrators. Miner. Eng. 20(14):1315–1319.
Majumder, A.K., Bhoi, K.S., and Barnwal, J.P. 2007. Multi-gravity separator: An
alternative gravity concentrator to process coal fines. Miner. Metallurg. Process.
24(3):133–138.
Menéndez, M., Gent, M., Torano, J., and Diego, I. 2007. Optimization of multigravity
separation for recovery of ultrafine coal. Miner. Metallurg. Process. 24(4):253–263.
Mohanty, M.K., and Honaker, R.Q. 1999. Evaluation of the Altair centrifugal jig for
fine particle separations. Coal Prep. 20(1):85–106.
Mohanty, M.K., Honaker, R.Q., and Patwardhan, A. 2002. In-plant evaluation of the
Altair centrifugal jig for fine coal cleaning. Miner. Eng. 15:157–166.
Paul, B.C., and Honaker, R.Q. 1993. Production of Illinois Basin Compliance Coal Using
Enhanced Gravity Separation. Final Technical Report. Report Number 93-1/5.1B-
1P. Illinois Clean Coal Institute.
Richards, R.G., MacHunter, D.M., Gates, P.J., and Palmer, M.K. 2000. Gravity
separation of ultra-fine (–0.1 mm) minerals using spiral separators. Miner. Eng.
13(1):65–77.
Riley, D.M., and Firth, B.A. 1993. Application of an enhanced gravity separator for
cleaning fine coal. In Proceedings, 10th International Coal Preparation Conference,
Lexington, KY. pp. 46–65.
Venkatraman, P., Luttrell, G.H., and Yoon, R.H. 1995. Fine coal cleaning using the
multi-gravity separator. In Proceedings of the High Efficiency Coal Preparation:
An International Symposium. Edited by S.K. Kawatra. Littleton, CO: SME. pp.
109–117.
Yoon, R.H., Lagno, M., and Luttrell, G.H. 1991. On the hydrophobicity of coal pyrite.
In Processing and Utilization of High Sulfur Coals. 4th ed. Edited by P.R. Dugan,
D.R. Quigley, and Y.A. Attia. Amsterdam: Elsevier Science Publishers. p. 241.
Design and Operating
Guidelines for Combined
Water-Only Cyclone and Spiral
Circuits
Gerald Luttrell, Zulfiqar Ali, Andy Dynys,
Larry Watters, and Robert Moorhead
ABSTRACT
The methods used to treat coarse and fine coal streams in modern coal preparation
plants have become largely standardized. For optimum efficiency, coal particles
larger than about 1 mm are upgraded using dense-medium separators, whereas
particles smaller than about 0.15 mm (100 mesh) are cleaned using froth flota-
tion. No single standard method has yet emerged as the most efficient option for
treating run-of-mine feeds in the size range between 1 mm and 0.15 mm. From
a historical perspective, gravity-based separators such as water-only cyclones have
been a popular commercial choice for treating this intermediate size fraction. In
recent years, however, the emergence of a new generation of two-stage compound
spirals has begun to largely displace water-only cyclones, generally because of the
higher efficiency offered by this multi-stage cleaning device. Despite this fact, several
recent studies have suggested that other gravity-based separators, such as water-
only cyclones, can be effectively incorporated into multi-stage circuits to improve
capacity, lower costs, and improve separation efficiency. This chapter discusses
application guidelines for configuring such circuitry and offers recommendations
for optimizing overall circuit performance.
INTRODUCTION
Modern coal preparation facilities incorporate a wide variety of solid–solid
separation processes for coal upgrading. Dense-medium processes, which
241
DM Vessel
Coarse Jig
DM Cyclones
WOC
Spirals
HydroFloat
Reflux Classifier
Teeter Bed
Fine Spirals
Froth Flotation
0.01 0.1 1 10 100

Particle Diameter, mm
Note: DM = dense-medium; WOC = water-only cyclone.
Figure 1 Approximate effective size ranges for different coal cleaning processes
include dense-medium vessels and dense-medium cyclones, have become the

preferred method for treating coarse coal in most new plants (see Figure 1).
The widespread acceptance of dense-medium technology can be attributed
to its large capacity, high efficiency, and operational flexibility. In contrast, a
variety of commercial flowsheet configurations exist for treating coal feeds
in the intermediate size range of 1 × 0.15 mm (Osborne 1988). These circuit
configurations may include various combinations of conventional water-based
density separators such as spirals and water-only cyclones as well as several types
of next-generation hydraulic classifiers. In many cases, the separation processes
are used in multi-stage circuits and integrated with various types of classifica-
tion processes (either before or after the cleaning step) in an attempt to improve
cleaning performance (Bethell and Arnold 2003). Differences in circuit layouts
are typically justified by flowsheet designers based on both technical and finan-
cial considerations, which attempt to balance the need to accommodate a spe-
cific raw-coal size distribution or washability against any undesirable increase
in capital, operating, or maintenance costs. Operator preferences and vendor
Combined Water-Only Cyclone and Spiral Circuits 243
Figure 2 Bank of industrial water-only cyclones
biases also appear to contribute to the large variations that are observed in how
fine coal is cleaned.
Two of the most popular methods for treating fine coal in the 1 × 0.15 mm
size range are water-only cyclones (WOCs) and spirals. A WOC is similar to
a classifying cyclone, but typically has a stubby, wide-angled conical bottom
(Figure 2). Separations occur in a WOC because of differences in the settling
rates of coal and rock in the centrifugal field within the cyclone. The separa-
tion does not utilize any external medium such as magnetite but is enhanced
by the formation of autogenous medium created by the natural fines already
in the feed slurry. WOCs typically utilize a truncated cone bottom and wider
included angle (usually 60–120°) that promotes the formation of a refuse bed
which prevents lighter coal particles from reporting to underflow. WOCs are
also typically operated with a relatively long vortex finder that “vacuums up”
the light coal particles off the top of the bed of autogenous medium. These
units are often employed in two stages or in combination with other water-
based separators to improve performance.
A spiral consists of a corkscrew-shaped conduit with a modified semicir-
cular cross section (Figure 3). During operation, feed slurry is introduced to
the top of the spiral and is permitted to flow by gravity along the helical path
to the bottom of the unit. Particles in the flowing film are stratified such that
lighter coal particles are forced to the outer wall of the spiral, whereas heavier
particles are forced inward to the center of the spiral. The segregated bands of
heavy to light materials are collected at the bottom of the spiral. Adjustable
diverters (called splitters) are used to control the proportions of particles that
report to the various products. A three-product split is usually produced, giv-
ing rise to three primary products containing clean coal product, refuse, and
misplaced “middlings.” Because of the low unit capacity, spirals are usually
arranged in groups or banks that are fed by an overhead radial distributor. To
save space, several spirals (two or three) may be intertwined along a single cen-
tral axis. Modern coal spirals typically incorporate two stages of cleaning along
a single support column to reduce misplacement. Typically, the first four (or
three) spiral turns are used to produce a throw-away reject product, after which
the clean coal and middlings products are remixed in a pulping box and then
re-cleaned using three (or four) additional spiral turns. The products from the
second stage include final clean coal, final reject, and middlings that can be sent
to clean coal, reject, or recycled back to be reprocessed. In most cases, recycling
of middlings is generally recommended to improve the overall sharpness of the
separation (Luttrell et al. 1998; Bethell and Arnold 2003).
O P E R AT I N G A N D D E S I G N PA R A M E T E R S
Water-Only Cyclones
The basic working features of a WOC are relatively simple, but numerous design
and operating variables can influence its performance. The most imporatant of
these variables are (1) the length of the vortex finder, (2) the included angle of
the truncated cone and apex, and (3) the apex-orifice diameter. Increasing the
length of the vortex finder extends the entry of the vortex finder into a region of
higher-density particles, rotating in the lower region of the cyclone. As such, a
longer vortex finder increases the specific gravity of separation (SG50). Increas-
ing the included angle of the truncated cone and/or the apex as well as decreas-
ing the apex-orifice diameter also increases the SG50 of the cyclone. All of these
adjustments increase the retention of particles in the autogenous zone of the
cyclone, thus making it more difficult for lower-density particles to penetrate
this zone, and, therefore, they are re-directed upward toward the vortex finder.
In addition, decreasing the apex diameter also increases the volumetric split
of the cyclone to the overflow. This results in greater drag forces, which pro-
mote directing a greater percentage of particles to the overflow. Generally, the
Low SG
Coal
High SG
Rock
Refuse Midds Clean
Figure 3 Bank of industrial compound coal spirals
apex diameter should be in the range of 50%–60% of vortex-finder diameter.

Increasing the vortex-finder diameter and the feed-inlet area also increases the
SG50, but to a lesser degree than changing vortex-finder length, apex diameter,
or cone angles. Both of these changes increase the throughtput capacity and
reduce the particle retention time.
Obviously, cyclone diameter is also an important design variable since this
dimension controls the relative intensity of the gravitational field within the
cyclone. Finer coals typically have to be treated in smaller-diameter cyclones
to prevent fluid drag forces from carrying unwanted higher ash fines into the
clean coal overflow. Therefore, most 1 × 0.15 mm coal applications utilize
38-cm- (15-in.-) diameter WOCs. Finally, the most important operating vari-
able directly under the control of the plant operator is the feed solids content.
An increase in the concentration of feed solids rapidly increases the SG50 and
leads to lower separation efficiency if not properly controlled. Ideally, the feed
concentration should be maintained within the range of 10%–15% solids by
weight when treating 1 × 0.15 mm feeds. Operation at the lower end of this
percent solids range will typically provide lower SG50 cutpoints. Likewise,
an operating pressure in the range of 69–103 kPa (10–15 psi) is normally
recommended to maintain good separation efficiencies for 1 × 0.15 mm feed

coals, with lower pressures recommended to achieve lower SG cutpoints.
Spirals
Spirals are capable of maintaining good recoveries of clean coal when properly
operated (Luttrell et al. 2007). One of the most important operating variables
is feed rate. Ideally, spirals should be configured to operate at a dry solids feed
rate of about 2.3 t/h (2.5 ton/h) per spiral start (Li et al. 1993; Subasinghe
et al. 1991, 1992; Holland-Batt 1994). The SG cutpoint increases rapidly as
the feed tonnage exceeds this value, making it difficult to maintain an SG
cutpoint that would best optimize those in the plant dense-medium circuits.
The Ep value also generally diminishes as the tonnage rate increases. Spirals
must be provided with an adequate and stable slurry flow rate of approximately
7–8 m3/h (35–40 gpm) to work properly. Too little flow can result in sluggish
movement of solids along the interior surface of the spiral that can eventually
lead to beaching/sanding, whereas too much flow can cause high-density rock
to report inadvertently to the low-density stream. A well-designed and properly
maintained feed distributor is essential to ensure that each spiral receives the
sample feed rate to avoid different SG50 cutpoints that would adversely impact
performance. Typically, the slurry level in the distributor should be maintained
at 30–45 cm (12–18 in.) to maintain the head necessary to obtain consistent
flow rates from the discharge ports. When the proper flows have been set, spiral
performance can be fine-tuned using the splitter positions. The degree to which
splitter positions influence the SG50 cutpoint is highly dependent on particle
size and washability characteristics (Mikhail 1988; King et al. 1992).
S E PA R AT I O N P E R F O R M A N C E
A typical set of size-by-size partition curves for a bank of industrial WOCs
are shown in Figure 4. Since the WOC technology separates particles based
on differences in effective mass, the performance of the process is affected by
particle size as well as particle shape (Bull et al. 1987). As a result, the expected
density cutpoint (SG50) varies considerably for each size fraction treated by the
process (i.e., finer particles are separated at much higher densities than coarser
particles). Very fine particles have a low mass and tend to be entrained into the
overflow stream by drag forces, regardless of density. Very large particles have
a high mass and tend to preferentially report to the underflow, even if they are
of low density. This characteristic drift in SG cutpoint is plotted in Figure 5 for
several sets of industrial WOC circuits. The decline in SG50 with increasing
particle size is undesirable because it results in an overall partition curve that is
1
1.7 × 1.0 mm
1.0 × 0.6 mm
0.9 0.6 × 0.25 mm
0.25 × 0.15 mm
0.8
0.7
0.6
Partition Factor
0.5
0.4
0.3
0.2
0.1
0
1.2 1.4 1.6 1.8 2 2.2 2.4
Specific Gravity
Figure 4 Typical size-by-size partition curves for a water-only cyclone
substantially less efficient than that obtained for any of the individual size frac-
tions (Abbott 1981; Luttrell et al. 2000). This phenomenon forces operators
to select between operating conditions that either (1) throw away a significant
portion of low-density coal particles in order to produce an acceptable clean
coal ash in the finer size fractions or (2) recover the coarser coal particles and
tolerate a relatively high ash content in the finer size fractions. Although nei-
ther of these options is ideal, the WOC system does provide a variable cleaning
option when targeting relatively low-density SG cutpoints for particles in the
coarser size fractions. In fact, the Ep values obtained in the coarser size fractions
are typically quite low and provide relatively efficient separations (Schlepp and
Schmidt 1988).
Figure 6 provides a typical set of partition curves for a modern spiral sepa-
rator. Much like WOCs, fine particles having a low mass tend to be hydrauli-
cally carried by the fluid drag forces into the clean coal product regardless of
their density. Since the bulk of the water flow reports to clean coal, so too does
the vast majority of the fine particles. Consequently, the density cutpoint for
2.7
WOC “A”
WOC “B”
WOC “C”
2.5
2.3
2.1
SG Cutpoint
1.9
1.7
1.5
1.3
0.1 1 10
Particle Size, mm
Figure 5 Effect of particle size on density cutpoint for three industrial installations
of water-only cyclones
spirals tends to increase as the particle size increases from the finest grain size
treated up to a critical particle diameter of approximately 0.4–0.6 mm. Above
this critical size, larger particles in the flowing film passing down the spiral
begin to experience body forces that act in direct proportion to their particle
size. These body forces effectively push the larger particles along in the flowing
liquid that reports to the clean coal product. Since the body forces depend on
size (and shape) and are independent of density, the density cutpoint for the
largest particles actually tends to increase as the particle size increases further.
The net effect of the combination of the settling and drag forces is a partition
curve that is minimum in the 0.4–0.6 mm size range and increases for larger
or smaller particles. A summary plot illustrating this phenomenon for serveral
industrial spiral operations is shown in Figure 7. The flattening of the SG50
versus particle size curve is actually desirable because it produces a composite
partition curve for the entire size range that is better than if a wider drift of
SG50 values had occurred. This inherent advantage is a major contributing
0.9
0.8
0.7
0.6
Partition Factor
0.5
0.4
0.3
0.2
1.7 × 1.0 mm
1.0 × 0.6 mm
0.1 0.6 × 0.25 mm
0.25 × 0.15 mm
0
1.2 1.4 1.6 1.8 2 2.2 2.4
Specific Gravity
Figure 6 Typical size-by-size partition curves for a coal spiral
factor to the historical shift in operator preferences from WOCs to spirals

over the last several decades in the efficiency-driven coal preparation industry.
Unfortunately, spirals also suffer from a major disadvantage in that low-density
cutpoints below about 1.60–1.65 SG are generally not attainable using current
spiral technology due to the intrinstic nature of flowing-film separators. Such
a capability would be highly desirable, especially in cases involving the upgrad-
ing of metallurgical coals that typically have more rigorous ash contraints that
require lower-density cutpoints.
COMBINED CIRCUITRY
One interesting approach for dealing with the inherent limitations of WOCs
and spiral separators is to integrate these two unit operations using multi-stage
circuitry. As early as two decades ago, Mikhail et al. (1988) demonstrated
experimentally that a combination of WOCs and spirals can help achieve
optimum recovery of coal depending on the washability of the fine particles.
2.5
Spiral “A”
Spiral “B”
Spiral “C”
2.3
2.1
SG Cutpoint
1.9
1.7
1.5
1.3
0.1 1 10
Particle Size, mm
Figure 7 Effect of particle size on density cutpoint for three industrial installations
of coal spirals
More recently, Bethell and Moorhead (2003) reported plant operating data
showing that combined WOC and spiral circuits offered advantages in terms
of improved separation performance, higher capacity, and lower costs per ton.
Figure 8 provides a simplifed schematic of a typical two-stage WOC and
spiral processing circuit. In this circuit configuration, 1 × 0.15 mm feed coal
is pumped to a bank of WOCs to reject high-density particles. The upgraded
overflow is then directed to a bank of classifying cyclones to deslime and
remove most of the –0.15 mm (100 mesh) material. In this case, the WOCs
must be configured to provide a low-density cutpoint, which typically requires
that they be equipped with a relatively short vortex finder. Lower operating
pressures, e.g., 69–83 kPa (10–12 psi), may also be utilized to ensure that
a low cutpoint density is maintained in the WOCs. The coal-rich particles
intentionally lost to the WOC underflow are then recovered using a bank of
spirals. Spirals are ideally suited to recover the larger, low-density coal particles
contained in the WOC underflow. However, since the underflow stream also
Figure 8 Typical circuitry combining water-only cyclone and spiral technologies
typically contains a high proportion of rock compared to the original feed,

there is a tendency for some portion of this high-ash material to report with the
clean coal. The phenomenon responsible for the unavoidable misplacement of
high-density rock by spirals has been previously discussed (Luttrell et al. 2007).
To minimize this issue, compound recleaner spirals are generally recom-
mended for this type of circuitry. This special type of spiral integrates two
stages of spiral separation along a single low-profile assembly. Typically, these
units consist of three to four turns of primary spirals, which are followed imme-
diately by three or four turns of secondary spirals. The refuse from both the
primary and secondary spirals is rejected, whereas the clean product is taken
only from the secondary spirals. The clean product and middlings from the pri-
mary turns are remixed in a pulping box before being fed to the secondary turns.
This additional recleaning of the clean coal product from the primary spirals is
very important to ensure that misplaced rock is removed from the final clean
coal product. The secondary middlings product from the spiral units can be
discarded, sent to the clean coal product, or recycled back to the circuit feed.
Generally, the latter option is recommended to improve the overall sharpness of
the separation. The beneficial impact of and scientific rationale for utilizing the
middlings recycle has been described in detail elsewhere (Kohmuench 2000).
The positions of the spiral splitters should be optimized to ensure that an exces-
sive circulating load of middlings is not created within the two-stage circuitry.
Finally, the clean products from the spiral and classifying cyclone overflow
are passed across fine wire sieves to dewater the clean coal. This final process-
ing step also plays a critical role in efficiently removing any residual high-ash
ultrafine clay that would otherwise contaminate the final coal product (Barbee
and Nottingham 2007). In fact, one variation of this type of circuitry passes
the clean product from the spiral to the classifying cyclone sump to provide
an additional step of desliming before final dewatering on the fine wire sieves.
Although this strategy may have some merit, it is not recommended in cases
of high-grind coals that would be subjected to further size degradation in the
classifying cyclone circuit.
The intrinsic benefits of the two-stage WOC and spiral circuit can be
best understood using the illustration provided in Figure 9. The feed matrix
shown in the top left corner of the diagram represents the different types of
coal (black) and rock (gray) particles fed to the circuit. In the case shown, the
WOCs are configured to direct most of the high-density rock to the underflow
stream, with only the finest particles of high-ash material reporting to over-
flow. These particles are easily removed downstream using a combination of
classifying cyclones (to handle the large volume flow) and fine wire sieves (to
efficiently remove bypassed ultrafines). Because of the differential cutpoints in
the WOCs, a considerable amount of low-ash coal in the coarser size fractions
is misplaced to the WOC underflow. Fortunately, feed streams that have been
hydraulically classified by the WOC unit are ideally suited as feeds for flowing-
film separators such as spirals. Since the feed to the spiral contains a dispropor-
tionally large amount of rock, it is important to utilize two-stage compound
recleaner spirals in this application to miminize the unwanted misplacement of
high-ash particles to the clean coal product. The compound spirals efficiently
reject coarse rock without losing a significant amount of lighter coal particles
that are readily carried into the clean coal product by the body drag forces in
the flowing film. The clean product from the spiral contains the coarse low-ash
particles as well as a considerable amount of fines that are hydraulically carried
with the bulk of the slurry flow that reports to clean coal. These unwanted fines,
Feed WOC Overflow Classifier Overflow
Water-Only Classifying
Cyclones Cyclones
WOC Underflow Classifier Underflow
Spiral Reject Spiral Clean Sieve Oversize
Spiral Sieve
Separators Screens
Spiral Middlings Sieve Undersize

(Recycled to Feed) (Recycled or Flotation)
Figure 9 Simplified illustration of particle partitioning in a combined water-only

cyclone and spiral circuit
which contain a significant amount of high-ash clays, are readily removed using
one or more stages of fine wire sieves. The bottom right matrix shows the final
high-quality cleaned and classified product from the combined WOC and
spiral circuit.
P E R F O R M A N C E C O M PA R I S I O N
To better evaluate the capabilities of a combined WOC and spiral circuit, a
case study was undertaken using partition data collected from an industrial site.
Unfortunatley, a direct side-by-side comparision of the different technologies
was impractical because of natural fluctuations in the plant feed and random
errors in the sample collection and laboratory analysis procedures. Therefore,
the experimental data were subjected to mass balancing to obtain a consistent
set of experimental data and then used to obtain size-by-size partition curves
for the WOC and spiral units. The resultant partition curves were used to
simulate the separation performance for an identical set of feed coal washability
data. The simulations included the following types of circuitry:
• WOC only
• Spiral only (spiral second-stage middlings sent to reject)
• Spiral only (spiral second-stage middlings sent to clean coal)
• Spiral only (spiral second-stage middlings recycled to circuit feed)
• WOC followed by spirals (spiral second-stage middlings sent to reject)

• WOC followed by spirals (spiral second-stage middlings sent to clean
coal)
• WOC followed by spirals (spiral second-stage middlings recycled to
circuit feed)
When operating by itself, the WOCs were configured in the simulations to
provide a moderate to high SG cutpoint. When operating in tandem with the
spiral units, the WOCs were configured to provide a relatively low SG cutpoint
that allowed a much higher-quality clean coal product to be generated by the
WOCs.
The clean coal yield and ash obtained from the various simulations are plot-
ted in Figure 10. As expected, the WOC circuit operated as an independent
unit was clearly unable to provide the same level of separation performance that
could be attained using the compound spiral or any of the combined two-stage
WOC and spiral circuits. For this particular feed coal, the simulations showed
that similar levels of performance could be obtained when targeting ash values
below about 8.5% using either the compound spirals alone or combined WOC
and spiral circuits. For product ash contents above about 8.5%, the combined
WOC and spiral circuitry actually provided a slightly higher yield than the
spiral-only circuits. The combined circuit in which the spiral middlings were
recycled back to the feed provided the second highest clean coal yield of 78.3%
at an ash content very close to the plant product specification of 8.5% ash.
Moreover, the combined WOC and spiral circuitry offered a considerable ben-
efit in terms of throughput capacity and footprint.
G U I D E L I N E S A N D R E C O M M E N D AT I O N S
A close examination of the data from this investigation and other sites has
been used to develop several recommended guidelines for the installation and
operation of combined WOC and spiral circuits. The most important of these
include the following:
• The performance of the combined WOC and spiral circuit is very sen-
sitive to the particle size distribution of the feed stream. The presence
of gross oversize particles above the target topsize of 1 mm has a large
adverse impact on separation performance and should be minimzed
to avoid deterioration of product quality and yield. A slip-stream tell-
tell screen is recommended to assist plant operators in detecting this
unfavorable condition.
80
WOC & Spiral

(Keep Midds)
WOC & Spiral
(Recycle Midds)
78
WOC & Spiral
(Reject Midds) Spiral Only
(Keep Midds)
Spiral Only
(Recycle Midds)
76
Clean Coal Yield, %
Spiral Only
(Reject Midds)
74
72
WOC Only
WOC
Spiral
WOC & Spiral
70
7.6 7.8 8.0 8.2 8.4 8.6 8.8 9.0
Clean Coal Ash, %
Figure 10 Performance levels achieved by water-only cyclones, spirals, and

combinations of water-only cyclones and spirals
• Two-stage WOC circuitry incorporating spirals is superior to two

stages of WOCs because of the ability of the spirals to recover lower-
density particles in the coarsest size fractions. The twin process circuit
also requires less pumping and fewer sumps than two stages of WOCs
(i.e., the spirals can be fed by gravity).
• To maintain circuit performance, the WOCs should be equipped
with relatively short vortex-finder lengths. This equipment configura-
tion provides for a high-quality overflow from the WOC and allows
coarser coal to be diverted to underflow where it can be more readily
recovered by spirals.
• WOCs are effective in minimizing the amount of fine slimes passed
to the second stage of spirals. The efficient elimination of slimes has
the potential to further lower the overall density cutpoint in the spiral
units and simultaneously reduce the carryover of these high-ash par-
ticles in downstream dewatering systems (i.e., fine wire sieves).
• Compound recleaner spirals are highly recommended for use in com-

bined WOC and spiral circuits. The compound spiral is better able to
reduce the unwanted misplacment of rock to clean coal that would
otherwise occur because of the disproportionally large amount of rock
present in the spiral feed from the WOC underflow.
• The density cutpoints achieved in both the WOC and spiral units are
very sensitive to the solids content of the feed stream. Therefore, the
combined circuit must be carefully configured to ensure that the solids
content does not become too great due to the excessive recycle of spiral
middlings.
• The combined WOC and spiral circuit has the potential to reduce
total installation costs because spirals typically have a higher capital
cost per unit capacity than do WOCs.
SUMMARY
Two-stage circuits incorporating WOCs and spirals can offer an attractive
method for treating fine coal if properly designed and operated. From a capital-
cost standpoint, utilizing WOCs, with their higher unit processing capacity,
and recleaning the WOC underflow stream with spirals results in a reduced
plant volume for the fine coal circuitry. In addition, overall slurry processing
volumes are reduced when recycing spiral middlings (typically 35%–40% solids
by weight) versus the overflow from second-stage WOCs (typically 5%–10%
solids by weight). Typically in a two-stage WOC circuit, each secondary-stage
cyclone requires an additional primary-stage unit to process the volume of
overflow slurry being recycled. In contrast, recycling spiral middlings, with
their high solids concentration, can typically be handled by the primary-stage
cyclones without additional units.
From a performance standpoint, utilizing a “classifier” type device together
with a “flowing film” separator provides more consistent SG50 values for the
1 × 0.15 mm size fraction than if either unit was used for both the primary
and secondary stages of cleaning. Since maintaining a consistent SG50 by size
is critical to overall separation efficiency, mating a WOC and spiral in the
same circuit can result in improved performance. In addition, recycling spiral
middlings provides a significant improvement in Ep values for individual size
fractions and should always be utilized. Additional advantages in combustible
recovery can also be realized with the circuit. Because the top size of the fine
coal circuit increases as the raw coal screen panels wear, a two-stage WOC
circuit is extremely vulnerable to excessive combustible losses because the
additional coarser-sized particles are separated at successively lower SG50 val-

ues. In contrast to the WOC, a spiral has a response that is opposite, with the
additional coarser-sized particles being separated at a higher SG50. Although
neither response is optimum, the loss of low-density floatable coal, with ash
values below that of the incremental ash target, is less desireable than the addi-
tional recovery of some particles with ash values above that of the incremental
ash target.
From an operations standpoint, the combined WOC and spiral circuit
can provide a relatively stable and consistent long-term operation. Advanced
ceramics permits modern WOCs to operate for extended time with little main-
tenance and without internal geometry changes that would influence the SG50.
Given that the unit process capacity of the WOC is so high, less spiral starts are
required per ton of raw coal processed in this circuit. This reduces the number
of spiral starts (which typically require daily or even per-shift cleaning) and
results in lower labor requirements.
The combined WOC and spiral circuit also has a potential yield advantage
when making a high-quality metallurgical coal along with a middlings product.
Since a WOC can achieve a relatively low SG50 on the +0.25 mm size coal, this
circuit permits the option to send the WOC overflow to the clean coal classify-
ing cyclones and produce metallurgical-grade coal from that stream. The spiral
clean coal can then be directed to the middlings product. This option permits
producing at least a portion of the high-grade product from the 1 × 0.25 mm
size coal without an excessively high ash content that would otherwise force
the dense-medium circuits to be operated at a lower SG50. This can result in
a significant yield increase when the dense-medium circuit must otherwise
be operated at a relatively low SG50 where substantial near-gravity material is
present.
REFERENCES
Abbott, J. 1981. The optimisation of process parameters to maximise the profitability
from a three-component blend. In 1st Australian Coal Preparation Conference.
Edited by A.R. Swanson. Newcastle, Australia: Newey and Beath Printers. pp.
87–105.
Barbee, C.J., and Nottingham, J. 2007. Black Bear prep plant replacement of high
frequency screens with fine wire sieves. In Proceedings, 24th Annual Coal Prepa-
ration Conference and Exhibition, Lexington, KY. New York: Penton Media. pp.
113–122.
Bethell, P.J., and Arnold, B.J. 2003. Comparing a two-stage spiral to two-stages of
spirals for fine coal preparation. In Advances in Gravity Concentration. Littleton,
CO: SME. pp. 107–114.
Bethell, P.J., and Moorhead, R.G. 2003. Operating characteristics of water-only
cyclone/spiral circuits cleaning fine coal. In Advances in Gravity Concentration.
Bull, W.R., Pillai, K.J., and Spottiswood, D.J. 1987. An analysis of water-only cyclone
capabilities. Preprint No. 87-100. Littleton, CO: SME.
Holland-Batt, A.B. 1994. The effect of feed rate on the performance of coal spirals.
Coal Prep. 14:199–222.
King, R.P., Juckes, A.H., and Stirling, P.A. 1992. A quantitative model for the predic-
tion of fine coal cleaning in a spiral concentrator. Coal Prep. 11:51–66.
Kohmuench, J.N. 2000. Improving efficiencies in water-based separators using math-
ematical analysis tools. Ph.D. dissertation, Virginia Polytechnic Institute and State
University, Blacksburg, VA.
Li, M., Wood, C.J., and Davis, J.J. 1993. A study of coal washing spirals. Coal Prep.
12:117–131.
Luttrell, G.H., Kohmuench, J.N., Stanley, F.L., and Trump, G.D. 1998. Improving spi-
ral performance using circuit analysis. Miner. Metall. Eng. 15(4):16–21.
Luttrell, G.H., Catarious, D.M., Miller, J.D., and Stanley, F.L. 2000. An evaluation of
plantwide control strategies for coal preparation plants. In Control 2000. Edited
by J.A. Herbst. Littleton, CO: SME. pp. 175–184.
Luttrell, G.H., Honaker, R.Q., Bethell, P.J., and Stanley, F.L. 2007. Design of high-
efficiency spiral circuits for coal preparation plants. In Designing the Coal Prepa-
ration Plant of the Future. Edited by B.J. Arnold, M.S. Klima, and P.J. Bethell.
Mikhail, M.W., Salama, A.I.A., Parsons, I.S., and Humeniuk, O.E. 1988. Evaluation
and application of spirals and water-only cyclones in cleaning fine coal. Coal Prep.
6:53–78.
Osborne, D.G. 1988. Coal Preparation Technology. Vol. 1. London: Graham and Trot-
man. pp. 347–365.
Schlepp, D.D., and Schmidt, M.P. 1988. When to use water-only cyclones. In Indus-
trial Practice of Fine Coal Processing. Edited by R.R. Klimpel and P.T. Luckie.
Subasinghe, G.K., and Kelly, E.G. 1991. Model of a coal washing spiral. Coal Prep.
9:1–11.
Subasinghe, G.K., and Kelly, E.G. 1992. Predicting the cut specific gravity of a coal
washing spiral. Miner. Eng. 5(2):193–203.
Moisture Reduction
and Special Topics 4
Performance, Operation, and
Maintenance Experience of Coal
Ultrafines Filtration with Modern
High-Speed Disc Filters
Jürgen Hahn
ABSTRACT
After a successful modernization and revamp of standard design disc filters in
several coal handling and preparation plants (CHPPs) in Australia (e.g., Saraji
mine), Billiton Mitsubishi Alliance (BMA) decided to go with modern high-speed
disc filters in their new coking coal preparation plant at Blackwater, Australia.
These two Bokela L4 disc filters have been in operation for 4 years, and further
high-speed disc filters have been ordered in the meantime by BMA and Anglo
for new coking coal applications because of the outstanding performance and easy
operation and maintenance. Layout tests for coking coal applications (–0.25 mm)
in the Republic of South Africa (RSA) show similar results compared to Austra-
lia with specific solids throughput rates of 500–1,000 kg/m²/h and moistures in
the low 20 wt% range. Furthermore, the filtration results of the tailings are very
promising as well with specific solids throughput rates of about 500 kg/m²/h and
moistures around 20 wt%. This would allow preparation plants to run high-speed
disc filters for both coal concentrates and tailings and use spare filters for both pur-
poses. In the future, high-speed disc filters may even be used for thermal coal fines
(–0.125 mm).
D E F I C I E N C I E S O F O L D D I S C F I LT E R C O N S T R U C T I O N S
A P P L I E D F O R C O A L F I LT R AT I O N
Vacuum disc filters are the most compact rotary filter type and are comprised
of a series of vertical discs, usually up to 10 or 12 per filter. The conventional
discs, each being composed of up to 24 radial segments, are arranged on a
horizontal shaft. Each segment of the discs is covered with filter media, usually
261
262 Moisture Reduction and Special Topics
exchangeable filter cloth. Disc filters can have more than 240 m2 filter area in a
single machine. Main characteristics of disc filters are
• Low footprint area
• Relatively low capital costs (around Euro 2.500–3.000 per m2)
• Low residual moisture content in filter cake
• Low operating costs
• Low maintenance
• Very effective cake discharge system (up to 100%)
• Variations in feed flow rate acceptable
• High specific solids throughputs
• Can handle fairly large particle size distribution of feed material
• Operational problems if no level control available
• Inadequate discharge without snap-blow system
• Inadequate for cake washing
• Older type filters require more maintenance and attendance
Old Design
Looking to old disc filter constructions of 1960s technology, many remarkable
deficiencies can be observed. These weak points can be categorized and sum-
marized as follows.
Process
• Poor cake pickup—the cake is dropped back into the slurry bath as it
emerges out of the bath.
• Cake moistures were higher than expected.
• Poor discharge of the cake caused 50% or more of the cake to fall
behind the scraper into the slurry bath.
• The vacuum was always lower than required for good performance.
Maintenance
• Cloths were tearing on the scrapers, which drove the change to wire
mesh filter media.
• Temporary platforms were required to exchange bats (segments),
which has generated a culture of cloth changes only during plant
shutdowns and weekends no matter how many cloths were damaged.
Coal Ultrafines Filtration with Disc Filters 263
• The design of bat fixing system was cumbersome, which required tie
rods to keep bats in place.
• Bats were heavy, weighing around 40 kg, making the re-clothing task
difficult.
• High wear of filtrate piping and control head due to holes in the filter
media caused high solids in the filtrate.
• Filtrate pipes with 12 to 14 discs were made in two halves with a welded
joint in the middle. This joint breaks within 2 months because of fatigue.
This joint loses vacuum and introduces solids into the filtrate.
• Agitator maintenance is always problematic and often replaced by
compressed air.
I M P R O V E M E N T O F D I S C F I LT E R S B Y S Y S T E M AT I C
REDESIGN
Based on the consequent transfer of fundamental knowledge from the filter
theory and the experience gained in more than 250 filter optimization/revamp-
ing projects for all major filter types and manufacturers, Bokela developed the
Boozer disc filter, a new generation of disc filters incorporating many innova-
tive changes to conventional design practice. Most of these innovations have
been made to resolve capacity and performance problems related to hardware
bottlenecks and/or poor hardware design. The most critical filter parts that
were addressed in the new design are
• Filter discs and segments,
• Filtrate pipes,
• Center barrel and bearings,
• Filter trough,
• Control head, and
• Cake discharge.
The second development focus has been to minimize maintenance require-
ments and time, which has lead to the following results:
• Low wear segments with massive hydraulic capacity to process the
large filtrate flow
• Lightweight segments with quick release bayonet connections for ease
of installation and removal—no tie rods!
• Quick fit filter bag system to allow cable ties to seal the neck
• Filter designed to use poly bags, which are easier to replace and more
cost-effective
• Permanent walkways to allow easy access to replace segments with
holed cloths
• Removable filtrate pipes
• Center barrel and bearings designed for the high loads encountered
with high capacity
• Gearbox and motor designed for high loads at low speeds
• Filter trough designed to eliminate agitators by being self-agitating
• Level control system to prevent overflow back to feed tank
• Control head with low pressure losses (low wear) at high capacities
• Back suck on the filter cloth to prevent damage on the scrapers during
cake discharge.
Consequently, all the previous changes have resulted in
• Drastically improved cake pickup due to the high vacuum achieved
inside the disc,
• Better cake moistures than other filters of the same area at the same
tonnage,
• Excellent discharge of the cake with 95% to 100% reporting to the
product, and
• Vacuum always at the level required for good performance.
F E AT U R E S O F N E W VA C U U M D I S C F I LT E R
G E N E R AT I O N A P P L I E D F O R C O A L F I LT R AT I O N
With the high-performance disc filter Boozer, Bokela Company has developed
a new generation of large-diameter disc filters, which have set a new standard
for seed filtration in the alumina industry and in the dewatering of coal slurries.
BMA decided to go with these modern high-speed disc filters in their new coking
coal preparation plant at Blackwater, Australia. Two Bokela L4 disc filters have
been in operation for 4 years, and additional high-speed disc filters have been
ordered in the meantime by BMA and Anglo for new coking coal applications
because of the outstanding performance and easy operation and maintenance.
The outstanding hydraulic characteristics of the Boozer disc filter were
achieved by improving each detail of the filter design leading to extraordinarily
high performance capacity, high operational safety and reliability, as well as
1. Filter Segments
2. Center Barrel
3. Control Head 1
4. Filtrate Receiver 2
5. Blowback Tank 3
4
6 5
6. Snap-Blow Valve
7 7. Filter Trough
8. Drive Unit
Figure 1 Boozer disc filter with three discs
low maintenance and operation costs. The main features of this new disc filter
generation are listed as follows:
• Minimized pressure drop leading up to 100% higher pressure differ-
ence at the filter cloth compared to conventional disc filters
• Double capacity compared to conventional disc filters
• High-filter speed of 6 rpm
• High operational reliability and flexibility
• Easy maintenance
• Fully automatic and safe operation due to superior process philosophy
realized in a programmable logic controller system
Disc diameters range from 1.7 m to 5.6 m. For filtration of large slurry feed
rates such as tailings, a Boozer filter with a large disc diameter of 5.6 m (L-type)
is the appropriate filter size that is available with one to four filter discs.
The Boozer disc filter is designed by optimizing each detail consequently
according to flow requirements. In this way, pressure loss in the whole flow
route from the filter segments up to the filtrate receivers has been minimized. In
the following sections, the main components of the Boozer disc filter (Figure 1)
are described.
Robust Filter Segments

Filter discs consist of 30 segments with robust construction and a low weight of
less than 20 kg per segment. By having 30 segments per disc leads to small-sized
Figure 2 Pre-separation control head of Boozer disc filter
segments, ensuring optimal hydraulic conditions for cake formation, fast filtrate
drainage, and complete cake discharge in less than 0.2 seconds without filtrate
blowing back (i.e., without cake re-wetting). On the other side, this number
of low-weight segments is ideal to facilitate re-clothing and maintenance. The
segment bell is made for a fast, easy, and secure filter bag fixing and is designed
in a shape that allows fixing of the filter bags with cable ties. This fast and
secure method simplifies filter cloth fixing and reduces the time needed for
re-clothing.
Pre-Separation Control Head

The two-phase flow of filtrate and air, which enters the pipe system in the cake
dewatering zone, together with the high flow velocity are the important factors
for pressure loss. Therefore, the control head of the Boozer disc filter (Figure 2)
is designed to pre-separate filtrate and air already inside the control head. The
outlet nozzles have a specially streamlined shape to minimize flow restrictions
and pressure loss. The control head is pressed against the wear plate via a central
tightening. It is not supported on the trough construction (as is standard filter
design) but on the shaft center that leads to a short, closed linkage.
Filter Arrangement
Between the control head and filtrate receiver, the previously mentioned
two-phase flow of filtrate and air can lead to a high pressure loss, Δp, of up to
10 times higher than a one-phase flow. Hence, the arrangement of the filter and
filtrate receivers influences filter operation and filter performance. Therefore,
the Boozer disc filter and the filtrate receiver are located as close and straight
as possible to keep this flow section very short (see Figure 1). The layout of the
pipes between control head and receiver avoids any bend as far as possible, and
the receiver inlet is carried out tangentially.
Joint Single Trough Without Agitator

The “joint single” trough (Figure 3) of the Boozer disc filter has been designed
to combine the advantages of the common trough design (all discs run in a
common trough) and the single trough design (every filter disc runs in its own
single trough) without the respective disadvantages and operation failures as
described by Bott et al. (2004). The joint single trough is constructed like a
common trough with installations between the discs at the slurry inlet side,
which subdivide the trough in narrow compartments for each disc. Thus, it
works without an agitator (like the standard single trough) given that the
stirring effect of the discs rotating in narrow compartments homogenizes the
slurry. Similar to the common trough design, all discs run with the same slurry
level because of a large sized, common overflow. The free exchange of slurry
between the single trough compartments ensures a common level in the trough
for all discs. The design and arrangement of the feed manifold is such that the
incoming slurry supports the stirring effect of the rotating discs.
Cake Discharge and Cake Deflector Boxes

The filter cake is discharged by a compressed air blowback, which is precisely
timed by a snap-blow valve. The small sector volume and the hydraulic design
provide for quick filtrate drainage and for a quick and intensive blowback
impulse with only 0.25–0.35 bar overpressure and without blowing filtrate
back into the cake. Even at high filter speed when thin filter cakes of only
4–5 mm in thickness are produced, the cake is totally discharged. The Boozer
disc filter can therefore be satisfactorily operated with thin filter cakes at a high
filter speed of up to 6 rpm.
Cake deflector boxes with anti-friction tops are made of ultrahigh modulus
polyethylene (UHMPE) (Figure 4). These prevent accumulation of cake and
ensure an exact adjustment of the distance between cake deflector and filter disc
(in conjunction with disc guide rollers [Figure 5]; i.e., no filter cake falls back
into the trough and filter performance keeps constantly high).
C O M M I S S I O N I N G I S S U E S A N D D E S I G N A D A P TAT I O N S
AT B L A C K W AT E R
At the coking coal preparation plant at Blackwater, two Boozer L4-type disc
filters have been in operation for 4 years. During commissioning of these two
Common Part for Level Control
Single Compartments
Figure 3 Joint single trough (all discs run with the same slurry level)
filters, operational problems occurred, which were eliminated by some design

modifications (i.e., adaptation of special issues to the requirements of coal
filtration).
Antifriction Inlays for Discharge Chutes

During commissioning, the discharge chutes were permanently blocked, which
required hourly removal of sticking filter cake from the discharge chutes by
hose cleaning. The discharged cake kept sticking at the discharge chutes because
of rust, which was formed even on the stainless-steel material of the discharge
chutes because the filters were installed some months prior to start of opera-
tion. Although this problem would have disappeared after some months of
operation, Bokela decided to solve this problem with a design modification.
The discharge chutes of the filters were equipped with antifriction inlays made
of UHMPE, and sticking filter cake was no longer an issue of operation.
Exchangeable Weir Plates in Control Head

In the control head, wear occurred at any position where air streams pass
through with high speed. To solve this problem, the control heads of the two
filters have been equipped with exchangeable weir plates to prevent destruction
of the control heads.
Figure 4 Complete cake discharge (left) and advanced scraper design (right)
Figure 5 Antifriction inlays for discharge chutes made of UHMPE (left), and disc
guide rollers (right)
Form Zone Vacuum Control Valve

The supplied filters were equipped with an operation control system, which
initially worked with one control circuit basing on measurement of the slurry
level in the filter trough and adaptation of filter speed, n, accordingly between
nmin and nmax to keep slurry on optimum level for operation.
With coarse feed slurries, however, filtration performance of the filters was
too high to keep the slurry level within the range of operation even when run-
ning with minimum filter speed, nmin. Therefore, the level dropped below the
minimum set point and the filters sucked air, which affected the vacuum line,
and filter operation was no longer possible. A typical solution for such cases is
the reduction of the cake form zone by insertion of bridge blocks in the control
disc. This remedial action, however, requires about 8 hours of maintenance
work, and the same amount of time is required for removing the bridge blocks
when fine slurry is again fed to the filters.
To solve this problem, Bokela installed a control valve in the form zone
vacuum of each filter. This allows reducing of form zone vacuum in the case
of coarse slurries by a secondary control circuit as described in the next sec-

tion. In this way, filter operation is possible over the whole range of feed slurry
characteristics.
PROCESS AND CONTROL PHILOSOPHY

The design and outstanding hydraulic capacity of the Boozer disc filter is the
basis of the Bokela operation philosophy for improved filter operation without
continuous slurry overflow. An automatic filter operation control adapts filter
performance to changing slurry and process conditions and avoids emergencies.
A primary and a secondary control circuit ensure a self-regulating and secure
filter operation by using the slurry level in the trough as the control variable,
and the filter speed and pressure difference as the control output. The pos-
sibility to run the Boozer disc filter with a high filter speed of n = 6 rpm and
the high pressure difference, which is available at the filter cloth, allow filter
operation in a wide performance range. All important functions and periph-
eral components—such as vacuum pump, filter drive, trough level, blower, the
vacuum in the receivers, pressure in the cake blow-off tank, lubrication pump,
and so forth—are supervised and integrated in the interlock schedule. In case
of an emergency, the programmable logic controller activates an alarm, and an
adequate procedure is started that depends on the respective emergency case.
In any case, the drain valve of the trough opens, the filter is put off from the
vacuum, and the feed valve is closed to prevent the slurry from flooding the
filter floor.
Primary Control Circuit—Trough Level Control by Adjusting Filter Speed

The slurry level is measured continuously by a proven radar measurement
device, which is suitable even for foam on the slurry surface and mist/vapor in
the air. The signal of the level measurement is used to adjust the variable filter
speed by a variable frequency controller (Figure 6).
At normal operation, the filter is fed with a more or less constant slurry
flow, and the Boozer operates with a medium filter speed of approximately
2.0–3.0 rpm and a fully open vacuum control valve in the vacuum line (form
zone) between the control head and filtrate receiver. The best performance of
the filter is achieved if the slurry level in the filter trough is close to the over-
flow level (approximately 50% submergence of the filter disc). In this case, the
cake formation angle is maximum, and an even filter cake with a constant cake
thickness over each segment is formed. This results in very constant operation
conditions.
Loop 1
Filter Speed Controlled
by Slurry Level
Loop 2
Vacuum Controlled
by Slurry Level
LIC
Figure 6 Filter operation control system of the high-performance Boozer disc filter
with fully automatic operation
If the slurry level in the filter trough drops or increases because of a chang-
ing slurry flow or changing product characteristics, the filter speed is automati-
cally reduced or increased to maintain a constant high slurry level in the trough
(level set point). If the filter runs at minimum permitted speed and the slurry
level in the trough still drops, then the second control circuit becomes active
and the vacuum in the filter cake formation zone is adapted accordingly.
Secondary Control Circuit—Trough Level Control by Adjusting Form

Zone Vacuum
Control of the vacuum in the cake formation zone with the vacuum control
valve is the second control circuit (Figure 6). The target is to maintain a con-
stant and high slurry level in the trough, even at minimum filter speed. This
control circuit becomes active if the filter runs with minimum speed and the
slurry level in the trough still drops. Then the vacuum control valve in the
vacuum pipe (form zone) between filter and filtrate receiver is closed as needed
to maintain the level set point in the trough. If the form zone vacuum control
valve is closed to the limit and the filter speed is already at the minimum but
the slurry level still drops, then the trough level can be reduced to a mini-
mum permitted set point to further reduce filter performance. The individual
vacuum control in the form zone also enables the required cake moisture to be
maintained.
F I LT E R O P E R AT I O N AT B L A C K W AT E R C H P P
The most important data for the two disc filters are listed as follows:
• Filtration area of 176 m2 each
• Four discs per filter
• Disc diameter of 5.6 m
• Thirty light segments (24 kg each) per disc with bayonet segment fix-
ing for fast change-out
• Filter barrel with internal filtrate pipes
• Joint single trough that works without agitator and allows perfect level
control
• Permanent walkways between the discs
• Snap-blow system for cake discharge (snap-blow valve, control system,
and feed tank)
• Cloth washing system as wash bars underneath scraper with solenoid-
operated shut-off valve
• Automatic level control so there is no overflow recirculation required
• Filter speed n = 0.5 to 4 rpm (typical operating range for coal slurry
filtration)
Slurry Data and Operation Results

The slurry to be filtered is an ultrafine clean coal slurry (flotation concentrate)
with particle sizes <300 µm and a nominal feed solids concentration of 38 wt%.
Each filter has to match varying feed conditions requiring a specific solids
throughput of 517 kg/m2/h (nominal) up to 812 kg/m2/h (maximum) with
varying feed concentrations ranging from 35 wt% to 56 wt% (see Figure 7 and
Table 1). For this performance, the filters will run with a filter speed of 0.5 rpm
up to 1.7 rpm. The filters are also capable to match worst-case feed conditions
with peak feed flow, minimum feed solids concentration, and particle sizes of
100% less than 200 µm as well as coarse slurries.
Maintenance
With the realized design adaptations described previously, the filters run with
an availability of 99%. Scheduled maintenance effort is 31 h/a (1.5 days per
year). Maintenance is required for re-clothing (two times/a, i.e., 3 h for re-
clothing of one disc every 1.5 months), exchange of roller guides (4 h/a), and
maintenance of the snap-blow valve for cake discharge (one per year). In 4 years
Expected Solids Throughput of L4-Type Disc Filter

with Comet Ultrafine Clean Coal (<300 micron)
320.0
300.0 35 wt% Feed Solids All <300 micron
Typical Operating Range of Filters
240.0
Maximum Solids Throughput Required
220.0
Solids Throughput, t/h
Minimum Solids Throughput Required

200.0
180.0
160.0
140.0
120.0
100.0
80.0
60.0
40.0
20.0
0.0
0 0.5 1 1.5 2 2.5
Square Root of Filter Speed, rpm 0.5
Figure 7 Solids throughput vs. square root of filter speed for different feed
concentrations—layout performance for Boozer L-type disc filter with Blackwater
ultrafine clean coal slurry
Table 1 Performance of Boozer L-type disc filter at BMA Blackwater CHPP (ultra-
fine clean coal slurry with solids particle size <300 µm)
Feed Specific Solids Filtrate
Concentration, Flocculant, Throughput, Solids,
wt% g/t DS Moisture, % kg/m2h g/L
Layout 35 minimum <15 <25 517 nominal <10
performance 38 nominal 812 maximum
56 maximum
of operation, only one unexpected filter stop was necessary due to holes in the
control head, but no other abrasion has occurred since then.
R E S U LT S W I T H R S A C O K I N G C O A L — F L O TAT I O N C E L L
OVERFLOW
Coking coal from RSA (ash contents around 10 wt%) is showing good filtra-
tion behavior. The pure flotation cell overflow with a typical feed solids content
of 14–18 wt% is already filtering with a specific solids throughput of 400 kg/
m²/h, which relates to 70 t/h on typically used L4 vacuum disc filters.
Furthermore, the RSA coking coal is very responsive to flocculant dosage.
With 10 g/t flocculant dosage, the specific solids throughput doubles, as shown
1,200
dp = 60 kPa, No Floc Dosage
Specific Solids Throughput mS, kg/m2h
dp = 60 kPa, 5 g/t Floc Dosage

1,000
800 Feed Solids = 15 wt%
600
400
200
0
0.0 0.5 1.0 1.5 2.0
1
n
min
Figure 8 Solids throughput vs. filter speed for different flocculant dosages (South
African coking coal—flotation overflow)
in Figure 8 (140 t/h per standard L4 disc filter). The solids throughput further
increases when more flocculant is added.
In addition, filtration rates can be further increased with the use of a coal
thickener. If the flotation overflow is thickened from 15 wt% to about 30 wt%,
the specific solids throughput triples, as Figure 9 shows. The specific solids
throughput increases from about 800 kg/m²/h to more than 2,500 kg/m²/h.
This means that one standard L4 disc filter is able to filter up to 450 t/h.
However, high solids throughput is usually at the expense of higher mois-
ture. Figure 10 shows this very well. A typical ratio of dry time to form time of 2
and a flocculant dosage of 10 g/t (typical setting of standard L4 disc filters used
in coal filtration) is considered for the comparison. Under these conditions, the
nonthickened flotation overflow is filtered with a moisture content of about
20 wt%, while the thickened flotation overflow still contains about 25 wt%
moisture, however, with three times the throughput.
R E S U LT S W I T H R S A C O K I N G C O A L C E L L U N D E R F L O W
Similar to the coking coal concentrate, the coking coal tailings from RSA are
showing good filtration behavior. However, tailings always require thickening
in combination with the addition of flocculant to achieve reasonable through-
put figures.
3,000
15 wt% Feed Solids

2,500 30 wt% Feed Solids
dp = 60 kPa
2,000 10 g/t Floc Dosage
1,500
1,000
500
0
0.0 0.5 1.0 1.5 2.0 2.5 3.0
1
n
min
Figure 9 Solids throughput vs. filter speed for thickened and nonthickened flotation
overview (South African coking coal)
30
30 wt% Feed Solids
28
15 wt% Feed Solids
26 Operation Range
24
Moisture Content, wt%
22
20
18
16
14 ∆p = 60 kPa
hc = 15 mm
12 10 g/t Floc Dosage
10
0 1 2 3 4 5 6 7
Dry Time/Form Time [–]
Figure 10 Comparison of moisture content of nonthickened and thickened flotation

overflow (South African coking coal)
In this case, 30 g/t flocculant was added and the flotation cell underflow
was thickened to a typical thickener underflow solids content of 30 wt%. In this
condition, a specific solids throughput of 460 kg/m²/h (see Figure 11) can be
achieved, which relates to about 80 t/h on vacuum disc filters as used for coal
500
450 30 g/t Floc Dosage
400 ∆p = 60 kPa
30 wt% Feed Solids
350
300
250
200
150
100
50
0
0.0 0.2 0.4 0.6 0.8 1.0 1.2
1
n
min
Figure 11 Solids throughput versus filter speed for RSA coking coal tailings
(flotation underflow)
28
30 g/t Floc Dosage
26 Operation Range
24 ∆p = 60 kPa
30% Feed Solids
Moisture Content, wt%
22
20
18
16
14
12
10
0 0.5 1 1.5 2 2.5 3 3.5 4
Dry Time/Form Time [–]
Figure 12 Moisture content versus dry time for RSA coking coal tailings (flotation
underflow)
concentrates. Further thickening and more flocculant addition can increase the
capacity by another 30%–60%.
Within the typical operating range of L4 vacuum disc filters of 1.3 to 4.0
(cake dry time/cake form time), the moisture content of coking coal tailings
700
33–36 wt% Moisture

600 24–28 wt% Moisture
500 ∆p = 60 kPa
Feed Solids = 30 wt%
400 Floc Dosage: 80–120 g/t
300
200
100
0
0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4
1
n
min
Figure 13 Test results from different RSA thermal coals—filtration rates
is fairly stable, as Figure 12 shows. The moisture remains in the range of 18

to 22 wt%. At this moisture content, the tailings are easy to handle, either on
conveyor belts or trucks.
R E S U LT S W I T H R S A T H E R M A L C O A L C O N C E N T R AT E
Vacuum disc filters can also be used for the filtration of thermal coal from RSA.
But thermal coal requires thickening and addition of flocculant to generate a
feed flow suitable to be filtered on vacuum disc filters.
Figure 13 shows two test results from different RSA thermal coals. In both
cases, the coal was thickened to about 30 wt% (typical coal thickener underflow
concentration), and flocculant in the amount of 80–120 g/t was added. This
preparation of the feed flow then allows filtration rates of 300–600 kg/m²/h.
This is a solids throughput of 50–100 t/h on standard L4 disc filters in coal
preparation plants. The moisture depends on the thermal coal properties and
can range from low 20-wt% values to high 30-wt% values, as the test results
show.
REFERENCE
Bott, R., Langeloh, T., and Hahn, J. 2004. Latest state of the art of Al-hydrate seed
filtration. ICSOBA, St. Petersburg, Russia, June 15–18.
Dewatering Fine Coal and
Tailings with a Filter Press
G. Prat
ABSTRACT
The coal industry is facing new challenges with the treatment of ultrafine
(0.044 mm × 0) coal slurries. The need to comply with today’s demanding envi-
ronmental rules, the interest of the industry to recover all valuable products, the
lower moisture requirements, and the need to minimize water consumption have
pushed the industry to develop more efficient dewatering technologies capable of
addressing these challenges. This chapter focuses on filter press technology and its
applications in the coal industry. Three typical applications are presented: raw coal
fines or refuse dewatering, froth flotation concentrate dewatering, and screen-bowl
effluent dewatering. Moreover, representative data from test work carried out in
South Africa in a metallurgical coal plant are presented.
Different options are available when designing a filtration plant utilizing
a filter press as the dewatering method; the selection of the right equipment and
auxiliaries will guarantee a quality product, produced at the lowest moisture and
cost. An overview of the different types of filter presses and auxiliaries is detailed.
INTRODUCTION
Filtration has been and is today a common dewatering method employed for
coal slurries with ultrafine solids. In the past, different types of dewatering
technologies were employed, with rotary drums and disc filters being the most
popular. Today, two technologies are widely established for ultrafine coal dewa-
tering: the filter press and the belt press.
The main application of the belt press in the coal industry is the dewatering
of coal tailings, and this technology is widely used around the world. The belt
press has proven to be a technology capable of treating large volumes of coal
tailings slurry, but on the other hand, it is not able to consistently produce a low
moisture product. In addition, the need for a large amount of polymer dosage
279
represents a high operating cost per ton of dry coal produced. The demand by
the coal industry for lower-moisture products, reduced treatment cost, and
fewer environmental impacts is putting the filter press in the forefront when
designing a coal tailings dewatering plant. Filter presses have proven to be the
most effective and reliable technique to meet today’s industry requirements.
Filter presses are, however, more expensive in terms of capital costs than belt
presses.
Another field of application for the filter press in a coal preparation plant
is recovery of the filtrate, which exits a screen-bowl centrifuge. This ultrafine
product, sometimes below 10 μm, is especially valuable in metallurgical coal
plants because of its high price. The combination of a screen-bowl centrifuge
and filter press is an ideal solution for total capture of clean coal and to close a
plant’s water circuit (de Korte 2008).
To have a cost-effective, reliable, and trouble-free dewatering plant, it is
critical to make the right selection of filter press and auxiliaries. Various options
are available in the market; an overview of these is detailed in the following
sections.
A P P L I C AT I O N S F O R F I LT E R P R E S S E S AT A C O A L
P R E PA R AT I O N P L A N T
Dewatering 0.15 mm × 0 Raw Coal or Refuse
Dewatering of 0.015 mm × 0 raw coal or refuse (tailings) has been a typical
installation of filter presses in the coal industry around the world (see Figure 1).
This circuit allows the fine raw coal to be recovered to the clean coal product
or for the fine raw coal or refuse to be discarded and disposed along with the
coarse coal refuse. The final moisture of the product ranges from 18% to 30%
(surface moisture) and is largely dependent on size distribution of the feed; the
finer the product, the more moisture retained.
Dewatering Froth Flotation Concentrate

Another application for the filter press has been dewatering froth flotation
concentrates when the flotation feed has not been deslimed (see Figure 2). A
filter press will capture more than 99% of the fines, not losing a portion of the
ultrafine material as typically occurs with centrifugal technologies. Recovering
this ultrafine coal is especially important for metallurgical coal plants where the
product has a very high value.
Dewatering Fine Coal and Tailings with a filter press 281
0.15 mm × 0 Refuse Filter Press

or Raw Coal Fine
Clear
Water
20%–30%
Solids
Cake
18%–30%
Thickener Buffer Slurry Moisture
Tank Pump
Figure 1 Circuit for dewatering 0.15 mm × 0 raw coal or refuse
0.15 mm × 0
Froth Flotation
Filter Press
Concentrate
0.15 mm × 0
>15% Solids
Refuse Slurry
Clear
Water
20%–30%
Solids
Clean Coal
Product
Thickener To Impoundment Buffer Slurry 15%–20%
Tank Pump Moisture
Figure 2 Circuit for dewatering froth flotation concentrate
Dewatering Ultrafine Material from Screen-Bowl Effluent

A filter press can also be used for dewatering the 0.044 mm × 0 fraction of coal
that is lost in the main effluent (and also the screen drain portion) from dewa-
tering clean coal in screen-bowl centrifuges (see Figure 3). Screen-bowl centri-
fuges have been applied in coal preparation plants to dewater 1 mm × 0 coal to
lower moisture levels. If froth flotation feeds have not been deslimed at about
0.044 mm, then about 40%–50% of the 0.044 mm × 0 entering the screen-bowl
centrifuge is lost through the main effluent. This stream is quite dilute, requir-
ing a thickener to be installed prior to feeding this ultrafine clean coal to the
filter press. Alternatively, this dilute slurry can also be fed directly to the filter
press, representing longer feeding times and thus larger units required to treat
the dry solids. As in dewatering froth flotation concentrates, this is especially
important for recovering high-value metallurgical coal. Testing of this material
in a laboratory plate-and-frame filter press has indicated that moisture levels of
about 15%–20% can be achieved with greater than 99% solids capture.
1.5 mm × 0
Clean Coal
Slurry Clean Coal
Screen-Bowl
Centrifuge
Filter Press
Clean Coal
Product
Clear
Water
20%–25%
Solids
Clean Coal
Product
Thickener Buffer 15%–20%
Slurry
Tank Moisture
Pump
Figure 3 Circuit for dewatering screen-bowl effluent
Figure 4 shows the results from extensive test work performed on metal-
lurgical coal in a South African coal preparation plant. The test was carried
out with different products, both on concentrate of different size distributions
and on tailings from flotation and run-of-mine refuse. The graph in Figure 4
shows the relationship between “filtration rate” (kg/m2/h) versus the “surface
moisture” (wt%). It is shown that for all applications, the filtration rate was
reduced if the surface moisture was lower. To achieve low moisture, longer air
drying time is needed (see Figure 5), increasing total cycle time and decreasing
productivity. The graph in Figure 5 shows that low moisture can be achieved,
both on concentrate and refuse applications.
Figure 5 shows the influence of air drying time on the surface moisture
level. It can be seen that for longer air drying times, lower surface moistures
were achieved. It also shows the influence of particle size distribution on mois-
ture level. At a higher percentage of –37 µm material, higher moisture contents
were produced for the same drying time. This confirms that size distribution of
the product has a great influence on final product moisture.
T Y P I C A L F I LT E R P R E S S D E W AT E R I N G C I R C U I T
Figure 6 illustrates the typical equipment used in a filter press dewatering cir-
cuit, summarized as follows:
1. The filter press is the main piece of equipment of a dewatering circuit.
2. The slurry pump provides the pressure that enables dewatering. The
selection of the slurry pump, its volume, and type are key to the dewa-
tering process. The slurry pump is responsible for the slurry transfer
32
Flotation Conc. Ultrafines
30
Flotation Conc. Normal Fines
28 Flotation Tails
ROM T/U
Surface Moisture, % w/w
26
24
22
20
18
16
14
0 20 40 60 80 100 120 140 160 180 200 220 240 260 280
Filtration Rate, kg/m2h
Figure 4 Surface moisture vs. filtration rate for a South African metallurgical coal
Flotation Conc. Ultrafines

Flotation Conc. Normal Fines
Flotation Tails
32 ROM T/U
30
99% - 150µ
28 94% - 74µ
Surface Moisture, % w/w
81% - 37µ
26
96% - 150µ
86% - 74µ
24
72% - 37µ
22 90% - 150µ
78% - 74µ
20 63% - 37µ
79% - 150µ
64% - 74µ
18 50% - 37µ
16
14
0 40 80 120 160 200 240 280 320 360 400 440 480 520
Drying Time, seconds
Figure 5 Surface moisture vs. drying time for a South African metallurgical coal
into the filter press chambers and for providing the necessary pressure
that produces the solid/liquid separation.
3. The polymer unit is used for preparation and dosing of the flocculant,
which may be needed in some cases when dewatering coal tailings.
Filter Press
Filtrate
Cake
Polymer
Unit
Compressor Receiver Cloth Wash Slurry Buffer

Tank and Pump Pump Tank Sump and
Pump
Figure 6 Typical filter press installation
4. The capacity of the buffer tank will depend on the required storage
time and the available space in the plant.
5. The compressor and receiver are used to provide the air for additional
cake drying. Sizing of the compressed air system is especially impor-
tant when low cake moistures are required.
6. The cloth wash system is used for washing the filter cloth.
7. A collecting sump and pump is used to evacuate the potential slurry
that falls to the ground floor, keeping the floor clean and safe.
Filter Press
The filter press is employed for solid/liquid separation using the principle of
pressure provided by a slurry pump. Many options are available in the market,
and it is important to choose the right equipment focusing on
• Simplicity of design and reduced number of moving parts,
• Strong construction,
• Easy operation and maintenance,
• Optional features such as cloth wash and cake available compression,
• Proven design in the coal industry, and
• References.
Types of Filter Presses

Various types of filter presses are available and include the following:
• Overhead beam or side bar design
• Pull-to-close or push-to-close design
Figure 7 Overhead beam push-to-close filter press
• Quick discharge/slow discharge design

• Quick cloth washing/slow cloth washing design
Two variations of these designs are shown in Figures 7 and 8.
Coal mining and processing is a very demanding industry, often requiring
a 24-hour operation all year long. The filtration equipment designed to work
in the coal industry requires heavy-duty construction with the simplest design
possible and has to perform in this environment. The fewer moving parts, the
less time and parts spent on maintenance and repairs. This equipment needs
to have very high availability for continuous operation. With these factors in
mind, equipment manufacturers do their best to offer the industry the equip-
ment they need. It is important to choose the right equipment for every specific
process, because not all the filter presses are suitable for all solid/liquid separa-
tion applications.
Types of Filter Press Plates

Polypropylene plate. Filter presses can have fixed or variable volume depend-
ing on the plate types selected. Figure 9 shows a fixed-volume polypropylene
filtration plate. The following list provides some of the main characteristics of
these plates.
• Material of construction: polypropylene
• Light weight
• Require stay bosses to avoid plate breakage due to differential pressure
• Cloth wear in the stay boss area
Figure 8 Pull-to-close side bar filter press
Figure 9 Recessed polypropylene plate
• Require cloth washing to remove solids in the sealing area

• Have a limited life, so they need to be replaced due to wear or breakage
• Many models and brands available in the market
Membrane plate. Figure 10 shows a variable-volume polypropylene mem-
brane plate. The following list provides some of the main characteristics of
membrane plates.
• Material of construction: plate from polypropylene; membrane can be
made from polypropylene, rubber, EPM, NBR, TPE
Figure 10 Membrane polypropylene plate
• Light weight
• Welded or detachable membranes
• Can use either air or water as the inflation medium
• Require stay bosses to avoid plate breakage due to differential pressure
• Cloth wear in the stay boss area
• Require cloth washing to remove solids in the sealing area
• Have a limited life, so they need to be replaced due to wear or breakage
• Many models and brands available in the market
Variable plate. Figure 11 shows a variable filter plate. The following list pro-
vides some of the main characteristics for these plates.
• Material of construction: steel core, polypropylene drain plate, and
perimeter rubber seals
• Heavy-duty construction
• Unlimited life
• Watertight chambers
• Flat surface, so no stay bosses required
• Resistant to maximum differential pressure, therefore possible to iso-
late and work with empty chambers
• No cloth required except when dealing with very sticky products
• Replaceable rubber seals
• Variable chamber volume
• Cake compression possible
Figure 11 Steel plus polypropylene plus rubber seals plate
Slurry Pumps
Filter press manufacturers have different alternatives when selecting the slurry
pump that feeds the filter press. The most common pumps are given in the fol-
lowing sections.
Centrifugal Pumps
Figure 12 shows two centrifugal pumps used for feeding a filter press. Centrifu-
gal pumps are a common method of slurry pumping in a filter press application.
These units offer a lower cost compared to other pumping methods, which is
the main driving force for manufacturers to select it. The following list provides
some of the main characteristics of a centrifugal pump:
• Electric drive
• Lower cost compared to positive displacement pumps
• Different configurations and linings available
• Wide range of brands available
• Need of variable-speed drive to provide variable flows and pressure
• Alters chemical treatment of the slurry due to floc shear
• Lower filtration pressure compared to positive displacement pumps
• Low energy efficiency
• High wear due to principle of operation
• High operating expense in terms of energy consumption and wear
parts
Figure 12 Centrifugal pumps
Diaphragm Pumps
Diaphragm pumps are an effective method of slurry pumping and as a pres-
sure provider in a filter press application. Compared to centrifugal pumps,
diaphragm pumps provide higher filtration pressures (up to 16 bar), are less
sensitive to cavitation, provide gentle pumping of materials sensitive to shear,
and are more energy efficient. Figures 13 and 14 show two types of piston dia-
phragm pumps.
Piston diaphragm pump. Some of the main features of these pumps are as
follows:
• Electric drive
• Require multiple pistons for high volumes
• Require a variable-speed drive to provide variable flow rates
Hydraulic piston diaphragm pump. Some of the main features of these types
of pumps are as follows:
• Hydraulic drive by the same hydraulic power unit of the filter press, so
low installed power
• Low operating expense; only one moving part
• Variable flow and pressure, so no need for a variable-speed drive
• Operate with upward or downward suction
Figure 13 Piston diaphragm pump
Figure 14 Hydraulic piston diaphragm pump
Slurry Preparation/Polymer Preparation and Dosing Units

The preparation of large flocculated particles rather than small flocculated par-
ticles tends to have a positive influence on filtration processes in coal tailings
application (Alam et al. 2010). Large floc sizes significantly increase the perme-
ability of the cake. Coal tailings, especially those containing a large percentage
of ultrafine particles and having a high clay content, need flocculant addition.
This is done with the equipment shown in Figure 15.
Anionic and cationic flocculants can be used to enhance the filterability
of coal tailings, and the selection of the right type is based on experience and
testing. After the proper flocculation has been achieved, it is critical to maintain
the size and morphology of them; otherwise, the benefits for filtration will dis-
appear. To do that, it is compulsory to use a positive displacement pump, which
is very gentle with slurry sensitive to shear.
Compressed Air System

As seen in Figure 5, the longer the air drying time, the lower the product mois-
ture. The sizing of the compressed air system depends on the cake permeability;
Figure 15 Polymer preparation and dosing unit
this information is obtained during filtration test work. Alternatively, if test

work is not available, the air consumption has to be based on experience with
previous similar applications.
The sizing of the compressed air system becomes more difficult when mul-
tiple filter presses are in operation in the plant; therefore, an experienced part-
ner has to be selected for this duty. Air pressure, together with correct volume,
is an important factor when designing the compressed air system. Installing a
pressure regulator at the outlet of the receiver is recommended for setting the
air pressure to the specific need of each application. Figure 16 shows a compres-
sor unit that is used in this application.
CONCLUSION
The use of filter presses is a proven method for clean coal and tailings dewater-
ing, achieving low moisture consistently with reduced operational costs. To
obtain low operational costs and consistent performance, it is important to
select the right type of filter press and auxiliaries (Bickert 2012). On coal tailings
applications, there might be the need to flocculate the slurry to enhance/enable
filterability. If this is the case, a positive displacement pump is compulsory as
the filter press feeding mechanism to maintain the floc morphology and size.
To obtain low moisture in the final product, drying by means of compressed air
Figure 16 Compressor unit
is required. Filter presses can recover virtually all the solids in the feed, which
make them especially suited to closing the water circuits in plants.
REFERENCES
Alam, N., Ozdemir, O., Hampton, M.A., and Nguyen, A.V. 2010. Dewatering of Coal
Plant Tailings: Flocculation Followed by Filtration. School of Chemical Engineer-
ing, The University of Queensland, Brisbane, QLD, Australia.
Bickert, G. 2012. Filter Presses for Coal and Tailings—Design Details Make Them
Work. ACPS Symposium, Queensland, Australia.
de Korte, G.J. 2008. Dewatering of Ultra-fine Coal with Filter Presses. Report number
CSIR/NRE/MIN/ER/2008/0062/A. Coaltech.
Deep Cone Thickener at Lone
Mountain Processing Plant
B.K. Gupta and Peter Bethell
ABSTRACT
The Lone Mountain coal processing plant has successfully operated a deep cone
thickener (DCT) since 2007 to concentrate underflow from conventional high-rate
thickeners with 20%–25% solids to 45%–50% solids (by weight). The concen-
trated paste is discharged through a gravity pipeline into the impoundment. Ini-
tially, Lone Mountain discharged the underflow from the conventional thickeners
with 20%–25% solids directly into the impoundment, and subsequently used four
belt presses to concentrate the slurry to 45%–50% solids and trucked the paste to the
impoundment. Replacement of belt presses by the DCT has resulted in savings of
chemical cost, as well as improvement in the plant’s feed rate and operating hours.
The DCT has performed satisfactorily with varying feed density, volumes, and size
distribution of the slurry.
BACKGROUND
The Lone Mountain coal processing plant located in Lee County, Virginia, has
a capacity of 1,200 stph (1,088 mtph). It uses a dense-medium vessel, dense-
medium cyclones, spirals, and column flotation to wash the coal. The coal is
mined from three seams, named Kellioka, Darby, and Owl in Kentucky and
belted to the plant through a tunnel in the Little Black Mountain bordering
Virginia and Kentucky. The slurry from the underflow of two 50-ft (15.2-m)
conventional high-rate thickeners was initially pumped into the impound-
ment. In 1998, slurry from the impoundment leaked through old, unmapped
underground works. To reduce the risk of such an occurrence, Lone Mountain
started concentrating the thickener underflow with 20%–25% solids to 45%–
50% solids (by weight) by belt presses before discharging it into the impound-
ment. The paste from the belt presses was hauled and dumped by truck into
the impoundment. The refuse from the dense-medium vessel, dense-medium
293
cyclones, and spirals was deposited on top of the paste by trucks and compacted
in layers. Because of high chemical consumption, high maintenance costs, and
frequent bottlenecks associated with belt presses, alternatives to the belt press
were evaluated.
One alternative was the deep cone thickener (DCT), which has been
used successfully to make paste in diamond, gold, and iron ore tailings since
the 1970s but had not been used in coal tailings. After an extensive literature
review, visits to diamond mines in South Africa, and lab tests, a pilot test was
conducted at the Lone Mountain plant. It indicated that a DCT could provide
45% to 55% solids paste and very good quality clarified water for reusing in the
plant’s wash circuit, with a small fraction of chemicals compared to the belt
press operation. After a successful pilot test, the decision was made to construct
a DCT to replace the belt presses.
D E E P C O N E T H I C K E N E R L AY O U T
The Lone Mountain plant, which is located at an elevation of 1,950 ft (594 m),
is approximately ½ mile (0.81 km) from the impoundment. A DCT measur-
ing 50 ft (15.2 m) in diameter and 60 ft (18.3 m) in height was constructed
at an elevation of 2,335 ft (712 m) adjacent to the impoundment by Westek
and was commissioned in mid-May 2007. The underflow from two existing
conventional high-rate thickeners with 20%–25% solids, which was pumped to
the belt presses before, is now pumped to a small sump in the plant at 1,980 ft
(604 m) elevation. It feeds two 6/4 high head Warman centrifugal pumps
working in series. They pump the slurry to the top of the DCT to an elevation
of 2,415 ft (736 m), through a pipeline consisting of 10 in. (254 mm) and 8 in.
(203 mm) diameter polyethylene pipes. Liquid anionic flocculant and liquid
cationic flocculant are added in the feed flume and the center well of the DCT.
Four 7.5 hp (5.6 kW) motors drive a slow moving rake in the DCT. The solids
in the slurry are essentially concentrated by the weight of the slurry. As the rake
rotates, water released from the slurry rises through the space behind the pick-
ets, which support the rake. Clarified water created in the DCT comes back to
the plant by gravity for reuse.
The paste is discharged into the impoundment by gravity through a 10-in.-
(254-mm-) diameter polyethylene pipeline laid at a 14-degree decline. The pool
elevation in the impoundment is approximately 2,170 ft (661 m) at present. It
will rise over time up to 2,290 ft (698 m). A pond was dug approximately 130
ft (40 m) above the DCT and receives water pumped from the underground
mines across the hill in Kentucky through a pipeline in the 2-mile- (3.2-km-)
Deep Cone Thickener at Lone Mountain Processing Plant 295
Figure 1 Deep cone thickener circuit
Figure 2 Deep cone thickener
long conveyor tunnel. This water is used for the mixing, purging, and makeup
at the DCT and the conventional thickeners at the plant. Figure 1 gives a sim-
plified circuit of the DCT layout. Figures 2 and 3 show the DCT and the slurry
and clarified lines, respectively.
The process at the DCT is operated and remotely controlled from the
plant operator’s control room through a programmable logic controller (PLC)
Figure 3 Slurry and clarified water pipelines
system. No one is located at the DCT location. Several cameras also help the
operator to monitor the DCT operation. The slurry feed line has a nuclear
density gauge and flow meter to continually provide the slurry density and
flow rate for the required dosage of liquid flocculants. The PLC calculates and
controls the chemical dosages required. A pressure gauge at the bottom of the
DCT is utilized for the operator to set the pressure limits between which the
DCT needs to operate. The gravity line for paste disposal to the impoundment
also has a flow meter and nuclear density gauge to measure the flow rate and
density of the paste being discharged from the DCT to the impoundment.
Figure 4 shows a simplified circuit of the chemical feed system at the DCT. Fig-
ures 5 and 6 show the gravity discharge pipeline and liquid flocculant pumps,
respectively.
D C T O P E R AT I O N
The underflow streams from two 50-ft- (15.2-m-) diameter conventional high-
rate thickeners are pumped by two centrifugal pumps with variable frequency
drive into a small sump in the plant building. The particle size distribution in
the underflow tailings averages 5% each for 1 mm × 0.25 mm (60 mesh) and
0.25 mm × 0.15 mm (100 mesh), 12% for 0.15 mm × 0.045 mm (325 mesh),
and 78% for 325 mesh × 0. Two high-head centrifugal pumps, working in
series, pump the slurry into the flume of the DCT. The flow rate and density
of the feed slurry are measured continually by a flow meter and nuclear density
Figure 4 Deep cone thickener chemical system
Figure 5 Gravity discharge pipeline
Figure 6 Liquid flocculant pumps
gauge inputted into the PLC system. The slurry flow rate averages 1,300 gpm
(295 m3/h), and its specific gravity averages 1.13, which equates to 23% solids
(by weight) and 84.5 stph (dry) (76.6 mtph). Anionic and cationic chemicals
for use in the DCT are stored in 5,500-gal (20.8-m3) tanks located beside the
DCT. A 0–3 gpm (0–0.68 m3/h) transfer pump draws anionic chemical from
the storage tank and pumps it to a 3,500-gal (13.2-m3) mix tank after diluting it
with water flowing through a flow meter and dole valve at 80 gpm (18.2 m3/h)
and 85 psi (586 kPa). A 0–20 gpm (0–4.5 m3/h) flocculant feed pump draws
the diluted anionic chemical from the mix tank and pumps it to the top of the
DCT into the flume and the center well. As mentioned previously, the feed
rates of both the anionic and cationic chemicals are varied by the PLC accord-
ing to the incoming slurry’s flow rate and density. Several adjustable windows in
the center well allow the clarified water to flow into the center well and dilute
the incoming slurry to 8%. It was determined that the anionic chemical is most
effective at 8% solids concentration in the slurry.
The solids in the slurry gravitate to the bottom of the DCT where a slow-
moving rake concentrates it to about 50% solids. Water released from this
concentration flows upward through the space created in the paste behind the
vertical pickets by which the rake is suspended. The cationic chemical fed into
the center well clarifies the water sufficiently for reuse in the plant. When the
bed level for solids set by the operator is reached, the PLC automatically opens
the pneumatically operated automatic knife valve in the gravity discharge pipe-
line. A flow meter and nuclear density gauge measure the flow rate and density
Figure 7 Deep cone thickener mass balance
of the paste discharged, and it is recorded continually in the PLC system. The
paste with 45%–50% solids reaches the pool area of the impoundment through
a 10-in. (254-mm) polyethylene pipeline laid at a 14-degree slope. The friction
in the discharge pipeline reduces the shear stress of the paste, and by the time
it comes out of the pipe, it flows like water and has almost zero degree angle
of repose. In the pool area, the DCT paste further consolidates and dries up
by capillary action and evaporation, unlike the paste made by the belt presses,
which forever stayed in paste form because the large amount of chemicals in
the belt press operation inhibited capillary action and thus prevented drying
and consolidation.
After a few months for fine-tuning and tweaking, the DCT has since
worked well and requires very little attention and maintenance. All DCT
operation is PLC controlled remotely without any operational labor. Introduc-
tion of the DCT in the plant circuit has resulted in improvement of the overall
$0.30
$0.25
$ per Plant Feed, raw tons
$0.20
$0.15
$0.10
$0.05
$0.00
Jul-05
Oct-05
Jan-06
Apr-06
Jul-06
Oct-06
Jan-07
Apr-07
Jul-07
Oct-07
Jan-08
Apr-08
Jul-08
Oct-08
Jan-09
Apr-09
Jul-09
Oct-09
Jan-10
Apr-10
Jul-10
Oct-10
Jan-11
Apr-11
Jul-11
Oct-11
Jan-12
Apr-12
Month-Year
Figure 8 Monthly chemical cost at the Lone Mountain plant for the two conventional
thickeners and deep cone thickener
operating time of the plant compared to belt presses, which had frequent down-
time and caused frequent reduction in coal feed rate to the plant. The capillary
action and consequent drying of the DCT paste in the pool area has resulted
in additional increase in % solids and consequently a lesser volume of the paste.
A paste of up to 1.38 specific gravity equivalent to 55% solids has been pro-
duced by the DCT. However, it is normally operated to produce paste of 1.31
specific gravity (48% solids) to keep the gravity discharge operation trouble free
and to minimize chemical consumption. If there is shortage of makeup water,
the chemical dosage can be readily increased to produce a higher % solids paste.
Figure 7 shows the mass balance of the DCT.
CHEMICAL CONSUMPTION AND COST

After the belt presses were discontinued and the DCT operation was started,
a dramatic reduction in the consumption of chemicals was experienced. Over-
all, chemical consumption for the plant for the conventional and the DCT is
less than a quarter of the consumption with belt presses. Total chemical usage
rate at the plant for both anionic and cationic chemicals is estimated at 174 g/
metric ton of fine refuse with the DCT, compared to 808 g for the belt press
operation.
The chemical cost for the plant for both the conventional thickeners and
DCT has averaged about $0.087 per raw short ton fed to the plant compared to
$0.25 for the belt press, in spite of multiple price adjustments in the unit cost of
the chemicals. Figure 8 shows the cost of chemicals at the Lone Mountain plant
in dollars per raw short ton of plant feed for the conventional thickeners with
belt presses until May 2007 and for the conventional thickeners and DCT since
then. The yearly savings in chemical cost has been more than a million dollars.
Additional savings has been realized by discontinuation of four belt presses and
the truck, which was used to haul paste to the impoundment. Overall savings
with the DCT, which is estimated at $1.6 million per year, has resulted in a very
attractive rate of return on the capital invested for the DCT system.
CONCLUSIONS
A DCT to replace belt presses at the Lone Mountain coal processing plant
has performed satisfactorily since mid-2007. The unit is operated to produce
45%–50% solids paste from 100 mesh × 0 tailings, though up to 55% solids
paste has been obtained. Reduction in chemical consumption and maintenance
cost has resulted in a payback of less than 2 years. The paste from the DCT
further increases its % solids level due to capillary action and evaporation in
the impoundment. The paste in the pool area, which was dug by an excava-
tor after the DCT had operated for some time, showed that the top 3–4 ft
(0.91–1.22 m) had solidified to more than 80% solids, but the material below
it from the belt press was still soupy.
Centribaric Operations Update
J. Franklin, W. Schultz, and T. Estes
ABSTRACT
Decanter Machine began developing novel centrifuge technology in January 2008
with the manufacture of a prototype unit that could be used for field study of the
technology. The prototype unit was taken to Walter Energy in April 2008, where
field tests were encouraging at both the No. 7 and No. 4 preparation plants. The first
commercial unit of the technology was purchased by Walter Energy for the Mine
No. 7 preparation plant in September 2009. The unit was started up in February
2010. A single unit was selected to confirm the pilot testing and it required mini-
mal capital to install due to the existing plant layout. The successful installation of
the first unit led to subsequent purchase orders to process the entire effluent streams
for both preparation plants. This chapter discusses the process improvements result-
ing from the installation of the Centribaric units at both preparation plants.
INTRODUCTION
Walter Energy operates the Jim Walter Resources ( JWR) No. 7 mine, a deep
shaft coal mine near Brookwood, Alabama, that produces high-quality metal-
lurgical coal from the Blue Creek seam. Typical raw coal qualities are 90 to
95 HGI (Hardgrove grindability index), and a resulting plant feed particle
size distribution with 30% –1 mm. Raw coal from the mine is hoisted to the
surface where it is stockpiled before being conveyed to the preparation plant.
The original JWR No. 7 preparation plant was designed and constructed by
McNally in the late 1970s and consisted of four parallel circuits processing raw
coal with dense-medium cyclones and conventional froth flotation. Feed to
the dense-medium cyclones was deslimed at 0.5 mm. Flotation cells processed
the –0.5 mm raw coal. Clean coal was dewatered using horizontal vibrating
basket centrifuges and vacuum disc filters. Typical total plant moistures were
10% to 12% moisture, and the vacuum filters typically produced moistures
between 20% and 30%. Contractual requirements for total moisture were often
exceeded using the original process flowsheet. In the late 1980s, the plant was
303
Raw Coal Met Cyclone

Cyclones
Steam
Cyclone
Spirals Float Cells
Refuse
Distributor
Belt
Box
Screen Bowl Screen Bowl
Met Met
Product Product
Raw Fine
Middlings Effluent
Coal Met
Sump Sump Thickener
Sump Sump
Slurry
Pond
Figure 1 JWR No. 7 preparation plant flowsheet of fine coal circuit before
installation of the Centribaric screen-bowl centrifuges
expanded by McNally with the addition of two identical circuits, bringing the
plant feed capacity to 1,200 t/h in six circuits.
In 1996, Sedgman designed and constructed a fine coal circuit upgrade
with the addition of spirals and replacement of the vacuum disc filters with
screen-bowl centrifuges. In the current configuration, raw coal is sized at 1 mm
and 0.15 mm (100 mesh). Raw coal coarser than 1 mm is processed by twelve
711-mm (28-in.) dense-medium cyclones and is dewatered with horizontal
vibrating-basket centrifuges. The 1 mm by 100 mesh size fraction is classified
using cyclones and then processed using thirty triple-start single-stage pri-
mary spirals with the middlings reprocessed in three triple-start single-stage
secondary spirals. Raw coal finer than 100 mesh is processed by six banks of
conventional froth flotation cells. Clean coal from the spirals and flotation cells
are combined and dewatered using five screen-bowl centrifuges. The current
plant feed capacity is 1,400 t/h. The total plant moisture after the Sedgman
fine coal upgrade was completed was approximately 10%, and the screen-bowl
centrifuges typically produce between 12% and 14% moisture. Figure 1 depicts
the JWR No. 7 preparation plant flowsheet prior to the installation of the
Centribaric centrifuges.
Centribaric Operations Update 305
40 100
80
30
Tons per Hour
60
Percent
20
40
10
20
0 0
Refuse Decanter Primary Cyclone Spiral Cell
Stream Effluent Refuse Refuse Tails
TPH 30.0 5.2 2.5 1.8
Percent 75.9 13.2 6.3 4.6
Cum % 75.9 89.1 95.4 100.0
Figure 2 Coal tonnage lost to the refuse stream prior to Centribaric installation
The updated flowsheet and process conditions designed by Sedgman

offered improved recoveries in the spiral and flotation circuits and improved
total plant moistures with the installation of the screen-bowl centrifuges. How-
ever, some of the improved plant moistures were at the expense of discarding
some of the ultrafine (–45 μm) clean coal through the screen-bowl effluents
to the refuse thickener. The main effluent from the screen-bowl centrifuges at
JWR No. 7 typically contains 4%–6% solids by weight and has an ash value
of approximately 15%. This material is processed in the static thickener and
pumped to the slurry impoundment. Total loss through the screen-bowl efflu-
ent at JWR No. 7 is 25–30 t/h, which on an annual basis exceeds 100,000 t.
Figure 2 depicts the coal losses by circuit prior to the installation of the Centri-
baric screen-bowl centrifuges at the No. 7 preparation plant.
C E N T R I B A R I C C E N T R I F U G E I N S TA L L AT I O N
In March 2011, the Centribaric centrifuge upgrade was complete at the No. 7
preparation plant. The flowsheet calls for the collection of the main effluents
and screen effluents from the standard screen-bowl centrifuges, which are
pumped to a distributor for gravity feeding of the Centribaric units. The main
effluent from the Centribaric units reports to the refuse thickener. The screen
drain from the Centribaric units is recycled. Figure 3 shows the modified flow-
sheet that incorporates the Centribaric centrifuges.
Middlings
Raw Coal Cyclone
Fine Clean
Cyclones
Cyclone
Secondary
Spirals
Float Cells
Distributor
Spirals
Box
Centribaric
Screen Bowls
Refuse Distributor
Belt Box
Screen Bowls Screen Bowls
Met Met
Product Product
Raw Fine
Middlings Effluent
Coal Met
Sump Sump Thickener
Sump Sump
Slurry
Pond
Figure 3 JWR No. 7 preparation plant flowsheet after installation of Centribaric

centrifuges
The most immediate and noticeable change with installation of the Cen-
tribaric centrifuges was the immediate reduction in the amount of stable froth
on the refuse thickener. Significant economic benefit of the installation was the
reduction in the amount of solids reporting to the refuse through the effluents
of the centrifuge. The effluent percent solids by weight reporting to the thick-
ener have ranged between 0.5% and 1.0% and have an ash value between 30%
and 50%. This is compared to the standard screen-bowl centrifuge effluents of
4% to 6% solids and 15% ash. Figure 4 depicts the coal losses by refuse stream
after installation of the Centribaric units.
The amount of coal lost to the thickener has decreased from 25 to 30 TPH
to less than 5 TPH of dry solids. The removal of the coal from the refuse stream
has resulted in a decrease in the dry solids pumped to the fine slurry disposal
impoundment by 15%. This equates to an annual volume of approximately
1.02 × 105 m3 (3.6 × 106 ft3) of future disposal capacity per year. The refuse
thickener chemical treatment costs have decreased because of the reduced ton-
nage reporting to the thickener. Figure 5 shows the impact of the Centribaric
unit installation. The thickener underflow ash value increased, reflecting the
removal of the fine coal lost in the effluents of the standard screen-bowl units.
Centribaric Operations Update 307
14 100
12
80
10
Tons per Hour
8 60
Percent
6 40
4
20
2
0 0
Refuse Primary Cyclone Decanter Spiral Cell
Stream Refuse Effluent Refuse Tails
TPH 5.2 4.5 2.5 1.8
Percent 37.1 32.1 17.9 12.9
Cum % 37.1 69.3 87.1 100.0
Figure 4 Coal tonnage lost to the refuse stream after Centribaric installation
80
Underflow
Cell Tails
70
60
Ash Value, %
50
40
Centribaric Screen Bowls

Installed
30
20
June 1, 2011 July 14, 2011 August 31, 2011
Sample Date
Figure 5 Comparison of ash values of the thickener underflow and flotation tails
before and after installation of Centribaric screen-bowl centrifuges
CONCLUSION
The installation of the Centribaric units at the Walter Energy preparation
plants provided significant economic return with the capture of the fine coal
lost to the refuse stream. In addition to the increased coal recovery, other
benefits that can be attributed to installation of the Centribaric units include
extending the life of an existing slurry disposal facility with the capture of fine
coal and reduced chemical consumption for the refuse thickener.
Belt Filter Press in Coal Tailings
Dewatering—A Comprehensive
Economic, Design, and Process
Analysis
Matt S. Fenzel
ABSTRACT
Increasing environmental regulations have created new and changing demands
for coal preparation plant operators. Critical among these is the disposal of fine coal
tailings. Belt filter presses have been utilized as a cost-effective option for dewater-
ing fine coal tailings for ultimate disposal for decades. As regulations have become
more stringent regarding disposal options, plant designers, operators, and equip-
ment suppliers have developed comprehensive economic, design, and process evalua-
tion techniques to ensure proper equipment selection and system design. Plant and
tailings disposal design, material characterization, site-specific dewatering options,
process performance capabilities, and capital and life-cycle costs are important
considerations to ensure a successful tailings dewatering system and disposal design.
INTRODUCTION
Global changes in coal tailings (refuse) disposal, requirements, regulations, and
coal washing processes have required the industry to evaluate the numerous
variables impacting dewatering technology to ensure a sustainable, economic
decision in mine waste disposal. As of 2001, there have been at least 713 active
fresh-water and slurry impoundments in the United States that are monitored
by the Mine Safety and Health Administration. This is in addition to an aver-
age of 70–90 million tons of fine coal tailings that are reported per year (NRC
2002). Considerations of rehabilitation costs, difficulties, and time associated
with slurry impoundments have also caused further evaluations (Harriman
2012). These statistics along with other variables contribute to why tailings
309
treatment remains the most difficult and expensive area of coal production
(Budge et al. 2000).
Today, it has become increasingly important to provide detailed and
comprehensive analysis in prefeasibility and economic evaluations of not
only whole plant designs, but individual equipment and components, so as
to ensure plant optimization and proper handling of tailings. Current trends
in coal preparation practice and development are influenced by three major
factors of legislation: changes in coal mining practice, quality of raw coal, and
pressures on commercial performance (Budge et al. 2000). In the case of belt
filter press technology, it has been a practical option for decades in fine coal tail-
ings dewatering as well as many other industries worldwide and has assisted in
eliminating problems associated with process waste streams. In the recent past,
dewatering technology options such as recessed chamber and membrane filter
presses have been introduced. However, the belt filter press, when compared
to other mechanical dewatering options, is a low capital cost option (Bickert
2004), with similar operating costs of alternative tailings technologies such as
filter presses (Mathewson et al. 2007). As technology suppliers, design firms,
and end users adjust with the changing variables introduced by new mining
practices, fine coal recovery technology, and regulatory issues, it is important to
perform a site-specific economic and process analysis when selecting the most
appropriate fine tailings dewatering and refuse disposal option.
B E LT F I LT E R P R E S S T E C H N O L O G Y
The belt filter press (Figure 1) is a mechanical dewatering device extensively
used for dewatering solids suspended in water. Since the 1980s, belt filter
presses, also referred to as twin belt filters, belt filters, or multi-roll filters, have
been used to dewater ultrafine coal tailings and other thickened mineral slur-
ries. The goal of the technology is to separate liquids from solids while creating
a product, referred to as “cake,” in a stable, easy-to-handle, stackable solid phase.
Key considerations in the selection of belt filter presses can include
• Continuous operation,
• Sufficient dewatering capabilities under variable conditions,
• Relatively low capital and operating costs, and
• High mechanical reliability and equipment availability.
Operating Principles
The belt filter press is considered a mechanical, pressure filtration, dewatering
device. However, prior to the introduction of feed slurry and the mechanical
Belt Filter Press in Coal Tailings Dewatering 311
Source: Phoenix Process Equipment Company 2012.
Figure 1 Multiple belt filter press installation
dewatering phases, the tailings slurry requires preconditioning (mixing) with

flocculating chemicals (polymers). Optimal feed densities (in mining applica-
tions) for mixing slurry and polymer are in the range of 25%–35% weight by
weight (w/w), which is near the terminal underflow densities produced from
sedimentation thickening devices conventionally used in coal preparation
plants. The mixing of slurry and polymer results in the agglomeration of solids
and free water release as the feed enters the belt filter press. The free water is
allowed to drain through the belt as it moves the material through a series of
moderate to high compression sections before it can be discharged in a surface-
dry, stackable state. The filtrate that has been drained and separated from the
material is collected in a common sump and is typically recirculated to a clarifier/
thickener for plant reuse.
The seven distinct operational stages in a belt filter press are shown in Fig-
ure 2 and defined as follows:
1. Flocculation: The slurry is dosed and mixed with the correct dewater-
ing chemicals (polymers).
2. Feed: The feed box distributes the flocculated slurry evenly across the
drainage (lower) belt’s width.
3. Gravity drainage: Free water drains through the drainage (lower) belt,
leaving a mat of solids. In many cases, plow devices are used to enhance
water release.
Untreated Slurry 2. Feed 3. Gravity 4. Wedge

Stage Drainage Stage Stage
Lower Belt
1. Flocculation Upper Belt

Stage
6. Discharge
5. Stage
High-Pressure
Stage
7. Belt Wash
Stage
Filtrate Water Filtrate Sump
Dewatered Cake
Figure 2 Belt filter press showing the seven operational stages
4. Wedge: Upper and lower belts come together to converge on the solid
material to initiate pressure dewatering. External wipers wick away
moisture.
5. High pressure: The solids are compressed and sheared between the
belts while traversing a series of decreasing-diameter S-rolls, releasing
additional water.
6. Discharge: The cake is scraped from the belts, releasing suitable mate-
rial for transport by conveyor.
7. Belt wash: The upper and lower belts cycle through a wash-water box
to ensure a clean dewatering filter belt for the next dewatering cycle.
Process Principles
The belt filter press is typically used to process slurries of fine solid particles
with nearly all particle sizes smaller than 600 µm. The belt filter press is particu-
larly effective over other dewatering technologies when slurries are ultrafine, or
have a high degree of colloidal solids, or have solids that do not settle over time
without polymer addition (Shields 1992).
In belt filter press applications, the small fine particles must be flocculated
during pretreatment (flocculation stage). For fine coal tailings, the typical pre-
treatment requires a dual polymer regiment of an anionic flocculant followed
by a cationic coagulant with a degree of mixing occurring in between stages.
The flocculated structure created by the introduction of an anionic flocculant is
introduced to a small dosage of a cationic coagulant that will bind any remain-
ing finely dispersed clays, while breaking down and tightening the existing
flocculated structure that will aid in releasing the remainder of its interstitial
water when subjected to pressure from the belt filter press (Budge et al. 2000).
The anionic flocculant is typically a medium-weight, medium-charge-density
polyacrylamide chained molecule that is received in either dry (powdered)
form or as an emulsified liquid (emulsion).
It is also necessary for the anionic polymer to be correctly prepared. It must
be educed and made with clean water, then stirred, separated, and aged by a
polymer preparation system to ensure maximum effectiveness and efficiency
of the polymer. This preparation will require time, suitable water quality, and
an efficient polymer system to provide polymer/water blending, mixing, and
aging. Too much shear or too fast activation will result in either poor floccula-
tion or unnecessary cost of operation (Alderman 2010).
Sizing and Performance Considerations

Belt filter presses are sized on the capacity at which they can handle on a per-
meter-of-belt-width basis. In coal preparation, commercially available unit filter
belt widths typically range between 1.2 m and 3.0 m and the capacities can
range from 3 to 14 dry t/h/m (short tons per hour per meter) of belt width,
depending on feed conditions. Along with the capacity or throughput of a
machine, other critical performance measures are the discharge cake, measured
in terms of moisture content (% w/w) or solids content (% w/w), and the
amount of material that has been recovered (e.g., >90% yield).
As with any coal fines dewatering technology, there are many variables that
have a direct impact on belt filter press performance. Variables associated with
feed conditions are given in Table 1.
Operating variables and conditions can include the following:
• Proper polymer application
• Mixing (polymer with fines slurry)
• Belt speed
• Belt tension
• Belt washing
• Proper mechanical maintenance
In thickened coal tailing slurries, one of the most common variables attrib-
utable to performance characteristics is the particle size distribution (Bickert
et al. 2007). In general terms, feeds that contain higher percentages of coarser
particles tend to achieve higher capacities, lower cake moistures, and lower
chemical demands. A summary of performance levels corresponding to feed
particle size distribution are found in Table 2.
Table 1 Variables of feed slurry and performance impact

Variable Performance Impact
Particle size distribution Coarser correlates to increased process performance measures.
Specific gravity of solids More dense solids correlate to increased process performance
measures.
Volatile content Higher levels of volatile content that are not attributable to coal
correlate to reduced process performance measures.
Ash content Higher levels of ash content correlate to reduced process
performance measures.
Feed density Higher levels of feed moisture correlate to reduced process
Homogeneity of slurry More homogeneous slurry correlates to increased process
Feed consistency More consistent feed conditions correlate to increased process
Importantly, many of the variables, performance characteristics, or operat-

ing conditions are considered dependent or interrelated when looking at one
specific variable and its impact. With such a wide range of variables and particle
size distributions that can impact not only belt filter press processing rates but
nearly all mechanical dewatering technologies, it is necessary that plants take
critical steps to ensure that the belt press filter system design takes a comprehen-
sive approach to ensure success.
PLANT DESIGN AND SIZING CRITERIA

C O N S I D E R AT I O N S
Comprehensive plant economic and process evaluations are becoming increas-
ingly critical to accurately identify, address, and assess decisions on reject fines
circuits. Plant designs first require two critical analyses or assumptions: the
maximum rate of slurry and solids to be processed and the rate at which a
particular piece of equipment can adequately process the material. From this, a
plant will be able to determine at least the quantity and type of belt filter press
it would need to dewater the waste stream.
During prefeasibility and evaluation phases, it is also vital to have a clear
understanding of what the plant’s goals are for the dewatering stage to be a sus-
tainable contributor to the success and overall operation of the facility. Over-
all plant designs, life-of-mine economic analysis, environmental studies and
regulations, in addition to bench-scale and pilot testing should be conducted
regardless of whether the project is for a new plant or an expansion. However,
Table 2 Nominal belt filter press capacity for coal fines

Coarse Medium Medium/Fine Fine Ultra-Fine
high content mostly silt- mostly silt- a mixture of some silt-sized
of silt-sized sized grains, sized grains, silt-sized (63 (63 µm × 5
grains and and little to no with a small µm × 5 µm) µm) with high
Description larger, and clay content but significant and clay-sized clay-sized
of Grain-Size little to no clay clay content (<5 µm) (<5 µm)
Distribution content content content
Typical <50% passing >50% passing >75% passing >90% passing >90% passing
Grain Size % 63 µm 63 µm (230 63 µm (230 63 µm (230 63 µm (230
Passing, µm (230 mesh) mesh) and µm) and <20% mesh) and mesh) and
(mesh) <20% passing passing 5 µm 20%–40% >40% passing
5 µm passing 5 µm 5 µm
Typical 55–65 58–68 60–70 65–75 70–85
Range of
Feed Mois-
ture, % w/w
Typical 35–45 32–42 30–40 25–35 15–30
Range of
Feed Solids,
% w/w
Typical 10–14 9–12 8–11 7–10 4–7
Range of
Maximum
Capacity,
t/h/m
Typical 60–120 60–120 50–110 50–100 30–90
Range of
Maximum
Capacity,
gpm/m*
Typical 22–30 25–35 30–40 35–45 40–55
Range of
Cake Mois-
ture, % w/w
Typical 70–78 65–75 60–70 55–65 45–60
Range of
Cake Solids,
% w/w
Typical 90–98 90–98 90–95 85–95 60–95
Range of
Solids Yield
Rate, % w/w
Source: Alderman 2010.
*Gallons per minute (gpm) per meter.
in many cases the selection of dewatering equipment is usually governed by

capital expenditures set aside for the project (Budge et al. 2000). Experience
indicates that a site-specific evaluation yields the best opportunity for a prop-
erly designed and workable dewatering filter system and tailings disposal plan.
Economic and Life-Cycle Analysis

New plants or expansions typically evaluate the option of constructing slurry
impoundments to handle thickened fine coal tailings. Traditionally, this
has been the low-cost alternative for tailings. However, costs and regulation
requirements have increased for this disposal method. Consider the rehabilita-
tion costs that are required for coal preparation on a per-area basis. Peabody
Coal estimated that rehabilitation costs for tailings dams are approximately
US$100,000/ha (per hectare) compared with rehab costs of only US$20,000/
ha and US$50,000/ha for filtration and co-disposal options, respectively, in
Australia (Harriman 2012). Once this disposal method has been determined to
be at substantial costs, belt filter presses are typically considered as an alterna-
tive disposal solution and can be economically evaluated when compared with
traditional disposal methods.
As an example, a study conducted through the Australian Coal Associa-
tion Research Program (ACARP) in 2004 performed an economic analysis on
indicative capital and operating costs for a variety of tailing dewatering technol-
ogies, including belt filter presses. The calculations from this study (see Tables 3
and 4) are under the following assumptions:
• Flocculent and coagulant cost: $1.4 per pound.
• Power cost: $0.056/kWh at 70% of installed kilowatt.
• Exclusive of civil and project costs.
• Utilization of 6,350 h/a (144 hours × 49 weeks × 90% utilization).
• Does not consider tax regime, variable capital costs, and inflation.
• Maintenance—actual plant data used, where available, otherwise
5%–20% of capital costs used.
• Labor: costs were not considered.
• Capital: 15% of capital/installation costs are calculated per year. This
covers interest and depreciation.
• Pricing is in Australian dollars (A$) unless otherwise noted.
• Cost per ton ($/ton) is based on feed tonnage or throughput of the
machine.
Capital costs for the belt filter press indicate averages of A$0.44/t, whereas
operating costs were higher (due to chemical demand) at A$1.67/t. With this
or any study relating to supplied technical data on belt filter presses and other
tailing technologies, it is important to note the following:
• Information is from different sources and based on different practices.
Table 3 Economic considerations for a belt filter press

Size 2.5–3.0 m
Source of Data Supplier, Operating Plants
Performance Minimum Typical Maximum
Throughput, t/h 7.70 18.7 27.6
Cake Moisture, % w/w 25.0 35.0 45.0
Consumption
Chemicals, lb/t 0.10 0.90 2.00
Power, kW/t 0.05 0.06 0.16
Maintenance, $/t 0.25 0.38 0.91
Capital per ton, A$1,000/t 9.07 18.1 58.1
Source: Bickert 2004.
Table 4 Total economic considerations for a belt filter press

Total Costs Capital Operating Total
Minimum A$/t 0.22 0.44 0.64
Maximum A$/t 1.38 3.80 5.17
Average A$/t 0.44 1.67 2.09
Source: Bickert 2004.
•
Suppliers will sometimes underestimate operating costs.
•
Plant operations may not be optimized.
•
Export data sometimes cannot be directly compared.
•
Coal tailings differ significantly and thus performance and other char-
acteristics will also change depending on the material type (Bickert
2004).
By considering some of the expenditures associated with dewatering equip-
ment that include capital costs and operating expenses such as wear parts, labor,
power consumption, and chemical consumption, a study will be able to deter-
mine a close estimate of how much it would cost to install, operate, and main-
tain a belt filter press at a specific site. In conjunction or even before performing
economic analysis of belt filter press technology, a careful plant and process
design analysis should be performed to accurately size the dewatering system.
Plant Process Analysis

In coal preparation plants, it is typical for designs to include a sedimentation
thickening (clarification) unit to treat reject fines and recover water from
upstream washing and separating processes. The management and disposal

is increasingly critical to continued mining operations with changes in water
availability and costs (Brown 2003). Depending on the wash circuit, it may
include effluent and fines from cyclones, screens, spirals, centrifuges, teeter-
bed classifier tailing, and other upstream classifying equipment. Traditionally,
thickened material, or thickener underflow, ranging from 25%–35% w/w, is
then discharged into ponds or coal impoundments that have been constructed
on the mine site or alternatively report to mechanical dewatering systems. As
noted, the reject coal fines generally have a particle size distribution under
600 µm (100% passing) and, in many cases, a large percentage of particles pass-
ing 38 µm, all of which require chemical treatment for sufficient agglomeration
and settling of fines. Recently, plants have found a high presence of sub-38
µm, especially if the mine mineralogy contains expandable clays, resulting in
numerous issues in the impoundments, thickening, and/or dewatering systems
installed. This along with other contributing factors cause preparation plants
to revaluate how they dispose of these thickened fines, which include the
following:
• Increased discharge regulations,
• Environmental regulations,
• Limited impoundment space,
• Overall impoundment costs, and
• Need for water recovery and reuse.
After a plant has evaluated and concluded that an impoundment disposal
method is no longer viable, the option for a belt filter press would usually be
considered for disposing of tailings in a conveyable, easy to handle, stackable
solid phase.
Figure 3 represents a typical plant schematic that includes a belt filter
press. Thickened coal tailings from a thickener report to a belt filter press and
are dewatered into a solid phase. The filtrate, typically containing less than
10% of fines captured discharged along with free water and belt wash water,
are collected and recirculated back to the thickener in most installations. This
effectively creates a closed loop recovery system for water and solids.
Knowing this information, the plant designer would need to first deter-
mine the amount of slurry needed to be processed along with the quantity
of machines needed to adequately handle the waste stream. As a generalized
process sizing example, the following should be considered.
Maximum (or worst case) rate of slurry and solids reporting to belt filter
press(es):
Thickener Polymer
Addition Fine Coal
Tailings
Clarified Water
(<200 ppm)
Dilution Water Pump
Pump Belt Filter Press
Polymer
Addition
Dewatered Fines
(Cake Conveyor)
Pump
Figure 3 Plant schematic with a belt filter press
• 60 dry t/h
• 580 gpm
Established laboratory or pilot demonstration equipment capacity results for
specific material:
• 8 dry t/h/m of belt width
• 77 gpm/m of belt width (assumed density of 35% w/w)
In the assumption that this particular plant would utilize a 3.0-m belt filter
press, the capacity of the machine would be calculated by
 t t gal
 8.0 h   3.0 m  24
h
230
min
 m   machine  = machine or machine
 
 
As the capacity of one 3.0-m machine has been measured at 24 t/h or

(225 gpm), the total number of machines required to handle peak solids or
worst-case capacity would be determined by dividing the total amount of solids
reporting to the belt filter press by the capacity of the machine as
   
 t   gal 
 60 h   580 min 
  = 2.5 machines or   = 2.5 machines
 24 t   230 gal 
 h   min 
 machine   machine 
Rounding both results to the nearest whole machine, this application would
need three 3.0-m belt filter press units to meet peak capacity, without consid-
erations given to system redundancy or maintenance plans. After determining
the quantity of machines, it will also become important to determine the plant
layout to optimize floor space, auxiliary equipment usage, and overall process
for the belt filter press area of the plant. Typical coal preparation plant designs
of tall structures with multiple levels, low portal frame building structures,
or modular plants is a function of available budget, mine life, power cost,
and operating philosophy (Budge et al. 2000). These design philosophies can
influence the selection criteria and decision for what type of filter layout can
be incorporated. Figure 4 shows a cluster of four belt filter presses in line and
discharging onto a common conveyor.
With the installation of a belt filter press or any other mechanical dewa-
tering unit, the number of machines must be sufficient for the plant’s needs
for disposal of reporting waste fines. However, depending on the economic,
production, and process climate of a plant, it may require management and
operators to look at adjusting processes to achieve the necessary result for
waste disposal, especially where cake moisture is important for compaction
and disposal. One approach has been the blending of coarse and fine rejects.
This is commonly referred to as co-disposal and the advantages are as follows
(Bickert 2004):
• Maximizing storage space
• Improved water return from impoundments
• Facilitation of impoundment rehabilitation
• Reduction in the risk of spontaneous combustion of coarse rejects
Alternative and less common approaches have been used to improve cake
moistures and fine tailings dewatering results. Blending or bleeding a small
portion of coarser product rejects into waste fines streams has been used to
provide improved dewaterability. This is achievable through decreasing the effi-
ciency of cyclones or other separating equipment upstream to create a coarser
feed reporting to the waste fines circuit. In general, lower cake moistures are
Figure 4 Installation of four belt filter presses
typically obtained with more hydrophobic feeds, and increased levels of coarse
material lead to lower cake moistures and higher capacities (Mathewson et al.
2007). However, a plant must again take overall economic considerations of the
small percentage loss in production to the operating costs associated with dif-
ficult processing conditions where this modification would prove to be benefi-
cial. Another alternative that has been used is blending concrete, lime, or waste
by-products (i.e., fly ash) with the discharge cake. These agents typically assist
in the stabilization and handling for coal tailings. Many of these techniques
can be utilized when there is a large quantity of ultrafine, expandable clays in a
plant’s process.
In plant design, whether new or existing, it is also important to accurately
size and/or build redundancy whenever possible in the tailings dewatering sys-
tems. During feasibility studies or evaluations, thorough empirical analysis and
laboratory simulations on the most difficult material to process are necessary to
give proper sizing for the plant to adequately handle this material type. With
correct sizing from laboratory simulations, sound decisions can be formulated
as to whether the plant should install additional machines or if there is an
alternative disposal method, such as a small impoundment, that can be utilized

during upset or maintenance conditions. If this is not feasible with respect to
economic, regulatory, or process conditions, there is a risk that the operation
will be restricted to the tailings dewatering capacity, which could decrease plant
production or stop it completely.
Historically in the United States, belt filter presses have been sized to
handle the maximum allowable throughput or capacity (typically 10 t/h/m of
belt width when no sample or analysis is available). Though belt filter presses
have little issue processing these capacities with typical feed conditions, the
machines are running at high output nearly 24 hours per day, which can have
increased strain on wear parts and overall process. As the global demand for
belt filter presses has increased—particularly in South Africa, Australia, and
India—suppliers, design firms, and coal companies have seen a change in pre-
ferred sizing where they typically size more conservatively on the minimum
end of an applications-rated capacity. For example, if a coal tailing sample had
a range of 7–10 t/h/m of belt width, the trend is to size on the minimum of
the range (i.e., 7 t/h/m of belt width). The observed result has been a net posi-
tive impact on cake moisture, operator attention, maintenance, wear parts, and
overall operation of the belt filter presses.
There has also been a trend with installing redundant machines, due to the
low capital cost, to allow for maintenance and/or plant upset conditions. For
example, after thorough evaluation in 2009, an Alpha Natural Resources facil-
ity in Rockspring, W.V., required only six 3.0-m belt filter presses to handle
peak production; however, they purchased eight machines so that two could
be utilized during maintenance and other conditions where excess production
may be required (Cousins 2012). Prior to equipment requisition, laboratory
simulations had been performed on Alpha Natural Resources thickener under-
flow to determine that six machines were needed, which contributed to the
success of the installation and is another key evaluation component in plant
design and process analysis.
Bench-Scale Laboratory Simulations

The need for accurate bench-scale and laboratory analysis for proper sizing of
belt filter presses and other dewatering technologies is vital from a plant design
standpoint. Sizing can be determined in an analytical environment with a well-
equipped laboratory capable of running bench-scale simulations. First, most
laboratories will evaluate key slurry and water characteristics that can impact
belt filter press performance (see Figure 5), which can include
• Ash and volatile content,
Figure 5 Laboratory preconditioning of slurry prior to belt filter press simulations
• Moisture (or slurry) content,

• Particle size distribution,
• Specific gravity of solids, and
• pH.
After material and slurry characteristics have been defined and the sample
appears to be representative of expected tailings, the next step is to perform
bench-scale simulations to determine accurate sizing and model selection type.
Key results that suppliers and plants use as performance indicators are polymer
dosage, cake moisture, capacity, and capture rate. Table 5 shows five coal tailing
thickener underflow samples with both slurry characteristics and performance
results displayed. Though there are some attributes such as particle size distri-
bution that can assist in sizing a proper belt filter press or other dewatering
device, the results seem counterintuitive in some cases. Focusing on Sample 3
and Sample 4, both have similar feed densities and particle size distributions;
however, the belt filter press results are significantly different, most likely attrib-
utable to geology. Without the use of a well-equipped laboratory and bench-
scale dewatering simulation, there is little evidence to say that these two samples
Table 5 Laboratory characterization and bench-scale belt filter press simulation

results for coal tailings
Sample 1 Sample 2 Sample 3 Sample 4 Sample 5
Date of simulation Apr. 2010 Nov. 2011 Nov. 2010 July 2011 Aug. 2011
Feed:
Solids, % w/w 43 31 34 34 31
Density 1.24 1.15 1.17 1.14 1.16
Solids, specific gravity 1.84 1.76 1.74 1.55 1.77
Ash, % 72 41 42 29 37
% Passing 38 µm 37 78 77 66 36
D50, µm 76 9.8 14 16 83
Dosage:
Anionic, lb/t 0.21 0.34 0.50 1.40 0.54
Cationic, lb/t 0.02 0.35 NA 0.09 0.00
Cake:
Solids, % w/w 72 64 71 62 70
Density 1.49 1.38 1.43 1.28 1.44
Moisture, % w/w 28 36 29 38 30
Capacity, t/h/m (gpm/m) 11.2 (84) 9.3 (105) 9.6 (96) 7.1 (73) 10.0 (110)
Capture/Recovery, % >90 >90 >90 >90 >90
with similar characteristics would have dewatered differently. The importance

of incorporating laboratory testing during feasibility studies is ever more crucial
to accurately size and design all components of a plant.
Pilot Demonstrations
In cases where site specifics generate concerns regarding fine tailing dewater-
ability and compactability for disposal, commercial-scale dewatering equip-
ment pilot studies are recommended (see Figure 6). Pilot programs can vary in
length depending on the number of variables considered. One to two weeks is
typical for generating enough data to establish process and dewatering equip-
ment selection; although in cases where multiple seams are mined, demonstra-
tions can last much longer.
The data shown in Figure 7 are typical of the fines dewatering informa-
tion generated in a commercial-scale pilot study. The overall results indicated
an average capacity range of 5–7 t/h/m to maximum capacity rates of 10–12
t/h/m. Cake moistures were well below 35% w/w moisture, even at maximum
Figure 6 Pilot-plant belt filter press
capacities with minimum moisture recorded of 23% w/w. The recommen-

dations from this study were to base sizing on nominal feed conditions to
allow for some redundancy in the system, while also verifying that belt filter
press technology was a suitable, cost-effective alternative. Final plant design
recommendation is to incorporate eight 3.0-m belt filter presses for tailings
dewatering.
CONCLUSIONS
Belt filter press technology has been a practical, cost-effective option for
mechanically dewatering coal tailings for decades and is still considered an
excellent solution for a broad range of application conditions. It is important
that equipment suppliers, design firms, and coal companies perform in-depth
evaluations on plant design, especially for the tailings streams. Not only is
this area already one of the most costly operating expenses of a plant, it could
restrict plant production or further increase the cost of handling waste fines if
not properly designed. Both process and economic evaluations should be per-
formed to accurately determine whether a belt filter press is the most suitable
technology for the application or process. As long as designs take a comprehen-
sive and detailed approach in determining equipment selection, coal prepara-
tion plants will be able to adequately and efficiently process coal tailings circuits
and meet disposal requirements.
39.0
Seam 1
37.0
Seam 2
Seam 3
35.0 Seam 4
Seam 5
Total Moisture, %
Seam 6
33.0
31.0
29.0
27.0
25.0
0 2.0 4.0 6.0 8.0 10.0 12.0 14.0
Dry Feed, tph
Figure 7 Capacity and cake moisture results relating to a coal seam in Queensland,
Australia
REFERENCES
Alderman, K. 2009. Belt filter presses. In Coal Prep Primer. Pittsburgh: Coal Prepara-
tion Society of America.
Bickert, G. 2004. Review of Tailings Dewatering Technologies for Australian Coal Pro-
cessing Plants. Final Report to Australian Coal Association Research Program.
ACARP Project C14012.
Bickert, G., Selomulya, C., Liao, J.Y.H., and Amal, R. 2007. Coal fines filtration—The
relevance of coal floc micro-properties. In Proceedings, Eleventh Australian Coal
Preparation Conference. Paper C2. Sunshine Coast, QLD, Australia.
Brown, E.T. 2003. Water for a sustainable minerals industry—A review. In Proceedings,
Water in Mining 2003. Melbourne: The Australasian Institute of Mining and
Metallurgy.
Budge, G., Brough, J., Knight, J., McNamara, L., and Woodruff, D. 2000. Review of the
Worldwide Status of Coal Preparation Technology. Final Report to the Department
of Trade and Industry’s Cleaner Coal Technology Programme. Report No. Coal
R199 DTI/PUB URN 00/1205.
Cousins, B. 2012. Alternatives to coal mine tailings impoundment—Evaluation of
three dewatering methods at Rockspring coal mine. In International Symposium
on Water in Mineral Processing. Englewood, CO: SME.
Harriman, A. 2012. ACPS Dewatering Workshop 2012: Background and Fundamen-
tals. Australian Coal Preparation Society, Newcastle, NSW, Australia.
Mathewson, D.J., Norris, R., and Dunne, M.J. 2007. Tailings dewatering, dry screening
and water clarification for the coal industry. In Proceedings, Eleventh Australian
Coal Preparation Conference. Paper C3. Sunshine Coast, QLD, Australia.
NRC (National Research Council). 2002. Coal Waste Impoundments: Risks, Responses,
and Alternatives. Washington, DC: National Academy Press.
Sanders, G.J. 2007. The Principles of Coal Preparation. 4th ed. Newcastle: Australian
Coal Preparation Society.
Shields, G. 1992, Reducing the Moisture Content of Clean Coals. Vol. 3: Belt Filter Press.
CQ for Electric Energy Research Corporation.
Fine Coal Drying and Plant
Profitability
M.K. Mohanty, H. Akbari, and G.H. Luttrell
ABSTRACT
This chapter describes the increased plant revenue that can be achieved by coal
preparation plants by using a suitable drying technology to significantly (by ~50%)
lower the product moisture content of the fine clean coal fraction. The equalization
of incremental inert (i.e., combined ash and moisture) content approach and real
plant feed washability data for an Illinois Basin coal have been utilized to estimate
the potential increase in plant yield that can be achieved for a simple two-circuit
plant by the above-mentioned reduction in moisture content of the fine clean coal
product. Computer simulation results indicate that the overall plant yield of a
typical coal preparation plant can be potentially increased by 5.74% if the mois-
ture content of the clean coal product can be further reduced by 50% (i.e., from
the ~18% level commonly achieved by the best available mechanical dewatering
processes, such as screen-bowl centrifuges, vacuum disc filters, or plate-and-frame
filter presses, to ~9% with the use of a suitable fine coal drying technology). With
the substantial reduction in moisture content and thus the inert content of the fine
clean coal product, the specific-gravity cut of the coarse coal circuit can be increased
from 1.47 to 1.69, resulting in the majority of the above-mentioned yield improve-
ment for the overall plant while still maintaining the desired heat content specifica-
tion of the total clean coal product.
The Parsepco Drying Technology (PDT) and Nano Drying Technology
(NDT), two recently developed drying technologies, have shown the feasibility of
providing the above-mentioned moisture reduction of the fine clean coal product.
The PDT dewaters the moist coal on a woven steel belt fitted with medium-wave
infrared radiators in a negative-pressure environment, whereas the NDT utilizes
molecular sieves to absorb the excess moisture from fine coal. These two emerging
drying technologies along with several other commercially available coal drying
technologies, more commonly used for low-rank coal, have been reviewed in this
329
chapter for their potential integration into higher rank coal preparation plant
circuits.
COAL DRYING PRINCIPLE

Coal drying generally refers to evaporating the water trapped in coal particles
and thereby lowering the moisture content of coal. Commonly, moisture con-
tent in coal is categorized as surface moisture and inherent moisture. However,
a comprehensive list of five of these categories, as reported by Osman et al.
(2011), includes
1. Interior adsorption water, located in the micropores and microcapil-
laries within each coal particle;
2. Surface adsorption water, located on individual particle surfaces;
3. Capillary water, located in the capillaries of coal particles;
4. Inter-particle water, located in small crevices found within particle
aggregates; and
5. Adhesive water, contained in the film around the surface of particle
aggregates.
The first three categories constitute what is commonly referred to as inherent
moisture, whereas the last two are included in the surface moisture category.
Most of the surface water can be removed using some type of mechanical
dewatering systems, whereas removal of inherent moisture requires some type
of drying method.
The majority of the surface moisture content of fine coal is usually removed
in coal preparation plants by mechanical dewatering systems, such as screen-
bowl centrifuges, vacuum disc filters, horizontal belt filters, or plate-and-frame
filter presses; however, removal of inherent moisture from fine coal is not a
usual practice. In very specific cases needed to satisfy the overall moisture speci-
fication, convective thermal drying is the only method commonly utilized to
remove residual surface moisture and, at times, a part of the inherent moisture
content of fine coal. The thermal drying system used most is the fluidized bed
dryer, which uses coal, oil, or natural gas as the fuel source to heat the intake
air stream (Pratton et al. 2012). The amount of fuel required depends on the
amount of water content of the fine coal fed to the dryer and the desired mois-
ture content of the product. However, thermal dryers require a substantially
large capital and installation cost and also a large operating and maintenance
cost; for this reason, they are rarely used in coal preparation plants.
Fine Coal Drying and Plant Profitability 331
Drying Rate
Moisture
Drying
Change
Rate Drying Rate
Drying Time
Initial
Period Constant Rate Period Falling Rate Period
Figure 1 Typical drying curve for fine coal
Irrespective of the drying method used, each fine coal type will have a
drying characteristic curve as a function of drying temperature, hot air (gas)
velocity, and pressure environment. A typical drying curve (shown in Figure 1),
as described by De Korte and Mangena (2004) indicates three distinct time
periods: the initial increasing drying rate period followed by a constant drying
rate period and a falling drying rate period. As shown, the moisture reduction
rate significantly decreases after the removal of all surface moistures in the con-
stant drying rate period.
During the initial period, the wet coal is heated from the ambient tem-
perature to the process temperature maintained inside the dryer. Heat energy
is transferred to the coal particles, resulting in evaporation of the contained
moisture. The rate of evaporation and thus the drying rate increases rapidly
with the removal of most of the surface moisture during this initial period.
This causes the exponential decay in the moisture content during this period.
At the end of this initial period, when the heat transferred from the source
(hot air, gas, etc.) becomes equal to the the cooling caused by evaporation of
surface water from coal particles, the drying rate stops increasing and continues
at a constant rate throughout this second period. This is shown as the constant
drying rate period with a horizontal line for the drying rate and and a straight,
though inclined line with the same slope over this entire second period for the
moisture content change. When all the the surface moisture from coal particles
is evaporated, drying of the inherent moisture begins, which is marked by a
substantially reduced drying rate in the falling rate period, shown in Figure 1.
The cost of drying the inherent moisture becomes prohibitively expensive due
to the exteremly slow drying rate observed while removing moisture from
the micropores and microcapillaries of individual coal particles. The authors
believe that a suitable drying technology can be used to remove almost all the
surface moisture content of the mechanically dewatered fine coal product in a
coal preparation plant at a relatively low cost. This is how the moisture content
of the fine clean coal can be brought down to almost the same range of values
as that of mechanically dewatered coarse coal. It is true that such change in
commercial practice will require significant capital investment and increase
operating costs for coal preparation plants. However, the resulting reduction in
moisture content of the fine clean coal product would allow suitable increase
in the specific-gravity cut achieved in both coarse and fine coal cleaning circuits
and thus significantly increase the plant clean coal yield while maintaining the
heat content of the plant product at the original level. It is believed that the
resulting increase in plant revenue will far offset the additional cost of integrat-
ing the suitable drying technologies to a conventional coal preparation plant
and, thus, increase plant profitability.
POTENTIAL BENEFITS OF DRYING TECHNOLOGY

I N T E G R AT I O N T O A C O A L P R E PA R AT I O N P L A N T
Coal mining companies are usually paid on the basis of the heat content of coal
they deliver. Therefore, most of the mining companies utilize coal preparation
plants to remove both in-seam and out-of-seam dilution from the run-of-mine
coal to produce clean coal having a low ash value. Given that most coal prepara-
tion plants today operate with wet separation processes, the addition of water
to the run-of-mine coal is a common practice followed in these plants. Today’s
coal preparation plants attempt to dewater the clean coal product using the best
available mechanical dewatering processes based on the particle size distribu-
tion of the fine coal and the desired ultimate moisture content. However, due
to the inherent difficulty of removing water from fine coal, the final moisture
content of the fine clean coal product is usually two to three times higher than
that of the coarse clean coal product generating from the same plant, although
the product ash values are in a similar range. This is why, although the fine
coal proportion of the total clean coal tonnage produced from a plant is in the
10%–15% range, as much as one third of the total moisture content in the clean
coal product is contributed by the fine coal fraction. Through this chapter, the
authors attempt to show the potential plant profitability that can be achieved
by using some of the emerging fine coal drying technologies to further lower
the moisture content of the mechanically dewatered fine coal product of a coal
preparation plant to nearly the same level as that of the mechanically dewatered
coarse coal product.
To explain the potential benefits that could be realized in a realistic plant
environment, a computer simulation exercise has been conducted using the real
feed washability data obtained from a coal preparation plant operating in the
Illinois Basin. To keep the calculations relatively simple, only two cleaning circuits
(i.e., coarse and fine coal circuits of the plant) have been considered for this analy-
sis. The plant cleans the +2 mm size coarse coal and 2 mm × 75 μm size fine coal
using dense-medium cyclones and coal spirals, respectively. The two-circuit plant
cleans 900 tph (tons per hour) of raw coal having an overall ash value and a mois-
ture content of 21.85% and 5.65%, respectively. The feed washability data for
both coarse and fine coal are listed in Table 1. The as-received (ar) heat content
of the feed coal has been calculated based on the strong correlation between the
heat content (Btu/lb) and the combined ash and moisture content of a variety of
Illinois Basin coals, shown in Figure 2. The product moisture contents of coarse
and fine clean coal for the simulation exercise have been assumed to be 6% and
18%, which are quite comparable to the moisture content that is commercially
achieved in the coal preparation plants in the Illinois Basin.
The computer simulation exercise was targeted to maximize the plant
yield while producing an overall product heat content of 11,900 Btu/lb (on
as-received basis). The equalization of incremental inert content (combined
ash and moisture) approach of plant optimization was pursued for two differ-
ent cases: the first case being the conventional plant without using any drying
technology, whereas the second case made use of a suitable drying technology
to lower the moisture content of mechanically dewatered fine coal by 50% (i.e.,
from 18% to 9%). For both cases, the coarse clean coal moisture content was set
at 6%, which is quite comparable to the moisture content that is commercially
achieved in coal preparation plants in the Illinois Basin.
Figures 3 and 4 illustrate the clean coal yield versus the overall inert con-
tent and incremental inert content relationships obtained for both case 1 and
case 2, respectively. For case 1, the incremental inert content for both coarse
coal and fine coal circuits was equalized at 25.5% to achieve the desired heat
content >11,900 Btu/lb for the overall plant product. As indicated in Table 2,
the heat content of the individual products from the coarse coal and fine coal
circuits were 12,081 Btu/lb and 10,799 Btu/lb, respectively, and the total clean
tonnage produced was 673 tph. The significantly lower heat content of the
fine coal product was the result of the high moisture content of 18% of this
product. Case 2 simulations show the increase in clean coal production that
can be achieved by lowering the fine coal moisture content by 50% (i.e., from
Table 1 Washability data for the coarse and fine coal obtained from a coal
preparation plant operating in the Illinois Basin
16,000
14,000
12,000
Heat Content, Btu/Lb
10,000 Y = –155.39X + 14375

R2 = 0.9847
8,000
6,000
4,000
2,000
0
0 10 20 30 40 50 60 70 80 90 100
% Ash + % Moisture
Figure 2 Strong correlation of the heat content of a variety of coal samples

collected from Illinois Basin coal with the combined ash and moisture content
18% to about 9% level) by using a suitable fine coal drying technology. As

shown in Table 2, using the plant optimization approach, clean coal tonnage
can be increased to nearly 725 tph, while still maintaining the heat content
100
A
90
80
Clean Coal Yield, %
70
60
50 Fine Circuit
Coarse Circuit
40
10 20 30 40 50 60 70 80 90
Incremental Inert Content, %
100
B Fine Circuit
Coarse Circuit
90
80
Clean Coal Yield, %
70
60
50
40
10 12 14 16 18 20 22 24 26 28
Average Inert Content, %
Figure 3 Clean coal yield vs. incremental inert (moisture + ash) and overall inert
content relationships for the coarse and fine coal circuits of a coal preparation plant
not using any coal drying technology
of the overall clean coal product at the original level of 11,937 Btu/lb. Thus,
the additional 52 tph of clean coal can be produced by suitably increasing
the specific-gravity cuts of both coarse and fine coal circuits to 1.69 and 1.74,
respectively. In spite of the significant increase in the incremental inert content
of both circuits to a 39.1% level, the overall inert content of both products and,
thus, the overall heat content could be maintained at the original level. At a
90
A
80
Clean Coal Yield, %
70
60
50
Fine Circuit
Coarse Circuit
40
10 20 30 40 50 60 70
Incremental Inert Content, %
90
B
80
Clean Coal Yield, %
70
60
50
Fine Circuit
Coarse Circuit
40
0 5 10 15 20
Average Inert Content, %
Figure 4 Modified clean coal yield vs. incremental inert (moisture + ash) and overall
inert content relationships for the coarse and fine coal circuits of a coal preparation
plant simulated with the use of drying technology to further lower the moisture
content of the fine clean coal product from 18% to 9%
rate of $50/t of Illinois Basin clean coal and 6,000 working hours per year, the
additional clean coal production would result in an increase in annual revenue
of $15.6 million. This will require capital investment in a suitable technology to
dry nearly 100 tph mechanically dewatered fine clean coal from 18% moisture
Table 2 Potential increase in clean coal tonnage that can be achieved by integrating
a drying technology into the fine coal circuit of a simple two-circuit plant preparing
Illinois Basin coal
tph tph Moisture, Ash, Btu/lb Inert, Incremental
Coal (dry) (ar) % (ar) % (dry) (ar) % (ar) Inert, % (ar)
Without Drying Technology
Coarse 561.4 597.2 6 8.76 12,081 14.8 25.5
Fine 62.2 75.9 18 5.01 10,799 23.0 25.5
Total clean coal 623.6 673.1 7.4 8.39 11,937 15.7 25.5
With Drying Technology
Coarse 588.6 626.19 6 9.53 11,961 15.5 39.1
Fine 89.7 98.6 9 7.68 11,783 16.7 39.1
Total clean coal 678.3 724.8 6.4 9.29 11,937 15.7 39.1
to a 9% level. This can be achieved by three full-scale Parsepco dryers having a

capital cost of $1 million each (Buisman 2012). Clearly, this increase in annual
revenue would pay off the additional capital investment for the coal dryers in
a few months’ time.
C O M M E R C I A L LY AVA I L A B L E D R Y I N G T E C H N O L O G I E S
Most of the commercially used coal drying technologies are of convective ther-
mal drying type, in which drying occurs when a hot gas (air) is allowed to be in
contact with moist coal. The circulating hot gas (air) also acts as a carrier for the
removal of evaporated moisture from the dryer (De Korte and Mangena 2004).
Drying of coal is more widely practiced commercially for low-rank coals having
very high inherent and surface moisture contents. Pikon and Mujumdar (2006)
provide a detailed discussion on various commercially used coal dryers in the
Handbook of Industrial Drying. Jangam et al. (2011) provides a comparative
analysis (Table 3) of the advantages and disadvantages of state-of-the art coal
drying technologies. Various past studies (Pikon and Mujumdar 2006; Bongers
et al. 1998; Wilson et al. 1997; Suwono and Hamdani 1991; Mujumdar 1990)
indicate the significant advantages associated with superheated steam drying.
These include reduced risk of fire hazard/spontaneous combustion due to the
absence of oxygen, increased drying rate and energy efficiency, and reduction in
dust emission. These types of superheated steam dryers could be very suitable
for large-scale fine coal drying applications at coal preparation plants.
Table 3 Comparative analysis of conventional drying technologies available for

coal drying
Coal Dryer Type* Key Advantages Key Disadvantages
Belt dryer (Pikon and Compact construction; simple Limited capacity; large
Mujumdar 2006; Li 2004) design; lower temperature footprint
drying
Fluid bed dryer (Karthikeyan Intensive drying due to good High pressure drop; attrition
et al. 2009; Li 2004) mixing
Horizontal agitated bed dryer Indirect heating through shaft High power requirement; high
using jacket or screw heating and jacket; very low drying maintenance
(Mujumdar 2006) medium flow rate needed
Pneumatic dryer (Li 2004) Simple construction Attrition
Pulsed combustion dryer Short drying time; high drying Noise problem; scale-up
(Ellman et al. 1966) efficiency; environmentally issues; fire hazard
friendly operation
Rotary dryer (Clayton et al. Drying along with disintegra- High maintenance
2007; Li 2004; Hatziylberis tion; internal heating with
2000) coils; no fire hazard
Rotary tube dryer (Pikon and Indirect heating; no fire Capital intensive
Mujumdar 2006; Li 2004) hazard; high efficiency
Spouted bed dryer Very good heat and mass Scale-up issues; limited
(Karthikeyan et al. 2009) transfer rate particle size
Superheated steam using High thermal efficiency; no Only suited for high-capacity
various types (Bongers et danger of fire or explosion; applications; lots of heat loss
al. 1998; Wilson et al. 1997; energy efficient; suitable for in the exhaust
Suwono and Hamdani 1991; high-capacity continuous
Mujumdar 1990) operation
Vibrated bed dryer Low gas velocity required for More moving parts
fluidization
Source: Jangam et al. 2011.
*Listed in alphabetical order.
EMERGING FINE COAL DRYING TECHNOLOGIES/

METHODS
Extensive studies conducted in the food processing industry in the past indicate
the superiority of radiation-based drying methods (Chou and Chua 2001; Das
et al. 2004; Wang and Sheng 2006). One of these studies indicates that drying
time can be reduced by 50% to 60% by the use of an infrared radiation–based
drying system in comparison to that of the convective air-based system. It is
known that some of the infrared radiation (IR) systems can evaporate more
than 3 L of water in a vacuum environment while consuming only 1 kWh of
energy in comparison to only 0.5 L of water that is evaporated using the ther-
mal dryers utilizing a convective heating mechanism (Fisher 2012).
Pin Mixer Product or

Plant Dewatered Fine
Coal with Up to 30%
Emitter Moisture
Boards
To Blower
Vacuum/Blower
Dry Coal
Condensate
Source: Fisher 2012.
Figure 5 Parsepco dryer schematic
IR drying systems transfer thermal energy to the materials to be dried in

the form of electromagnetic waves. Past studies indicate medium-wave infra-
red radiation (MIR) to be more effective than short-wave or long-wave IR in
removing water from a substrate (Buisman 2010). A new highly efficient drying
system has recently emerged in the form of the Parsepco dryer with the combi-
nation of MIR, a steel-belt dryer and a pin mixer.
This Parsepco Drying Technology (PDT) dries fine coal/mineral feed
from ~25% moisture to below 10% level utilizing ceramic MIR emitter boards
on the negative pressure environment of a steel belt.
Parsepco Drying Technology

As shown in the schematic diagram in Figure 5, the PDT consists of a woven
steel belt on which the moist fine coal is dried using MIR emitter boards from
the top and a vacuum system from the bottom of the belt. Steel belts are unaf-
fected by variances in temperatures and can easily accept high temperatures.
Where normal plastic or material belts would stretch or deform at elevated
temperatures, steel belts can easily withstand high temperatures. While poly-
esters and polyamides are hydrophilic, the steel is neutral; thus, the steel belts
provide better water drainage than the plastic belts. In addition, because of the
nonhydrophilic property of steel, cake does not tend to adhere to the steel belts
at the discharge end, unlike the plastic belts. The negative pressure environment
on the belt created by vacuum greatly assists the transfer of the IR energy into
the substrate removing all vapor downward and away from the MIR emitters.
Feed
Product
Source: Buisman 2010.
Figure 6 Pin mixer
In cases of extremely fine clean coal below 75 µm, a pin mixer (shown in Fig-
ure 6) is used to prepare the feed for the dryer in the form of 3–6 mm size
microgranules. More details about the PDT is available elsewhere (Buisman
2010), which reports some of best drying results achieved from the Parsepco
dryer for extremely fine (–45 µm) coal tailings. Product moisture contents of
9.51% and 13.73% were achieved by drying the dewatered tailings product
obtained from a plate-and-frame filter press.
Nano Drying Technology

The Nano Drying Technology (NDT drying system, a patent-pending process;
Bland et al. 2011), uses molecular sieves to extract the majority of the remain-
ing moisture from a mechanciallly dewatered fine clean coal product from
about 25% in the feed to less than 10% in the product. The molecular sieves
are mixed with fine coal particles at a desired sieve-to-coal ratio for a suitable
retention time to absorb almost all the surface moisture from the fine coal
aggregates. After this step, as shown in the schematics of batch- and pilot-scale
process steps in Figures 7 and 8, the soaked molecular sieves are screened off,
leaving behind the nearly dry fine coal product. The pores of the molecular
sieves are sufficiently large to draw in and absorb water molecules, but too small
to allow any of the fine coal particles from entering the sieves. Some molecular
sieves can absorb up to 42% of their weight in water (Bland et al. 2011). In the
subsequent step, the water absorbed in the molecular sieves is evaporated using
a heating system to regenerate those for their reuse in the next cycle. Greater
details of the the NDT drying system is available elsewhere (Bratton et al.
Dry high-moisture
Separate dewatered Reuse molecular
coal fines by
coal fines from sieves after thermal
combining with
molecular sieves. regeneration.
molecular sieves.
Rotary Mixer Laboratory Sieve Microwave Dryer
Source: Bratton et al. 2012.
Figure 7 Batch-scale NDT process steps
Feed Coal/Sieve
Coal Contactor
Make-Up Coal/Sieve Dry

Sieves Screen Coal
Water Sieve
Vapor Regenerator
Source: Bratton et al. 2012.
Figure 8 Pilot-scale NDT process steps
2012). The same study reported product moisture contents in the range of 5%
to 10% for both –0.6 mm and –0.15 mm coal having feed moisture contents in
the range of 22% to 28%.
Dielectric Heating and Drying

Dielectric heating refers to heating by high-frequency electromagnetic radia-
tion; that is, microwave frequency and radiofrequency waves (Menendez et
al. 2010). To avoid the interference with microwave bands used for telecom-
munications, the wavelengths for industrial heating and other applications are
regulated by national and international authorities. Thus, the main operating
microwave frequency for industrial/domestic applications in the majority of
countries is 2.45 (±0.05) GHz (Meredith 1998). Microwave drying is well

known for its advantages, such as volumetric heating and faster drying rates.
The Drycol process, developed by DBAGlobal Australia, is based on the use of
microwave drying for coal. A 15-tph plant operates to dry low-rank coal from
28% to 12% moisture content product (Graham 2008). The presence of micro-
wave-absorbing impurities in coal can result in both hot spots and fire hazards
during drying. Intermittent microwave drying is a possible option to remove
moisture efficiently during the final stages of coal drying ( Jangam et al. 2011).
CONCLUDING REMARKS
The wide-scale practice of drying coal to lower its moisture content has been
restricted mostly to low-rank coal having a significantly higher proportion of
inherent moisture, which cannot be removed by mechanical means. However,
results reported in this chapter, based on a detailed plant optimization analysis
conducted using the approach of equalization of incremental inert content of
each cleaning/dewatering circuit of a two-circuit plant, indicates that suitable
coal drying technologies should also be integrated to the bituminous coal and
anthracite preparation plants of the future. Some of the emerging drying tech-
nologies, based on superheated steam, infrared heating, microwave heating,
and/or molecular-sieve-based nanotechnology drying may be quite useful in
lowering the moisture content of the mechanically dewatered fine clean coal
product by removing only its residual surface moisture content. Attempting to
lower inherent moisture content of fine coal may not be quite viable in most
cases, due to the extremely slow drying rate achieved during the last stage of
drying period (i.e., the falling rate period). However, the residual surface mois-
ture content of fine clean coal could be nearly eliminated by the use of high-
efficiency emerging drying technologies while adding to the profitability of the
coal preparation plant and, thus, mining operations.
REFERENCES
Bland, R.W., Harsh, P., Hurley, M., Jones, A.K., Vinod, K., and Sikka. K. 2011. U.S.
Patent Application Publication, Pub. No. US 2011/0078917 A1 Pub. April 7.
Bongers, G.D., Jackson, W.R., and Woskoboenko, F. 1998. Pressurized steam drying
of Australian low rank coals Part I. Equilibrium moisture contents. Fuel Process.
Technol. 57:41–54.
Bratton, R., Ali, Z., Luttrell, G.H., Bland, R., and McDaniel, B. 2012. Nano drying
technology: A new approach for fine coal dewatering. In 29th Annual Interna-
tional Coal Preparation and Exhibition and Conference Proceedings. Lexington,
KY. pp. 97–110.
Buisman, R. 2010. Agglomeration and drying of mineral fines. Fines Beneficiation,

Dewatering and Agglomeration Conference. The South African Institute of Min-
ing and Metallurgy. Misty Hills, South Africa, November 16–17. pp. 1–7.
Buisman, R. 2012. Director, Parsepco Ltd., personal communication.
Chou, S.K., and Chua, K.J. 2001. New hybrid drying technologies for heat sensitive
foodstuffs. Trends Food Sci. Technol. 12:359–369.
Clayton, S.A., Desai, D., and Hadley, A.F.A. 2007. Drying of brown coal using a super-
heated steam rotary dryer. In Proceedings of the 5th Asia-Pacific Drying Conference.
Hong Kong, August. pp. 179–184.
Das, I., Das, S.K., and Bal S. 2004. Drying performance of a batch-type vibration aided
infrared dryer. J. Food Eng. 64:129–133.
De Korte, G.J., and Mangena, S.J. 2004. Thermal Drying of Fine and Ultra-Fine Coal.
Coal Tech 2020. Report No: 2004-0255. Division of Mining Technology, CSIR,
South Africa.
Ellman, R.C., Belter, J.W., and Dockter, L. 1966. Adapting a pulse-jet combustion
system to entrained drying of lignite. In Fifth International Coal Preparation Con-
gress. Pittsburgh. October 3–7. pp. 463–476.
Fisher, B. 2012. Fine coal drying. Presentation at the Illinois Clean Coal Institute.
March.
Graham, J. 2008. Microwave for coal quality improvement: The Drycol Project. Mil-
ton, Queensland, Australia: DBAGlobal.
Hatzilyberis, K.S., Androutsopoulos, G.P., and Salmas, C.E. 2000. Indirect thermal dry-
ing of lignite: Design aspects of a rotary dryer. Drying Technol. 18(9):2009–2049.
Karthikeyan, M., Zhonghua, W., and Mujumdar, A.S. 2009. Low-rank coal drying tech-
nologies—Current status and new developments. Drying Technol. 27(3):403–405.
Li, C.Z. 2004. Advances in the Science of Victorian Brown Coal. Elsevier: Oxford.
Li, C.Z., Su, W., Wu, Z., Wang, R., and Mujumdar, A.S. 2010. Investigation of flow
behaviors and bubble characteristics of a pulse fluidized bed via CFD modeling.
Drying Technol. 28(1-3):2071–2087.
Luttrell, G.H. 2012. Nano drying technology: A new approach for fine coal dewater-
ing. Presented at Coal Prep 2012. April 30–May 3.
Meredith, R. 1998. Engineers’ Handbook of Industrial Microwave Heating. London:
Institution of Electrical Engineers.
Mohanty, M.K. 2003. Evaluation of a High-Efficiency Fine Coal Dewatering Technology.
Final Technical Report: ICCI Project Number: 02-1/4.1A-3. Illinois Department
of Commerce and Economic Opportunity/Illinois Clean Coal Institute.
Mujumdar, A.S. 1990. Superheated Steam Drying: Principles, Practice and Potential for
Use of Electricity. Report No. 817, U 671. Montreal, Quebec: Canadian Electrical
Association.
Mujumdar, A.S. 2006. Handbook of Industrial Drying. 3rd ed. Boca Raton, FL: CRC
Press.
Mujumdar, A.S., and Jangam, S.V. 2011. Drying of Low Rank Coal. Minerals, Metals
and Materials Technology Center (M3TC) Report-M3TC/2011/01. National
University of Singapore.
Osman, H., Jangam, S.V., Lease J.D., and Mujumdar, A.S. 2011. Drying of Low-Rank
Coal—A Review of Recent Patents and Innovations. Minerals, Metals and Materials
Technology Center (M3TC) Report-M3TC/TIPR/2011/02. National Univer-
sity of Singapore.
Pikon, J., and Mujumdar, A.S. 2006. Drying of coal. In Handbook of Industrial Drying.
3rd ed. Edited by A.S. Mujumdar. Boca Raton, FL: CRC Press. pp. 993–1016.
Suwono, A., and Hamidani, U. 1991. Upgrading the Indonesia’s low rank coal by super-
heated steam drying with tar coating process and its application for preparation of
CWM. Coal Prep. 21:41–54.
Wang, J., and Sheng, K. 2006. Far-infrared and microwave drying of peach. LWT-Food
Sci. Technol. 39:247–255.
Wilson, W.J., Walsh, D., and Irvin, W. 1997. Overview of low rank coal drying. Coal
Prep. 18:1–15.
Nano Drying Technology—A
New Approach for Fine Coal
Dewatering*
R. Bratton, Z. Ali, G. Luttrell, R. Bland, and B. McDaniel
ABSTRACT
The removal of moisture from fine coal has been a long-standing problem in the
coal preparation industry. Although coal fines often represent as little as 10% of
the total run-of-mine feed, this size fraction may contain more than one third
of the total moisture in the final marketed product. Existing thermal dryers can
effectively reduce moisture; however, these massive units require very large capi-
tal expenditures and have become a target of increased environmental scrutiny.
Likewise, existing mechanical equipment for fine coal dewatering tends to produce
unacceptably high moistures that often cannot be tolerated on existing coal con-
tracts. In light of these issues, an innovative mechanical-thermal dewatering pro-
cess known as Nano Drying Technology (NDT) has recently been developed. This
chapter (1) reviews the working features of this novel drying process, (2) presents
experimental results obtained from recent laboratory and pilot-scale test programs,
and (3) discusses the potential advantages of the process over existing thermal dry-
ing and mechanical dewatering systems.
INTRODUCTION
Essentially all coal supply agreements impose strict limitations on the amount
of moisture contained in the shipped product. Residual moisture lowers heat-
ing value, increases transportation costs, and can create downstream handling/
freezing problems for customers. To meet the moisture specification, a variety
of solid–liquid separation processes are used in modern coal preparation plants.
* All rights, title, and interest to this work, including the copyrights, belong to Nano
Drying Technologies, LLC. Nano Drying Technologies, LLC have given permis-
sion to SME, Inc., to include this work in this book.
345
Available methods for reducing surface moisture can be broadly classified into
three main groups: sedimentation, filtration, and thermal drying (Wills and
Napier-Munn 2006). Sedimentation methods make use of static or induced
centrifugal forces to separate solids from water based on differential settling/
compaction, whereas filtration methods trap particles against a mesh or porous
medium to separate solids from water. Equipment such as vibrating screening
systems and various types of centrifugal dryers (stoker, screen-scroll, and vibra-
tory centrifuges) are commonly used to dewater coarser coal particles. Finer
coal particles (<0.5–1 mm topsize) are typically dewatered using more complex
dewatering equipment such as screen-bowl centrifuges and various types of
vacuum disc and belt filters (Luttrell et al. 2007). Unfortunately, existing fine
coal dewatering processes are inefficient in terms of moisture reduction, solids
recovery, and/or energy consumption (Osborne 1988; Le Roux et al. 2005;
Keles 2010).
It is widely recognized that the moisture content attainable by mechanical
dewatering systems is strongly dependent on coal particle size. For example,
Figure 1 shows the approximate lower limit on moisture than can be attained
using mechanical coal dewatering equipment. The inverse relationship between
particle size and moisture content should be expected because of the sharp
increase in surface area as particle topsize is reduced. The finest coal fraction can
account for as little as a few percent by weight of the total run-of-mine coal but
may represent one third or more of the total moisture in the final coal product.
In some industrial operations, fine (<100–200 µm) or ultrafine (<40–50 µm)
coal particles may be intentionally removed by classification circuits and dis-
carded at the plant site to avoid an unacceptably high product moisture. This
loss represents a waste of valuable coal resources and a potential environmental
liability when discarded into waste impoundments (Orr 2002).
Historically, thermal dryers have been utilized in the coal preparation
industry to reduce clean coal moisture to single-digit values whenever mechani-
cal dewatering processes were incapable of meeting contract specifications.
The most popular design is the fluidized bed dryer, which uses coal, oil, or
natural gas as the fuel source to heat the incoming air stream. The amount of
fuel required depends on the amount of water that must be evaporated, which,
in turn, depends on the amount of coal fed to the dryer and the percentage
of water in the dewatered product (Miller 1998). When operating correctly,
thermal dryers can reduce the clean coal moisture to less than 6% by weight
(Meenan 2005). Unfortunately, thermal dryers involve a substantial investment
of upfront capital funds when installed and large annual costs for equipment
maintenance and repair throughout their life span. Operating costs for thermal
dryers have also greatly increased in recent years in response to higher fuel and
Nano Drying Technology: A New Approach 347
100
80
60
40
30
Vacuum/Pressure Filters
Product Moisture, %
20
Screen-Bowl Centrifuges
Screen-Scroll
10 Centrifuges (Fine)
8
Vibratory Centrifuges
6
Sreen-Scroll
4 Centrifuges (Coarse)
3 Vibratory Stoker Vibrating
2 Centrifuges Screens
0
325M 200M 100M 48M 28M 14M 8M 6M 4M ¼″ ½″ 1″ 2″ 4″ 8″
Particle Size, inches or mesh
Figure 1 Comparison of dewatering alternatives for different particle size ranges
labor costs. Thermal dryers can also suffer from emission problems associated
with fugitive dust and poor opacity. In fact, the opacity standard for coal dryers
was recently reduced from 20% to 10% as a result of recent legislative action.
Emissions of nitrous oxides, sulfur dioxide, volatile organic compounds, and
particulate matter may also present issues for some sites seeking operating
permits. Moreover, thermal drying of combustible particles of coal can present
safety hazards resulting from accidental fires or dust and gas explosions.
The development of an innovative, efficient, and low-cost technology for
removing moisture from fine coal is an important need for the coal prepara-
tion industry. In light of this necessity, a novel thermal-mechanical dewatering
process known as Nano Drying Technology (NDT) has recently been devel-
oped for the coal preparation industry. In the current study, an experimental
test program was undertaken to evaluate the dewatering performance of the
NDT process. This chapter provides a brief description of the new dewatering
technology and presents experimental results obtained from recent bench- and
pilot-scale test programs.
NANO DRYING TECHNOLOGY

NDT uses molecular sieves to wick water away from wet fine coal particles and
does not require crushing or additional finer sizing of the wet coal to dry it.
These molecular sieves are a form of nanotechnology-based particles, which are
used for extracting moisture from airborne, aerosol, and liquid environments.
Molecular sieves contain pores of a precise and uniform size, typically in the
range of 3 to 10 angstroms (Ramakrishna et al. 2011). These pores are large
enough to draw in and adsorb water molecules, but small enough to prevent
any of the fine coal particles from entering the sieves. Some molecular sieves can
adsorb up to 42% of their weight in water (Bland et al. 2011). Molecular sieves
are used in the drying process because these are reusable after the absorbed
water is removed from the sieves by heating.
Molecular sieves often consist of aluminosilicate minerals, clays, porous
glasses, microporous charcoals, zeolites, active carbon, or synthetic compounds
that have open structures through or into which small molecules such as nitro-
gen and water can diffuse (Breck 1964). When the molecular sieves are mixed
with wet coal fines, these sieves quickly draw water away from the wet solids.
To maximize surface contact between molecular sieves and coal particles,
the mixture is contacted/mixed/agitated for a short period of time (typically
2–6 minutes depending on the characteristics of the coal feed and types of
materials utilized). After contacting, the molecular sieves are recovered from
the dry coal by simple screening given that the sieves are substantially larger in
size than the topsize of the dried coal particles. When the separation occurs, the
remaining coal particles have a substantially reduced moisture content, which
can reach low single-digit values regardless of coal particle size. The molecular
sieves are then regenerated by simple heating to drive off moisture and are
recycled back through the process. It is important to note that the regenera-
tion occurs after the deeply dewatered coal particles have been removed (i.e.,
no portion of the coal is ever subjected to heating). Consequently, this process
is considered by the inventors to be an advanced dewatering process and not a
thermal drying process, which offers many advantages in terms of operational
cost and environmental compliance.
BENCH-SCALE TESTING
Experimental Procedure
A bench-scale experimental test program was performed to evaluate the
performance of the NDT process in removing water from fine coal. For all
experimental tests, the wet feed sample consisted of either 0.6 mm or 0.15 mm
topsize clean metallurgical coal (filter cake) collected from an industrial plant.
During testing, a weighted sample of as-received fine feed coal was mixed with
a predetermined weight of molecular sieves. The mixture was then contacted
together in a small bench-scale rotary mixer for a defined period of time (see
Dry high-moisture
Separate dewatered Reuse molecular
coal fines by
coal fines from sieves after thermal
combining with
molecular sieves. regeneration.
molecular sieves.
Rotary Mixer Laboratory Sieve Microwave Dryer
Figure 2 Bench-scale NDT process
Figure 2). After contacting, the mixture of molecular sieves and coal fines
was separated by using a laboratory sieve. The dewatered coal particles passed
through the sieve and were collected as an underflow product, whereas the
molecular sieves were retained on top of the sieve and were collected as an
overflow product. After separation, the coal particles and molecular sieves were
individually weighed and the reduction in the percentage moisture of the coal
sample was calculated. The last step in the experimental procedure was drying
the molecular sieves. To speed the regeneration process, a microwave oven
was used to evaporate the moisture held in the pores of the molecular sieves.
The regenerated molecular sieves were then reused in the testing program. No
significant difference was observed in the effectiveness of the moisture removal
using either newly manufactured or regenerated molecular sieves.
Five independent “groups” of statistically designed bench-scale experiments
were performed using the NDT process (Table 1). The type (size) of molecu-
lar sieves and weight of coal sample was kept constant for each experimental
group, while the weight of molecular sieves and time of contact were varied
over a range of predetermined values as dictated by the statistical parametric
test matrix. Duplicate test runs (a minimum of three to four) were conducted
at each test point to assess the degree of variability and level of reproducibility
in the test data. The first group of tests (Group A) were comprised of explor-
atory tests designed to identify the suitable ranges of experimental conditions
for testing. This group of test runs involved the processing of five batches of
sample with eight experimental test runs per batch. Groups B and C consisted
of two sets of central composite designs of 39 tests each (15 central point
tests). These groups were identical except for the range of variables examined.
Table 1 Overview of parametric tests conducted using the NDT process

Particle Media Batches/
Experimental Experimental Topsize, Type, Group, Tests/Batch,
Group Design Type mm size number number
A Exploratory 0.6 I 5 8
B Central composite 0.6 I 1 39
C Central composite 0.6 I 1 39
D Uniform grid 0.15 II 4 12
E Central composite 0.15 II 1 52
Groups D and E were conducted using a different type (size) of molecular sieve.
The test matrix for Group D consisted of a uniform grid with four batches of
experiments involving 12 test runs each, while Group E consisted of a central
composite design encompassing a single batch of 52 test runs (20 central point
tests). After completing each test matrix, the data were evaluated using standard
statistical techniques.
Results and Discussion

A target moisture of 9% was selected with a range of 8% to 10% as the operat-
ing parameter for the NDT process. When 100 mesh × 0 product coal gets
below 8% moisture, dust problems become a concern, and, if dried further,
then explosion hazards must be considered. If the moisture is more than 10%,
then the potential benefits of adding this size material to the clean coal are
reduced. Therefore, the tests were designed to determine whether the NDT
process could produce a 9% moisture product with a 95% confidence level. The
central points for Groups B, C, and E were specifically selected, and each cen-
tral composite design was statistically configured to see if this 95% confidence
level could be obtained for a 9% product moisture. Maximum drying tests con-
ducted during the bench- and pilot-scale testing showed that moisture levels in
the 1.5% to 2.5% range could be easily produced if desired.
Table 2 shows the overall performance of the NDT process in terms of
average moisture contents of products for each batch/group. The data indicates
that the technology can readily provide single-digit moistures over a wide range
of operating conditions. In fact, moisture values in excess of 10% were only
obtained when using very short contact times or when low-weight fractions
of molecular sieves were utilized. To fully demonstrate the impact of these fac-
tors, one group of tests from Type I (i.e., Group B) and one group of tests from
Type II (Group E) were selected for further discussion in this publication.
Table 2 Summary of performance data obtained using the NDT process

Group A Group B Group C Group D Group E
Particle 0.6 mm 0.6 mm 0.6 mm 0.15 mm 0.15 mm
Topsize
Feed 22.6+0.07% 21.9+0.11% 21.2+0.12% 27.6+0.70% 26.2+0.19%

Moisture
Batch Product Contact Product Contact Product Contact Product Contact Product Contact
Number Moisture, Time, Moisture, Time, Moisture, Time, Moisture, Time, Moisture, Time,
% min % min % min % min % min
1 10.44 2 9.84 2.6 9.53 3 12.84 2 10.25 2.1
2 9.19 3 10.25 3 9.42 3.6 12.30 2 10.46 2.5
3 8.26 4 8.45 3 8.94 3.6 12.01 2 10.95 2.5
4 7.67 5 9.61 4 9.34 4.6 12.11 3 9.55 3.5
5 5.02 5 8.90 4 8.74 4.6 11.89 3 8.43 3.5
6 2.50 5 7.57 4 8.12 4.6 11.20 3 8.74 3.5
7 — — 8.85 5 8.76 5.6 10.78 4 8.81 4.5
8 — — 7.48 5 7.39 5.6 10.87 4 6.88 4.5
9 — — 8.18 5.4 7.43 6 10.22 4 6.38 4.9
10 — — — — — — 8.91 5 — —
11 — — — — — — 8.34 5 — —
12 — — — — — — 7.81 5 — —
Figure 3 shows the central composite text matrix used in the Group B
test program on the 0.6 mm × 0 feed. As discussed previously, a total of 39
individual test runs were performed in this group using Type I media. The tests
included nine combinations of experiments based on contact time and media
factor. The media factor is a dimensionless number representing the relative
amounts of coal and molecular sieves used in the test run (i.e., a larger media
factor represents a greater addition of feed coal relative to sieve weight, whereas
a smaller number represents less coal relative to sieve weight). The central test
conducted at 4 minutes of contact time and media factor of approximately 0.3
was randomly repeated 15 times throughout the test matrix to evaluate the sta-
tistical reproducibility of the process. Also, each of the satellite tests conducted
around the central test was performed in triplicate to further evaluate the data
reproducibility and to assist in the identification of statistical outliers. All test
runs performed in Group B utilized a constant sample weight of as-received
feed coal. The average moisture content of the feed coal samples used in this
group of tests was 21.9±0.11%.
As shown in Figure 3, all but one of the test runs conducted for the
Group B test matrix gave single-digit moistures in the final 0.6 mm × 0 prod-
uct. The product moistures decreased with either an increase in contact time or
1.0
3 tests 3 tests
10.23% 8.85%
(0.30) 3 tests (0.55)
9.61%
(0.06)
0.5 3 tests 3 tests
9.84% 8.18%
(0.37) (0.41)
15 tests
8.90%
Media Factor
(0.14)
0.0 3 tests 3 tests
8.45% 7.48%
(1.05) (0.01)
3 tests
7.57%
(0.32)
–0.5
Legend
Number of Tests
Average Product Moisture
(Standard Deviation)
Feed Moisture = 21.9±0.11%
–1.0
1 2 3 4 5 6 7
Contact Time, min
Figure 3 Parametric test matrix and performance data for Group B (0.6 mm × 0 size
feed)
a decrease in media factor (i.e., less coal per unit weight of sieve media). The
standard deviations for each set of conditions varied from a low of 0.01 to a
high of 1.05, which indicated that the data were generally reproducible. In fact,
as shown in Figure 4, the 15 replicate tests conducted at the central point of the
test matrix (i.e., 4 minute contact time and 0.3 media factor) showed little vari-
ability in the product moisture despite significant variations in the feed mois-
ture. The average moisture content for the feed sample used in the 15 central
point tests was 21.8±0.16% with a standard deviation of 0.90. After contacting
with the molecular sieves, the product moisture dropped to an average value
of 8.90±0.02% with a standard deviation of 0.14. The very small confidence
interval and low standard deviation values associated with the data obtained
for the dewatered product indicate that a high degree of reproducibility can be
achieved using the bench-scale version of the NDT process.
A similar trend in moisture removal was observed for the tests conducted
for Group E having a topsize of 0.15 mm. These experiments were conducted
using Type II molecular sieves over a similar range of contact times and a
lower range of media factors. Each satellite test conducted around the cen-
tral test point was repeated four times to assist in identifying outliers and
25
20 Average = 21.8±0.16%
Std. Dev. = 0.90
15
Moisture, %
10
Average = 8.90±0.02%
Std. Dev. = 0.14
5
0
0 10 20 30 40
Replicate Test Number
Figure 4 Replicate tests conducted at the central test condition for Group B
(0.6 mm × 0 size feed)
evaluating reproducibility. The central test point, which involved a contact time
of 3.5 minutes and media factor of –0.63, was repeated 20 times in random
order throughout the test matrix. For this particular group of tests, the aver-
age moisture contents of the as-received 0.15 mm × 0 feed was 26.2±0.10%.
After contacting with the molecular sieves, the 0.15 mm × 0 product moistures
were reduced to single-digit values for all tests conducted at contact times of
3.5 minutes or longer (see Figure 5). The lowest product moisture content of
6.38% was achieved for the longest contact time of 4.9 minutes. Tests con-
ducted with contact times less than 3.5 minutes did not achieve single-digit
moistures, but at 10.2%–10.9% moisture they were not far from breaking this
meaningful barrier.
One noteworthy difference in the Group E test series was the greater
degree of scatter in the experimental data. Standard deviation values greater
than 1 were observed for the vast majority of the test points and a value as high
as 4.67 was obtained for one of the satellite tests. The increased data scatter
was clearly observed in the 20 replicate tests conducted at the central test point
(see Figure 6). The confidence interval of ±0.19 obtained for the Group E set
of replicate tests was considerably larger than the ±0.02 value obtained for the
0.0
Legend Feed Moisture = 26.2±0.10%
Number of Tests
Average Product Moisture
(Standard Deviation)
–0.2
4 tests 4 tests
–0.4 10.46% 8.81%
Media Factor
(1.91) (1.32)
4 tests
4 tests 9.55% 4 tests
10.25% (0.53) 6.38%
–0.6 (0.60) (1.06)
20 tests
8.43%
(1.26)
4 tests 4 tests
–0.8 10.95% 6.88%
(4.67) 4 tests (1.56)
8.74%
(1.49)
–1.0
1 2 3 4 5 6 7
Contact Time, min
Figure 5 Parametric test matrix and performance data for Group E (0.15 mm × 0 feed)
Group B tests. Because the moisture of the feed was not determined before
each bench-scale test, the standard deviations for both the feed and product
moistures would be expected to agree, which in fact they do (i.e., 1.29 standard
deviation for the feed moisture and 1.26 standard deviation for the product
moisture). A refinement for future testing would be to determine the moisture
of each feed sample, then calculate the correct media ratio before conducting
the bench-scale test. This is in fact how the pilot-plant tests, described in the
following section, were conducted.
P I L O T - S C A L E D E M O N S T R AT I O N
In light of promising bench-scale data, a decision was made to construct a pilot-
scale NDT plant to demonstrate the capabilities of this new technology in
continous mode. The flowsheet for the facility is shown in Figure 7. The com-
pleted facility, which was largely assembled using off-the-shelf components, was
designed with an effective throughput capacity of 1,000 lb/h (0.5 tph). The
self-contained facility included unit operations for handling, contacting, and
separating the coal and media. An advanced gas-fired dryer was used to regen-
erate the molecular sieves such that the entire process operated in a closed-
circuit loop. The prototype facility was designed, constructed, and successfully
30
25
Average = 26.2±0.19%
Std. Dev. = 1.29
20
Moisture, %
15
10
5 Average = 8.43±0.19%
Std. Dev. = 1.26
0
0 10 20 30 40
Replicate Test Number
Figure 6 Replicate tests conducted at the central test condition for Group E
(0.15 mm × 0 size feed)
Feed Coal/Sieve
Coal Contactor
Make-Up Coal/Sieve Dry

Sieves Screen Coal
Water Sieve
Vapor Thermal Dryer
Figure 7 Simplified flowsheet for the pilot-scale NDT processing facility
commissioned over a period of approximately 10 months. During this time,

shakedown tests were completed and the process circuit was refined, modifed,
and optimized to provide a demonstration facility that operated smoothly and
efficiently.
Table 3 provides an overview of test results obtained with various coals
using the pilot-scale NDT facility. As shown, the prototype facility success-
fully achieved single-digit product moistures for a wide range of feed coal
Table 3 Examples of pilot-scale NDT test results

Particle Size Capacity, Feed Product
Coal Source Class, mm lb/h Moisture, % Moisture, %
A 1 1,600 17.88 7.63
B 1 1,200 10.41 5.38
1 1,200 10.41 7.13
1 1,200 10.41 6.84
C 0.15 600 27.28 2.52
0.15 550 27.28 7.46
D 0.15 1,000 31.83 3.18
0.15 1,000 31.83 5.86
0.15 1,000 31.83 8.27
applications. Engineering criteria developed from bench-scale testing, such

as contacting (retention) times and coal-to-sieve loadings, were also validated
using the pilot-scale plant. More importantly, the pilot-scale test runs success-
fully demonstrated that the molecular sieves could be regenerated and recyled
back through the process without incurring significant losses due to media
degredation and at a lower heating/evaporation cost than traditional thermal
drying.
DISCUSSION
The removal of unwanted moisture from fine coal has historically been con-
sidered one of the most challenging technical problems in the coal preparation
industry. The NDT process was developed specifically to address this issue
by providing effective moisture removals, efficient energy utilization, and
enhanced environmental performance. Experimental data collected from both
bench- and pilot-scale operations show that single-digit moisture values can be
readily achieved from fine coal feeds containing 30% moisture or more. The
process is highly flexible in that the product moisture can be “dialed in” by
varying contacting time and coal-to-seive media loadings. Also, unlike existing
mechanical processes, the product moisture from the NDT process is largely
independent of the particle size distribution of the feed stream.
The removal of moisture in the NDT process occurs at ambient tempera-
ture. As such, the coal particles are never subjected to high temperatures, which
greatly reduces the emissions of criteria pollutants that are normally associated
with conventional coal drying systems. Recent estimates by an environmental
consulting group indicate that emission reductions as large as 90% or more
1,200
PM
CO
NOx
1,000 SO2
VM
800
Criteria Pollutants, t/a
600
400
200
0
Thermal Dryer NDT
Figure 8 Comparison of air emissions for thermal drying and the NDT process
when compared to a thermal dryer are possible using the NDT process. The
emission projections from one such case study is shown in Figure 8. In this case,
emissions of volatile matter (VM), sulfur dioxide (SO2), and particulate matter
(PM) would be essentially eliminated (>99% reduction) using the NDT pro-
cess. Projected emissions of carbon dioxide (CO2) and nitrous oxides (NOx)
would be reduced by 91% and 84%, respectively. It is particularly important to
note that the projected total emissions of 59.4 t/a (tons per annum) of criteria
pollutants is likely to be less than the threshold value that would trigger the
need for a Title V Air Quality permit in many states. For example, no such per-
mit would be required in West Virginia because the threshold value is 100 t/a
of criteria pollutants. The NDT process also requires no added chemicals and
generates no other by-products that could potentially be released into the
environment.
Finally, the NDT process is very efficient in terms of energy utilization.
Given that only the molecular sieves are thermally dried, the drying step can
be fully optimized in the absence of coal-imposed contraints associated with
dryer temperature levels, gas-solid contacting systems, and coal dust explosions.
As such, the system provides the highest possible energy efficiency at the lowest
possible fuel cost. Because the process treats only the fine coal fraction, which
120
Fuel
Chemicals
Drying Media
100 Electricity
Maintenance
Labor
80
Relative Cost, %
60
40
20
0
Thermal Dryer NDT
Figure 9 Comparison of relative costs for thermal drying and the NDT process
is generally between 10% and 15% of the total clean coal product (and not the
entire clean coal product treated by conventional thermal dryers), the required
footprint for the facility is only a fraction of that demanded by a large-scale
coal thermal dryer. Also, because of fewer operational complexities, significant
cost savings are also expected for ancillary items such as electricity, chemicals,
maintenance, and labor. Cost estimates conducted in cooperation with a com-
mercial engineering firm are plotted in Figure 9. Although such economic
calculations tend to be site specific, the costing figures for this site do suggest
a relative operating cost of less than half of that required to operate a conven-
tional thermal dryer.
SUMMARY
The removal of surface moisture from fine coal has been a long-standing prob-
lem in the coal industry. To address this need, an innovative process based on
nanotechnology has been developed. Bench-scale studies indicate that NDT is
an effective method for coal drying. The NDT process can effectively dewater
fine (1 mm × 0) coal from slightly more than 30% surface moisture to single-
digit values. Test data obtained using a pilot-scale NDT plant further validated
this impressive capability using a continous prototoype facility. It was also
observed that, unlike existing fine coal dewatering processes, the performance
of the NDT process is not dictated or constrained by particle size (i.e., it works
equally well on 1 mm × 0 coal as it does on 325 mesh × 0 coal). The NDT pro-
cess overcomes problems associated with other techniques for fine coal drying
because dewatering occurs at ambient temperature and low airflow. Only the
molecular sieves have to be dried, which reduces energy. Moreover, this process
requires no chemicals, produces no damaging contaminants, and has a very
small installed footprint and environmental impact.
REFERENCES
Bland, R.W., Harsh, P., Hurley, M., Jones, A.K., and Sikka, V.K. 2011. U.S. Patent
Application Publication, Pub. No. US 2011/0078917 A1 Pub. Date Apr. 7.
Breck, D.W. 1964. Crystalline molecular sieves. J. Educ. 41(12):678–689.
Keles, S. 2010. Fine coal dewatering using hyperbaric filter centrifugation. Ph.D. dis-
sertation, Virginia Polytechnic Institute and State University. Blacksburg, VA. pp.
4–18.
Le Roux, M., Campbell, Q.P., Watermeyer, M.S., and de Oliveira, S. 2005. The optimi-
zation of an improved method of fine coal dewatering. Miner. Eng. 18(9):931–934.
Orr, F.M. 2002. Coal Waste Impoundments: Risks, Responses and Alternatives. Washing-
ton, DC: National Research Council.
Osborne, D.G. 1988. Solid-liquid separation. In Coal Preparation Technology. Vol. 1.
London: Graham and Trotman. pp. 478–542.
Ramakrishna, S., Ma, Z., and Matsuura, T. 2011. Chapter 1. In Polymer Membranes in
Biotechnology. London: Imperial College Press.
Wills, B.A., and Napier-Munn, T.J. 2006. Chapter 15. In Wills Mineral Processing Tech-
nology. 7th ed. Amsterdam: Butterworth-Heinemann.
Development of a Fine Coal
Recovery Operation at the
Centralia Mine Coal Slurry
Impoundment Structures
C. David Henry
ABSTRACT
Coalview Recovery Group, LLC, is in the final stages of establishing a fine coal
recovery operation at the TransAlta-Centralia mine located near Centralia, Wash-
ington. The operation will extract coal slurry refuse from three impoundment struc-
tures identified as “Series 3 Ponds” containing an estimated 18 to 20 million tons
of slurry refuse material. The coal slurry refuse will be delivered to the Coalview
Recovery Group, LLC, fine coal recovery plant by use of a 200-tph all-electric
hydraulic dredge. The fine coal plant will process the nominal +74 µm coal slurry
refuse and produce a fine coal product to be transported to the adjacent TransAlta-
Centralia coal-fired power plant. The slurry refuse developed by Coalview Recovery
Group, LLC, fine coal recovery plant will be treated through a deep-cone paste
thickener and pumped to a below-grade, incised impoundment for final deposition.
PROJECT DESCRIPTION
Coalview Recovery Group, LLC (CRG), will enter into a contract with
TransAlta-Centralia Mining to develop a fine coal recovery operation at their
mine site located near Centralia, Washington. The purpose of this project is to
• Evacuate the majority of coal slurry refuse (CSR) from the three
impoundment structures;
• Process the CSR through the CRG fine coal recovery (FCR) plant to
produce a fine coal product;
361
Figure 1 Site plan of Centralia mine’s power plant complex
• Transport the fine coal product from the FCR plant to the TransAlta-
Centralia coal-fired power plant within a mile of the recovery opera-
tion; and
• Treat the slurry refuse from the CRG-FCR plant through a deep-cone
paste thickener and pump the resulting paste into an on-site, below-
grade, incised impoundment for final deposition.
Impoundment Structures
The CSR, contained within the impoundment structures, was deposited from
the TransAlta-Centralia mine coal preparation plant and operated from 1971
to 2006. The impoundment structures to be evacuated are part of a system of
five structures known as the “Series 3 Ponds.” The impoundment structures
containing CSR are identified as Pond 3B, Pond 3C, and Pond 3D and contain
3.5 million, 3.9 million, and 10.5 million tons of solids, respectively. Pond 3E
or the North Hanaford Pit, is the impoundment structure that will contain
the slurry refuse from the FCR plant containing an estimated volume of 59.1
× 106 yd3 (45.2 × 106 m3). Pond 3A is included in the Series 3 Ponds but will
not be included in the FCR project. Pond 3A is located at the toe of Pond 3B
and receives overflow water from Pond 3B before discharge into the regulated
NPDES (National Pollutant Discharge Elimination System) outflow point.
Pond 3A has an approximate surface area of 70 acres (28.3 ha) with a total
depth of about 50 ft (15.2 m). Pond 3B was the initial impoundment developed
by the Centralia mine for the CSR disposal and filled to capacity in 1986. The
3B pond covers about 112 acres (45.3 ha) at a maximum depth of about 130 ft
(39.6 m) and contains approximately 8.3 × 106 yd3 (6.3 × 106 m3) of CSR and
3.3 × 106 yd3 (2.5 × 106 m3) of water. Pond 3C covers about 91 acres (36.8 ha)
Development of a Fine Coal Recovery Operation 363
Top of
Embankment First Downstream
Embankment Addition
Coal Slurry Refuse
Second Downstream
Embankment Addition
Starter
Embankment
Downstream Valle
y Flo
or C
ontour
Figure 2 General design of a downstream-constructed impoundment
at a maximum depth of about 160 ft (48.8 m) and contains approximately 8.3

× 106 yd3 (6.3 × 106 m3) of CSR with very little water. Pond 3D has an area of
about 180 acres (72.8 ha) at a maximum depth of about 220 ft (67.1 m) and
contains approximately 24.9 × 106 yd3 (19.0 × 106 m3) of CSR. The locations
of the impoundment structures and other facilities are shown in Figure 1.
Impoundment Design and Coal Slurry Refuse Deposition

The impoundment embankments were constructed of compacted overburden
spoil and other select soil materials from the surrounding area. The construc-
tion of the impoundment structures at this site are commonly known as a
downstream design, where a starter embankment is placed at the mouth of the
valley. The elevation of the impoundment is raised by adding compacted fill
material on the downstream side of the dam with a continuous common slope
on the interior of the impoundment structure. This type of impoundment is
considered to be the most stable design and construction, when compared to
upstream- or centerline-designed impoundment structures. The general design
of a downstream-constructed impoundment structure is shown in Figure 2.
The CSR was pumped through a high-density polyethylene (HDPE) pipe
from the Centralia mine preparation plant and deposited into and behind the
impoundment structure. As the valley area behind the impoundment fills, the
discharge pipe is moved to permit unhindered flow of the CSR and continuous
deposition into the impoundment.
As the CSR is being deposited into the impoundment, a general sizing
of the discharge materials is occurring and creates general zones, identified as
follows:
• Near discharge: The CSR material is more coarse and dense that cre-
ates the delta area, having a moderate recovery at a higher ash value.
• Sweet zone: This material is most often considered the “best” for
recovery. It contains the most lower-ash coal and is located at the
outermost areas of the delta zone and within the slurry water interface.
• Far discharge: This is the finest sized with the lowest density; the mate-
rial in this zone consists of clays/slimes and may have considerable
amounts of ultrafine coal.
Sampling and analysis results indicating the discharge point(s) for the vari-
ous impoundment structures are listed as follows:
• Pond 3B is discharged from the upper valley area at the southeastern
and southern end of the impoundment.
• Pond 3C is discharged from potentially two to three points along the
southwestern side of the impoundment.
• Pond 3D is discharged from one point at the northwest corner of the
impoundment.
Impoundment Sampling and Analysis

Extensive sampling and analysis has been performed on the impoundment
structures at the TransAlta-Centralia mine. The samples were taken from the
impoundment structures by CRG using the Vibracore technique, with 25
samples each from Pond 3B and Pond 3C and 50 samples from Pond 3D. In
addition, deep-hole samples were taken by Norwest from each of the impound-
ment structures; three samples from Pond 3B, two from Pond 3C, and three
from Pond 3D at maximum depths of 113 ft (34.4 m), 123 ft (37.5 m), and 160
ft (48.8 m), respectively. Split-spoon samples were extracted at 10-ft (3.05-m)
intervals to evaluate the size distribution of the refuse slurry material.
Pond 3B and Pond 3C analysis. The analysis for each of the impoundment
structures included the following:
• Head samples were analyzed for ash% and %solids, and the bulk den-
sity and specific gravity of each sample were determined.
• Samples were sizing at +600 µm (28 mesh), 600 × 150 µm (100 mesh),
150 × 88 µm (170 mesh), and –170 mesh.
• Sink–float analysis was performed on the +28 mesh and 28 × 100
mesh size fractions.
• Microtrac size of the –170 mesh was analyzed, reporting the sizes
between 88 µm and 0.818 µm.
Summaries of the analyses for Pond 3B and Pond 3C are shown in Table 1
and Table 2, respectively.
Table 1 Pond 3B recovery and analysis summary
Table 2 Pond 3C recovery and analysis summary
Pond 3D analysis. The analysis of this impoundment included the following:

• Head samples were analyzed for ash% and %solids, and the bulk den-
sity and specific gravity of each sample were determined.
• Samples were sizing at +28 mesh, 28 × 100 mesh, 100 × 170 mesh,
and –170 mesh.
• Sink–float analysis was performed on the +28 mesh and 28 × 100
mesh size fractions.
• Microtrac size of the –44 µm (325 mesh) was analyzed, reporting the
sizes between 44 µm and 0.818 µm.
The summary of the analysis for Pond 3D is shown in Table 3.
FCR Plant Design

CRG will use and incorporate within the FCR plant the most advanced equip-
ment and proven technologies available to the mining industry. The FCR plant
will be constructed in an existing pre-engineered building that previously housed
the TransAlta-Centralia preparation plant. All equipment, substructure steel,
and the majority of foundations have been removed from the pre-engineered
building. Exterior conveyors, load-out bin, static thickener, chemical treatment
building, and other related structures will remain and be incorporated into the
new FCR plant system. The flowsheet is shown in Figure 3.
The CRG-FCR facility will consist of four major systems: dredge feed, siz-
ing/processing, product dewatering, and slurry refuse treatment/paste. General
descriptions are as follows:
1. Dredge feed. The CSR will be delivered to the FCR plant by use of
the primary 200-tph all-electric hydraulic dredge, pumping the mate-
rial through a 10-in.- (254-mm-) diameter to 14-in.- (356-mm-) diam-
eter HDPE pipe. A secondary diesel dredge will pump an estimated 40
to 50 tph of lower-grade, lower-recovery CSR directly to the static/
paste thickener system, bypassing the FCR plant and converting this
CSR to a paste material.
2. Sizing/processing. Initial sizing of the CSR will use a sieve/vibrating
screen separating at 28 mesh; material >28 mesh will be discharged
from the system, and material <28 mesh will be sized in primary
classifying cyclones making a nominal sizing at 28 mesh × 74 µm
(200 mesh). The 28 × 200 mesh will be processed through a two-stage
spiral system and the material <200 mesh will be discharged to the
slurry refuse treatment system. The spiral product will be pumped
to the clean coal classifying cyclone circuit to rid the product of
Table 3 Pond 3D recovery and analysis summary
misplaced fine/ultrafine noncoal material. The clean coal classifying

cyclone overflow material will recirculate to the primary classifying
cyclone, and the underflow product will be washed and screened
through a series of Stack Sizer vibrating screens before delivery to the
screen-bowl centrifuge. The material passing through the Stack Sizer
screens will be discharged to the slurry refuse treatment system.
Figure 3 CRG fine coal recovery system flowsheet
3. Dewatering. The screen-bowl centrifuge is the main dewatering unit

removing the majority of surface water from the coal product. Two
additional discharges occur from the screen-bowl centrifuge: the water
effluent pumped to the slurry refuse treatment system and the screen
effluent recirculated to the clean coal cyclones circuit. The dewatered
clean coal product is discharged to the product conveyor and stored
in the existing load-out bin. The clean coal product is discharged into
TransAlta-Centralia power plant (TCPP) dump trucks to be trans-
ported to the adjacent TCPP stockpile.
4. Slurry refuse treatment/paste. All slurry refuse developed by the
FCR processing will be pumped through HDPE pipelines to a two-
stage thickener system. The first stage of the process will be treated
through a 250-ft- (76.2-m-) diameter static thickener and will produce
a slurry refuse underflow discharged at 12% solids. The slurry refuse
underflow will be pumped to the 115-ft- (35.1-m-) diameter and 26-ft
(7.9-m) deep cone paste thickener to develop a clarified water, for use
within the FCR plant system or discharge to other locations as needed
or required, and a paste material. The paste material will be discharged
at 39% solids into Pond 3E for final deposition.
Notes
1. The water level of the dredge hole should be dropped to a
depth of at least 20′ to 40′ below the surface of the slurry.
2. Slurry material should be dredged at a depth of at least 20′.
3. A water cannon mounted on the dredge and/or on the bank
should be used to saturate and wash the exposed slurry material.
4. The saturation and washing actions will cause the following effects:
a. the failure and wash of the exposed slurry face and the
creation of a beach area providing a safety zone between
the dredge and exposed slurry face; this beach area should
be maintained at no less then 20′ and
b. additional weight to the exposed slurry material above the
water level, enhancing the downward action and movement
of the slurry material toward the cutterhead.
5. The slewing arc should be maintained at the longest allowable
Top of Slurry length and as specific site conditions allow.
Beach Area
20′ to 100′+
20’ to 40’
Water Cannon
Top of Water
Line of Movement/Force
of Slurry Material to
Dredge Cutterhead
20’ ±
Figure 4 CRG general dredging plan for the removal of coal slurry material
P R O J E C T O P E R AT I O N S
The recovery operation is anticipated to begin processing during the third quar-
ter of 2013. The FCR facility will operate 24 hours per day, 7 days per week for
50 weeks per year and deliver a minimum of 200 tph of CSR to the processing/
refuse treatment system. The facility is expected to operate for 12 to 15 years
with earliest completion expected in 2025. The average anticipated clean coal
yield is estimated to be 24% with an estimated 1,428,000 tons of CSR for pro-
cessing and producing 342,000 tons of fine coal product to be delivered to the
adjacent TCPP.
Mining/Dredging Plan
The recovery operation will begin in Pond 3C, then move to Pond 3B and Pond
3D, respectively, as the operations advance. All slurry refuse developed from
the CRG-FCR plant will be deposited into Pond 3E. The dredging units and
pipelines will be relocated from impoundment to impoundment as the FCR
operation proceeds. The general dredging plan is depicted in Figure 4.
The mining/dredging plan will be developed by dividing each impound-
ment into a series of dredge cells. The dredge cells are established around
the Vibracore samples extracted for analysis, with the sample point being in
the center of the cell area. The quality of each dredge cell will be determined
by interpolating the quality between each sample point and establishing the
overall average quality for the dredge cell. As the dredging continues to greater
depths of the impoundment, added Vibracore samples will be extracted and
new dredge cells/quality established. This pattern will continue to the maxi-
mum depths of the impoundment structures. The layout of the dredge cells in
Pond 3B is shown in Figure 5.
The operation will be a two-dredge system; the primary dredge will operate
in dredging cells designated to have adequate recovery rates, feeding the FCR
facility a minimum of 200 tph, and the secondary dredge will deliver no/low-
yield CSR directly to the static/paste thickener system at a rate of 40 to 50 tph.
FCR Facility Labor

The plant will be operated for four 12-hour shifts based on an 8-day week. The
shift staffing will include one site manager (to oversee complete operation),
one working foreman, and four to five plant/dredge operators. All workers at
the facility will multitask and be trained to operate, maintain, and repair all
dredges, equipment, and facility systems being utilized in the FCR plant.
Special Operation Requirements and Processes for the FCR Facility

In addition to recovering the fine coal from the three impoundment structures
and delivering this fine coal to the TCPP, two other special operation require-
ments will be performed:
1. Removal of the no/low-yield “overburden” from specific areas of the
impoundment structures and bypassing the FCR processing facility,
and
2. Processing the no/low-yield CSR and refuse from the CRG-FCR
facility through the static/paste thickener system before final dis-
charge into Pond 3E.
Removal of No/Low-Yield Overburden

The coal that was mined and processed previously at the Centralia mine con-
tains a kaolin clay material that can remain suspended in water for extremely
Actual Vibracore Drill Location

Slurry Ponds 3B
TransAlta-Centralia Mining LLC
Figure 5 Mining/dredging plan for Pond 3B
long periods of time. The Centralia mine treated the CSR from their process
in a static thickener, attempting to cause the clay material to settle out, with
little effect. The clay that was discharged from the thickener underflow into
the impoundment structures remained suspended as long as slurry material was
pumped into the impoundment structure. The clarified water from the static
thickener also contained substantial amounts of the clay, creating problems
within the Centralia preparation process. As the mining and preparation activi-
ties continued, the concentration of the clay in the impoundment and clarified
water continued to increase, causing problems in the processing system.
Table 4 Deep hole sample, Pond 3D-1 size distribution

Sample Depth, ft Percent >100 Mesh Percent >200 Mesh
2.5 0.5 1.6
12.5 4.5 9.1
22.5 2.2 3.8
42.5 10.7 26.7
62.5 9.1 14.9
82.5 10.2 15.5
106 10.1 17.8
132 11.6 19.5
146.5 7.4 19.3
As the CSR material was discharged into the impoundment structures, the
majority of more dense materials and larger particle coal settled and extended
out and under the suspended clay materials. When the impoundment struc-
tures reached capacity and/or the mining and preparation operations ceased,
the clay slowly settled and covered the underlying materials, developing an
“overburden” on the more coarse and higher recoverable reserves. This is
confirmed in the evaluation of a series of deep core samples taken from the
impoundment structures and shown in one of the deep core sample summaries
shown in Table 4. The dredge cells of no/low yield will be identified and the
secondary dredge will pump this material directly to the static/paste thickener
to be developed into a paste for final deposition into Pond 3E.
Deposition of Paste into Final Cut, Pond 3E

This project was developed to not only create a fine coal product to sell into the
TCPP but to relocate the CSR material in the valley fill impoundment struc-
tures to a below-grade final cut, Pond 3E. The Centralia mine site is located
near Mount Rainier to the east and Mount Saint Helens to the south. Federal
regulations require valley fill dams to be reclaimed when in active seismic areas
to prevent the possibility of liquefaction and long-term structural issues.
The development of a paste refuse material is considered the best method
to deposit the CSR from the CRG-FCR plant. Pond 3E is a final cut area,
currently filled to capacity with clear water and is a link in the NPDES for the
mine site. It is essential to maintain and minimize suspended solids within this
structure. Previous operations of the paste thickener have indicated that paste
discharge will remain in a paste form and will not break down or return to
slurry form with suspended solids.
PROJECT SUMMARY
The development of recovery operations at slurry waste impoundment struc-
tures provides a variety of benefits. These benefits include but are not limited to
• Regaining minerals or energy from the waste facility and returning to
end users as a usable product,
• Reduction of the quantity of slurry materials and water volumes
within the impoundment structure or evacuating all materials and
relocating them to another area,
• Decreasing the reclamation costs and enhancing the final reclamation
plan, and
• Eliminating the Mine Safety and Health Administration hazard risk
ranking of the impoundment structures.
The installation of the fine coal recovery system at the TransAlta-Centralia
mine site will incorporate all the benefits previously listed. CRG will process an
estimated 16.2 million tons of raw coal slurry to produce 3.42 million tons of
fine coal product to be transported directly to the adjacent TransAlta-Centralia
complex, a 1,404-MW two-unit coal-fired power plant. The coal slurry refuse
resulting from the CRG-FCR plant will process and treat this refuse through
the two-stage static-paste thickener system and develop a more stable paste
material for final deposition into Pond 3E, a final-cut, below-grade, incised
impoundment. An additional 1.86 million tons of no/low-yield material will
bypass the CRG-FCR plant into the thickener system for treatment before final
deposition into Pond 3E.
This project will evacuate nearly all the coal slurry refuse and water from
the three downstream-constructed impoundment structures—Ponds 3B, 3C,
and 3D—and deposit the more stable paste into Pond 3E. This recovery opera-
tion will reduce the reclamation cost to TransAlta-Centralia mine by
• Eliminating the requirement to create a positive drainage from the
upstream portion of the impoundment, across the slurry surface to
discharge through the impoundment berm;
• Reducing the requirement of soil cover on the slurry surface to create
a stable surface area and one that will sustain a vegetative growth of
natural and native flora; and
• Doing away with the maintenance and repair of soil cover erosion and
exposure of the slurry subsurface.
The CRG recovery operation is a benefit to the following entities:

• TransAlta-Centralia mine—removing an environmental liability and a
reduction to their reclamation cost
• TransAlta-Centralia complex power plant—gaining an enhanced
quality, low-cost coal product
• Labor force—recovery activities that create employment in a labor-
depressed area
• Environment—will consolidate refuse materials to one below-grade
area and reduce surface areas exposure of the CSR, soil cover, and
reclamation
Pressurized Fluidized-Bed
Combustion Technology for
Fine Coal Utilization*
Deborah A. Kosmack, Michael Fenger,
and Esko Polvi
ABSTRACT
Waste coal, otherwise known as “culm” or “gob,” has been stored in large quantities
throughout the coalfields of Pennsylvania and other Appalachian states. Some of
this legacy waste material and much of the continuing production of waste coal is
in the form of wet fine particles stored in impoundments. Pressurized fluidized-
bed combustion (PFBC) technology can use these impounded fines for highly
efficient electric power generation. PFBC plants in the 100–400 MWe (megawatt,
electric) size range have been in successful commercial operation since 1990 with
demonstrated efficiencies of more than 40% higher heating value and availabili-
ties of 85%–90%. A small footprint and inherently low emissions make PFBC
combined-cycle technology an attractive choice for repowering dated and difficult-
to-permit pulverized coal power plants. The PFBC process is very robust and it can
efficiently burn a wide range of fuels. It is well suited for high-moisture fine coals
and it has been tested with a variety of fuel mixes with low and high Btu values,
including a wet waste coal/biomass mix, without significant detrimental impact
on the combustion behavior or emissions. To generate the project-specific data
necessary for designing commercial power plants, PFBC Environmental Energy
Technology, Inc., and Consol Energy, Inc., operate a 1-MWt (megawatt, thermal)
PFBC pilot plant. This chapter will describe the process test facility and selected
results of its operation on fine wet waste coal material.
* All rights, title, and interest to this work, including the copyrights, belong to
PFBC-Environmental Energy Technology, Inc., and Consol Energy, Inc. PFBC-
Environmental Energy Technology, Inc., and Consol Energy, Inc., have given
permission to SME, Inc., to include this work in this book.
377
PRESSURIZED FLUIDIZED-BED COMBUSTION

TECHNOLOGY
PFBC Environmental Energy Technology, Inc. (EET), and Consol Energy,
Inc., are developing pressurized fluidized-bed combustion (PFBC) technology
to turn fine, wet waste coal efficiently into clean electricity. The PFBC-EET
PFBC technology is licensed from Alstom; this is the same technology on
which the existing PFBC plants in operation are based. The technology is a
combination of the Rankine steam cycle and Brayton gas turbine cycle with the
result of achieving high cycle efficiency and low emissions.
The PFBC system process (shown in Figure 1) employs a pressurized
fluidized-bed combustor, which operates at a bed temperature of 840°–880°C
(1,544°–1,616°F) and a pressure of 12–16 bars (174–232 psi). The combustor
generates pressurized flue gases from the combustion of coal in the bed and
steam from the fluidized-bed boiler consisting of tube bundles immersed in
the 3.5-m (11.5-ft) deep bed of hot ash. The coal is injected dry or pumped as
paste along with the sorbent into the bed and burned in the fluidized bed. The
sorbent is fed to capture sulfur from the coal in the bed, resulting in low sulfur
emission. The larger sorbent particles remain in the pressurized combustor to
become fluidized-bed ash. The pressurized flue gas exits the fluidized bed at
a low velocity of 1 m/s to be cleaned of the suspended particulate by means
of two-stage cyclones. The pressurized flue gas is then expanded through a
gas turbine connected to an air compressor and a generator. A Brayton cycle
gas turbine expander drives the air compressor supplying air required for the
combustion process; the combustion air requirement sets the limitation on the
gas turbine output. The combustion air requirements are maintained at about
20% excess air. However, there is a limit on gas turbine sizes available, so the
original Alstom PFBC plant size was determined using all these design criteria.
On the steam cycle side, heat transfer surfaces are immersed in the fluidized
bed, and steam generated passes through the conventional steam cycle operat-
ing on the Rankine cycle. The split between the electric power generated by
the steam cycle and that generated by the gas turbine is on the order of 80:20.
Thus, a combination of Rankine cycle and Brayton cycles results in high cycle
efficiencies, which are projected to be higher than conventional steam plants
by 4% to 5%.
PFBC POWER PLANT EMISSIONS

The current PFBC power plants operating throughout the world include the
following:
Pressurized Fluidized-Bed Combustion Technology 379
Figure 1 PFBC system process diagram
Table 1 Typical flue gas emissions

PFBC Emissions* Flue Gas Properties (typical values for Stockholm†)
Low particulates 98% of ash removed by two-stage cyclones (0.005 lb/MBtu‡)
Low carbon ash 98%–99% carbon combustion efficiency
Low SOx 90%–95% sulfur removal in the fluidized bed (0.14 lb/MBtu)
Low NOx 1/3 of New Source Performance Standards (0.07 lb/MBtu)
Low CO 80% less than pulverized coal combustion (0.01 lb/MBtu)
Less CO2 produced 20% less than pulverized coal combustion
*CO = carbon monoxide; CO2 = carbon dioxide; NOx = nitrogen oxides; SOx = sulfur oxides.
†Source: R. Farmer, World’s first direct coal-fired PFBC plant goes commercial. Gas Turbine
World, September-October 1991.
‡Emissions are reported in pounds per million British thermal units.
• Cottbus, Germany—1999 Model P200, combined heat and power

(80 MWe )
• Stockholm, Sweden—1991 Model P200X2, combined heat and
power (135 MWe )
• Karita, Japan—2001 Model P800 (360 MWe )
These plants have operated at thermal efficiencies of 40% and low emissions as
shown in Table 1.
The PFBC process has inherently low nitrogen oxides (NOx) emissions
because the combustion temperature is well below the range of thermal NOx
formation. Use of selective non-catalytic reduction allows PFBC to meet

all applicable NOx levels. Up to 99% SO2 (sulfur dioxide) removal can be
achieved by adjusting the calcium-to-sulfur ratios. Some fuel-bound mercury is
captured in the cyclones, and the remainder is captured in the fabric particulate
filter. Because PFBC produces a pressurized gas stream, the partial pressure of
carbon dioxide (CO2) is naturally elevated, making it easier to separate down-
stream. The higher cycle efficiency also reduces the amount of CO2 required to
be removed by burning less fuel per unit output. The bed ash and fly ash—col-
lected from the bottom of the bed with the cyclones and fabric filter—consists
of calcium sulfite and sulfate, calcium carbonate, silica, and very little carbon.
PFBC ashes have low leaching characteristics and have been proven and
accepted for beneficial use.
P F B C F O R F I N E C O A L U T I L I Z AT I O N
The fine waste coal from active coal preparation plants represents a significant
resource. Given that the cost of cleaning fine coal in the preparation plant can
be three to four times higher than the cost of cleaning larger size coal, the fine
coal and refuse are often discarded to a waste coal impoundment as slurry.
The discarded wet fines often contain 50% or more coal. These fines can be
dewatered to 25%–30% moisture to produce a pumpable paste, which can be
injected in a fluidized-bed combustor. Fluidized-bed combustors are inherently
fuel flexible. High pressure (12 to 16 bars) increases the density of the air/
oxygen relative to fuel particles. The turbulence, the deepness of the bed, and
the low gas velocity (leading to longer residence time) all contribute to nearly
complete oxidation of carbon in the fuel. Wet and high-moisture fuels create
additional water vapor mass flow through the gas turbine expander, increasing
the power output. Therefore, cycle efficiency is only slightly reduced by wet
fuels, unlike pulverized-coal-fired plants or atmospheric fluidized-bed units
where the water vapor is simply exhausted. To reduce the amount of waste coal
pumped to impoundments from active coal preparation plants, the PFBC plant
can be integrated with a coal preparation plant.
Pennsylvania Department of Environmental Protection records show that
the coal refuse piles in Pennsylvania which were abandoned prior to 1978
included 819 separate sites in 35 counties. Those sites contain an estimated
170 million tons of material that covers 8,021 acres of surface land. These
figures include only those refuse piles abandoned before 1978 yet exclude
many others, such as a 40-to-70-million-ton pile in Washington County,
which was still active in 1978. Some of these refuse sites contain substantial
concentrations of coal, and the material may have a heating value of 8,000 Btu/lb
or more; others have much lower energy values, about 2,000 Btu/lb. When
compared to coal on an energy-content basis (e.g., lb/MBtu), coal refuse often
contains higher concentrations of sulfur, mercury, and other deleterious ele-
ments. Thus, whatever technology is employed to consume the coal refuse must
be highly effective in controlling emissions of pollutants and must be able to
burn low-grade fuel of highly variable quality.
P F B C P R O C E S S T E S T FA C I L I T Y
PFBC-EET and Consol Energy joined together in 2006 to
• Design and reconstruct a 1-MWt input PFBC Process Test Facility
(PTF) in South Park, Pennsylvania (pressure vessel and fluidized-bed
combustor relocated from Alstom’s PTF in Finspong, Sweden);
• Demonstrate the operation of PFBC technology using fine wet waste
coal as the fuel; and
• Evaluate performance of PFBC on waste coal/biomass with CO2
capture.
The purpose of the PTF pilot-plant research test program is to provide the
necessary fluidized-bed combustion and emissions data needed to design and
build commercial-scale pressurized fluidized-bed combined gas-steam cycle
generating units that will operate on fine waste coal. The operating pilot-scale
PTF at Consol Energy Research & Development was constructed to conduct
tests to determine
• Fuel properties and sorbent selection;
• Fluidized-bed temperatures and heat transfer;
• Flue gas emissions and emissions sensitivity;
• Bed ash and fly ash distribution, handling, and utilization samples; and
• Ash samples for lab analysis and utilization tests.
The data from the PTF testing will be used to prepare
• Operations and maintenance projections;
• A fuel impact model;
• A fluid dynamic model for boiler, combustor, and cyclone design;
• Operator training;
• Risk assessment and mitigation for equipment design; and
• Define scope for full-scale plant design and cost estimates.
The pilot plant consists of a wet and dry fuel handling system, pressurized
combustion vessel, and ash handling system. Coal and limestone sorbent are
mixed with water and fed as a paste into a pressurized fluidized-bed combus-
tor. The combustion process occurs in a fluidized bed that is characterized by
a low gas velocity and high residence time, which provides high combustion
efficiency and low pollution emissions.
Process Flow Diagrams of the PTF

The PTF was not designed or built to have the capability of generating electric-
ity. The pilot plant does not include a steam turbine, gas turbine, generators,
condenser, economizer, or deaerator. Instead there are heat exchangers and an
air compressor to simulate inputs and outputs representative of a commercial-
scale power-generating unit. The PTF includes fuel preparation equipment
(Figure 2) to produce and pump the fuel/sorbent as a paste; a pressure vessel
(Figure 3), which contains the 6.7-m- (22-ft-) high fluidized-bed combustor
vessel, 3.0-m- (9.8-ft-) high water-cooled tube bundle inside the combustor ves-
sel, and C1 and C2 cyclones to collect fine ash; and flue gas cooling and ash col-
lection equipment (Figure 4). The C0 cyclone shown in Figure 3 was blocked
during all the test runs to date because ash recirculation was not required.
The fluidized bed of ash in the combustor is 329 mm (14.3 in.) in diameter
by 3.5 m (11.5 ft) deep, the same depth as a full-scale combustor. The PTF is
started up by preheating a 0.5-m- (1.6-ft-) deep bed of ash to 650°C (1,202°F)
and then injecting fuel and stored bed ash to increase the bed depth to 3.5 m
(11.5 ft). The bed temperature is raised to 840°–880°C (1,544–1,616°F) by
adjusting the fuel and air flows. All incoming and outgoing flows of gas and
solids are monitored and sampled to perform material balances and analysis.
The heat removed by the tube bundle immersed in the fluidized hot bed ash is
monitored.
C L E A N C O A L A N D F I N E W A S T E C O A L P T F T E S T D ATA
C O M PA R I S O N
A series of test runs were conducted to commission and calibrate; the PTF on
three different coals. The coals included a Bailey mine clean coal; a coal (Pol-
ish) used at the Vartan PFBC plant in Stockholm, Sweden; and a fine waste
coal from a Consol preparation plant. The test runs were designated as follows:
• Run 14, Bailey clean coal with Maple Grove Quarry (MGQ) dolomite
limestone sorbent
• Run 15, Vartan coal with Vartan dolomite limestone sorbent
Figure 2 PTF waste coal/sorbent fuel preparation process diagram
Figure 3 PTF pressure vessel, bed vessel, and cyclones process diagram
Figure 4 PTF flue gas cooling and ash collection process
• Run 16, Vartan coal with Vartan dolomite limestone sorbent

• Run 20b and 20c, fine waste coal with MGQ dolomite limestone
sorbent
The Bailey and fine waste coal are Pittsburgh-seam bituminous coals. The
MGQ dolomite limestone was from MGQ Aggregates of Old Fort, Ohio. The
fuel particle size is shown in Table 2. The paste fuel (coal/sorbent) properties
are shown in Table 3. Test runs 14, 15, and 16 were conducted to verify and
calibrate the operation of the PTF relative to earlier test runs performed by
Alstom in Finspong, Sweden. Test run 20 was primarily conducted to produce
flue gas for a CO2 capture pilot-plant test. Each test run included collecting
operating data, solid samples, and gas samples during 2-hour test periods after
stabilizing at targeted oxygen levels. The following sections summarize the sig-
nificant results from the test runs.
PTF Test Data Comparison—Combustion Temperatures

Figure 5 shows combustion temperatures (°C) that were measured in the
3.5-m- (11.5-ft-) deep fluidized bed and freeboard above the bed for the three
different coals during separate test periods of each test run. Compared to the
Bailey clean and Vartan coals, the waste coal with higher amounts of fine ash
appeared to influence the temperatures throughout the fluidized-bed combus-
tor. The higher amount of fine coal and ash particles appeared to raise the fluid-
bed mean temperature above the fluid-bed bottom temperature. The fluid-bed
freeboard temperature also tended to rise to be closer to the bed temperatures.
The bed temperatures were set to achieve stable emissions and other operat-
ing conditions. Inspection of bed ash and cyclones showed no sign of bed ash
Table 2 Coal particle size

Run No. Coal* % >2.0 mm % >0.3 mm % >0.05 mm
14 Bailey clean coal 90 50 25
15 Vartan coal 90 50 25
16 Vartan coal 90 50 25
20b Fine waste coal 95 80 50
20c Fine waste coal 95 80 50
*5-mm topsize was limited by paste injection lance requirements.
Table 3 Paste fuel (coal/sorbent) properties

Run No. Coal Ash, % Sulfur, % Sorbent, % Ca/S Ratio Water, %
14 Bailey clean coal 13.6 1.6 13 1.7 25
15 Vartan coal 12.0 0.4 9 4.5* 28
16 Vartan coal 12.0 0.4 9 4.5* 28
20b Fine waste coal 30.0 3.4 26 1.8 30
with 3% sawdust
20c Fine waste coal 30.0 3.4 26 1.8 30
with 3% sawdust
*Ca/S (calcium/sulfur) ratio required to maintain bed level.
900
880
860
Temperature, °C
840
820
800
Fl Bed Btm Temp

780 Fl Bed Freeb Temp
Fl Bed Mean Temp
760
#14 Bailey CC
#14 Bailey CC
#14 Bailey CC
#14 Bailey CC
#15 Vartan
#16 Vartan
#16 Vartan
#16 Vartan
#20b W Coal
#20b W Coal
#20b W Coal
#20c W Coal
#20c W Coal
#20c W Coal
#20c W Coal
Figure 5 Fluidized-bed combustor temperatures
0.40 8
SO2 lb/MBtu
0.35 NOx lb/MBtu 7
O2 %
0.30 6
NOx or SO2, lb/MBtu
0.25 5
O2, %
0.20 4
0.15 3
0.10 2
0.05 1
0.00 0
#14 Bailey CC
#14 Bailey CC
#14 Bailey CC
#14 Bailey CC
#15 Vartan
#16 Vartan
#16 Vartan
#16 Vartan
#20b W Coal
#20b W Coal
#20b W Coal
#20c W Coal
#20c W Coal
#20c W Coal
#20c W Coal
Figure 6 Combustion emissions
sintering or cyclone ash deposits during any of the test runs. This indicated
that the temperatures were not excessive. Further commercial test programs
will verify this temperature pattern. The temperature pattern will influence the
design of the boiler, cyclones, and gas turbine expander.
PTF Test Data Comparison – Combustion Emissions

The emission data shown in Figure 6 show the expected behavior of rising SO2
and falling NOx as the excess oxygen was lowered during the test runs. The
SO2 emissions for the PTF tended to be lower than what will occur in the full-
scale combustor because of the higher solids gas contact in a small-diameter
combustor. The carbon monoxide (CO) during the test runs averaged 3.2 ppm
and varied from 1 to 10 ppm, depending on excess air. The carbon combustion
efficiency remained very high at 98% to 99%.
PTF Test Data Comparison—Combustion Ash Sulfur Distribution

The purpose of the sorbent is to absorb sulfur and provide the fluidizing media
for the bed. The bed ash removed from the fluidized bed, C1and C2 ash col-
lected by the C1 and C2 cyclones produced during PTF test runs, was analyzed
to determine the performance of the sorbent. Figure 7 shows a distribution
100
C2 Ash %
90 C1 Ash %
26 25 Bed Ash %
31
80 34
48 45 47 45
51 50
70 57
60
78
81 79 72
Sulfur, %
50
40
72 73
30 64 68
51 53 52 54
47 49
20 41
10 22 19
16 17
0
#14 Bailey CC
#14 Bailey CC
#14 Bailey CC
#14 Bailey CC
#15 Vartan
#16 Vartan
#16 Vartan
#16 Vartan
#20b W Coal
#20b W Coal
#20b W Coal
#20c W Coal
#20c W Coal
#20c W Coal
#20c W Coal
Figure 7 Distribution of sulfur in C2, C1, and bed ash
of sulfur flow in the three ashes collected during separate test periods of each
test run. Test run 14 shows that most of the sulfur was removed with the bed
ash. Test runs 15 and 16 results show the effect of a very high calcium/sulfur
(Ca/S) ratio. Test run 20 at 26% sorbent in the fuel shows lower sulfur in the
bed ash compared to test run 14 at 13% sorbent in the fuel. The lower sulfur
flow during test run 20 was most likely due to the larger amount of sorbent in
the paste fuel (coal/sorbent). The increased flow of sorbent into the fluidized
bed will increase the production of bed ash, which reduces the residence time
of the sorbent in the bed and thereby reduces sulfur absorption. The sorbent
properties, particle sizing, and particle attrition in the bed will influence the
distribution of the sulfur in the three ashes.
PTF Test Data Comparison—Combustion Ash Calcium Utilization

The Ca/S molar ratio in the ash indicates the utilization of the calcium to
remove the sulfur. Stoichiometrically 1 mole of calcium is required to remove
1 mole of sulfur (i.e., a 1.0 molar ratio). Figure 8 shows the Ca/S ratio in the
three ashes collected during separate test periods of each test run. Test run 14
shows that the best utilization of calcium occurred with Ca/S ratios of 1.6 in
the bed ash. Test runs 15 and 16 results show very high Ca/S ratios because of
12
11 Bed Ash Ratio

C1 Ash Ratio
10 C2 Ash Ratio
7
Ca/S ratio
0
#14 Bailey CC
#14 Bailey CC
#14 Bailey CC
#14 Bailey CC
#15 Vartan
#16 Vartan
#16 Vartan
#16 Vartan
#20b W Coal
#20b W Coal
#20b W Coal
#20c W Coal
#20c W Coal
#20c W Coal
#20c W Coal
Figure 8 Calcium/Sulfur ratio in C1, C2, and bed ash (1.0 is optimum)
the excess amount of calcium needed to sustain the bed level and the very low
sulfur in the Vartan coal. Test run 20 shows poorer calcium utilization com-
pared to run 14, most likely due to the larger amount of sorbent in the paste
fuel, which reduced the residence time of the sorbent in the bed and raised the
bed ash removal rate. Adjustment of the limestone sorbent particle sizes may
be necessary to improve calcium utilization and thereby reduce the amount of
sorbent required. The Ca/S ratios of 1.7 and 1.8 in the fuel during test runs 14
and 20 could potentially be lowered and still achieve acceptable sulfur removal.
The third test period during run 14 shows an advantageous condition where all
three ashes are at the lowest ratios.
PTF Test Data Comparison—Tube Bundle Heat Removal

The heat removed by the tube bundle as a percentage of the input heat from
the coal is shown in Figure 9. The presence of larger amounts of fine ash in the
waste coal appeared to lower the heat transferred to the tube bundle. The fine
ash particles were most likely taking heat out of the bed, which corresponded
with the rise in mean bed and freeboard temperatures shown in Figure 5. The
full-scale fluidized boiler would be designed for these conditions.
60
58
56
54
52
50
%
48
46
44
42
40
#14 Bailey CC
#14 Bailey CC
#14 Bailey CC
#14 Bailey CC
#15 Vartan
#16 Vartan
#16 Vartan
#16 Vartan
#20b W Coal
#20b W Coal
#20b W Coal
#20c W Coal
#20c W Coal
#20c W Coal
#20c W Coal
Figure 9 Tube bundle heat removal—% of input heat from coal
CONCLUSIONS
Fine coal utilization involves many processing steps, starting at the coal prepa-
ration plant or impoundment and finishing with combustion at a power plant
boiler. Based on the early results of test runs at the PTF, fine wet waste coal can
be burned efficiently and cleanly using PFBC technology. The major conclu-
sions from the current work are as follows:
• The fine waste coal was successfully burned in the 1-MWt PFBC PTF
in South Park, Pennsylvania. The low fluidized-bed superficial gas
velocity of 1 m/s is adequate to maintain combustion efficiency of
98% or more.
• The sorbents selected for the test runs performed well, but further
testing of potential sorbent sources and size distributions need to
be completed to improve calcium utilization and thereby reduce the
amount of sorbent required.
• The fine coal and larger amounts of ash do not appear to influence
emissions as compared to the coarser and cleaner coals.
• The amount of fine ash influences the temperature profile in the com-
bustor, which influences the fluidized boiler design.
Index
Agglomeration, 18–19 Belgium, and dense-medium cyclones,
Alabama 140–141
Centribaric centrifuges in capture of fine Billiton Mitsubishi Alliance, 261–277
coal lost to refuse stream (No. 7 Blackwater, Australia, 261
mine), 303–308 high-speed disc filters in coal ultrafines
Jim Walter Resources No. 7 mine processing for coking coal, 261–277
(Brookwood), 303 Bokela Boozer disc filters, 261–277
Anglo American Thermal Coal South Briquetting, 18–19
Africa, 95–98, 120–121 Brookwood, Alabama
briquetting, 100–102, 108–109, Centribaric centrifuges in capture of fine
118–119, 120 coal lost to refuse stream (No. 7
dewatering and filtration, 99–100, mine), 303–308
104–106, 114–117, 120 Jim Walter Resources No. 7 mine, 303
dual cell flotation, 110–114, 119
flotation, 98–99, 102–103, 106–107, Cavitation-tube sparging technology, 195
109–110 Centralia, Washington, 361
multicell flotation, 103 Coalview Recovery Group in processing
Arch Coal, Inc., 67–68 of slurry refuse from TransAlta-
fine pyrite rejection by cyclone and Centralia mine, 361–375
sieve classifiers, compound spiral Centrifugal flow-film concentrators,
concentrators, and reflux classifiers 231–234
(Leer Mining plant, W. Va.), 68–78 Centrifugal fluidized-bed concentrators,
Australia 234–236
clay issues, 126–133 Centrifugal jigs, 236–238
coal grain analysis, 133–134 Centrifuges, 280–281
coal markets, 124–126 Centribaric centrifuges in capture of
and dense-medium cyclones, 145–147 fine coal lost to refuse stream ( Jim
high-speed disc filters in coal ultrafines Walter Resources No. 7 mine,
processing for coking coal, Brookwood, Ala.), 303–308
261–277 Chemical separation, 15–19
mechanical flotation cell usage, 221–224 China
regional coal types, 123–124 and dense-medium cyclones, 147–149
University of Newcastle and Ludowici mechanical flotation cell usage, 224
Australia in development of reflux Coal
classifiers, 159–183 clays in, 59–62
391
392 CHALLENGES IN FINE COAL PROCESSING
high-reject, 42 and filtration (Anglo American Thermal

inorganic components, 57–62 Coal South Africa), 99–100,
macerals in, 49–57, 62 104–106, 114–117, 120
minerals in, 58–62, 63–64 of fine coal and tailings with filter press,
organic components, 49–57 279–292
poorly liberated, 33–37 froth flotation concentrate, 280
pyrite in, 62 typical filter press equipment and circuit,
pyrite-rich, 42 282–291
rank, 49–57, 63 ultrafine material from screen-bowl
Coal jigs, 7–8 effluent, 281
Coalview Recovery Group, 361 0.15 mm × 0 raw coal or refuse, 280
in processing of slurry refuse from See also Drying
TransAlta-Centralia mine, Distributed control systems (DCSs),
361–375 25–26
Computer technology in process control, Dry separation, 19–21
25–28 Drying, 345–347
Cones, 8 current technologies, 337–338
deep cone thickener (Lone Mountain dielectric heating and drying, 341–342
coal processing plant, Lee County, emerging technologies, 339–344
Va.), 293–301 Nano Drying Technology, 329–330,
Cyclones 340–341, 347–359
in fine pyrite rejection, 68–78 Parsepco Drying Technology, 329–330,
See also Dense-medium cyclones; Water- 339–340
only cyclones potential benefits to preparation plants,
332–337
Decanter Machine, 303 principle of, 330–332
Deep cone thickeners, 293 See also Dewatering
at Lone Mountain coal processing plant
(Lee County, Va.), 293–301 Electrical separation, 21–22
Dense-medium cyclones, 13–15, 139, Elutriation “teeter-bed” separators, 11
156–157 Engineering developments, 28
Australia, 145–147 Eriez Manufacturing Company, 187–209
Belgium, 140–141 Europe, and mechanical flotation cell
China, 147–149 usage, 219
diameter and geometry, 149
feed pressure, 150–151 Falcon concentrator, 231–234
history of use with coal, 140–149 Filter presses
and magnetite consumption, 151–152 belt type, in dewatering of coal tailings
and magnetite sizing, 149–150 for disposal, 309–327
and particle size, 152–155 belt type, technological description of,
South Africa, 140, 145, 147 310–314
United States, 141–144 in dewatering of fine coal and tailings,
Dewatering, 22–25, 345–347 279–292
by filter press, of tailings for disposal, typical equipment and circuit, 282–291
309–327
Index 393
Filtration particle size, 216

and dewatering (Anglo American particle size distribution, 199–200, 216
Thermal Coal South Africa), residence time (mechanical cells), 214
99–100, 104–106, 114–117, 120 SlamJet technology, 194–195
high-speed disc filters in coal ultrafines StackCell technology, 81–94, 195–199
processing for coking coal,
261–277 Gravity separation, 227–228, 238–240
Fine coal processing, 3–7 centrifugal flow-film concentrators,
at Anglo American Thermal Coal South 231–234
Africa, 95–121 centrifugal fluidized-bed concentrators,
Australian challenges and developments, 234–236
123–136 centrifugal jigs, 236–238
challenges in, 33–44 Falcon concentrator, 231–234
Coalview Recovery Group in processing Knelson continuous variable discharge
of slurry refuse from TransAlta- concentrators, 235–236
Centralia mine, 361–375 spiral concentrators, 228–231
with dense-medium cyclones, 139–158
future directions, 28–29 Heavy-medium cyclones. See Dense-
and high-reject coals, 42 medium cyclones
high-speed disc filters in coal ultrafines
processing for coking coal, Knelson continuous variable discharge
261–277 concentrators, 235–236
methods, 7–28
and plant water shortages, 37–42 Leer Mining plant (Taylor County, West
and poorly liberated coals, 33–37 Va.), 67–68
pressurized fluidized-bed combustion fine pyrite rejection by cyclone and
in utilization of impounded coal sieve classifiers, compound spiral
fines for electric power generation, concentrators, and reflux classifiers,
377–389 68–78
and pyrite-rich coals, 42 Lone Mountain plant, Lee County,
Flotation, 15–18 Virginia, 293
aeration, 202–204 experience with deep cone thickener,
Anglo American Thermal Coal South 293–301
Africa (dual and multicell), 95–121 Ludowici Australia, 159–183
carrying capacity, 199–200
cavitation-tube technology, 195 Magnetic separation, 21
cell technology, 193–199 Materials science, 28
circuits, 190–192
column and nonconventional systems, Nano Drying Technology (NDT),
187–209 329–330, 340–341, 347–348
froth washing, 200–202 bench-scale testing, 348–354
frother rates and handling, 204–207 pilot-scale demonstration, 354–359
hydrodynamics of, 214–215 Newcastle, University of (Australia),
mechanical circuits, 211–226 159–183
394 CHALLENGES IN FINE COAL PROCESSING
Parsepco Drying Technology (PDT), TransAlta-Centralia Mining (Centralia,

329–330, 339–340 Wash.), 361
Pressurized fluidized-bed combustion Coalview Recovery Group in processing
(PFBC), 378 of slurry refuse from Centralia
in utilization of impounded coal fines mine, 361–375
for electric power generation, Tribo-electric separation, 21–22
377–389 Troughs, 8
Programmable logic control (PLC), 25–26
Pyrite United Kingdom, and mechanical flotation
in coal, 62 cell usage, 219
coals rich in, 42 United States
fine, rejection by cyclone and sieve and dense-medium cyclones, 141–144
classifiers, compound spiral and mechanical flotation cell usage,
concentrators, and reflux classifiers, 218–219
67–78
Virginia
Reflux classifiers experience with deep cone thickener
development of, 159–166, 182–183 (Lone Mountain plant), 293–301
in fine pyrite rejection, 68–78 Lone Mountain coal processing plant
theoretical basis for, 166–183 (Lee County), 293
SCADA systems, 25 Walter Energy

Shaking tables, 9–11 Centribaric centrifuges in capture of fine
Sieve classifiers, in fine pyrite rejection, coal lost to refuse stream (No. 7
68–78 mine), 303–308
SlamJet bubble generation systems, 194–195 Jim Walter Resources No. 7 mine
South Africa (Brookwood, Ala.), 303
and dense-medium cyclones, 140, 145, Washington
147 Coalview Recovery Group in processing
mechanical flotation cell usage, 220–221 of slurry refuse from TransAlta-
Spiral concentrators, 8–9, 228–231 Centralia mine, 361–375
compound spiral concentrators in fine TransAlta-Centralia mine, 361
pyrite rejection, 68–78 Water-based separation, 7–15
and water-only cyclones in combination Water-only cyclones, 12–13
(design and operating guidelines), and spiral concentrators in combination
241–258 (design and operating guidelines),
StackCell system, 81–83, 93–94, 195–199 241–258
agitation, 85–87 West Virginia
evaluation, 88–93 fine pyrite rejection by cyclone and
froth washing, 87 sieve classifiers, compound spiral
multistage technology, 83–85 concentrators, and reflux classifiers
(Leer Mining plant, Arch Coal,
Tailings, filter presses in dewatering of, Inc.), 68–78
279–292, 309–327 Leer Mining plant, 67–68
Challenges in Challenges in
Challenges in FINE COAL Processing, Dewatering, and Disposal

FINE COAL
Processing, Dewatering, and Disposal
Edited by Mark S. Klima, Barbara J. Arnold, Peter J. Bethell FINE
new sources of energy, increased environmental awareness, and more
stringent regulations from the U.S. Environmental Protection Agency and other
organizations are changing the way coal is found, extracted, and used. As
a result, fine coal cleaning, dewatering, and refuse disposal are now at a
COAL
major crossroads.
and Disposal
techniques to address the need for acceptable moisture levels in plant products.
Moreover, the larger volume of fine refuse being generated, coupled with

Symposium, sponsored by the Coal Preparation Society of America; the
Pittsburgh Section of the Society for Mining, Metallurgy, and Exploration,
Inc.; and the Pittsburgh Coal Mining Institute of America.

general knowledge and in-depth discussion on the current challenges
facing the industry, techniques for designing more efficient plants, and new
cleaning and dewatering technologies. The book is a practical yet cutting- Mark S. Klima
edge resource for plant designers, engineers, and other practitioners, and Barbara J. Arnold
for university students and faculty. Peter J. Bethell
The Society for Mining, Metallurgy, and Exploration,

Inc. (SME), advances the worldwide mining
and minerals community through information
exchange and professional development. SME
is the world’s largest association of mining
and minerals professionals.
!SME_FineCoalPDD_FULL_CV_L2.indd 1 9/10/12 11:18 AM

Fine Coal: Challenges in

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Fine Coal: Challenges in

Uploaded by

Copyright:

Available Formats

Challenges in Challenges in

Challenges in FINE COAL Processing, Dewatering, and Disposal

This book is a compilation of information presented at the 2012 Fine Coal

Provided by international coal companies, major research organizations,

The Society for Mining, Metallurgy, and Exploration,

!SME_FineCoalPDD_FULL_CV_L2.indd 1 9/10/12 11:18 AM

Copyright © 2012 Society for Mining, Metallurgy, and Exploration, Inc.

All Rights Reserved. Printed in the United States of America.

No part of this publication may be reproduced, stored in a retrieval system, or transmitted

Library of Congress Cataloging-in-Publication Data

Part 1 Perspective on Coal Fines

Nano Drying Technology—A New Approach for Fine Coal

Barbara J. Arnold G.J. de Korte

Peter J. Bethell Andy Dynys

J. Franklin Gerald Luttrell

Fred Stanley Asa Weber

Chris Swanepoel Keith Wilkes

Larry Watters Eric Yan

In a previous chapter on this subject (Osborne 1986), it appeared that the

New Materials, etc. Intensified

First Jigs Earliest Early Trough Baum Water Only

Figure 1 Development pathway for fine coal beneficiation

Troughs, Spirals, and Cones

Hoyois Slack Screens

included many changes to the traditional mineral spirals, specifically improv-

Single Stage Primary-Secondary Primary-Secondary Primary-Secondary

Feed Feed Feed Feed

Refuse Clean Refuse Clean Refuse Clean Refuse Clean

Source: After Bethell 2011.

Figure 3 Alternative spiral configurations

and 25 mm (although 15 mm is the norm) with capacities of up to 13.6 metric

Elutriation “Teeter-Bed” Separators

Figure 4 Krebs water-only cyclone

Reject Clean Coal

Source: Driessen 1945, reprinted with permission from DSM.

variety of other dual-purpose, high-frequency screens have also emerged via

Reprinted with permission from DSM.

Figure 6 Original DSM sieve-bend design

U.S.A. Belgium Holland South Africa

8. Feed Rate (t/h) 3.3 6.5 12.0(100) 8 15.0(200) — 10–12 — — — — 5 — 36

10. Organic Efficiency — 96.7 96.8 77.9 97.4 — — — — — — — — 93.2 97.68 —

Table 2 Comparison of the probable error (Epm), specific gravity of separation

800 mm 660 mm 500 mm 500 mm 200 mm Two- Two-

the United Kingdom (Mineral Separation), Holland (DSM), Germany (Ekof

Agglomeration and Briquetting

Courtesy of K.R. Komarek, Inc.

Figure 7 Commercial briquetting and compacting roll press

• Pre-dominant dust issues and resultant increased government legisla-

Magnetic and Electrical Separation

of unit, which has already shown great promise in commercial applications

products by a combination of screens, centrifuges, and filters. Thermal drying is

Computers and Process Control

Figure 8 Plant control center with integrated mimic (circa 1980s)

limited in terms of system reliability, and with the introduction of distributed

• Sampling and blending devices, including belt scales (weightometers)

Materials Science and Engineering

Product Supply Process Know-How

Courtesy of SMS Demag.

Figure 9 Development path for equipment improvement

Figure 1 Liberation of an Indian Gondwana coal at different top sizes

make processing economics attractive at marketable quality by reducing plant

Feed Inlet Chute (2)

Fuidization Nozzles Fluidization Chamber