United States Industrial Environmental Research EPA-600/7-79-241
Environmental Protection Laboratory November 1979
Agency Research Triangle Park NC 27711
Devolatilization Kinetics
and Elemental Release in
the Pyrolysis of
Pulverized Coal
Interagency
Energy/Environment
R&D Program Report
-------
RESEARCH REPORTING SERIES
Research reports of the Office of Research and Development, U.S. Environmental
Protection Agency, have been grouped into nine series. These nine broad cate-
gories were established to facilitate further development and application of en-
vironmental technology. Elimination of traditional grouping was consciously
planned to foster technology transfer and a maximum interface in related fields.
The nine series are:
1. Environmental Health Effects Research
2. Environmental Protection Technology
3. Ecological Research
4. Environmental Monitoring
5. Socioeconomic Environmental Studies
6. Scientific and Technical Assessment Reports (STAR)
7. Interagency Energy-Environment Research and Development
8. "Special" Reports
9. Miscellaneous Reports
This report has been assigned to the INTERAGENCY ENERGY-ENVIRONMENT
RESEARCH AND DEVELOPMENT series Reports in this series result from the
effort funded under the 17-agency Federal Energy/Environment Research and
Development Program. These studies relate to EPA's mission to protect the public
health and welfare from adverse effects of pollutants associated with energy sys-
tems. The goal of the Program is to assure the rapid development of domestic
energy supplies in an environmentally-compatible manner by providing the nec-
essary environmental data and control technology. Investigations include analy-
ses of the transport of energy-related pollutants and their health and ecological
effects; assessments of, and development of, control technologies for energy
systems; and integrated assessments of a wide range of energy-related environ-
mental issues.
EPA REVIEW NOTICE
This report has been reviewed by the participating Federal Agencies, and approved
for publication. Approval does not signify that the contents necessarily reflect
the views and policies of the Government, nor does mention of trade names or
commercial products constitute endorsement or recommendation for use.
This document is available to the public through the National Technical Informa-
tion Service, Springfield, Virginia 22161.
-------
EPA-600/7-79-241
November 1979
Devolatilization Kinetics and Elemental
Release in the Pyrolysis of Pulverized Coal
by
V.H. Agreda, R.M. Felder, and J.K. Ferrell
North Carolina State University
Department of Chemical Engineering
Raleigh, North Carolina 27650
Grant No. R804811
Program Element No. EHE623A
EPA Project Officer: N. Dean Smith
Industrial Environmental Research Laboratory
Office of Environmental Engineering and Technology
Research Triangle Park, NC 27711
Prepared for
U.S. ENVIRONMENTAL PROTECTION AGENCY
Office of Research and Development
Washington, DC 20460
-------
ABSTRACT
The evolution of volatile matter and trace elements from
pulverized coal during pyrolysis in an inert atmosphere has been
studied in batch and laminar flow furnace reactors. Five coals
were investigated, ranging in rank from lignitic to anthracitic.
It was determined that at any pyrolysis temperature ash
losses may have a significant effect on calculated extents of
devolati1ization, making the commonly used ash tracer technique
a potential source of error in all experimental pyrolysis stu-
dies. A technique to correct estimated weight losses for this
effect has been developed.
Data on transient and equilibrium elemental release and
volatile yields were obtained in a batch furnace reactor, under
slow heating rates (5-45°C/sec) , over a wide range of tempera-
tures (100-1200°C) and residence times (1-20 minutes). Weight
losses of all coals increased significantly vith temperature.
Sm, Cr, Th, Sc, Fe, and Co were retained completely in the chars;
As and Se showed intermediate volatility (<50% release at 1200°C);
and S, Pb, Hg, and Cl were found to be highly volatile (-50#
release at 800°C).
An empirical mathematical model has been developed to cor-
relate the equilibrium release of Hg, Pb, Cl, As, and Se, as a
function of temperature, for the five coals studied. The same
model was found to correlate sulfur release data for coals with
-------
rank up to bituminous.
Data on devolatilization kinetics were obtained in a lami-
nar flow reactor for two lignites and a subbituminous coal,
under rapid heating conditions (3X10 -10 °C/sec), over a low
to intermediate range of temperatures (300 to 900°C) and rapid
quenching (^10 °C/sec) conditions, at residence times between
150 and 1500 msec. Weight losses of the three coals increased
significantly with time and temperature and approached different
final asymptotic values at different temperatures. Hg, Pb, S,
As, Se, and La were found to be evolved in significant quanti-
ties in these experiments. The rate at which sulfur is released
from coal was found to be directly proportional to the rate of
dry ash-free weight loss under all pyrolysis conditions (includ-
ing transient and equilibrium batch pyrolysis).
A minor and trace element balance was carried out around
a steam oxygen coal gasification pilot plant.
m
-------
TABLE OF CONTENTS
ABSTRACT ii
LIST OF FIGURES vii
LIST OF TABLES x
ACKNOWLEDGMENTS . xii
1. SUMMARY 1
2. INTRODUCTION 5
2.1 Background 5
2.2 Literature Review ... 7
2.2.1 Coal Petrography and Chemistry 7
2.2.2 Coal Pyrolysis 12
2.2.3 Experimental Methods and Results 15
2.2.4 Pyrolysis Models 21
2.2.5 Elemental Release During Gasification ... 29
2.2.6 Elemental Release During Pyrolysis 37
2.2.7 Chemistry of Elemental Release
During Pyrolysis 49
2.2.8 Conclusions from Literature Review 53
3. DEVOLATILIZATION APPARATUS AND PROCEDURE 56
3.1 Selection of Apparatus 56
3.2 Description of Batch Reactor System ........ 63
3.3 Experimental Procedure for Batch
Reactor Experiments 66
3.4 Description of Laminar Flow Reactor System 67
3.4.1 Gas Supply and Utilities Subsystem 70
3.4.2 Feeder Subsystem 71
3.4.3 Furnace Reactor and Gas Heaters 72
3.4.4 Char Collection Apparatus 74
3.4.5 Exhaust and Suction Subsystem ....... 75
3.5 Experimental Procedure for Laminar Flow
Reactor Experiments ... 76
3.6 Coals and Sample Preparation 77
4. DESIGN CALCULATIONS AND DATA REDUCTION EQUATIONS 83
4.1 Design Calculations for Batch Reactor System .... 83
iv
-------
4.2 Design Calculations for Laminar Flow Reactor .... 86
4.2.1 Particle Velocities and
Residence Times 8?
4.2.2 Heating and Cooling of the
Coal Particles 96
4.3 Coal Composition and Weight Loss Variables log
4.4 Determination of Weight Loss 113
5. ANALYSIS OF RESULTS FROM BATCH EXPERIMENTS 122
5;1 Preliminary Experiments 122
5.2 Equilibrium Batch Experiments 12g
5.2.1 Analysis of Weight Loss Results 12g
5.2.2 Analysis of Elemental Release Results .... 135
6. ANALYSIS OF RESULTS FROM LAMINAR FLOW REACTOR
EXPERIMENTS 164
6.1 Weight Loss Estimation Errors 168
6.1.1 Precision and Accuracy of Weight
Loss Estimations 17g
6.1.2 Particle Classification Errors 182
6.1.3 Chemical and Physical Ash Losses 19Q
6.1.4 Mechanical Ash Losses 191
6.1.5 Tar Condensation . 192
6.1.6 Effects of Coal and Gas Feed Rates
and Residence Time Effect 1Q3
6.2 Calculational Procedure for Estimation
of Weight Loss 195
6.3 Modeling of Coal Pyrolysis 206
6.4 Analysis of Elemental Release Results 220
7. APPLICABILITY OF RESULTS TO A PILOT PLANT GASIFIER .... 248
7.1 Plant Description 248
7.2 Elemental Balances 250
7.3 Evaluation of Data 257
8. CONCLUSIONS 264
8.1 Trace and Minor Element Release 264
8.2 Volatile Yields and Kinetics of Devolatilization. . 266
8.3 Experimental Methodology 267
9. RECOMMENDATIONS 269
LITERATURE CITATIONS 272
-------
APPENDIX: CHEMICAL ANALYSES 282
C.I Proximate Analysis 282
C.2 Ultimate Analysis 283
C.3 Trace/Minor Element Analyses 284
C.3.1 Neutron Activation Analysis 284
C.3.2 Atomic Absorption Analysis 285
C.3.3 Assessment of Trace Analyses 290
vi
-------
LIST OF FIGURES
2.1
2.2
2.3
3.1
3.2
4.1
5.1
5.2
5.3
5.4
5.5
5.6
5.7
5.8
5.9
5.10
5.11
5.12
5.13
Major Elements Retained In Montana Lignite
Major Elements Retained In Pittsburg Seam
No. 8 Chars From Kobayashi (1976)
Comparison of Char Compositions from
Pyrolysis and Hydro-pyrolysis of Litnite
From Suuberg et al. (1978) ........
Temperature-Time Histories of Batch Samples .
Weight Loss in Transient Batch Experiments
Sulfur Loss in Transient Batch Experiments
Correlation Between Sulfur and A.R. Weight
Loss in Transient Batch Experiments ....
A.R. Weight Loss in Equilibrium Batch
Weight Loss of Partially Dried Montana
Lignite From Suuberg ejt al. (1978) ....
Comparison of D.A.F. Weight Loss Data for
Scandium Mass Fraction in Batch Chars . . * .
Mercury Mass Fraction in Batch Chars ....
Chlorine Mass Fraction in Batch Chars ....
40
41
43
64
68
84
124
126
128
132
133
135
142
143
144
145
146
147
153
vii
-------
5.14 Sulfur Retention in Batch Chars 154
5.15 Lead Retention in Batch Chars 155
5.16 Mercury Retention in Batch Chars ...... 156
5.17 Comparison of Experimental Data With
Sulfur Release Model .... 158
6.1 D.A.F. Weight Loss by Ash Tracer Method -
MRS Coal 169
6.2 Mass Fractions in LFR Chars, MRS 800°c
Runs 172
6.3 Mass Fractions in LFR 800°C MRS Chars -
Volatile Elements 175
6.4 Mass Fractions in LFR 800° MRS Chars -
Nonvolatile Elements 177
6.5 Rosin Rammier Plot of Montana Rosebud
Coal 184
6.6 Ash Content of MRS Size Fractions 185
6.7 D.A.F. Weight Loss of MRS Coal 201
6.8 D.A.F. Weight Loss of NB8 Coal 202
6.9 D.A.F. Weight Loss of BZN Coal 203
6.10 D.A.F. Weight Loss of Texas Lignite
From Nsakala (1976) 205
6.11 Comparison of Isothermal Models'
Predictions 208
6.12 Comparison of lion-1 so thermal models'
Predictions 211
6.13 Comparison of LFR Asymptotic D-.A.F.
Weight Losses with Batch D.A.F.
Weight Losses ............... 213
6.14 Comparison of Wire-Screen and LFR Quasi-
Equilibrium D.A.F. Weight Loss Data . ... 216
V111
-------
6.15 Iron Mass Fraction in Chars 232
6.16 ASTM VM, C and H Retentions in MRS Chars . . 234
6.17 Lanthanum Retention in LFR Chars 235
6.18 Arsenic Retention in LFR Chars 236
6.19 Selenium Retention in LFR Chars 237
6.20 sulfur Retention in LFR Chars 238
6.21 Lead Retention in LFR Chars 239
6.22 Mercury Retention in LFR Chars 240
7.1 Gasifier and Particulate, Condensables and
Solubles (PCS) Removal System Showing
Process Variable Monitoring Points .... 249
7.2 GO-15 Run Summary . . . •• 253
-------
LIST OF TABLES
2.1 Weight Loss During Pyrolysis (From Kuhn
et a_l., 1977) 45
2.2 Preliminary XES Data For Pyrolysis of Six
Coals (From Kuhn et ajL., 1977) 45
2.3 Results of Analysis River King (From
Kuhn et al., 1979) 47
2.4 Results of Analysis Crown No. 2 (From
Kuhn et al., 1979) 48
3.1 Coal Characterization Data .......... 79
3.2 Elemental concentrations and Organic Affinity
of Elements in The Rosebud Coal From
Montana (From Fiene et al., 1978) ..... 80
5.1 Moisture and ash content of batch coals . . . 123
5.2 Equilibrium Batch Weight Loss Experiments . . 130
5.3 Significant Element - Temperature
Correlations for batch Pyrolysis 138
5.4 Equilibrium Elemental Release Model
Parameters 152
5.5 Significant Element - Element Correlations -
Batch Experiments 161
5.6 Significant Element - Element Correlations -
Batch Experiments 162
5.7 Significant Element - Element Correlations -
Batch Experiments 163
6.1 Summary of LFR Run Conditions ........ 165
6.2 Room Temperature LFR Runs with MRS Coal ... 186
6.3 Comparison of CY-1 to CY-2 A.R. Char
Compositions ........... 189
-------
M * ~*
6.5
6.6
6.7
6.8
6.9
6.10
7.1
7.2
7.3
7.4
7.5
C.I
C.2
Significant Correlations in MRS -
Significant Correlations in NBb -
Significant Correlations in BZN -
Significant Element - Time Correlations
at 800°C - MRS Runs 25, 27, and 29
Significant D. A. F. Weight Loss-
Significant Element - Element Correlations
A
Percent Error (Instrumental) NAA in Coals . .
Summary of Atomic Absorption Analysis
223
225
226
227
244
247
251
255
256
258
263
412
415
-------
ACKNOWLEDGMENTS
This report is a modified version of the Ph.D. dissertation
of Dr. V. H. Agreda. The research was performed at North Caro-
lina State University, under the direction of Professor R. M.
Felder, and the dissertation was published in 1979.
The dissertation author (VHA) expresses his appreciation
to all persons who have contributed advice and assistance in the
research, notably the members of his advisory committee, Profes-
sors Felder, J. K. Ferrell, R. W. Rousseau, and W. L. Switzer.
Special gratitude is expressed to Ms. K. Steinsberger and Mr.
L. Hamel, who assisted in the chemical analyses. Finally, the
author expresses his thanks to his wife, Carl a, for her assis-
tance in the preparation of the thesis, and to his son, Vic,
for his patience and sacrifice during the course of this study.
xii
-------
1. SUMMARY
The evolution of volatile matter and minor and trace
element constituents was studied for five coals under
inert gas pyrolysis conditions in a batch-crucible re-
actor and a laminar flow reactor. The specific objec-
tives of this project were as follows:
a. To design and build an experimental apparatus to study
the devolatilization of coals at thermal equilibrium,
i.e., at long residence times at a given temperature.
b. To design and build an experimental apparatus to study
the kinetics of devolatilization during coal pyrolysis
with high heating rates and small residence times.
c. To extend to low temperatures the study by Kobayashi
(1976) of the problems involved in the use of ash as
a tracer for the estimation of weight loss in laminar
flow reactors.
d. To determine the rates at which total volatile matter
is released (i.e., d.a.f. weight is lost) as a function
of temperature and residence time during the pyrolysis
of three coals.
e. To compare measured devolatilization rates with exist-
ing models of pyrolysis kinetics, and to modify the
models or propose new models, as the results may
dictate.
-------
f. To determine which trace and minor elemental con-
stituents of a given coal are evolved to a measurable
extent.
g. To determine the equilibrium extent to which volatile
trace and minor elements are evolved as a function of
temperature.
h. To attempt the development of a mathematical model to
describe the extent of elemental release at thermal
equilibrium.
i. To measure and describe in a qualitative manner, the
extent and rate at which minor and trace elements are
evolved as a function of temperature and residence
time during pyrolysis.
j. To carry out a trace element balance around a steam-
oxygen coal gasification pilot plant, and to use the
results of the pyrolysis studies in the analysis of
the elemental balance.
A batch tube furnace and a laminar flow furnace were
used for the experiments. Nitrogen was used as the bulk
reactor gas for all the experimental runs with both reactors,
Particles of only one average particle size, 41.5 microns,
were pyrolyzed in the laminar flow reactor. Residence
times in the batch reactor ranged up to 20 minutes, and
reaction times in the laminar flow reactor ranged from
170 to 1500 msec. Temperatures ranged from 300 to 1200°C in
-------
the batch reactor and from 300 to 900°C in the laminar
flow reactor. The following results and conclusions were
reached:
a. Volatile material yields and devolatilization rates agreed
reasonably well with those found in other studies.
Kobayashi's two first-order parallel reactions model
has been found to predict fast pyrolysis dry ash-free
weight loss reasonably well in the 300 to 900°c tem-
perature range.
b. It was determined that at any pyrolysis temperature
ash losses may have a significant effect on calculated
extents of devolatilization, making the commonly used
ash tracer technique a potential source of error in
all experimental pyrolysis studies. A technique to
correct estimated weight losses for this effect has
been developed.
c. Equilibrium yields of volatile matter and volatile
trace elements generally increase with temperature.
Sm, Cr, Th, Sc, Fe, and Co are retained completely in
the chars up to 1200°C. As and Se exhibit intermediate
volatility (<50% release at 1200°C), and B, Pb, Hg,
and Cl are highly volatile (>50% release at 800°C).
d. The rate at which sulfur is released from coal is di-
rectly proportional to the rate of dry ash-free weight
loss under all pyrolysis conditions studied.
-------
e. Kg, Pb, S, As , Se, and La are evolved in significant
quantities during fast pyrolysis.
•f. There appear to be three modes of trace-minor element
•release from coal during pyrolysis in an inert gas:
some elements (e.g., sulfur) are released together
with the volatile matter; others appear to be released
with the ash (e.g., Sm); and a third group of elements
appear to be released at a much faster rate than
either volatile matter or ash (e.g., Hg).
g. Several pyrolysis models, including a first-order
model featuring temperature dependent asymptotic weight
loss, were tested and found to provide reasonable corre-
lations of the experimental data.
h. An empirical mathematical model has been developed to
correlate the release of Hg, Pb, €1* As, and Se, as a
function of temperature, for coals ranging in rank
from lignitic to anthracitic. The same model was
found to correlate sulfur release data for coals with
rank up to bituminous.
i. A minor and trace element balance was carried out
around a steam-oxygen coal gasification pilot plant.
The pyrolysis studies proved useful in analyzing the
results of the mass balance, and they may be more use-
ful when the plant is operated with New Mexico No. 8
coal instead of Western Kentucky No. 11 chemical grade
coke.
-------
2. INTRODUCTION
2.1 Background
The main approaches to the conversion of coal to
gaseous fuels are steam-oxidant coal gasification, hydro-
carbonization of coals, and coal pyrolysis. In all three
processes, coal devolatilization or pyrolysis plays an im-
portant role. When coal is heated in any atmosphere it
begins to release volatile products as the temperature and/
or reaction time increase. This release can be very rapid
and violent when the coal is heated rapidly to high tem-
peratures . This is the case during flash pyrolysis in en-
trained bed gasifiers. When coal is subjected to very high
heating rates, most of the volatile matter must be released
before reactant gases, if there are any, begin to enter the
pores of the coal particle and start to react.
In addition to the organic gases and vapors that con-
stitute the principal products of pyrolysis, measurable
quantities of minor and trace elements present in coal are
evolved. Some of these species, such as sulfur, arsenic,
and mercury, pose health hazards or are catalyst poisons
which render the evolved gas unsuitable for subsequent
catalytic combustion or synthetic fuel production. It is
difficult to measure the extent to which such pollutants
are emitted as pyrolysis effluents since they are present
at low concentrations, and it is even more difficult to
-------
determine the rates at which they are evolved, since the
pyrolysis process is complete in times typically on the
order of tens to hundreds of milliseconds, depending on
the temperatures. Nevertheless, an understanding of the
kinetics and thermodynamics of trace element evolution
is an essential step in the development of the technology
needed to control emissions of these species from coal
conversion plants.
Many trace element studies have been carried out on
entire plants and on particular reactors. Studies have
been carried out on the occurrence and distribution of
trace elements in different coals (Gluskoter et al., 1977),
trace element measurements at coal-fired steam plants
(Lyon, 1977) , and trace and minor element balances around
coal gasification plants (Forney e_t al., 1975; Gasior et
al., 1978). The data obtained in these studies provide a
good qualitative picture of the behavior and fate of trace
and minor element constituents of coal during gasification;
however, they do not provide the detailed information about
reactor conditions needed to model the behavior and predict
the fate of those elements during coal gasification ope-
rations. Some of the problems arise because of the diffi-
culty of obtaining representative samples during steady
state operation of a coal gasifier. However, the most
serious problems appear to be posed by the need to obtain
representative homogeneous samples and to analyze them
-------
precisely for trace elements. No trace element balances
yet attempted appear to have been closed satisfactorily.
The purpose of this research was the quantitative
determination of the extent and rate of evolution of
selected minor and trace elements during the pyrolysis
of coal. The results have relevance to processes based
on the pyrolysis of coal and indirectly to all coal gasi-
fication processes.
2.2 Literature Review
2.2.1 Coal Petrography and Chemistry
The organic material in coal is a heterogeneous mix-
ture of organic minerals known as macerals. Fourteen mace-
ral groups have been identified (Spackman, 1975). The
three principal groups of macerals, called microlythotypes
are vitrinite, exinite, and inertinite. Exinite has the
highest hydrogen content, volatile matter content, and
heating value of the three microlythotypes, while inerti-
nite has the least of all three. Inertinite has the highest
density and the greatest degree of aromaticity, while exi-
nite is the lowest in both properties. Vitrinite usually
exhibits chemical and physical properties between those of
the other two groups. The most abundant of the three micro-
lythotypes is vitrinite. The different microlythotypes
exhibit different behavior under pyrolysis: the total yield
-------
of volatiles is usually in the order exinite > vitrinite
> inertinite. No information appears to be available on
the trace and minor element content of the three micro-
lythotypes.
Coal has a highly aromatic, cross linked micromole-
cular network structure. The aromatic rings form clusters;
the number of rings per cluster, and therefore the aroma-
ticity, increases with increasing rank of the coal (Hirsh,
1958), ultimately approaching a fully condensed graphitic
structure. The aromaticity can be as low as 40% for
subbituminous coals/ which contain significant amounts of
polycyclic aliphatic rings. The aromatic rings are thought
to be linked by hydrocarbon and O - N - S chains of widely
differing bond strengths.
Coals contain varying amounts of inorganic impurities,
most of which are present in the form of ash. Ash, the
inorganic mineral matter in coal, comprises about 5-20%
of the mass of coal. The principal minerals found in coals
include kaolite,.pyrite, illite, calcite, and quartz.
Mineral matter is distributed in coal more or less uni-
formly as small inclusions of variable composition and size.
Typically, the ash inclusions appear to be about 1 vim; how-
ever, they can be as small as 0.1 ym and as large as the
20 to 60 urn pyritic particles observed in x-ray scans of
coal (Solomon, 1977).
8
-------
Because of its organic origin and its intimate con-
mixture with crustal formationsf coal contains a large
number of elements in major, minor, and trace quantities.
The organic matter in coal consists primarily of carbon,
hydrogen, oxygen, nitrogen, and sulfur. Minor and trace
quantities of most other elements are also found in coal.
Out of 92 known non-transuranic elements, only 14 have
not yet been found in coal (Loran and O'Hara, 1977) .
Gluskoter et al. (1977) have done extensive research
on the occurrence and distribution of trace elements in
coal. Their results show that the geometric mean concen-
trations of four of the elements that they investigated
are greater by a factor of six or more than the geometric
mean concentration of those elements in the earth's crust
(Clarke values of the elements). Boron, chlorine, and
selenium are enriched in coals of the Illinois Basin; ar-
senic, chlorine, and selenium are enriched in eastern coals;
and selenium is the only element enriched in western coals.
A larger number of elements are depleted in coals; that is,
they are present at less than one-sixth of the Clarke
value. The elements depleted in coals of the Illinois
Basin are Al, Ca, Cr, F, Hf, Lu, Cu, Mg, Ha, P, Sc, Si, Sr,
Ta, and Tl. All other elements were found to be within the
range of one-sixth to six times the Clarke value.
Gluskoter el: al. (1977) also reported that many ele-
ments are positively correlated with each other in coals.
-------
The most highly correlated are Zn and Cd (correlation co-
efficient r = 0.94 for coals of the Illinois Basin).
Chalcophile elements (As, Co, Ni, Pb, and Sb) are all
mutually correlated, as are the lithophile elements (Si,
Ti, Al, and K). Other significant correlations are Ca:Mn
(r = 0.65) and Na:Cl (r = 0.48).
Van Krevelen and Schuyer (1957) have pointed out that
virtually all of the nitrogen in coal exists as part of
the organic coal substance. It is well known (Lowry, 1963)
that sulfur occurs in coal in three forms: in organic com-
bination as part of the coal substance, as pyrites or mar-
cansite, and as sulfates. The amount of organic sulfur is
normally not over 3%, but in exceptional cases it may be
as much as 11%. The sulfates, mainly of calcium and iron,
rarely exceed a few hundredths of a percent except in
highly weathered or oxidized samples. It has been reported
(Yurovskii, 1977) that in certain coals, elemental sulfur
may be present in amounts up to 0.15%.
Duck and Himus (1951) concluded that most of the ar-
senic in coal occurs in the form of arsenopyrite. Horton
and Aubrey (1950) found that phosphorus is associated with
the inorganically-combined mineral matter in some coals,
but the organic affinity of phosphorus seems to be rather
high in other coals (Gluskoter et al., 1977).
10
-------
Goldschmidt (1935) developed the concept of organic
and inorganic affinity for elements in coal. Gluskoter
e_t al. (1977) produced an "organic affinity index" in an
attempt to quantify the information presented in coal
washability curves and histograms of washability data.
Such curves and histograms are effective means of indi-
cating whether the elements are associated with the or-
ganic or inorganic fractions of the coal. Values for
the organic affinities of the elements were defined by
normalizing the washability curves, removing from them a
component that represents the contribution from the in-
separable mineral matter, and then calculating the areas
under the corrected curves. Values of this property range
from 0.08 to 2.02 for the elements determined in the coals
analyzed by Gluskoter e£ al. (1977) . The variability in
organic affinities between coals from different geographic
locations is sufficiently large that a prediction of the
value of organic affinity of an element in a sample is
necessarily imprecise; however, it is safe to say that
Ge, B, and Br generally are among the elements with the
highest organic affinities, and As is among the elements
with the lowest organic affinities. The organic affinities
of 53 elements in several coals have been determined by
Gluskoter et al. (1977).
11
-------
A total separation of the mineral matter from the or-
ganic matter in coal cannot be made by gravimetric methods
alone. Kuhn et. al. (1977) have removed the mineral matter
from cleaned coal by means of selective chemical disso-
lution in which the organic fraction of the coal was rela-
tively unaltered. Their results show that Ge, Be, Sb, and
Br have high organic association in coal; Ni, Cu, Cr, and
Hg tend to be present in both organic and inorganic com-
bination; and Zn, Cd, As, and Fe are primarily associated
with coal mineral matter. It is apparent that correlation
with organic sulfur is not an indicator of the organic
association of other elements. Data in the same reference
also imply that most of the organically bound elements
are weakly bound; no more than a few parts per million can
be considered an inherent part of the organic molecules.
2.2.2 Coal Pyrolysis
The pyrolysis of coal, variously termed thermal de-
composition, carbonization, and devolatilization, is a
chain of decomposition reactions wherein the linkages
between aromatic clusters are broken and volatile decom-
position products escape. Coals exhibit more or less de-
finite decomposition temperatures, as indicated by melting
and rapid evolution of volatile products, over a wide
range in rank.
12
-------
There are apparently five principal phases of devolati-
zation (Suuberg el: al., 1978). The first occurs at very
low temperatures, about 100°C, and is associated with
moisture evolution. The second phase occurs between 350°C
and 450°C and is associated with the evolution of a large
amount of carbon dioxide and a small amount of tar. The
third phase involves evolution of chemically formed water
in the range 500-700°C. The only other significant product
evolved in this phase is carbon dioxide. The fourth phase
involves a final rapid evolution of carbon-containing
species at temperatures from 700-900°C. Carbon oxides, tar,
hydrogen, and hydrocarbon gases are rapidly evolved in this
phase while little water is produced. The fifth phase is
the high-temperature formation of carbon oxides.
Pyrolysis is extremely rapid. Equilibrium is reached
in tens to hundreds of milliseconds, depending on the tem-
perature, for most pyrolytic reactions. However, the
nature of the equilibrium is complex; the process involves
many parallel and series reactions with rates that vary by
orders of magnitude. The study of those complex reactions
is hampered by the fact that coal is not a homogeneous
material; different portions (both microscopic and macro-
scopic) of a single coal sample exhibit widely differing
chemical compositions and physical properties. In addition,
samples from different portions of a mine are/not identical.
13
-------
Furthermore, there is a broad range of coal types, each of
which decomposes in a slightly different manner. However,
some useful (though sometimes contradictory) generali-
zations have been made.
The physical nature of coal devolatilization depends
to a great extent on whether the coal is plastic or non-
plastic. Plastic coals are also referred to as caking
coals, since in the plastic state they are viscous liquid
masses capable of coalescence and, upon resolidification,
formation of a cake. The thermoplasticity of caking coals
is manifested by softening, deformation, and resolidifi-
cation upon heating. Plastic coals often devolatilize with
the formation and eruption of bubbles, leaving a highly
porous char containing entrapped bubbles. In extreme
cases, the particles may swell to many times their original
size, forming hollow char particles called cenospheres.
The temperature limits on the region of plasticity depend
on the heating rate: at low heating rates, the plastic
region is typically 420 to 500°C with some variation among
different coals, and at high heating rates, the plastic
region extends to 2000°C or higher. The growth and escape
of gas filled bubbles constitutes an important mode of
volatiles transport in plastic coals.
14
-------
2.2.3 Experimental Methods and Results
Variables known to be important to the pyrolysis pro-
cess include coal type, composition, and source, particle
size distribution, heating rate, final temperature, du-
ration of heating, type and duration of the quenching pro-
cess, and composition and pressure of the ambient gas.
Virtually all experimenters have relied upon the
collection and analysis of quenched samples of the gas
and/or char from a pyrolysis experiment. Measurements rele-
vant to pulverized-coal pyrolysis have been obtained from
four types of experiments.
In the first type of experiment, the coal dust is
placed inside a crucible in a furnace and heated, and an
inert gas flows past the crucible, sweeping the devola-
tilization products. Low heating rates and high residence
times are obtained in this type of experiment. The largest
uncertainty in the experiment is that the crucible must be
heated first, thus giving rise to an unknown lag time and
temperature difference.
In the second experiment type, the coal dust is em-
bedded in the pores of a wire screen that is heated elec-
trically. Heating rate, final temperature, and ambient
atmosphere can be controlled and are not dependent on the
pyrolysis process in this case. This approach is subject
to the same problems as the first type.
15
-------
The third type of experiment involves injecting the
coal particles into a preheated gas. Thus, the ambient
atmosphere, final temperature, and coal dust concen-
tration are controlled. However, the heating rate of the
particles is not known precisely because it is dependent
on the ambient conditions experienced by the particle and
the mixing of the carrier and main stream gas flows.
In the fourth type of experiment, coal dust is burned
in a flame. The principal disadvantage of this experiment
is that the only independent variables are those of the
feed stream. Heating rate, final temperature, and ambient
atmosphere are then all determined by the resultant flame;
the reaction time is not known with any certainty because
the reactions cannot be quenched simply by cooling.
Crucible experiments are usually carried out for the
slow heating of coal (10-600°C/sec). Typically, this pro-
cess is characterized by long residence times (minutes to
hours) of the solids in the reactor zone. Experimental
and theoretical studies of slow coal pyrolysis have focused
on the plastic behavior of coals, optimization of coke
yields, evolution of volatiles, mechanisms of primary de-
composition, and, more recently, trace element studies.
Van Krevelen et al. (1961) observed two different
stages of devolatilization undergone by coals being heated
at 2°C/min: primary reactions which took place between 400
16
-------
to 500°c, producing primarily tar, and secondary reactions
(above 500°C) producing gases rich in hydrogen. Van
Krevelen concluded that the two stages are governed by the
amounts of aliphatic and aromatic hydrogen in coal. The
primary devolatilization is a depolymerization process in
which aliphatic bridges are ruptured with simultaneous
transmission of hydrogen (disproportionation). The struc-
tural units to which this hydrogen is transmitted evaporate
as tar, or recondense and yield semi-coke. The formation
of tar terminates completely when the original aliphatic
hydrogen atoms in the reaction mixture have been used up.
Reactive oxygen groups such as OH groups, which are richer
in low rank coals, decrease tar yield by consuming available
hydrogen through dehydration, thus promoting condensation
of aromatic nuclei. Low tar yields of high rank coals are
explained in terms of the structural units being too large
to evaporate.
Nsakala (1976) found that weight loss during batch
pyrolysis decreased with increasing coal weight (i.e.,
increasing bed depth) for the pyrolysis of an HVA Ohio #5
coal. Kobayashi (1976) found the same effect (although to
a fairly small extent) for an HVA Pittsburgh seam bituminous
coal; however, he also found no bed depth effect for a
Montana lignite. Therefore, it appears that the effect of
bed depth decreases with decreasing coal rank.
17
-------
The mechanism of rapid devolatilization in a dispersed
phase was first investigated by Chukhanov (1952), and
Shapatina et al. (1960) . Augmentation of volatile yield
under rapid heating conditions (greater than 10s°C/sec)
has been observed by different researchers using various
experimental techniques. These include entrained flow
reactors (Nsakala, 1976; Kobayashi, 1976; Coates et al.,
1974; Stickler et al., 1974; Badzioch and Ilawksley, 1970;
Kimber and Gray, 1967; Eddinger et al., 1966) and electrical
screen heating (Suuberg et al, ,1978; Menster e_t al., 1974;
Anthony e_t al., 1974; Loison and Chaubin, 1964).
At high heating rates (1,000 - 50,000°C/sec), such as
those typically attained in continuous fluidized bed and
entrained bed gasifiers, the yield of volatiles at a given
temperature and the tar-to-gas ratio of the product are lar-
ger than at low to moderate heating rates (l-200°C/sec).
This effect decreases with decreasing coal rank (Badzioch
and Hawksley, 1970), becoming relatively small in the case
a lignite (Kobayashi, 1976) . Furthermore, Kobayashi (1976)
found that, for a bituminous coal and a lignite, little in-
creases in volatile yields (above the ASTM volatile matter
yield) could be expected by increasing the heating rate to
peak temperatures below 1000°K.
Many different explanations have been given for the
higher volatile yields from higher rank coals. However
18
-------
since only lower rank coals are used during the fast pyroly-
sis experiments in this study, the reader is referred to
Kobayashi (1976) for a thorough review of the different
proposed explanations, including his own which was based
on his experimental results with crucible, free fall, and
laminar flow reactors. In all cases, it was assumed that
the ultimate volatile matter yield would be that measured
during fast pyrolysis plus the ASTM volatile matter found
in the chars. This implies a very fast set of reactions
during fast pyrolysis and a much slower second set during
the ASTM test. Typically, the residual volatiles in the
char, as determined by proximate analysis, decrease ex-
ponentially with residence time in the reactor. The rate
of pyrolysis is less at lower temperatures and heating
rates, while the amount of residual volatiles decreases
with increasing reactor temperature. As indicated above,
both the rates of pyrolysis and the amount of residual
volatiles also depend on coal type.
Anthony's (1974) results with coal particles sus-
pended in a wiregrid show no discernable effect of heating
rate on the volatile yield for a lignite, and only a 2%
increase in volatile yield for bituminous coal, when the
heating rate was increased from 600 to 10,000°C/sec for
a final temperature of 1000°C. However, Kobayashi (1976)
showed that the heating rates covered by Anthony were
19
-------
below the critical rates for the coals used, and so were
in the range where little heating rate effect would be
expected. An excellent discussion of the estimation and
applicability of critical heating rates is given in
Kobayashi (1976). For the purpose of this discussion,
suffice it to say that if the characteristic heating time
is much shorter than the characteristic reaction time
during the heating period, only a small amount of reaction
occurs during this period and the rest proceeds isother-
mally at the final temperature. Changes in heating rates
in this range, therefore, should not make large differences
in the devolatilization behavior.
It is generally agreed that rapid heating influences
not only the amount of volatiles generated but also the
product composition. The product distribution is a strong
function of both the final reaction temperature and the
heating rate. Depending on the reaction conditions, the
volatiles may emerge as tars, repolymerize and deposit on
the char, or crack to form low molecular weight hydrocar-
bons.
Anthony (1974) reported that a lower ambient pressure
favors the liberation of a greater mass of volatiles for a
Pittsburgh seam bituminous coal, lie observed no effect of
pressure for Montana lignite.
The effect of particle size on volatile yield and pro-
duct distribution is unclear, as many contradictory results
20
-------
have been obtained by different researchers. Badzioch
and Hawksley (1970) and several other researchers found
no particle size effect, while Anthony e_t al. (1975) and
others did find an effect, albeit a small one up to par-
ticle sizes of 1000 ym (Anthony et al^, 1976). Ksakala
(1976) found a very strong size effect between coal size
fractions with average diameters of 64 ym, 86 ym, and
179 ym. However, the excellent theoretical analysis of
that same data by Reidelbach and Algermissen (1978) shows
that the results are due simply to the difference in
heating rates between small and large particles.
2.2.4 Pyrolysis Models
Thfe simplest and most commonly used model for corre-
lating the kinetics of devolatilization entails treating
coal pyrolysis as an equilibrium-limited first-order re-
action occurring uniformly throughout the particle. The
rate law is usually expressed as
8 = Be-E/RT (V.-V) (2-1)
where
V = volatiles lost from the particle up to time tf
expressed as a fraction or percentage of the
original coal weight.
21
-------
T = temperature
Vo,, = volatiles lost from particle up to t = °°
(ultimate yield), approximated by extrapo-
lation of measurements at long reaction
times, expressed as a fraction or percentage
of the original coal weight.
t = time
B = frequency factor
E = activation energy
The determination of the unknown parameters B, E, and
V,,,, has been the focus of most kinetic studies. Discrepan-
cies of as much as several orders of magnitude in rates
are evident from the parameters obtained by different in-
vestigators, even when the same coal was studied (Kobayashi,
1976). Some of the discrepancies may be attributed to the
differences in the structure of coal and physical factors;
however, the extent of the discrepancies appears to be too
large to be explained solely by these factors. It appears
evident that the parameters obtained by different investi-
gators are strongly dependent on the experimental apparatus
employed and the manner in which the data were analyzed.
Kobayashi (1976) showed that differences between one
bituminous coal and another can cause one to two orders of
magnitude difference in the rates of devolatilization, and
rate differences as large as four orders of magnitude may
be observed if the coals differ widely in rank. Kobayashi
22
-------
also found that rates measured in a laminar flow experiment
(reaction time 0-200 msec) were about one to two orders of
magnitude larger than those in free fall experiments
(reaction times of 1 sec to 10 min). He concluded from
these observations that differences in the data reduction
can cause as much as two orders of magnitude difference in
the rates.
Many kinetic models have been tried, besides the simple
single first-order scheme, ranging from series reaction
schemes to complex series-parallel competing mechanisms,
transport process-controlled reaction schemes, and empiri-
cally developed models. The most successful models de-
veloped to date, are the empirical model of Badzioch and
Kawksley (1970), the two first-order competing reactions
model of Kobayashi (1972), the infinite parallel first-
order reactions model of Anthony et al. (1976), and the
ten reactions model of Reidelbach and Summerfield (1975
Badzioch and Hawksley (1970) used the following equations
to correlate their data:
AW*=Q VM* (1-C) (1-exp {-A|exp(-B/T)|tj.}) (2-2)
C = exp [-Kx (T-K2)] (2-3)
where
VMQ - proximate volatile matter of coal
tj = isothermal reaction time
Q, A, B, Klr K2 •* empirically determined constants.
23
-------
This model was also used by Nsakala (1976) . In both cases,
the model was used to correlate data from laminar flow
reactors assuming isothermal reaction conditions. Ex-
cellent fit of high dry-ash-free (d.a.f.) weight loss data
between 750 and 1000°C was obtained. However, Horton (1979)
indicated that this model lacks the flexibility required to
describe much of the experimental data available, and may
even be inadequate to describe nonisotherroal pyrolysis.
Kobayashi (1972) proposed — and later used successfully
(1976) — a model consisting of the following pair of
parallel, first-order, irreversible reactions,
(2-4)
R2
where
al» °2 = mass stoichiometric coefficients
vl» V2 ~ volatile yields
^1 ' K-2 ~ cnar yields
Kj_ and K2 denote Arrhenius rate constants.
-Ei/RT .„ _.
Kl = Ble (2~ 5)
-E2/RT
K2 = B2e (2-6)
By assumption, E^ < E2.
24
-------
At relatively low temperatures, the first reaction is
assumed to be dominant, leading to an asymptotic volatile
yield of a-,. At high temperatures the second reaction
becomes faster than the first one, resulting in larger
volatile yields. The rate equations are:
dC*
= (K + K)c (2-7)
and
dv*2)
where ^
C = dry-ash-free mass of coal
V* = dry-ash-free mass of volatiles
*
C = original d.a.f. mass of coal
Integration of equation (2-8) results in the following
expression for the overall fraction d.a.f. weight loss,
AW*:
f
t
-/ (K1+K2)dt
02K2)e dt (2-9)
where C « original d.a.f. mass of coal.
Under isothermal conditions, equation (2-9) can be
integrated to give.
25
-------
+ K2)t
ct2K2 (1"e } (2-10)
The asymptotic value of AW under isothermal conditions is
given by
* _ alKl + a2K2
AW~ ' K! + K2 . (2.11}
This equation shows that the asymptotic weight loss is a
weighted average of a^ and a2, the asymptotic values of
weight loss at low and high temperatures, respectively.
Hence, a value a2=l may be assigned as complete weight loss
is expected at extremely high temperatures, while a^ may be
approximated by the ASTM volatile matter or by a characte-
ristic volatile yield at temperatures low enough for the
second reaction to be considered negligible.
Kobayashi (1976) integrated equation (2-9) numerically
using a simple empirical model to calculate the temperature
rise of the coal particles. He obtained excellent corre-
lations of data for a bituminous coal and a lignite in the
temperature range between 1000°K and 2100°K, using the same
parameters for both coals. The model is conceptually sound
in that the variation in volatiles yield with temperature i
explained by a second reaction rather than by a correlating
parameter. One of the advantages of the model is that the
competing reactions reduce to a single reaction when the
second reaction is much slower than the first one, so that
26
-------
kinetic parameters obtained at relatively low temperatures
could be utilized for the first reaction.
Perhaps the best approach developed to date for the
modeling of the pyrolysis reactions of coal is based on
the assumption of a large number of parallel decomposition
reactions. The appearance of product i is modeled as a
reaction first-order in the amount of i yet to be produced;
C - » Vi + R (2-12)
dV.j
(2-13)
. (2-14)
Anthony (1974) noted that the single reaction model is
applicable only to a single set of experimental conditions
and therefore proposed this model with a Gaussian distri-
bution of activation energies. The model is then ex-
pressed as;
* * t t
v~ ~ v
i f f
1 {Jexp[(-jKdt)f (E)dE]} (2-15)
o o
with f(E) = [o(2Tr)JsJ"1 exp[-(E-E0)2/2a2] (2-16)
K . Be~E/RT (2-17)
27
-------
where EO = mean activation energy
a = standard deviation of the activation
energy distribution.
For simplicity, Anthony (1974) assumed that the Ki's dif-
fered only in activation energy; therefore, a single pre-
exponential factor can be used.
This approach provided an excellent correlation of the
data from Anthony et. al_. (1975); however, Kobayashi (1976)
argued that estimation based on Anthony's results suggested
that the single reaction model may provide as good a corre-
lation as the multiple reaction model in the experimental
range in which the latter model was applied. Moreover,
Horton (1979) indicated that probably this model cannot
correlate certain types of fast pyrolysis data.
nevertheless , the statistical model provides an ex-
planation of a number of observed phenomena. Observed
values of E in the range of 10-20 kcal/mole have been erro-
neously attributed to diffusion control of the observed
rate. It can be shown that a distribution of activation
energies in the range of 30-70 kcal/mole leads to an
apparent single step activation energy of 10-20 kcal/mole.
Suuber et al. (1978) determined the kinetics of many
pyrolysis reactions and then used the model pro-
posed by Anthony to fit the data, allowing the preexpo-
nential factor, B, to assume a different value for each
28
-------
reaction. The distribution of activation energies obtained
by Suuberg ej: al. (1978) from product composition was simi-
lar to that obtained by Anthony et. al. (1976), whose re-
sults were based on weight loss. Both studies were made
on the same lignitic coal.
Reidelbach and Summerfield (1975) proposed a complex
reaction mechanism whereby a set of ten reactions was used
to describe the process of pyrolysis. The model was later
refined by Reidelbach and Algermissen (1976). As yet, only
limited comparison has been made between this model and
experimental data, but the results have been favorable.
This is to be expected since the model also contains the
reaction steps of Kobayashi's (1975) two parallel reactions
model, which by itself quite successfully correlates much
experimental data. Because of its complexity, only the
first five reactions have usually been used (Reidelbach
and Algermissen, 1978).
2.2.5 Elemental Release During Gasification
Many elemental mass balances have been made around pro-
duction and pilot coal gasification plants. There appears
to be no difficulty in closing major element balances:
recoveries better than 95% are common for C, H, N, S, and
O. Trace and minor element balances appear to be quite
difficult to close, however, recoveries range from 4%
(usually for Hg) to 1000%.
29
-------
Trace element balances around the Synthane Process
Development Unit (PDU) gasifier (Forney et al. 1975) point
out the need for more precise sampling and analytical
methods. Sixty-five elements were analyzed by spark source
mass spectrometry in that study. The results show that the
trace elements remained primarily in the chars and dusts
emanating from the gasifier. Some elements, such as boron,
chlorine, fluorine, and selenium were found in the water;
others such as arsenic, lead, and cadmium, were in the tars.
Most of the mercury appeared in the tar and water with
little remaining in the char or dust.
A study on trace element disposition for the Sasol
(South Africa) facility (Bennet, 1976) followed the
partitioning between solid residues, liquid streams and
gases. Lead, arsenic, and beryllium were found mainly in
the ash, selenium and tellurium primarily in the liquid
streams, fluorine two-thirds in the ash and one-third in
the liquids. Mercury was found in all streams but concen-
trated mainly in the gas. The usual problems with data
reliability were encountered; for example, 50% of the mer-
cury could not be accounted for.
A coal hydrogasification study by Attari and Mensinger
(1976) at the Institute of Gas Technology followed the con-
centrations of several trace elements in the feed and resi-
due samples from the gasification of a Montana lignite
30
-------
and Illinois No. 6 coal. They found that the concentrations
of several volatile trace elements (based on amounts fed)
decreased appreciably in the residues of the different units
of the process. It was also determined that under the same
gasifier conditions, one coal suffered significant losses of
one set of trace elements, but not of another set, while the
other coal exhibited the reverse behavior. It was postu-
lated that this behavior could be due to the chemical forms
in which those elements occur in the coals studied having
different tendencies to form hydrides in the reducing atmos-
phere of the hydrogasifier. Alternatively, it could simply
be a case of different volatility of the compounds in the
two coals. It was also pointed out that such behavior could
be due to one of the coals having been pretreated, and there-
fore subjected to mild oxidation, prior to hydrogasification.
One of the better elemental balances available in the
literature is that of Gasior et al. (1978). Major, minor,
and trace element balances were made for the Synthane PDU
gasifier operating witli Illinois No. 6 coal. A several-fold
improvement in recovery and balance of a selected group of
trace elements was achieved relative to the work of Forney et
al. (1975). The primary reasons for the improvement appeared
to be that truly representative samples of the various solid,
gas, and liquid streams were obtained, followed by
31
-------
meticulous care in preparing the samples to prevent contami-
nation. Recoveries for minor elements appeared to be good
except for chlorine (12% recovery). Most of the minor and
trace elements were recovered in the char. However, in
spite of all the sampling, sample preparation, and analysis
improvements, the trace element balances were not closed
satisfactorily. The worst case was, as expected, mercury
for which only a 12.5% to 20% recovery was obtained. It
was assumed that because of its volatility mercury leaves
the process with the gas, the only stream that was not
analyzed. Gasior et al. (1978) concluded that the trace
element mass balances made were satisfactory within the
degree of precision of the analytical methods used.
Trace element measurements at coal-fired power plants
(Kaakinen, 197b; Klein, 1975) have shown that
in general, the elemental constituents of coal can be di-
vided into two groups: low volatility elements, which appear
mainly in the bottom ash; and higher volatility elements,
which appear in or with the fly ash. It has been shown
(Lyon, 1977) that the most volatile elements, such as mer-
cury and selenium, can actually escape in their elemental
forms with the flue gas.
To summarize, environmental assessment studies on
toxic elements have emphasized elemental material balances
around the gasifier and quench system. For many elements.
32
-------
closing of the material balance is difficult since a signi-
ficant fraction of the material may be part of the quenched
product gas. Further, the analytical techniques often used
for trace elements are not of high accuracy (Anderson et al.,
1979). The problem is compounded by the imprecision intro-
duced by the sampling of nonrepresentative materials,
sample preservation problems, and the microscopic inhomo-
geneities of coal and char samples. In most cases, the
trace element content of the quenched product gas is esti-
mated by difference. Anderson e£ al. (1979) pointed out
that trace element analysis in the gases is difficult be-
cause the form and approximate amounts of those elements
in the gas are not known. However, as he points out, if
one knew what compounds to look for and the approximate
amount present, some of the analytical problems would be
simplified.
The chemical reactions that C, H, and O undergo during
gasification, particularly steam-oxidant gasification of
coal, appear to be well understood. However, the chemistry
and kinetics of N and S are less well understood, and vir-
tually nothing is known about the chemistry and gasification
kinetics of minor and trace elements.
Conversion of coal sulfur to gaseous species is thought
to be a rate-limited phenomenon, generally promoted by con-
ditions that lead to high carbon conversion. The products
33
-------
are principally H2S and COS with some mercaptans and
thiophenes. A study of the fate of sulfur species during
the low Btu gasification of high volatile bituminous and
lignite coals (Page, 1977) shows that approximately 97%
of the bituminous coal sulfur was converted to H2S and
COS, while only 81% of the lignite sulfur was converted.
The variation was attributed to the alkalinity of the lig-
nite ash. The data indicate that the amount of COS formed
during the gasification of coal is approximately four
volume percent of the total gaseous sulfur species.
The gas-forming reactions involving sulfur are as
follows (Vestal and Johnson, 1969):
2 FeS + C •*• 2 Fe + CS2 (2-18a)
CS2 + H2 •* CS + H2S (2-18b)
CS2 + CO -»• CS + COS (2-18c)
CS + H2 •»• C + H2S . (2-18d)
Vestal and Johnson (1969) present rate constants and acti-
vation energies for these and other desulfurization re-
actions in hydrogenous atmospheres. A study by Jensen and
Austin (1977) gives plots of reaction rates vs. sulfur con-
version and presents an analysis that suggests diffusion
effects on the observed rates. Their work dealt with the
steam gasification of coal minerals obtained from solvent
refining of coal.
34
-------
Stinett et al. (1974) applied the principle of thermo-
dynamic free energy minimization to predict the effect of
operating variables in fuel gasification processes on the
equilibrium product gas , particularly with respect to sulfur-
containing species. For steam-air (or oxygen) gasification
of coal, they found that hydrogen sulfide is the pre-
dominant sulfur species. Small, but significant, amounts
of carbonyl sulfide are also shown to be present.
Nitrogen is partially converted during gasification to
ammonia, cyanide, and thiocyanate. Data for steam-air (or
oxygen) systems (Page, 1977) show that there can be signifi-
cant variation in the amount of ammonia formed during gasi-
fication of the same coal feedstock, depending on the amount
of steam used to gasify the coal, the surface moisture con-
tent of the coal, and the time-temperature history of the
coal particles in the gasifier. The first two parameters
affect the hydrogen partial pressure inside the gasifier,
which in turn directly governs the amount of NH3 formed,
and the third parameter affects the amount and characteris-
tics of nitrogen intermediates formed in the gasifier.
The average molar conversion of coal nitrogen to NH3, is
reported to be approximately 8%.
Anderson et al. (1979) performed a theoretical analysis
of the formation and disposition of compounds containing As,
Se, B, Pb, and Hg. Those elements were chosen because they
35
-------
were thought to be the elements roost likely evolved from
coal during gasification. This supposition agrees with the
conclusions of Jahnig e_t al. (1975), based on low tempera-
ture gasification of Pittsburg seam coal, that the follow-
ing percentages of volatile elements would be expected to
devolatilize and appear in the gas cleaning section of a
plant: Cl->90%? Hg->90%, Se-74%; As-65%; Pb-63%; and
Cd-62%.
According to Loran and O'Hara (1977), highly volatile
elements such as Be, Hg, and Pb, which do not form gaseous
hydrides, condense on cooling and are likely to be removed
in the aqueous condensates formed on gas cooling and/or
purification. As, Sb, and Se are less volatile but can
form covalent gaseous hydrides — arsine, stibine, and
hydrogen selenide. The authors point out, however, that
these hydrides have stability characteristics which pre-
clude their formation at the temperatures and pressures
prevailing in some commercial gasifiers.
A serious problem may be posed by metal carbonyls
formed by the reaction of carbon monoxide with free metals
in the 40-300°C temperature range. Carbonyls form with all
transition metals: higher pressures, of the order of 100
MPa (15,000 psi), and the presence of hydrogen favor their
formation, while oxygen represses it. They decompose
readily in air with half-lives estimated at 10-15 seconds
36
-------
for cobalt carbonyl, 10 minutes for nickel carbonyl, and
a few hours for iron carbonyl.
The theoretical analysis performed by Anderson et al.
(1979) suggests that the presence of arsine and its con-
centration in gasification process streams should be in-
vestigated further. Boric acid is projected to be the
major product of boron removal from the feed coal. Vola-
tile lead components should only exist in raw product
gases from high temperature gasification processes, such
as the Koppers-Totzek process. The thermodynamically pre-
ferred form of mercury in gasifier product gases has been
found to be the gaseous element.
2.2.6 Elemental Release During Pyrolysis
Major, minor, and trace elements are evolved during
the gasification of coal. The release is very rapid during
the devolatilization stage, becoming slower later, as the
coal continues to be heated or reacts with the surrounding
gas. Therefore, it is important that the rate and extent
to which different elements are evolved during the de-
volatilization of coal be determined.
Van Krevelen (1961) reported on the loss of C, H, and
0 during the slow pyrolysis of different coals at different
total weight losses. For lignite, oxygen was more easily
removed than hydrogen. For coals with carbon content around
37
-------
82%, more oxygen was retained compared to hydrogen at small
weight losses, but rapid loss of oxygen at the higher weight
losses exceeded the hydrogen losses. For coals above 90%
carbon content, relatively more oxygen was retained at all
values of weight losses, which was interpreted to mean
that oxygen becomes more strongly bonded as the rank in-
creases.
Kobayashi (1976) determined the retention of C, II, N,
S, and O in chars, following the pyrolysis of coal, in
crucible, free fall, and laminar flow experiments, as a
function of time and temperature. The retentions were cal-
culated from the original compositions of the coals,
measured overall weight losses, and the concentrations of
the elements in the chars. Both rates and final losses
were found to increase with temperature within the laminar
flow range. The results for sulfur showed more scatter
than those for the other elements. Kobayashi thought that
such behavior might be due to some interactions of organic
and inorganic sulfur; however, the scatter can also be
explained by the fact that sulfur has the lowest concen-
tration of the major elements in the coal, and its analysis
is less precise than those of the other elements. The prob-
lem is compounded by the fact that calculation of elemental
retention requires information on weight losses during the
devolatilization. Therefore, the scatter in the data
38
-------
represents, in part, the errors associated with the weight
loss measurements.
Among the elements studied by Kobayashi (1976), oxygen
and hydrogen were most easily removed. Nitrogen behaved
differently in the laminar flow runs: no appreciable loss
was found until about 800°K, but more than 30% was re-
tained even at the highest temperature (2200°K), which is
in contrast to the results obtained for hydrogen, oxygen,
and sulfur.
As shown in Figures 2-1 and 2-2, Kobayashi found that
both a lignite and a bituminous coal show similar trends.
At a given weight loss, more carbon, less hydrogen, and
less oxygen were retained in chars from the crucible runs
than in those from the laminar flow runs. Kobayashi in-
dicated that nitrogen did not evolve measurably until about
30% weight loss, and then the retention fell rapidly to
zero in the crucible runs. For the laminar flow runs, the
nitrogen retentions above 50% weight loss appear to be
close to those of carbon. The behavior of nitrogen sug-
gests that most of the volatile matter up to 30% weight
loss is aliphatic, since nitrogen is mainly incorporated
in the heterocyclic ring structures in the original coal
(Pohl, 1976), which are released in the latter stages of
pyrolysis. For lignite, sulfur retentions in the crucible
runs appeared to be higher on the average than those in the
39
-------
1OO
50
1OO
50
c
.3
4J
(U
JJ
100
50
100
100
t
50
x
O
0~
lormn
Frgw rpii
9 §—F>
J I CrucibU
1O
10'
101
»•
PESIOENCE TIME MILLISECONDS
V
< -
10
Figure 2.1 Major Elements Retained In Montana
Lignite Chars From Kobayashi (1976)
40
-------
c
o
•H
4J
a
z
100
f
o so
o
100
so
)
'• o
100
•
w
> so
: o
100
0
100
50
O
i
10
10*
ji
I!
II
i r
:v
«r
Fr*« roll
JL
now Fr*< roll CrueibM
RESIDENCE TIME MILLISECONDS
10*
Figure 2.2 Major Elements Retained in Pittsburg
Seam No. 8 Chars Prom Kobayashi (1976)
41
-------
laminar flow runs. The opposite trend was found for bitu-
minous coal.
Suuberg et al. (1978) found that although over 40%
by weight of a lignite is volatilized at a relatively high
temperature (1000°C), only 22% of the carbon is volatilized,
Therefore, most of the volatile material consists of hydro-
gen and oxygen. About 70% of the sulfur in the solid
material was found to have volatilized, but the nitrogen
content was reduced by only about 25%. The data from this
study, which were obtained with a wire screen furnace, are
summarized in Figure 2-3.
t
Kuhn et al. (1977) report that both organic and in-
organic coal constituents can be volatilized at low (<250°C)
and medium (250 to 650°C) temperatures. Their results from
batch experiments with long residence times show that:
1. Most coals exhibit similar behavior. Coals heated in
steps to 700°C show a reduction from 4.5% sulfur in the
raw coal to 1.5% in the char, a 66% loss of sulfur on a
whole coal basis.
2. Most of the sulfur was lost while the coal was heated
between 300°C and 400°C, coinciding with the temperature
range at which the coal char exhibited maximum Gieseler
fluidity and minimum internal surface area. Only a small
additional amount of sulfur was lost when the char was
heated to 700°C.
42
-------
100
50
£100
"5
•5 50
o
*- 0
°100
vT
| 50
I °
iioo
0>
V
50
0
50
00
CARBON
HYDROGEN
-o •
NITROGEN
SULFUR
OXYGEN
600 800
Temperature, °C
1000
Figure 2*3 Comparison of diar Compositions from Pyrolysis and Hydro-
pyrolysis of Lignite (o), 1 atm He, zero residence cine at
peak temperature; (A) 1 atm He, 5-20 sec. residence time;
(Q) 69 atm He, 5-10 sec residence time; (e>) 69 ttm H-,
residence tine 2-30 sec. From Suuberg el r1 """"1
43
-------
Shiley et al. (1978) report that elements which exhibit
significant devolatilization (>30%) for bituminous coals
heated in steps to 450°C and 700°C in a batch reactor under
a N2 atmosphere include P, Cl, S, As, Br, I, Pb, Sef Te,
and Zn. Another group that exhibits moderate to signifi-
cant losses (<30%) in some coals includes Cd, Cu, La, Li,
Rb, Sb, Sm, Sr, U, Yb, and possibly Al. Lignite is reported
to exhibit significant devolatilization losses of Si, K,
Ti, Dy, Ga, Hf, Ni, and Sc, in addition to most of the
elements listed for bituminous coals. Elements which appear
to be retained in the majority of chars, except those from
lignitic coals, include Si, Mg, Ca, Fe, Na, Ti, Ba, Ce, Co,
Cr, Cs, Dy, Eu, Ga, Hf, Lu, Mn, Ni, Sc, Ta, Tb, Th, and V.
Shiley et al. report inconclusive results, due to poor
statistics caused by extremely low concentrations, for Ag,
Au, In, Mo, w, and Sn. The weight and elemental losses
found by Kuhn et al. (1977) and Shiley et aJL. (1978) are
shown in Table 2-1 and Table 2-2. Shiley et al. (1978)
emphasize that only volatility ranges can be given because
the overall statistical errors involved in sampling,
pyrolysis, and analysis reflect significant uncertainty.
Percent relative standard deviation (%RSD) values for ele-
ments range from 5 to 10 in the case of minor elements to
20 to 40 for some trace elements. The %RSD includes cumu-
lative errors due to inhomogeneous samples, poor counting
44
-------
Table 2.1 Weight Loss During Pyrolysis (From Kuhn
et al., 1977)
Sample number
Seam and state
Percentage weight loss
at at
4500 700°
C 18857
C 18571
C18571F
C 18440
C 18185
C 18847
No. 6 Illinois
No. 6 Illinois
No. 6 Illinois
Lignite North Dakota
No. 5 Illinois
Blue Creek Alaska
32.2
27.5
30.3
33.9
27.0
8.4
34.3
21.5
40.5
44.1
37.1
17.0
Table 2.2 Preliminary XES Data For Pyrolysis of
Six Coals (From Kuhn et al., 1977)
C-18440
C-18571
C-1B571-T 0-18847
C-18857
C-U18S
Rl
tlMMitt ce
ca
In
Bn
Sb
ft
I
C*
Bt
La
C*
Zn
Br
•b
9r
IW
Ml
.3
.a
.0
.6
.3
.9
.2
37
.1
.7
.7
.3
.9
241.
4iO*C
2.3
0.72
1.0
2.0
4.4
8.5
26.9"
1205
10.7
13.6
14.3
2.8
6.4
245.1
Maw
Coal
3.3
1.7
6.9
5.2
0.7
2.7
2.7
44
8.8
9.2
84.5
7.2
12.1
27.5
4SO'C«
1.6
0.10
0.18
0.07
o.st
1.4
3.8
48.5
48.3
10.0
11.1
30.4
M*
Coal
1.8
1.9
5.1
S.8
0.9
1.4
2.9
34.3
6.3
9.7
21.8
10.1
10.3
21.7
45
0.8
<0.1
0.8
1.0
0.3
0.8
2.0
35.1
4.9
7.0
18.3
7.1
8.6
19. C
MM
C»al
1.4
.0
.9
.0
.5
.2
.6
202
13.5
20.4
13.7
2.7
14.0
68.2
4SO*C
0.9
<0.1
<0.1
<0.1
1.4
3.0
8.4
241
13.9
24.7
12.1
1.9
10.9
S9.3
MM
coal
1.9
0.9
2.2
1.9
0.8
1.8
3.3
51
10.8
10.0
3S.3
9.1
12.3
2*.B
4SO*C
0.75
<0.1
.68
<0.1
1.1
1.
3.
53.
8.
11.
45.7
5.7
10.0
a*. 4
MM
Co.1 4SO«C
7.0 7,0
0.8 <0.1
1.4 <0.1
3.3 <0.1
0.5 1.8
2.5 2.0
2.4 8.9*
40 302
4.9 8.1
8.9 13.3
323 246
4.4 5.0
9.2 9.4
12.0 17.S
All valiM* «xpr«»«d ** Ugr/gr.
*Av«r«q« of two d«t«rmin«tion«
blntarfwr«nc« frooi Ba
45
-------
statistics (in neutron activation analysis), contamination,
etc., for each coal or char.
Kuhn et al. (1978) and Shiley et al. (1977) used pri-
marily energy dispersive and wavelength dispersive x-ray
fluorescence, and instrumental neutron activation analysis.
They carried out an extensive research effort to optimize
the analytical conditions for the analysis of the coals and
resulting chars to achieve the best detection sensitivity
and to better quantify trace element losses.
More recently, Kuhn et al. (1979) have made a mass
balance of elements mobilized during pyrolysis using the
same bench scale pyrolysis system used in their earlier
studies. Two coals were studied. The original coals, the
chars, and the condensed volatile fraction were then sampled
and analyzed by x-ray and neutron activation analyses. Each
coal was pyrolyzed at 450°C and 600°C. The data thus ob-
tained are shown in Tables 2-3 and 2-4. The actual values
obtained for the materials are in one series of columns and
calculated values normalizing the data to the original coal
basis are presented in adjoining columns. The calculated
values can then be added for each series. Kuhn et. al. (1979)
indicate that a mass balance within ± 20% is obtainable for
many elements, a limit reported to be within the analytical
methods and sampling errors for most of the elements studied.
46
-------
Table 2.3 Results of Analysis River King
(From Kuhn et al.r 1979)
Raw Coal
C-2029?
£leaent Actual
Fc
K
st
Al
CM
Tl
S
Cl
Ha
hu
Ba
Br
c«
Co
Cr
c»
Eu
Ga
Hf
La
Lu
Hi
Kb
Sb
Se
Sa
SB
Sr
Ta
Tb
Th
n>
Zn
Book
Sn
I
Zr
Ho
2.73%
0.20%
3.95%
1.73%
.48%
.11%
3.68%
.03%
522
l.S
56
2.0
7.0
5.4
18
1.13
0.2
3.2
0.6
5.0
0.16
14
21
0.30
3.0
2.3
1.0
9
0.15
0.20
2.0
0.70
32
Ho.
3.0
1.6
19.8
10.7
450*C Char
Wt- 71.59
C-20300
Actual Caled.
2.53%
.28%
5.76%
2.39%
.74%
.09%
3.40%
.01%
720
l.S
105
2.5
14
7.2
29.
1.7
0.3
4.6
1.0
7
0.2
30 '
27
0.3
4.2
3.1
1.4
63
0.2
0.2
3
0.8
78
913-15
<1.0
l.S
29
14.4
1.81%
0.20%
4.12%
1.71%
.53%
.06%
2.43%
<.01%
515
1.07
75.07
1.79
10.01
5.15
20.73
1.21
0.02
3.29
0.71
5.00
0.14
21.45
19.3
0.21
3.00
2.22
1.00
45. 04
0.14
0.14
2.14
O.S7
55.77
450*C Oil
Wt- 13.839
C-20301
Actual Calcd.
38
5
LD
LD
LD
<2 pp>
2.22%
37 ppm
9
0.2
<10
3.2
<5
0.01
0.6
O.O4
0.002
<0.1
<0.01
0.006
0.007
-------
Table 2.4 Results of Analysis Crown Mo,
(From Kuhn et al., 1979)
Raw Coal 450 'C Char
Wt- 68.99
C-20239 C-20317
Element Actual Actual Calcd.
re
K
Si
Al
Ca
Ti
S
Cl
Na
As
Ba
Br
C*
Co
cr
Ca
Eu
Ga
ire
La
Lu
Hi
Rt
Sb
Se
Sa
Sa
Sr
Ta
Tb
Th
TO
Zn
Book
Sn
I
zr
Ho
2.38%
0.17%
2.43%
1.62%
.15%
.09%
3.43%
.07%
774
1.6
SO
3.6
12
2.8
13
1.2
0.18
3.1
0.4
5.4
0.14
14.0
13
O.5
2.4
2.3
1.0
38
0.11
0.15
1.7
0.6
21
No.
4.8
1.9
16.5
12.6
3.6%
O.20%
3.59%
2.36%
.26%
.08%
3.94%
.02%
1000
2.2
83
4
14
3.7
20
1.1
0.3
3.8
0.7
7
0.16
26
23
0.5
3.0
2.8
1.4
74
0.16
0.17
2.1
0.84
40
913-14
2.48%
0.14%
2.47%
1.63
.18%
.06%
2.71%
.01
689
1.51
57.19
2.75
9.64
2.55
13.78
O.76
0.21
2.62
0.48
4.8
0.11
17.91
15.85
0.34
2.07
1.93
0.96
50.99
0.11
0.12
1.45
O.SS
27.56
450*C Oil
Wt- 12.99
C-20318
Actual Calcd.
7 ppa
4 ppo
LD
LD
LD
<2 ppm
2.11%
61 ppa
6
2
0.90
0.52
LO
LD
LD
<2 ppa
.27%
0.7T
0.26
6OO'C Char 600*C Oil
Wt- 62.69 Wt- 15.919
C-20319 C-20320
Actual Calcd. Actual Calcd.
3.
0.
3.
2.
.
8%
24%
96%
53%
23%
.09%
3.84%
.01%
1112
2.
6
<10 <10 100
<0
0
0
0
<0
<0
<0
0
0
0
1
0
<0
<0
<0
6
.06
.02
1
-
.007
.09
.1
.1
.01
.3
<1
.13
.001
.2
.002
<1
.01
.01
.1
0.007
2
0.77
<0.08
0.0026
0.129
-
O.O009
0.012
^Oe 1
-------
2.2.7 Chemistry of Elemental Release During Pyrolysis
Relatively little information is available in the
literature concerning the chemical mechanisms and kinetics
of the release of sulfur and nitrogen from coal during de-
volatilization, and virtually no information is available
on the chemical mechanisms and kinetics of the release of
minor and trace elements during devolatilization.
The rates and extents of decomposition and volatili-
zation of nitrogen compounds depend on the thermal environ-
ment of the coal particles. Low heating rates yield mainly
ammonia and residual coke-nitrogen. As indicated earlier,
nitrogen is evolved late in the particle heating sequence,
indicating that most of the nitrogen is probably in the
strongly bonded aromatic structures, and that at sufficient-
ly high temperatures the relative yield of nitrogen exceeds
the yield of total volatiles. Malte and Rees (1979) report
that a rough, first-order fit of the data of Pohl and
Sarofim (1977) gave a rate constant of (93 x 103) [exp
(-11,400/T)]sec~l £0r OVerall coal-nitrogen pyrolysis, with
the reactant being the amount of the residual nitrogen. At
1500°Kf this expression gives a characteristic pyrolysis
time of 100 msec.
The behavior of sulfur in coal has been studied by
several researchers. Kuhn et al^ (1977) report a greater
loss of sulfur from the pyrite than from organic sulfur at
49
-------
low temperatures; whereas, the reverse was observed at
temperatures above 450°C.
The dominant form of organic sulfur in the coal mole-
cule is thought to be thiophene. Solomon (1977) indicates
that as much as 60% of the organic sulfur resides in such
heterocyclic rings. According to Attari e_t aJL. (1976), and
sulfur side chains (-S1I) and the linking sulfur chains
(-S-) rupture first in the heating sequence during pul-
verized coal pyrolysis, leading to early volatile sulfur.
The thiophene structure, however, is more stable and does
not decompose until temperatures of about 1200°K are
attained.
Through an x-ray scanning technique sensitive to S and
Fe, Solomon (1977) was able to examine both inorganic and
organic sulfur in coal char. He observed that the residual
organic matter was depleted of sulfur, while the sulfur con-
tent of the inorganic ash was increased.
Padia (1976) indicated that pyrite is quite unstable
at high temperatures. At 750°K, the oxidation of pyrite to
hematite occurs as follows:
2 FeS2 + 5.5 02 »• Fe203 + 4 S02 • (2-19)
In a reducing environment, FeS2 is transformed to FeS by
the following reactions (Malte and Rees, 1979):
50
-------
FeS2—*• FeS + h S2 (gas) (2-20a)
FeS—> Fe + h S2 (2-20b)
FeS2 + H2 —»• FeS + H2S (2-20c)
FeS + H2 —»• Fe + H2S . (2-20d)
Solomon (1977) established the following experimental rate
for the decomposition of FeS2 to FeS
[in(x-l)]/t = -480 exp (-8400/T) sec"1 (2-21)
where x pertains to EeSx.
Kuhn e_t al. (1977) report that the pyrite contained in
coal is converted to pyrrhotite and sulfur at 450°C or
lower in a nitrogen atmosphere. Their chemical analyses
also indicate a greater loss of sulfur from the pyrite than
from organic sulfur at low temperatures, whereas, the re-
verse is true at high temperatures (>450°C).
As discussed in Section 2.2.1, minor and trace elements
are present in coal in varying amounts and different degrees
of association with the organic and inorganic matters of the
coal. Most minor elements are generally associated with the
discrete mineral matter. The principal minerals found in
coals include kaolite, pyrite, illite, calcite, and quartz.
Upon heat treatment of the coal to elevated temperatures,
the minerals are changed chemically as described by Padia
51
-------
(1976). At high enough temperatures (>1300°K), the trace
and minor elements associated with the mineral matter may
be devolatilized and released from the coal, the extent
of the evolution increasing with temperature.
The major form of arsenic in coal is thought to be
arsenopyrite (Duck and Himus, 1951) . When temperatures
exceed 820°K, arsenopyrite begins to decompose into pyrr-
hotite (FeS) and metallic arsenic
FeAsS(s) --+ FeS(s) + As(s) . (2-22)
The decomposition proceeds rapidly at temperatures greater
than 1025C-K (Lukesh, 1940) .
Anderson et al. (1979) suggest that selenium is
initially present in coals as selenopyrite (FeSeS) which
decomposes upon heating, releasing the element according
to the reaction
FeSeS(s) heat, FeS(s) + Se(g) . (2-23)
Lead in coal is believed to exist initially as PbS
(Anderson, 1979) . However, it has been reported that
lead is associated almost entirely with the organic frac-
tions of coals. As indicated before, Horton and Aubrey
(1950) found that phosphorus is associated with the inor-
ganically combined matter in some coals, but the organic
affinity of phosphorus seems to be rather high in other
52
-------
coals (Gluskoter et aJL., 1977). Anderson (1979) suggests
that most of the boron in coal may be chelated. The chemi-
cal forms of mercury in coal are not known; however,
evidence suggests that metallic mercury is released from
coal during devolatilization.
Finally, Kuhn et al. (1977) concluded that the two
most important temperatures for which volatility data need
to be obtained are 450°C and 700°C. At 450°C, reactivity
is highest, and most volatile products are released (over
a very long period of time); at 700°C, virtually all
volatile products are released but the coal structure is
still intact. Heating above 750°C completely destroys the
original coal structure, and the internal surface area
decreases.
2.2.8 Conclusions from Literature Review
Significant progress has been made in the modeling of
the pyrolysis of coal. Several models appear to provide
fair correlations for weight loss as a function of time
and temperature; however, much work still remains to be
done to elucidate the actual mechanism of the pyrolysis
reactions. Furthermore, all efforts in this direction
are hampered by problems arising from the different be-
havior shown by different coals, inhomogeneities within a
given coal, and problems in the interpretation of data
53
-------
obtained with different types of equipment. This last
point was only discussed in the literature survey with re-
gard to the disagreement in first-order kinetic parameters.
Another extremely important problem, the interpretation of
data obtained with similar equipment but with different
operational assumptions, will be dealt with in the sections
on the selection of experimental apparatus and design cal-
culations.
Excellent mass balances around coal gasification plants
can be made on major elements, but attempts to close mass
balances on minor and trace elements are plagued with prob-
lems of irreproducibility and scatter of data. The causes
of those problems are varied. They include coal and char
inhomogeneity, sample contamination, low accuracy and pre-
cision of analytical techniques, and a lack of fundamental
information regarding the mobility of trace elements of
coal under gasification conditions.
Significant losses of major, minor, and trace elements
occur during pyrolysis. It appears evident that the yield
of a given element is a function of temperature. The
pioneering work of Kuhn et al. (1979) has provided an ex-
cellent start to the determination of the mobility of trace
elements during gasification. However, the pyrolysis data
available has been obtained only in batch reactors with
very long residence times, low heating rates, and at only
54
-------
two rather low temperatures. No studies have yet been re-
ported in the literature on the release of trace and minor
elements during fast devolatilization.
Finally, it should be pointed out that the foregoing
was not an exhaustive review of the literature on coal
pyrolysis, much less on coal gasification or on coal in
general. Such literature is very extensive. Only infor-
mation directly related to the work done in this research
has been covered, and then only in summarized form. Ex-
cellent reviews and summaries of the literature on coal
and coal gasification have been published by Lowry (1963) ,
Gould (1967), and Massey (1974). The best reviews on the
pyrolysis of coal are those of Kobayashi (1976), and Smoot
and Pratt (1979) . The subject of trace elements in fuel-
is reviewed by Babu (1975) .
55
-------
3. DEVOLATILIZATION APPARATUS AND PROCEDURE
3.1 Selection of Apparatus
Devolatilization has been shown to occur In a matter of
a few hundred milliseconds at low temperatures, and within
a few milliseconds at high temperatures. Therefore, it is
required that the experimental apparatus be able to resolve
times of this magnitude in order to observe the kinetics of
devolatilization. At the same time, most coal gasification
processes operate with residence times ranging from a few
seconds, as in the Garret Flash Pyrolysis Process (Sass,
1974), to hours, as in fixed bed reactors. Furthermore,
the analyses needed for the characterization of chars and
the determination of their trace element content require
reasonably large amounts of product char. Approximately
5 grams of sample are thought to be required. Even with
the restrictions imposed in this research, namely, the
study of the pyrolysis of pulverized coal in inert atmos-
pheres, widely divergent requirements are evident.
A diversity of apparatus types have been used for the
study of coal pyrolysis. Examples of equipment used are
crucibles, retorts, different sizes of batch reactors with
gas flow, differential scanning calorimeters , thermogravi-
metric balances, shock tubes, laser irradiation heating.
56
-------
wire screen furnaces, free fall reactors, and turbulent and
laminar entrained flow reactors.
Most experimental techniques were ruled out for the
present study because of inappropriate sample size, in-
adequate time resolution, or irrelevance to the conditions
that are encountered in real gasification systems. It was
decided that the use of a small bench scale batch reactor
with gas flow and a laminar flow reactor would provide the
information needed at long and short reaction times, and
the required sample sizes. In addition, the laminar flow
reactor provides high heating rates (103-105 °C/sec) com-
parable to those of entrained bed and fluidized bed com-
mercial processes.
Small batch reactors have been used by many researchers ,
most recently by Kuhn et al. (1977, 1979). The equipment is
simple and relatively easy to operate. However, the inter-
pretation of the results may be subject to error when high
rank coals are used. The volatile yields of high rank
coals have been shown to be highly dependent on the heating
rate and, to a lesser extent, on the bed depth in batch
experiments.
The laminar flow reactor was initially developed by
Sainsbury et al. (1966) and later used by many other re-
searchers. Even though, in principle, this reactor is
ideal for the study of coal pyrolysis, in practice its
57
-------
mechanical complexity makes it difficult to operate. Seve-
ral parameters cannot be measured directly. Therefore,
the reduction of the data has to be based on a mathemati-
cal model of the fluid and particle flow and heat trans-
fer phenomena in the reactor.
Among the mechanical and physical problems and limi-
tations of this type of reactor, the most important are
the following:
1. Quantitative collection of the char is very difficult.
A significant fraction of the char tends to miss water-
cooled collectors, and another fraction tends to stick to
the inner collector walls. For these reasons, weight
losses have usually been estimated using the ASTM ash con-
tent of the coals and chars as a tracer.
Kobayashi (1976) and Padia (1976) showed that the use
of ash as a tracer at high temperatures leads to signifi-
cant errors. This study shows that the error is also quite
significant at lower temperatues. Kobayashi (1976) solved
this problem to an extent by using a sintered metal filter
and collecting char at well above isokinetic suction flow
rates. However, he had to use water jets to quench the
pyrolysis. Besides increasing considerably the complexity
of the system, the use of this type of collector gave rise
to other problems. Ash and small coal particles were still
lost through the pores of the filter; the data had to be
58
-------
corrected for the formation of soot, and the washing of the
very hot chars may have caused chemical reactions between
the water and the constituents of the char. Kobayashi's
system has one more drawback, with respect to the pur-
poses of the present research: the quenching water may
change significantly the trace element content of the
chars by washing and dissolving its inorganic minerals.
2. Water-cooled collectors cannot handle highly caking
coals and quench adequately the pyrolysis reactions at the
same time. Nsakala (1976) had to restrict his work to non-
caking coals because of severe plugging problems in the
collector. The inside diameter of the collector and the
angle of the collector nozzle can be increased such that
caking coals can be handled. However, doing so decreases
the cooling rate of the particles in the collector. In
addition, lower suction rates are necessary to obtain well
defined laminar flow fields, which lead to lower particle
collection efficiencies.
3. The heating of relatively large volumes of gas around
water-cooled feeder probes is difficult. Nsakala (1976)
had to restrict his work to temperatures below 808°C.
Kobayashi was able to reach 2200°K using complex and ex-
pensive equipment which included an argon plasma gun and
a graphite muffle tube.
59
-------
4. The feeding of a small stream of coal dust well dis-
persed in a carrier gas at uniform rates is not a simple
matter. However, it can be accomplished.
5. The high temperatures required and the large number
of reactor internals and accessories make the materials
requirements of this type of reactor quite stringent.
6. The heating rates of the particles are highly dependent
on the heating of the cold carrier gas stream, which in turn
is a function of the carrier gas flow rate, feeder geometry,
and main stream gas flow rate. Estimation of the particle
heating rate requires complex mathematical analysis of mass
and heat transfer equations. The approaches taken have
been: (a) to attempt to measure the heating rate of the
carrier gas and (b) to solve simplifications of the mass,
momentum, and heat transport equations. Nevertheless, in
all cases the characteristic heating time of the coal par-
ticles ended up essentially as another fitted parameter.
This problem has been compounded by the different
assumptions made by different researchers regarding the
pyrolysis phenomena occurring during the heating period.
Dadzioch and Hawksley (1970) and Nsakala (1976) assumed
that no reactions occurred during the heating of the coal
particles and that the reactions are essentially isothermal.
The empirical model that they used and their calculated
kinetic parameters in essence reflect this assumption.
60
-------
However, Kobayashi (1976) and Reidelbach (1978) have demon-
strated that the heating period is indeed quite important.
7. The measurement of gas temperatures in the reactor
is difficult because of complex radiative interactions be-
tween the measurement devices and the hot walls and cold
spots of the reactor. Most researchers have resorted to
measurements with suction pyrometers under simulated run
conditions.
8. The residence times of the particles cannot be measured,
but are calculated from their velocity and distance between
the feeder and collector. However, the velocity of the coal
particles is dependent on the reactor gas velocity and the
particle size distribution. The gas velocity is, in turn,
a function of the gas mass flow rate, reactor temperature,
collector suction flow rate, and reactor geometry. Even
though it is obvious that the main gas has a developing
laminar flow velocity profile, fully developed laminar pro-
files have been assumed (Badzioch and Hawksley, 1970;
Nsakala, 1976). In addition, most researchers regarded
the free fall velocity of the particles as negligible (e.g.,
Badzioch and Kawksley, 1970; Nsakala, 1976). This assump-
tion is obviously not correct in the case of particles
larger than 100 ym such as some of those used by Nsakala
(1976).
Kobayashi (1976) measured the velocities of the par-
ticles at a point in the reactor using a laser doppler
61
-------
anemometer. He then used the data to fit the parameters of
a theoretical model of particle velocities and boundary
layer development coupled with an empirical equation to
describe the axial velocity of a developing laminar flow
profile. Such an effort obviously involved considerable
labor and added to the complexity of the apparatus.
9. Because of its complexity, control of the apparatus
is difficult. Several gas feed and exhaust flows have to
be maintained, coal feed rates must be reasonably uniform,
and temperatures must be controlled and monitored.
It can be seen that the problems aid drawbacks of
laminar flow reactors are sufficiently serious that re-
searchers (e.g., Tran, 1978) who could have built and used
such equipment decided not to do so, giving as reasons the
difficulties in the control and operation of the apparatus
and the complex mathematical analysis of mass and heat
transfer equations that are required for the data reduction.
Nevertheless, this type of reactor can be quite useful if
properly designed and if its limitations and the under-
lying assumptions made in the data analysis are carefully
considered and appraised.
The following sections provide a description of the
apparatus used, operating procedures, and the design cal-
culations used in the analysis of the reactor parameters.
62
-------
Of necessity, those descriptions and operating procedures
are brief and presented in summarized form. Detailed
operating procedures, calibration curves, and design
drawings of the equipment are available in an internal
technical report (Agreda, 1979) at the Chemical Engineering
Department, North Carolina State University.
3.2 Description of Batch Reactor System
A schematic of the batch reactor system is shown in
Figure 3.1. The system is built around a Lindberg Model
54032 single zone tube furnace and a Lindberg Model 59344
digital control console. The heating zone of the furnace
is twelve inches long. The reactor temperature is measured
with an Omega Model 2160A digital thermometer using a
grounded chromel-alumel thermocouple with a 316 SS sheath.
A quartz tube, 26 inches long and 1 inch in diameter is
used as the reaction tube. The samples are introduced in
porcelain boats. Two types of boats are used: glazed
porcelain boats with a capacity of 2 grains of coal, and un-
glazed porcelain boats with a capacity of approximately 0.5
grams of coal.
The temperatures inside the furnace are determined by
inserting the 1/16-inch thermocouple instead of a slide
wire. The temperature controller's feedback loop uses a
Platinel II thermocouple embedded in the heating element's
63
-------
HEATING ELEMENT
PYROLYSI8 MAT
HIAT TACING
lu wcni
SLIDE WIRE
^ ^ ^ ^
> "
FURNACE CONTROLLER
\
A
QUARTZ TUBE
SCRUBIIR
BATCH REACTOR SYSTEM
APRIL-28-1979 VICTOR H. A6REOA
Figure 3.1 Batch Reactor System
-------
ceramic support. Therefore, it was necessary to calibrate
the controller readings to the actual reaction zone tem-
perature. The temperature profile of the furnace was
determined experimentally. It was found that the peak
temperature occurred close to the center of the furnace,
and the profile is fairly flat in a 3-inch section at the
center of the furnace. The pyrolysis boats were placed in
this zone for the runs. Nitrogen was used as the sweep gas.
The purpose of the gas flow was to remove the products of
the pyrolysis without affecting the pyrolysis reactions.
It was determined that the gas flow rate did not affect the
reproducibility of the results, within experimental error,
for all flow rates below 700 cc/min. The gas flow rate was
kept at 300 cc/min in all runs.
Nitrogen flow to the pyrolysis tube is metered with
an Air Products rotameter. The pyrolysis tube is con-
nected to a scrubber bottle containing 70 ml of ION sodium
hydroxide. The partially cleaned exhaust gases are piped
to a vent. All connections are made with Swagelock con-
nectors with teflon ferrules. A 150 watt Fisher heating
tape is used to heat-trace the tube exhaust. A 1/16-inch
outer diameter (OD) stainless steel slide wire is used to
insert the pyrolysis boats into the furnace. The slide
wire holder is sealed with an 0-ring.
65
-------
Temperatures in this reactor can be varied between
200 and 1200°C. Residence times as low as 30 seconds can
be achieved with fair reproducibility.
3.3 Experimental Procedure for
Batch Reactor Experiments
The experimental procedure is as follows:
1. Select and set digital controller from temperature
calibration curve.
2. Monitor reactor peak temperature and wait until digital
temperature readout is at the desired level for at
least 30 minutes. Usually at least 3 hours are needed
for the reactor temperature to equilibrate.
3. Set gas flow at 300 cc/min.
4. Replace thermocouple with slide wire.
5. Turn heating tape on.
6. Weigh desired amount of coal in porcelain boat.
7. Place boat in mouth of tube. Connect slide wire in the
boat's hook.
8. Close tube and seal gas-tight with the Swagelock
fitting and the O-ring.
9. Purge with nitrogen for at least 5 minutes. The pur-
pose of this step is to ensure that no air remains
in the system, so that pyrolysis, not combustion, is
studied.
66
-------
10. Push boat, with slide wire, into the center of the
reaction zone.
11. Keep boat in furnace for desired length of time.
12. Pull boat out of the reaction zone.
13. Allow boat to cool for 5 minutes before opening the
tube to ensure that the coal does not combust.
14. Open mouth of pyrolysis tube and remove boat.
15. Place boat in desiccator. Wait until it cools com-
pletely (at least 15 minutes).
16. Weigh char and boat.
17. Empty boat into sample storage bottle.
18. Turn power off; cut gas flows off.
19. Clean pyrolysis tube after it has cooled to room
temperature.
3.4 Description of Laminar Flow
Reactor System
A detailed schematic diagram of the laminar flow
reactor system is shown in Figure 3.2. The reactor design
is similar to those of Badzioch and Hawksley (1970) and
Nsakala (1976). It is not a replica of either, however*
and can be operated over a wider range of conditions. The
coal feeder-hopper is patterned after the miniature system
used by Kobayashi (1976); it can be loaded with up to 25
grams of coal and reloaded as many times as necessary
throughout the run.
67
-------
«-».! i.
Cltl FEEIEI
•tin uciii
I" SIICE
III* IISTIIIUIO
!«< S1U1 «f»tfl
fill S1UICN1KI
1C FEEIEI
1C COUECHI
millt CEdHNT
HICI i»s»i«iie«
IFI-4
>»•«
JM-
Kill UK
Illll HIE
I \
1
n
9
A
=TL!
'
1
[
(
'
\
\
f
;
(-21
\— n
KRUI
~
»-M*~
FT
11 y
---L
— 1~
1
i
-
L-
• t
I"
i
i
w
n
H
r
i~ *
i
i
iiciini
iruni
~ — i
mrui
INtCt
IMUII
Ennitint
DISPIH
• I
DIAIN
UBIIU FLO" RUCTII I»STFM
Figure 3.2 Laminar Flow Reactor System
-------
The basic principle of operation of the laminar flow
reactor can be summarized as follows. Size-graded,
finely ground coal particles are introduced into the water-
cooled feeder tube from a partially fluidized vibrating
hopper. The particles flow into a preheated stream of
gas flowing downward through a vertical furnace tube at
a Reynolds number low enough to ensure laminar flow. The
furnace tube is held at the same temperature as the pre-
heated gas. The small carrier flow of cold gas mixes
rapidly with the hot gas stream, thus allowing the par-
ticles to be brought rapidly to the furnace temperature.
Because the flow is laminar, the particles travel in a
narrow streamline along the axis of the furnace and are
aspirated into a water-cooled collector. The collector
has a tapered entry so that the aspirated gases are
accelerated to a high-velocity turbulent flow and thus are
rapidly cooled. The cooling of the gases reduces the
temperature of the suspended particles and quenches the
decomposition. The transit time of the particles can be
varied by changing the gas flow rate or by altering the
distance between the feeder and the collector. Transit
times in this reactor can be adjusted from 50 to 2,000
milliseconds. The temperature of the furnace and gas stream
can be adjusted up to 1000°C. As indicated before, the
pressure can be adjusted from atmospheric to 2.4 atm.
69
-------
The operating ranges given are maximum obtainable
. Ml the parameters cannot be varied independently
within those limits, however, because they are inter-
dependent. For example, the residence time is a function
of yas flow rate, gas temperature, particle size and
density, and feeder-collector distance. At very high tem-
peratures, the gas velocity may be so fast that it cannot
be compensated entirely by lowering the gas flow and in-
creasing the feeder-collector distance. On the other hand,
the minimum velocity that the particles can achieve is
their free fall velocity in still gas. The pressure is a
function of the main, exit, and suction gas flow rates.
The following sections provide details of the appa-
ratus and equipment used.
3.4.1 Gas Supply and Utilities Subsystem
Nitrogen was used in all the experimental runs per-
formed in this study. Other non-flammable gases can also
be used, however. The gas supply consists of two A size
nitrogen cylinders fitted with regulators (PCVl and PCV2)
and piped such that the supply can be switched from one
cylinder to another during a run without disturbing the
run parameters. City water is used for cooling in the
feeder, water jacket, and collector. One 20 amp, 208 V
line and one 20 amp, 110 V line are installed to provide
electrical power for the equipment.
70
-------
3.4.2 Feeder Subsystem
The coal feeder hopper is constructed of 38.1 mm 00
lucite tube. It has a capacity of 25 grans of coal. A
small amount of carrier gas is injected into the hopper
(typically 1.0 £pm) through the hollow needle of the feed
rate control valve, V-25. The tip of the valve has four
small holes through which the gas is dispersed radially
into the hopper. Approximately 50 cc/min of the feeder
carrier gas rise through the hopper partially fluidizing
the bed. The bed itself is continuously shaken by an elec-
tromagnetic vibrator (MV-1) to impede agglomeration. The
remainder of the gas flows downwards through V-24 carrying
the coal particles. The fluidizing gas joins the carrier
gas stream through V-23. The feeder gas flow rate is con-
trolled at the set point with FCl. The fluidizing gas is
filtered through FR-4 and metered with R-2. V-21 is a
three way valve used to equalize pressure across the hopper,
which is necessary during start up or to stop the coal gas
flow temporarily. V-21 is also used to purge the coal Led
with nitrogen before a run. V-24 is closed and V-21 is set
on the vent position for that purpose. For further details
on the operation of the feeder system see Agreda (1979).
The coal particles, now well dispersed in the carrier
gas, are injected into the reactor through a water-cooled
feeder made of several concentric 316 stainless steel pipes.
71
-------
The two outer shells have fiberfrax insulation between them
to prevent excessive cooling of the main gas. The inner
tvjl.e LS 3.3 mm inner diameter (ID), and the outer shell is
2.54 cm OD. The end of the water-cooled feeder has a female
thread such that alumina tips can be screwed in to connect
with the 5 mm OD inner tube. Several feeder tins with
orifice sizes ranginq from 1.6 nm to 6.4 mm are
available. The velocity of the feeder gas can therefore be
controlled by increasing the gas flow, or, for a given gas
flow, by changing the feeder tip bore size. The maximum
Lore size attainable is that of the thread itself which is
7 mni. The maximum bore size was used for all the runs re-
ported in this thesis.
The coal feed rate can be adjusted by raising or lower-
ing the needle of V-25, the electromagnetic vibrator power,
and the ratio of fluidizing overflow gas to feeder gas. The
primary means of control is the needle of V-25. Coal feed
rates can be adjusted between 0.1 g/min to 2.0 g/min fairly
reliably; however, the feed rate tends to vary slightly as
the height of the bed decreases. Feeder gas flow rates can
be adjusted from 0.1 to 2.0 £pm.
3.4.3 Furnace Reactor and Gas Heaters
The furnace tube is made of alumina with an inside
diameter of 8 cm and a length of 1 meter. The
72
-------
liner tube is an alumina tube , 7 cni ID and 0.75
meters long. Heat is supplied by a three zone, 4000 watt,
230 volt Thermcraft furnace, with a heated length of 47
cm. The furnace temperature is controlled with a Lindberq
Model 59344 digital control console which uses a Platinel II
thermocouple as the sensing element. The furnace elements
can be heated from 200 to 1200°C. The temperature profile
of the heating elements is also monitored with a Pt-13% Uh
thermocouple connected to a display (Omega Model 250) on the
controller, and a chromel-alumel thermocouple multiplexed to
the other system thermocouples.
The main gas is heated in three stages. The preheater
consists of an 800 watt 19 mm Sylvania serpentine fluid
heater. The heater itself is placed in the reactor head.
It consists of two ARI 1000 watt heating coils separated by
a 3.2 mm thick sintered stainless steel disk and encased
in a stainless steel shell. The heaters are packed with
stainless steel wool to improve the heat transfer efficiency
to the gas. The power supplied to the preheater and heaters
is controlled with three variable transformers. Another
sintered stainless steel disc uniformly distributes the flow
leaving the second stage of the gas heater. A 12.7 mm
thick, 1.6 mm bore size alumina honeycomb disc placed
38 mm below the flow distributor, and resting on the
liner tube, ensures that the gas enters the reactor in
73
-------
streamline flow. The feeder tip is 38 mm below the
lit. t. tor: of the honeycomb. Ihe upper section of the gas
iicatur-i."ec-clcr reactor head has a v .ter jacket, provided so
that the viton gasket that seals the reactor against pres-
sure leaks does not nelt.
Two thermocouples are placed inside the reactor to
monitor the reaction zone temperature. One is placed at
the top of the reaction zone, close to the feeder tip, ana
the other is placed in the middle of the reactor. All
chromel-alumel thermocouples are rr.ultiplexed to an Omega
Kodol 2100A digital thermometer. The reactor pressure is
monitored with an Air Products 0-30 psig (0-1600 mm Hg) gauge (PI 12).
J.4.4 Char Collection Apparatus
The pyrolysed chars are collected with a 30-inch long
.stainless steel collector made of concentric shells. The
outer shell is 19 mm OD, and the inner tube is 0. 46 cm
ID. The r>outh of the collector is tapered, decreasing from
l.blj en to 0.46 cr. over a distance of 1.3 cm, so that the
mouth makes a 62° angle with a horizontal line at the base.
Two other collectors are available, which are designed to
handle caking coals. Their inner tubes are wider and the
collector angles with the base hroizontals are much larger.
The collector is fixeu at the desired position and
sealed pressure-tight with a packing gland. An
74
-------
electromagnetic vibrator (MV-2) prevents char particles from
sticking to the collector walls. The exit gas temperature
is monitored with a thermocouple (TC-5).
Most of the pyrolyzed chars are separated from the gas
stream using two high efficiency cyclones placed in series.
CY-1 is a 19 mm cyclone, and CY-2 is a 13 mm cyclone.
Both are manufactured by the Air Correction Division of
Universal Oil Products Company. The two cyclones ensure
collection efficiencies greater than 99% for particles
greater than 10 ym at the gas flow rates typically used in
this system.
The char that misses the water-cooled collector exits
the reactor through two ports located at the collector sup-
port base. This char is separated from the exit gas with
an Air Correction 13 mm cyclone (CY-3).
3.4.5 Exhaust and Guction Subsystem
The collector suction line is connected to a high flow
rate Lamert 03121 vacuum pump. Two water scrubbers and
several large surface area filters are used to clean the
erases exiting the cyclones. The water scrubbers were found
to be necessary because of the large amount of tars produced
in the pyrolysis of some coals. In addition, the main ex-
haust line has a Drierite drying column to remove the mois-
ture before the gases enter the exhaust rotameter, R-7.
75
-------
A gas sarr.pling port (GSP) is provided in the suction
line. The pressure at this point is usually close to the
reactor pressure, thus providing ^ positive pressure sample
which is obtained with a gas syringe. Preliminary work has
shown that sulfur compounds can be detected in this sample
using a Varian 3700 gas chromatograph with a dual flame
photometric detector. Such work was barely begun; there-
fore, it is not reported here.
All primary gas flows throughout the entire reactor
system are measured with wide-scale Fisher & Porter rota-
n>eters calibrated within 1% of full scale. Secondary flows,
e.g., heater purge flows, are measured with small Air
Products or f^atheson rotaneters.
The gas flow rate through the suction line is not
measured directly, but is determined as the difference be-
tween the flows through R-7 and R-5. Before a run is made,
V-13 is closed and a mass balance is r.ade (R-5 and R-7 must
measure the same flow) to ensure that there are no system
leaks.
3.5 Lxperimental Procedure for Laminar
Flow Reactor Txperiments
As indicated previously, a detailed experimental pro-
cedure is available in an internal technical report (Agreda,
1979). For the purposes of this report, the following
76
-------
brief summary suffices.
1. Set up cooling water and gas supplies.
2. Start cooling water circulation.
3. Heat reactor to desired temperature as measured by 13.
4. Load coal in feed hopper and purge with nitrogen.
5. Start gas flows and suction system.
6. Start heaters and heat gas until T3 returns to the de-
sired run temperature and reaches thermal equilibrium.
7. Start coal flow.
8. Maintain steady state operation by keeping reactor
temperature, pressure, and flow rates constant.
9. Gather data on flows, pressures, temperatures, and
levels every 10 minutes throughout the run.
10. Shut off feeder system when feed hopper is empty.
11. Turn heaters off.
12. Turn suction system off.
13. Stop all gas flows.
14. Collect chars from cyclone hoppers and weigh them.
15. Clean up and reset system.
16. When entire reactor system is below 50°C, turn cooling
water off.
3.6 Coals and Sample Preparation
Five coals and one coke were used in this study. The
coals used ranged in rank from lignite to anthracite. They
are:
77
-------
1. Beulah-Zap lignite from North Dakota.
2. North Barber No. 8 seam HVC coal, Navajo Mine, New
Mexico.
3. Montana Rosebud subbituminous coal.
4. western Kentucky No. 11 HVB coal.
5. Bottom Red Ash seam anthracite, Pennsylvania.
6. Chemical grade coke manufactured from Western Kentucky
No. 11 HVD coal (at 1600 to 2000QF).
The sample names were codified using the initials of the
coal names: BZN, NB8, MRS, WKll , and BRA.
Information about the coals was obtained from the North
Carolina Research Triangle Institute, Pennsylvania State
University - Coal Research Section, the U.S. Geological
Survey, and the Illinois Geological Survey. The information
obtained does not correspond necessarily to the specific
samples used in this research; nevertheless, such infor-
mation is useful in the interpretation of the findings of
this study. A summary of the information is presented in
Tables 3.1 and 3.2.
The coals were crushed to pass a No. 10 U.S. Standard
sieve using a large mortar and pestle and pulverized using
a Bico pulverizer with ceramic plates and a small porcelain
ball mill. They were then size graded with U.S. Standard
sieves and a mechanical sieve shaker.
78
-------
Table 3.1 Coal Characterization Data (from several sources, see text)
As Received Analyses:
Proximate
Analysis
Ultimate
Analysis *
Sulfur
Forms
Coal Code Rank FSI
Vol.
Moist. Ash Matter
II N
Pyrltic Sulfatic Organic
Beulah Zap
BZN Liq. A 0.0
New Mexico No. 8 NB8 HVC 0.5
Montana Rosebud MRS HVC 0.0
W. Kentucky No. 11 HK11 HVB 2.5
Bottom Red Ash BRA Anthr. 0.5
29.63 6.39 28.57
10.09 18.32 33.80
21.90 8.86 31.56
6.34 15.02 34.67
4.68 5.71 4.90
46.82 6.56 0.73
55.18 4.12 1.21
53.95 6.87 1.20
60.07 4.28 1.75
84.33 1.71 0.81
'0.01 0.02 0.54
0.31 0.00 0.42
0.21 0.17 0.21
2.63 0.14 1.87
0.15 0.00 0.58
excludes moisture
-------
Table 3.2 Elemental Concentrations and Organic Affinity
of Elements in The Rosebud Coal From Montana
(From Fiene et al., 1978)
Raw Coal
I'lenent Org . Aff.
% ppm
Al
Ca
Fe
K
Kg
Na
Ti
Si
LTA
HTA
Organic S
Pyritic S
Sulfate S
Total S
As
B
Ba
Be
Br
Cd
Ce
Co
Cr
Cs
Cu
Dy
t:u
F
Ga
Ge
Hf
Ikj
I
La
Li
Lu
Mn
.18
.82
.02
.02
.97
.88
.15
.06
.12
.07
1.10
.02
.02
.74
.03
1.24
.02
.73
.99
.06
.89
.80
.09
.03
.44
.77
.89
.76
.76
.74
.39
.03
.02
.90
.14
.68
.04
1.15
0.97
0.47
0.079
0.44
0.019
0.05
2.41
12.09
0.62
0.22
0.06
.90
0.69
100
808
0.47
1.6
0.22
10.3
1.2
6.2
0.43
8.8
0.6
3.3
0.90
1.2
0.06
0.3
5.2
14.4
0.06
85
80
-------
Table 3.2 continued
Element
Mo
Ni
P
Pb
Rb
sb
Sc
Ee
Sm
Sn
Sr
Ta
Tb
Te
Th
Tl
U
V
w
Yb
Zn
Zr
Org. Aff.
.83
.64
1.02
.04
.03
.95
.78
.05
.73
.04
.98
.61
.79
-
.56
.11
.58
.60
1.15
.74
.02
.04
Raw Coal
% ppro
7.1
3.1
121
4.6
3.3
1.6
0.93
0.86
8.1
103
0.13
0.11
<1
2.5
0.46
1.5
10.6
0.70
0.25
4.3
31
81
-------
The 325x400 mesh size fractions of MRS, NB8, and BZN
coals were used for the laminar flow reactor (LFR) runs,
and the 200x325 mesh size fractions of all five coals were
used for the batch runs. The particle size distribution
of only one coal (MRS) was determined. The different size
fractions of MRS coal were used to study the variation in
moisture, ash, and sulfur content with particle size. The
coke used in the pilot plant run was 10x80 mesh.
Kobayashi (1976) pointed out that particle size dis-
tribution is highly dependent on the pulverization method.
Reidelbach and Algermissen (1978) showed that the size of
the coal particles has a strong influence on the pyrolysis
time and recommended that pulverized coal not be charac-
terized only by a single mean particle size. Nevertheless,
for the sake of simplicity and because of the screening pro-
cedure used in this study, an average particle size of
41.5 ym was used for the three coals used in the laminar
flow reactor experiments.
No physical properties were determined for the dif-
ferent coals used. Average particle properties, also used
by Kobayashi (1976), were used. They are:
heat capacity - Cp = 0.25 cal/gm °K
density - pp = 1.25 gm/cm2
emissivity - cp = 0.9
thermal conducivity - Xp = 3.0 x 10~ cal/cm-sec'°K.
82
-------
4. DESIGN CALCULATIONS AND DATA REDUCTION EQUATIONS
4.1 Design Calculations for Batch Reactor System
There are only two controlled variables in the batch
system: the furnace temperature, and the residence time
of the coal in the furnace. The coal heating rate is a
function of the furnace temperature and the heat capacity
of the coal and boat.
The primary modes of heat transfer from the heating
elements to the coal are most likely radiation from the
top of the furnace and conduction from the heated boat at
the bottom and sides of the coal bed. The temperature-
time history of the coal was estimated in a crude manner.
It was assumed that the heating and cooling rates of the
coal were approximately the same as those of the surface
of the heated boat. Therefore, an empty boat with a 1/16
inch grounded chromel-alumel thermocouple touching its sur-
face was introduced into the reaction zone of the furnace.
The temperature rise was monitored as a function of time.
The heating and cooling curves are shown in Figure 4.1.
The heating rate is estimated by the equation:
0.95 Tf - T0
83
-------
1400
1300 -
8
16
10 \2 14
Time (mln)
Figure 4.1 Temperature-Time Histories of Batch Samples
18 2O 22 24 2B
-------
where T-. = final temperature
To = initial temperature
tv = time for heating to 95% of Tf.
Using this equation, the heating rates were found to range
from 5°C/sec at 300°C to 45°C/sec at 1200°C.
The time constant for the heating of the coal and boat
is estimated using the equation:
(4-2)
where t = time
T = temperature
TQ = initial temperature
Tf = final temperature
T = boat heating time constant.
Linear regression of t vs. the bracketed term yields TB as
the slope.
The gas velocity through the reaction tube is a function
of the furnace temperature, but the flow was laminar in all
runs. The Reynolds numbers ranged from 2.5 at 300°C to 1.6
at 1200°C.
85
-------
4.2 Design Calculations for Laminar
Flow Reactor
In order to obtain kinetic information on the devola-
tilization reactions from the laminar flow experiments,
temperature-time histories of the injected coal particles
must Le known accurately. However, introduction of coal
particles through a water-cooled injector makes actual
temperature and velocity fields quite coir.plicated.
Boundary layers grow near the walls of the muffle
tube and injector. However, since the gas and muffle tube
wall temperatures are the same, the thermal boundary layer
develops only along the injector. Overisokinetic gas
suction rates in the collector increase the gas velocity,
therefore reducing the particle residence time. In addition,
for larger particle sizes, the free fall velocities of the
particles may be significant. In view of these conside-
rations, the common assumptions that the particles move at
the axial velocity of a fully developed laminar flow profile
and that the reactor can be considered isothermal are in-
adequate.
The following sections describe the approach taken in
the analysis of the laminar flow reactor. A compromise has
been made between the simple assumptions made by Badzioch
and Ilawksley (1970) and Nsakala (1976) and the mathematical
rigor, coupled with experimental realism, used by Kobayashi
(1976) .
86
-------
4.2.1 Particle Velocities and Residence Times
Accurate knowledge of the velocity profile across the
entire furnace tube is not necessary, since the particles
are largely confined to a narrow streamline on the axis.
Because of the two sintered stainless steel disks and
steel wool used in the gas heater, the velocity profile
leaving the second distributor should be flat. The flow in
each hole of the flow straightener has a parabolic velocity
profile, since the maximum Reynolds number in each hole,
for the experimental conditions used, is about 5. Since
the voidage ratio of the honeycomb is approximately 0.5,
the maximum velocity of the parabolic honeycomb flow is
about four times as fast as the average main flow velocity.
The characteristic decay time of the parabolic flows may
be approximated by (Kobayashi, 1976)
Way - *2/v (4'3>
where
£ = spacing of the holes
v = kinematic viscosity of the gas .
At 800°C, this equation yields 0.1 msec as an order
of magnitude estimate of the decay time. The injector tip
was positioned 1.5 inches below the honeycomb to ensure flat
velocity profiles under all experimental conditions.
The discussion that follows is intended to show that a
flat velocity profile may be assumed at the feeder tip
87
-------
position in the reactor. The velocity of the coal feed
carrier gas at the feeder tip is given by
*£t
PfTTdf (4-4)
o
where m,. = feeder gas mass flow rate
P£ = feeder gas density at room temperature
and reactor pressure
df = feeder tip inside diameter.
Typically, the velocity of the cold feeder gas is set to
be greater than that of the main gas stream at the feeder
tip. Therefore, the coal particles enter at a velocity
greater than the main gas centerline velocity; however,
the theory of the fluid dynamics of jets indicates that
the core of the jet emerging from the feeder tip should
disappear within five feeder-tip bore diameters (Goldstein,
1965), i.e., the annular mixing region occupies the whole
jet.
Another factor that must be taken into account is the
density change due to the increase in temperature of the
entering feeder gas stream. The axial velocity of an iso-
thermal round jet in a uniform stream is given by (Pai,
1954)
ux * um_6-5 df/x
uf - u l-0.6(um/u
88
x < 30 df (4-5)
-------
where ux = axial gas velocity
u_ = main stream gas velocity
x = axial distance from feeder tip.
However, the velocity field near the feeder tip is compli-
cated by the formation of a cold boundary layer on the
surface of the water-cooled feeder. The velocity in this
region should be considerably lower than the average main
gas velocity.
In summary, there are three concentric annular regions
at the tip of the feeder: the emerging jet, the cold boun-
dary region, and the hot main gas. The boundary layer
thickness at the injector tip may be approximated by that
of forced convection over a flat plate.
(4-6)
where f> - boundary layer thickness
H = length of injector below honeycomb
Re = ^UjflPn/Mr Reynolds number
p = gas density
y = gas viscosity .
Temperature and velocity profiles in the boundary layer are
approximated by the following equation (Kobayashi, 1976) :
89
-------
_ TK r - rf
urn Tm - Tf 5(4-7)
where r^ = outer radius of feed injector tip
Tf = temperature at feeder tip
Tm = main gas temperature
u = velocity.
For the range of conditions used in this study, the
average velocity in the boundary region is on the order of
one tenth of the feeder gas velocity, and the average tem-
perature is only about 100°C hotter than the feeder gas
temperature.
For all the complexity of the system, when the thick-
ness of the feeder tip walls are considered, the continuity
equation applied, and the gas mixing effect due to the
movement of the coal particles in the feeder gas stream
accounted for, the conclusion is reached that the velocity
profile across the entire furnace tube at the feeder tip
may be considered flat and equal to the average main gas
velocity. At any rate, the uncertainties involved in the
full reactor analysis do not warrant a more complex mathe-
matical analysis.
The time that it takes for the coal particles to de-
celerate from uf to their terminal velocity u can be cal-
culated by making a force balance:
90
-------
2 2
du P g C_ u p 4ird
= 9 - g - ° P g P (4-8)
Pp 2 mp
where u = particle velocity
P
t = time
g = acceleration of gravity
m = particle mass
p = particle density
24
C = —— = drag coefficient assuming Stoke's
p law
Re = particle Reynolds number
d = particle diameter.
Assuming that p « p , the differential equation is solved
yielding:
1 uf ~ g//N
t ~ N An ( up - g/N > (4~9)
where N = li_Hl . {4_10)
From equation (4-9), it is estimated that the time necessary
for the particles used in this study to decelerate to uf is
less than 5 msec and therefore can be assumed to be negli-
gible compared to typical residence times.
In view of the above results, it is assumed that the
velocity of the particles at any point from the feeder to
the collector is given by
91
-------
u = u. + u (4-11)
p t x
where u = particle velocity
u = terminal velocity of particle falling
freely
u = main gas axial velocity.
The velocity distribution of the main gas across the re-
actor is that of a developing laminar flow. Kobayashi
(1976) curve-fitted the numerical results of Langhaar
(1942), yielding the following equations:
=^ = 1 + 0.34Y0*45 , 0 < Y < 6 (4.12)
um
— = 2 - 3.48Y'1'47 , Y > 6 (4-13)
Um
y = 40° ^ raj (4-M)
where u = centerline velocity of developing laminar
flow
u = average main gas velocity
X
x = distance from feeder ti^
D = tube diameter.
Therefore, to calculate the residence time of the particles,
fcR = ut + ux (4-15)
92
-------
I
where t = uncorrected particle residence time
ux = distance-averaged axial gas velocity
z = feeder-collector distance plus collector
mouth-to-throat length
it is necessary to calculate the average axial velocity
from the feeder to the collector
z
uv(x)dx
_
U ~
x ~ z (4-16)
or in terms of the dimensionless variable y defined in
equation (4-14),
y<6 y>6
/ uv(y<6)dy + / u (y>6)dy
u = 2 * fi . (4-17)
y(z)
Substituting from equations (4-12) and (4-13) yields
A — I l
x yTz)
z>6
-0.47
+ 7.4043 y
. (4-18)
To calculate the terminal velocities of the coal par-
ticles, it is necessary to determine their flow regime.
This is done by first calculating (McCabe and Smith, 1967)
93
-------
K = d.
Pg(Pp " Pg}
1/3
(4-19)
If K < 3.3, Stoke"s law applies, and
" V
(4-20)
If 3.3 < K < 43.6, the intermediate law applies, and
ut =
0.71 1.14 0.71
0.153 g d (P - P )
_ P _ _ P ?
0.29 0.43
(4-21)
However, the gas is accelerated by the overisokinetic
suction rate at the collector (necessary to obtain good
char collection efficiencies), so that a correction is
necessary. The following equations are used for this pur-
pose (Kobayashi, 1976) :
(tana) (ri~r0)
r,
tan a
(4-22)
o
m.
rl =
* pg ue
(4-23)
where At = decrease in residence tine due to acceleration
nL = mass flow rate of suction stream at collector
o
P = density of main gas
94
-------
u ^ = axial main gas velocity at collector mouth
entrance
r = radius of collector throat
a. = acute angle that the collector nozzle makes
with a horizontal line passing through its
throat.
In these equations, r, is the radius of a hypothetical ex-
tension of the collector mouth entrance, obtained by assuming
that the mass flov; rate entering the extension at reactor
conditions is the same as the true suction mass flow rate.
Under this assumption, integration of the mass conservation
equation yields equation (4-22) , and the particle residence
time is
t = - __ - At
K u + u ' (4-24)
The gas flow and particle velocity calculations are
carried out for each experimental run with the computer sub
programs shown in Appendix B.I. Subroutine GASPR calcu-
lates the gas properties as a function of temperature and
pressure; subroutine VTER calculates the terminal velocity
(ut) of the coal particles; subroutine VAXL calculates the
average axial gas velocity (u ) and the gas velocity at the
J^
collector mouth (u ) ; and finally, subroutine CORK calcu-
lates the time correction (At) due to the overisokinetic
suction rate.
95
-------
4.2.2 Heating and Cooliny of the Coal Particles
The coal particles are heated in the furnace by
radiation from the furnace walls and by convection from
the gas. The mathematical representation of the unsteady
heating of a coal particle suddenly plunged into hot gas
leads to the following partial differential equation,
assuming spherical symmetry (Reidelbach and Algernisseri,
1978) :
3TPi 1 3 2 3T -
—— - =• . — (r2-AD _-P£) (4-25)
8t CPr 3r P 8r
with boundary conditions
3T .
— = 0 for r = 0 (4-2G)
3r
8T . Ku A s e
— — = - - (T -T ) + _i-P (T4 _ T4 ) (4-27)
3 X . y pi xp w pi
.
for r = R.
where fiu = Nusselt number with respect to the diamotcr
T . = particle temperature
T = reactor wall temperature
T = gas temperature
A = gas thermal conductivity
96
-------
r = radial position
I'^ = averaye radius of i particle size
fraction
s = view factor
c7 = Doltznann constant
t = particle emissivity
X = particle thermal conductivity.
P
It is assumed that the coal is homogeneous and isotropic
and does not swell or shrink during the decomposition pro-
cess.
Rapid devolatilization is probably endothermic; however,
no heat generation term appears in the above equations.
Kobayashi (1976) pointed out that such an effect could be
accounted for, if necessary , by increasing the specific heat
of the coal particle. Furthermore, the enthalpy required to
heat the feeder gas is much larger than the reaction enthal-
py effect, thus making the later effect negligible.
It has been shown (Eadzioch and Hawksley, 1970;
Kobayashi, 1976; llsakala, 1976; Reidelbach and Algermissen,
1978) that heat transfer through the particle boundary layer
is the rate-determining step compared to thermal conduction
inside the particle for particles in the pulverized particle
size range (<200 um) . Hence, the particles can be treated
as spatially isothermal. Kobayashi (1976) found that
radiation heat transfer becomes dominant for particles lar-
ger than 'vlOO um, especially at high temperatures, while
97
-------
convection is dor.inant for small particles (%100 jjm) . Thus
equation (4-25) nay be replaced by an ordinary differential
equation, obtained by substituting (4irR- /3)C pdTn^/dt for
-L p i -*-
31', j_/3r in equation (4-27), and multiplyincj the right-hand
9
side by 4nR^ .
Particles of different sizes are heated at different
rates in the laminar flow reactor. Furthermore, due to con-
vective heat transfer, the temperature of the gas surround-
ing the particles also varies in time anu has to be cal-
culated. Therefore, v/e have as the governing heat transfer
equations (Reidelbach and Algermissen, 1978) :
c CJ
(TC " T} (4"28)
and
3"Pi Vg
— — -
where i=l,2,...,n = particle size class
Cn = gas heat capacity
$ = coal mass flow rate
P
m = gas mass flow rate.
Three thermodynamic parameters are important in this
analysis: reactor wall temperature, initial surrounding gas
98
-------
temperature, and mass flow ratio of coal to gas. Lach of
these parameters has a distinct influence on the temperature
history of a coal particle. The final temperature that can
Lc reached is the reactor wall temperature. Reidelbach and
Algermissen (1978) found that when the initial gas tempera-
ture is not too low compared to the wall temperature, the
convective heat flow is higher than the radiative flow.
The particle temperature at first rises rapidly until ther-
mal equilibrium with the gas is approached, after which a
slower temperature rise up to the wall temperature is ob-
served.
As has been shown, the analysis of the heat transfer
phenomena occurring when a coal particle is suddenly plunged
into hot gas in a reactor is fairly straightforward. In the
laminar flow reactor, however, the coal particles are in-
jected with cold feeder gas through a water-cooled tube.
The gas temperature near the particles increases at a rate
controlled by the mixing and diffusion between the carrier
gas and the hot main stream.
It has been shown by Kobayashi (1976), and implicitly
assumed by Badzioch and I'.awksley (1970) and Nsakala (1970),
that the heating of the coal particles as they leave the
feeder is controlled by the mixing of the feeder gas and
main gas. The characteristic gas mixing time is much lar-
ger than the characteristic convection time to the particles,
99
-------
This implies that detailed analysis of the temperature
fields in the furnace is necessary in order to get an accu-
rate temperature history of the injected coal particles.
Different approaches have been taken to solve, or yet
around, the problem of estimating the temperature history
of the coal particles. Badzioch and Hawksley (1970) de-
fined an isothermal reaction tine
tz = tR - tH (4-30)
where t = isothermal reaction time
tn = residence or transit time
I v
tjT = heating time.
They assumed that no reactions take place during t,j. The
best value of tp which allowed correlation of the data with
their model was found from statistical analysis of the data
tlsakala (1976) used a similar approach. He concluded that,
under certain assumptions such as the pyrolysis activation
energy being greater than 55 kcal/mole, the extent of
pyrolysis during the heating time is negligible. The analy-
sis of Badzioch and Ilawksley's (1970) and Nsakala's (1976)
data by Reidelbach and Algermissen (1978) shows that such
a conclusion is very likely invalid. Furthermore, there is
not one coal pyrolysis reaction, but a very large number of
complex simultaneous reactions with many different acti-
vation energies. Data analysis by the infinite parallel
100
-------
reactions model (Anthony, 1975) typically yields mean acti-
vation energies of 48.7 to 56.3 kcal/mole with standard
deviations ranging from 9.38 to 11.5 kcal/mole (Suuberg,
1978). For purposes of comparison with other researchers'
data, tjj in this study is defined as 95% of the particle's
heating time constant.
Kobayashi (1975) carried out a fairly rigorous analysis
of the temperature and velocity fields in his laminar flow
reactor. Three concentric regions — the center region,
boundary region, and main stream — were considered. All
the coal particles were assumed to remain in the center
region, and the local velocities and temperatures of gas
and particles were assumed to be the same. The momentum,
energy, and mass conservation equations were then solvea
with appropriate boundary conditions and assumptions con-
cerning the coal particles and their properties, shape and
size of the flow regions, and velocity and temperature
gradients at the interfaces. An approximate integral
method was used to simplify the analysis, which then
yielded a family of curves describing particle tempera-
ture as a function of distance from the feeder, time, and
the ratio of the momentum shape factor to the energy shape
factor (for a given set of reactor conditions)
101
-------
where K,. = momentum shape factor
KE = energy shape factor.
A theoretical analysis showed that 0 should lie between
1 and 3. Kobayashi (1976) used an indirect method based
on the observed weight losses at different temperatures
to determine the proper value. If a coal shows a given
asymptotic weight loss at a given temperature, it is assumed
that when the particle reaches this weight loss while being
heated to a higher temperature, the particle temperature at
that time is the same as the temperature at which the
initial asymptotic weight loss was determined. Thus the
coal weight loss itself provides the required temperature
measurements. The appropriate value of 6 was found to be
3.
Kobayashi (1976) pointed out that for practical pur-
poses, the temperature-time histories of the coal particles
may be approximated by the following exponential curve
T - T
TP . Tf = 1 - e-t TH (4_32)
R f
where T,: = feeder gas temperature
TR = reactor temperature
T,, = heating time constant.
102
-------
lie then used his mathematical analysis to generate values
of T r for 6=3, as a function of temperature, main gas
velocity, and carrier gas velocity. The calculated values
of T., were then used with equation (4-32) to calculate the
temperature rise of the coal particles.
Examination of Kobayashi's data for T.T suggests a
strong correlation between TJ,/ reactor temperature and
feeder gas velocity, whereas the main gas velocity does
not appear to influence T.. appreciably. Least squares mul-
tiple linear regression yields the following empirical fit
between T , T, and u^i
rn = exP [5.67238 - ^QQQQ R - 0.5348 in(uf)] (4-33)
where TJ, = particle heating time constant in milli-
seconds
TR = reactor temperature in degrees Kelvin
u,. = feeder gas velocity, at feeder tip, in
cm/sec.
2
The coefficient of determination for this fit is r =0.9941.
Equation (4-33) is used to calculate T., in the subsequent
analysis. The finding that um does not have an appreciable
effect on TJ, may be a consequence of the shielding effect of
the annular boundary region that develops around the feeder,
as discussed in Section 4.2.1.
Equation (4-29) shows that the coal mass flow rate
affects the heating of coal particles by increasing the
103
-------
enthalpy requirement. Furthermore, if large clouds of coal
are fed, the view factor S is reduced. Both effects tend
to reduce the heating rate of the coal particles and there-
fore the rate of devolatilization. On the other hand,
large numbers of particles in the laminar jet leaving the
feeder tend to make the flow turbulent. A turbulent jet
exiting the feeder could disturb the boundary region,
leading to faster mixing of the cold feeder gas with the
hot main gas and correspondingly faster heating rates, and
hence to higher devolatilization rates. Badzioch and
Hawksley (1970) found experimentally that the rate of de-
volatilization is independent of coal feed rate up to
0.5 g/min. The feed rates employed in their experiments
ranged from 0.25 to 0.5 g/min for feeder gas flow rates of
1 to 2 fc/min. Kobayashi (1976) used coal feed rates ranging
from 0.01 to 0.38 g/min for carrier gas flow rates ranging
from 15 to 608 cc/min. Nsakala (1976) used coal feed
rates ranging from 0.5 to 0.6 g/min for a feeder gas flow
rate of 1.98 £/min. Coal feed rates used in this study
range from 0.3 to 3 g/min for a feeder gas flow rate of
1.0 £/min.
The theoretical analysis of Reidelbach and Algermissen
(1978) shows that a single particle size cannot be assumed
for large size fractions of pulverized coal but that
equations (4-28) and (4-29) should be used. However, for
104
-------
the size fraction used in the present study (325x400 nesh) ,
their analysis suggests that the average particle size for
the fraction may be used as representative of the entire
size fraction.
In the reactor system used in this research, the
pyrolysis reactions are quenched with a water-cooled col-
lector. Heat is removed from the coal particles by con-
vection to the gas, radiation to the collector wall, and,
to a smaller extent, conduction by collision with the
collector wall.
The gas temperature decrease in the collector is cal-
culated by an iterative procedure (Howard, 1965) . For a
small section, Ax, in the collector
-1
T = T + (T-T) {£n[(T-T)/(T2-Tw)3} (4-34)
m d
— = L-|-^ (Ax)]*n[(T1-Tw)/
-------
The procedure is as follows:
1. Choose T, and ?2/ one of which must be at a known
point (e.g., T^ = Tm at probe entrance).
2. Calculate f.
3. Calculate h/Cg.
4. Calculate Ax to locate interval
5. Use previous ^2 as tne new TI and repeat the
procedure.
It has already been indicated that the rate-determining
step in the heating or cooling of coal particles is the
heating or cooling of the carrier gas. The high velocity,
turbulent flow in the collector leads to high heat trans-
fer rates between the gas and the particles. The particle
cooling rate calculated with the equations given above for
the collector probe and suction rates used in this study is
greater than 105 °C/sec. The quenching of the pyrolysis
reactions within a few milliseconds is thus ensured. These
conclusions are consistent with the findings of Badzioch
and Hawksley (1970) and I.sakala (1076).
Finally, the reactor temperature must be measured.
Suction pyrometers are typically used to measure gas tem-
peratures. Ladzioch and Hawksley (1970) and Nsakala (1976)
measured the axial temperature profile of the gas in their
reactors with such a probe. However, heat is transferred
to the coal particles by convection from the gas and by
106
-------
radiation from the wall. To avoid ambiguity, the gas tem-
perature and the reactor wall temperature should he the
same. This may not have been the case for the reactor
used by Nsakala (1976), in which the reactor walls were
at a higher temperature (about 100°c higher) than the
treasured gas temperature. Reidelbach and Algermissen
(1978) have shown that radiation heat transfer may be nore
important than previously thought because of the large
specific surface of small coal particles. Therefore, if
the gas and wall temperatures are not the same, the tem-
perature that should be associated with the experiment be-
comes uncertain.
The energy balance on a thermocouple in the laminar
flow reactor is
= hTAT(Tm - TT) + ET°AT(TW ~ TJ) (4-37)
where q^ = heat flow to the thermocouple
hT = convective heat transfer coefficient
AT = surface area of thermocouple
T™ = thermocouple temperature
£„, = thermocouple emissivity
o = Boltznan constant
T = raain gas temperature.
107
-------
Assuming that the grounded tip of a 1/16 inch thermo-
couple can be treated as a small sphere, the following
correlation can be usea (McAdanis, 1953)
hT = _£ 0.37 (Re)0'6 ; 25 < Re < 1Q5 (4-38)
r
where Re =
d = thermocouple characteristic diameter.
Therefore, assuming thermal equilibrium (q =0) , equation
(4-37) becomes:
Tm = TT + UTodT/0.37Xg) (TT - Tj (p/pu^J . (4-37)
The thermocouple emissivity may be assumed to have a value
of 0.22 (Kaskan, 1956). Therefore, if T and u^ are known,
the reactor gas temperature (Tm) can be estimated.
The procedure followed during most of the runs in this
study (also followed by Badzioch and Hawksley, 1970 and
Kobayashi, 1976) was to set TW = T with no gas flow,
assuming that h_, was small and Tm for the still gas inside
the furnace was close to T . Hence, when the gas flows are
Yt
started, TT drops to some intermediate temperature between
T.. and T . Heat is added to the gas until TT returns to
w
the original T reading. At that point, T should equal
w IU
108
-------
Tw (i.e., qT = 0). This may not be exactly true because
some spots in the cold ends of the reactor may change in
temperature, relative to the original measurements, as the
heated yas passes by them.
Sainsbury e_t al. (1966) studied this problem. Their
experimental results show that, when the procedure outlined
above is followed, the difference in the readings obtained
with a thermocouple and a suction pyrometer differ from
10 to 25°C, with the thermocouple always having the higher
temperature. Their study did not account for the heat trans-
fer between the reactor walls and the water-cooled collec-
tor as a function of feeder-collector distance, however,
this is a phenomenon that can cause the temperature profile
of the furnace walls to vary appreciably. Hence, it is
clear that there is at least a 5 to 12.5°C uncertainty in
the reactor temperature, under the best circumstances.
4.3 Coal Composition and Height
Loss Variables
For the purposes of this study, coal is taken to con-
tain three major fractions: moisture (M), ash (A), and
volatiles (V). V should not be confused with the ASTM
volatile matter (VM). Each coal element is expressed as
a fraction, by weight, of the coal according to the
following equation;
109
-------
weight of element in char
— (4_39\
as-received weight of char .
Different samples of a given type of coal may have
slightly different moisture and ash contents; furthermore,
the moisture content of a coal sample may change depending
on the environmental conditions to which the sample is sub-
jected. Therefore, it is sometimes advantageous to express
the weight fractions of coal components on moisture-free
(m.f.) and dry-ash-free (d.a.f.) bases, in addition to the
as-received (a.r.) basis. For the sake of simplicity, the
following nomenclature is used: superscript "+" denotes
as-received mass fraction (i.e., including moisture and
ash); superscript "T" denotes moisture-free mass fraction;
and superscript "*" denotes dry-ash-free mass fraction.
The mass fractions calculated on different bases are re-
lated to one another as follows:
t X v/eight of element
X = 1-K+ '' weight of dry char (4-40)
* X v/eight of element
X = -
_
-A^-M"1" ' weight of dry ash free char
X*
and X* = — JT - weight of element - (4_42)
1-A weight of dry ash free char .
110
-------
Throughout this report, subscripts are used as follows:
"C" denotes feed coal, "H" denotes char, "V" denotes
volatiles, and "A" denotes ash (e.g. X , X , etc.).
When WQ mass units of coal are pyrolyzed, W.. mass
units of char are produced. Each has its own moisture,
ash, and volatiles fractional content. The fractional
a.r. weight loss is given by:
.,+ .,+
AVI"1
(4-43)
The fractional m.f. weight loss is given by
AW =
or
t W
AW = 1 - -Ji
(4-44)
1-M
H
(4-45)
In the same manner, the d.a.f. fractional weight loss is
calculated as
Aw* =
or
AW = 1 -
w;
1-Mj-Aj
(4-46)
(4-47)
111
-------
All mass fractions can be expressed as percentages
simply by multiplying by 100. Throughout this thesis, mass
fractions and weight losses expressed as percentages will
* ±
be preceded by the symbol " £ " (e.g., £AC, %ATI;T, etc.).
The extent to which an element is retained in the
char can be expressed in different ways. Chemical analyses
yield the mass fractions of an element in the feed coal
(Xc) and in the char (X.j) . The mass fraction of that
element in the char is normalized to mass of element in
char per unit nass of feed coal with the following equation:
* = X (1-AVM - freight of element in char
VH V1 AM ' ( weight of feed coal > (4-48)
The fraction of that element retained in the pyrolyzed
coal is therefore:
weight of element in char _
; (weight of element in feed coal'
where 4> = fractional retention of element in char.
1 1
The elemental loss expressed as a fraction of its initial
weight in the feed coal is
AO - IY ,i, \/y • freight loss of element
A"ll ~ (XC~ H C ' (weight of element in feed5 (4-50)
coal
One advantage of $ and Afi is that they do not depend on
the basis, a.r., m.f., or d.a.f., in which X and \\>^ are
112
-------
fxprcsscd (Xp and \p must be in the same basis, of course) .
A disadvantage is that they do not show the actual mass
fractions upon which they are based. This makes it diffi-
cult to compare intelligently the relative retentions in
two coals which have very different mass fractions of a
given element.
4.4 Determination of VJeight Loss
V7eight loss can be determined directly by weighing the
feed coal and the resultant char, and using equation (4-43).
This procedure is used for the batch runs.
The determination of weight loss by direct weight
measurements is usually not feasible in laminar flow re-
actors v/ith water-cooled collectors, since a significant
fraction of the char tends to miss the collector mouth or
stick to the collector walls. Therefore, an indirect
method is required.
If a coal component is not released from the coal par-
ticles during devolatilization, it can be used as a tracer
for the indirect calculation of weight loss. The tracer
balance on an a.r. basis is:
YV = yjjw; (4-51)
where ^
y£ = a.r. tracer mass fraction in coal
Y?t = a.r. tracer mass fraction in char
113
-------
Combining with equation (4-43) yields
Y
H
It follows that
or
f y
AI; = l - _c
Y+
AV; =
Y,
(4-52)
(4-53)
(4-54)
and
or
AW = 1 - —r
(4-55)
(4-56)
or
AV7 = 1 -
y,
Y
H
(4-57)
It is obvious that, for the tracer method to be
effective, the tracer must have certain properties:
1. It must have zero volatility during pyrolysis.
2. it must be homogeneously dispersed throughout the coal
and char samples.
3. It must be amenable to highly accurate analysis.
114
-------
The first requirement rules out all major elements and
many volatile trace and minor elements. The third require-
ment probably rules out all trace and most minor elements.
Among the major fractional components of coal, moisture
and volatiles are immediately ruled out again by the first
requirement. This leaves the inorganic mineral matter in
the coal, or the ash.
Many investigators have used ash as a tracer in de-
termining weight losses of pulverized coals (Ksakala, 1976;
Kobayashi, 1976; Stickler et al., 1974; Eadzioch and
Ilawksley, 1970; Howard and Essenhigh, 1967). When ash is
used as the tracer, equation (4-56) reduces to
t
AW* = 1 -
A
1-A
£ (4-58)
1-A1
All researchers in the field have used the residue
left after ignition of the coals and chars at 700 to 750°C
(as in the ASTM ash test or its European equivalent) as
the measure of the ash. All researchers except Kobayashi
(1976) used the ash content (on an m.f. basis) to estimate
the d.a.f. weight loss using equation (4-58).
0'Gorman and Walker (1969) found from their studies of
mineral matter characteristics of American coals of various
rank that there is a significant loss of C02 from calcite
(CaCO3) at temperatures between 800 and 900°C, and of I^O
115
-------
fror. clay [specifically kaolinite (A^C^- 2Si02 • 2H20)1 at
500°C. Since most laminar flow reactors are operated at
temperatures in the range 800-1500°c, it is possible that
ash determinations of the feed and chars done at 750°C
could lead to errors if ash is used as a tracer. For
this reason, in the present study, ash content has been
determined at 950°C which is 50°C higher than the highest
temperature in any laminar flow reactor run.
The raw data of Badzioch e_t al. (1968) show consis-
tently negative d.a.f. weight losses (i.e., they show
d.a.f. weight gains) at low temperatures for most of the
coals in their study. Particularly in the case of low rank
coals, there appears to be a pattern whereby the weight gain
decreases as the temperature increases for a given residence
time. They indicated that such results were due to analyti-
cal errors in the determination of ash and to the scatter
that occurs because the ash particles are, to a great extent,
discrete from coal and tend to segregate on handling, e.g.,
in the vibrating feeder. Thus, samples collected after
passing through the furnace inevitably have a more widely
scattered ash yield than replicate samples of the feed coal.
Of all the investigators that have used ash as a tracer,
only Kobayashi (1976) examined in depth the accuracy of the
ash tracer method. However, his study covered only the
high temperature range of coal devolatilization (1000 to
2100°K). Good particle recoveries, achieved as a consequence
116
-------
of the well-designed collector and short residence tines
studied, enabled Kobayashi to compare weight loss directly
measured with that calculated by the use of ash as a tracer.
His results, coupled with an extensive study by Padia (1976)
on the behavior of coal ash under high temperature con-
ditions, enabled him to make a quantitative determination
of the extent of the error caused by the use of ash as a
tracer.
According to Kobayashi (1976) , possible problems
associated with the method are:
1. The percentages of ash in different size fractions
of pulverized coals vary in a manner that depends on
the method of classification.
2. Some of the ash particles are not embedded in the coal
particles but exist as separate entities.
3. The original mineral matter undergo different chemical
reactions upon heating. Hence, pretreatir.ents of coal
at temperatures even lower than the ashing tempera-
ture could cause errors in estimating weight losses.
4. Significant loss of ash occurs at high temperatures
due to vaporization and decomposition. Furthermore,
interactions between ash and carbon could contribute
to some of the weight losses of coal.
The first and second points are important in relation
to the sampling techniques used for char. Inertial
117
-------
.separation devices such as cyclones preferentially collect
larger and heavier particles, which may have different ash
contents compared to the average ash content of the original
coal. Such problems become more serious when a significant
fraction of the ash exists as separate particles, as has
been observed by Littlejohn (19G6).
Kobayashi (1976) used a bronze filter to collect the
char particles, and water jets at the collector mouth to
quench the reactions. Nevertheless, ash losses were still
significant because the high flow of quenching water
appeared to have forced some of the fine particles through
the bronze filters, which resulted in a significant under-
estimation of weitjht losses by the ash tracer method at low
decomposition rates. However, Kobayashi also reports that
even though the coals used in his research were size-
graded, under microscopic observation they showed some
smaller particles than the size grade should have. There-
fore, the direct measurement weight loss determinations
could have been biased (probably only slightly) towards
higher weight losses since small coal particles could also
have been lost through the filter.
Separate tests done by Kobayashi under simulated flow
conditions at room temperature revealed that the fraction
of ash in the coal samples from the test experiment was
about 15% lower than that of the original ash content.
Therefore, it was recommended that the ash fraction of coal
118
-------
used in the ash tracer method should be that of coal par-
ticles which have experienced the same sampling process at
room temperature as they would in the real experiment.
Kobayashi attempted to explain the experimental
observations that weight losses of a lignite by the ash
tracer method were about 5% lower than those measured di-
rectly, even when the devolatilization temperatures were
lower than the ashing temperature. He postulated (based
on the work of Padia, 1976) that this phenomenon was the
result of differences in CaSO. formation between ashes that
4
are produced from coals preheated in an inert atmosphere
and those obtained without devolatilization. Padia (1976)
conducted an extensive study of the behavior of ash under
pyrolysis and combustion conditions, and he concluded that
the ash produced by the ASTM method from chars pyrolyzed
in an inert atmosphere should weigh 4.75% less than the
ash produced from raw coal.
The fourth point does not pertain to the range of tem-
peratures used in this research. Kobayashi observed sig-
nificant ash losses due to vaporization and decomposition
only for temperatures above 1250°K.
Kobayashi found that in addition to ash losses, other
important sources of error or bias in weight loss deter-
minations in laminar flow and free fall reactors are par-
ticle losses (due to particles missing collectors or going
119
-------
through filters and cyclones) and the formation of soot and
tar. Particle losses cause high direct weight loss deter-
minations, v;hile soot and tar formation on the particles
themselves or on the collection devices cause low direct
weight loss determinations and compound the problem of ash
losses by "diluting" the ash. The following equation was
proposed to calculate the fractional a.r. weight loss of
coal during pyrolysis:
A/ . '"c - "R - "PL - "AL - "ST (4_59)
"c - KPL
,T+ + W+ (4-60)
H ST
" =
where V7_~ = weight of collected residue
V7* = weight of particle losses
PL
^AL = wei
-------
systematic examination of the effect of the ash content of
the coal and errors in the ash analysis on the accuracy of
the method, coupled with an experimental study of the prob-
lem/ could be quite useful. Experiments at low tempera-
tures would complement the work of Kobayashi (1976) .
121
-------
5. ANALYSIS OF RLSULTS FROM BATCH EXPERIMENTS
The results presented and discussed in this section
provide information pertaining to the devolatilization
behavior of five different coals at or near thermal equili-
brium. The five coals, described and characterized in
Section 3.6, range in rank from a lignite to an anthracite.
5.1 Preliminary Experiments
Data from batch experiments with extremely long re-
action times (hours or days) may not be applicable as a
direct extension of laminar flow reactor experimental data.
The objectives of these experiments were to determine the
minimum time required for devolatilization reactions and
elemental release to reach equilibrium in the batch reactor
and to qualitatively assess observed pyrolysis phenomena in
the same reactor.
Two parameters were monitored during each run - AW
and AS (see equations (4-43) and (4-50)). The runs were
performed using small unglazed porcelain boats loaded with
100 mg of the 200x325 mesh size fractions of the coals
studied. The moisture and ash content of these coals are
shown in Table 5.1. The weight loss curves obtained are
shown in Figure 5.1. Comparison of this figure with
Figure 4.1 suggests that there is a significant lag between
122
-------
Table 5.1 Moisture and ash content of batch
coals
Mesh = 200x325
Coal
BZN
NB8
MRS
WKll
BRA
% M+
28.14
10.33
15.90
5.07
2.22
%A+
5.55
19.60
7.64
5.59
9.14
123
-------
-
0
>o
£^
CO
in
O
_J
.c
0*
«
o:
<
BU
60
40
20
60
40
20
60
40
20
C /"\
oU
40
20
20
n
8D
o
* *
O
A
„
^
a a
-° x
0
A ^
Q.
Q y Q
O
—* A
>^
a a
o o
_o
A A A
a
S f ?
n ,=
^ _
A A -
A-400°C
^-600°C -
O-800°C
D-IOOO°C BZN
9s
^7
—
A A -
NB8
a
§ a
—
A _
MRS
a a
o ®
A A
WKII
a a ~
^ ^BRA
0
0
o
\j
5 10 15
Residence Time (min)
Figure 5.1 Weight Loss in Transient
Batch Experiments
20
124
-------
weight losses and the estimated temperature rise of the
coal beds. First-order pyrolysis time constants can be
estimated using the following equation:
AW+
- — )] (5-1)
where AW* = a.r. weight loss as t •* °°
T. + = devolatilization time constant.
Typical values of T w+ are 22 sec (NB8, 600°C) and 63 sec
(WK11), 1000°C) which may be compared with typical Tfi
values (calculated using equation (4-2)) of 0.4 sec at 600°C
and 0.1 sec at 1000°C. The comparison suggests that the
actual temperature rise of the coal bed may have been
slower than was estimated. Furthermore, the reaction times
reported in this section do not include the cooling time.
Because of these uncertainties, no attempt was made to es-
tablish any kinetic parameters from these data.
Nevertheless, qualitative analysis of the data yields
interesting results. All coals exhibit a dramatic increase
in AW+ between 400 and 600°C. Furthermore, it appears that
equilibrium is reached much more rapidly at the higher tem-
peratures. These observations are consistent with the data
of Kobayashi (1976) and other researchers.
The results of the transient sulfur analyses are shown
in Figure 5.2. AS was calculated using equation (4-50).
125
-------
in
O
a
•^
c.
E
a>
1 3
50
25
50
25
50
25
75
50
25
25
0
O
_D
J^
O
_Q
-o
D
O
y\
*".*.
^r
a
.0
A
0
1
0
*
D
A
O
V
a
A
9
*
A
£
A
I
5
1 1
8 8
A A
Q 9
a a
A A
®w
^1
° D
A A
8 B
A A
*-6<
* O-8
D-IC
9®
u
A A
i i
10 15
'
fi "
A -
BZN
©
a
—
NB8
8 -
-
AMRS
1
A ~
WKII
)0°C
DO°C
oo°c
)00°C
@
A
1 BRA
20
Residence Time (min )
Figure 5.2 Sulfur Loss in Transient Batch Experiments
126
-------
The plots show clearly the same trends as do those of AK"1",
indicating that during the devolatilization stage of coal
gasification, the rate of sulfur release is proportional
to the rate of volatiles released. The high correlation
between AS and AW+ shown in Figure 5.3, holds for all
coals studied. The intercept at the abscissa is close in
each case to the moisture content of the feed coal (%M+),
indicating that little or no sulfur evolves until the coal
moisture has been driven off.
An unexpected result is shown in Figure 5.2. For each
coal, the sulfur losses at 1000°C are lower than those at
8000C and in some cases at 600°C. This finding might be
attributable to systematic experimental error, but its
consistency from coal to coal and the rapid initial sulfur
loss shown during the transient heating period appear to
indicate the contrary.
The data for BRA coal are consistently scattered and
erratic. However, BRA is an anthracitic coal, and it shows
small devolatilization and sulfur losses. Because of the
small changes observed, the errors inherent in the experi-
ments and the chemical analyses are likely to propagate,
yielding the observed scatter in the data.
Finally, a reaction tine of 20 minutes was chosen for
subsequent experiments. After 20 minutes, devolatilization
and sulfur losses appear to have reached equilibrium at
127
-------
5
0
>0
tf^
V)
m
O
_|
o
emeni
Ul
f 3
50
/5 c
25
75
50
25
75
r ^
50
25
7*i
f *j
50
25
50
25
0
I 1 I I t 1 I
A- 400°C
*-600°C f\
O-eoo°c w 133O "
D-iooo°c o * a n
3-3O A an
%M = 28.14 ^A
/\ OtN
* °^
O
- —
%M*= 10.33
/Vi ^^ NB8
«6^
3
o
%M = I5.90
/\
A)fA ^ MRS
r\T©
**5 * *
%M= 5.07 Q
A
A^ WKII
-**
C&£ %M=2.22
i i 1 i 1 1 1 8RA
0
o
\J
0
0 10 20 30 40 50 60 70 80
A.R. Weight Loss (°/oi
Figure 5.3 Correlation Between Sulfur and A.R. Weight Loss
in Transient Batch Experiments
128
-------
temperatures greater than or equal to 600°C and to closely
approach equilibrium at 400°C.
The data from the experiments described above and some
statistical analyses are given in Appendix B.I.
5.2 Equilibrium Batch Experiments
All equilibrium batch runs were performed with pyroly-
sis times of 20 minutes, using the 200x325 mesh size frac-
tions of five feed coals. The moisture and ash contents
of the feed coals are given in Table 5.1. Glazed porcelain
boats were loaded with 2.0 grams of coal for each experi-
ment. Each experiment was repeated as many times as was
necessary to obtain the minimum amount of char (3 grams)
required for analysis. Typically, three coal batches were
pyrolyzed at each temperature. The exceptions were the 300°C
runs where only one batch of each coal was pyrolyzed (the
chars from these runs were analyzed only for Pb, Ilg, and
S). Weight losses and elemental retentions were determined
as functions of temperature at essentially thermal equili-
brium conditions. The results are analyzed in the following
sections.
5.2.1 Analysis of Weight Loss Results
A summary of run conditions and a.r. weight loss (AW+)
results is given in Table 5.2. The reproducibility of the
129
-------
Table 5.2 Equilibrium Batch Weight Loss Experiments
Run Mo.
B-l
B-2
D-3
B-4
B-5
B-6
n-7
B-8
B-9
n-io
B-ll
n-12
D-13
B-14
B-15
B-16
B-17
D-18
B-19
B-20
B-21
B-22
B-23
B-24
B-25
B-26
B-27
R-28
B-29
B-30
Coal
BZN
WK11
NB8
BRA
MRS
BRA
NB8
MRS
BZN
WK11
BRA
NB8
MRS
BZN
WKll
BRA
NB8
MRS
BZN
WK11
BRA
NB8
BZN
MRS
WK11
BRA
NB8
BZN
V7K11
MRS
T
°C
400
400
400
400
400
600
600
600
600
600
800
800
800
800
800
1000
1000
1000
1000
1000
1200
1200
1200
1200
1200
300
300
300
300
300
%AW+
40.81
16.63
18.15
2.77
24.79
3.34
36.36
41.45
52.68
39.70
4.45
41.34
47.81
63.57
44.98
6.11
44.93
51.72
76.19
47.73
7.04
51.29
89.87
64.27
49.29
2.66
11.66
31.32
6.55
18.40
S%AW+
0.25
0.36
0.14
0.03
0.89
0.08
0.15
0.03
0.07
0.47
0.21
0.23
0.70
1.78
0.25
0.02
0.35
0.84
2.63
0.08
0.02
0.85
2.62
0.76
0.11
no reps
no reps
no reps
no reps
no reps
130
-------
weight loss data was extremely good, as evidenced by the
standard deviations shown in the table. The average de-
volatilization a.r. v/eight losses are shown in Figure 5.4.
Weight losses calculated in m.f. and d.a.f. bases are listed
in Appendix B.I.
The weight loss data shown in Figure 5.4 depict the
characteristic devolatilization bahavior of coals pyrolyzed
in batch reactors. Moisture evolution occurs at 100°C,
devolatilization begins at about 350°C, and most of the
weight loss occurs between 400 and 750°C. The lignite
(BZN) exhibits the largest devolatilization weight loss, and
the anthracite (BRA) exhibits the smallest. Subbituminous
and bituminous coals (NB8, MRS, and WKll) exhibit inter-
mediate weight losses.
Data obtained by Suuberg et al. (1978) for the pyroly-
sis of Montana lignite in a strip wire-screen batch reactor
are shown in Figure 5.5. These data are quite similar to
the a.r. weight loss data of the MRS coal shown in Figure
5.4. The same pattern is apparent: moisture evolution
occurs first, devolatilization begins at about 350°C, and
most of the weight loss occurs between 400 and 750°C.
However, quantitative comparison of a.r. weight losses is
difficult because of the differing moisture and ash con-
tents of the feed coals used in the two studies. Dry ash-
free weight losses calculated from the data of Suuberg et al
131
-------
CO
ro
80
1C
0^
CO
V)
o
~ 40
o>
"J
3!
«' 2°
n
X =
O =
A =
D =
- *-
i i i i i i
BEULAH ZAP NORTH DAKOTA LIGNITE
WESTERN KENTUCKY-/?* II x
NORTH BARBER ^8 NEW MEXICO
MONTANA ROSEBUD SUBBITUMINOUS
BOTTOM RED ASH ANTHRACITE
JR=20.0 minutes
-
X
D
X D
x D A
§o
J(
*—*
x
^
X
D
D o .
A
0 0 ...***
*
i * 'f T 1 i i
200 400 600 800
Temperature ( °C)
1000
1200
1400
Figure 5,4 A.R. Weight Loss in Equilibrium Batch Experiments
-------
CO
co
50
40
o
>S 30
M
O
_J
£ 20
or
< "0
0
cP
o
0
°
0
o
200
400
600
800
1000
1200
1400
Peak Temperature Of Run (°C)
Figure 5.5 Weight Loss of Partially Dried Montana Lignite
From Suu berg etal. (1978)
-------
(1978) and from this study are compared in Figure 5.6. The
agreement is good considering that the two experiments are
fundamentally different. Suuberg'n experiments were
carried out with short residence times (about one second)
and high heating rates (10 °C/sec) (i.e. , fast pyrolysis
in a batch reactor), while this study's batch experiments
were carried out with long residence times (20 minutes) and
low heating rates (5 to 45°C/sec). It will be shown in a
later section that the lower batch fast pyrolysis values of
*
AW in Figure 5.6 are consistent with the LFR fast pyroly-
sis results of this investigation and with the findings
of other researchers.
The a.r. weight loss values at 450 and 700°C in Figure
5.4 agree qualitatively with those found by Kuhn e_t a_l.
(1977) for coals of similar ranks as those used in this
study (see Table 2.1). The residence times and heating
rates used by Kuhn ejt al. were comparable to those used in
this study for the batch experiments. Unfortunately, Kuhn
et al. do not report the moisture and ash content of the
coals in their study: therefore, the precise calculation of
their d.a.f. weight loss values for comparison purposes is
not possible.
Comparison of the batch equilibrium a.r. weight losses
with the asymptotic weight losses at each temperature in
the transient batch experiments reveals that WK11 and BRA
134
-------
OJ
tn
M
M
O
8O
60
40
20
D = data from Suuberg et ol (1978)
O = this study
0
r)
o i rP
CP
i
200 400 600 800
Temperature (°C)
1000 1200
Figure 5.6 Comparison of O.A.F. Weight Loss Data for MRS Coal
-------
coals had larger weight losses in the transient experiments
This effect is probably due in part to the change in bed
depth from several millimeters in the transient experi-
ments to over one centimeter in the equilibrium experi-
ments. This is consistent with that observed by many re-
searchers (e.g., Kobayashi, 1976; Nsakala, 1976) for coals
with ranks higher than or equal to bituminous (such as WK11
and BRA).
The d.a.f. weight loss (AW ) versus temperature curve
for each coal was curve-fitted with a fourth-order poly-
nomial. The resulting equations were used in the develop-
ment of a kinetic model for devolatilization in the laminar
flow reactor, as will be described in a later section. The
Statistical Analysis System (SAS) computer program used for
the curve-fitting is given in Appendix A.I. The statisti-
cal analysis of the regression and comparisons of the curve-
fits with the experimental data are in Appendix B.I.
5.2.2 Analysis of Elemental Release Results
Fifteen trace and minor elements were analyzed in
the feed coals and chars produced in these experiments.
In order to determine the behavior of trace elements as a
function of temperature, a simple statistical analysis was
carried out. The quantities XH, 4H» and A^n on as-
136
-------
received and moisture free bases (see Section 4.3) were
calculated for each coal and char elemental mass fraction,
and were then linearly regressed against T. Coefficients
of determination and Student's-t values (used to determine
whether the slope is significantly different from zero)
were also calculated. All calculations were done with
computer program BATCH, shown in Appendix A.I. The cal-
culated data are given in Appendix B.I.
The significant element-temperature correlations are
summarized in Table 5.3. The variations of the different
parameters (X , ^H, and 4>H) with T can be used to infer
certain types of elemental devolatilization behavior.
The interpretation of the regression results are as follows:
1. If X. versus T has a slope significantly greater than
zero (indicating that the char is becoming progres-
sively enriched in the element) , and i|>H versus T has
a slope equal to zero (indicating that the absolute
amount of element in the char is remaining constant),
it is deduced that the element is completely retained
in the char.
2. if the slope of X., versus T is greater than zero, and
t
that of i|> versus T is less than zero, the element's
H
release from the coal is significant. Furthermore, the
element is released at a proportionately lower rate
than the total volatile matter.
137
-------
Table 5.3 Significant Element-Temperature Correlations for Batch Pyrolysls
oo
oo
Coal
Parameter
Element
Sm
Sb
As
Se
V
Ci
La
Th
Cr
Sc
Fe
Co
Hg
Pb
S
NB8
X*
*+*
*
+
Xy
+^
+
XX
-r-
XX
X
XX
X
V*
*
XX
xx
0H
A
A
A
*
xx
**
WKII
X*
4-
^f
^
-f
9f
-*
X-^f
-<-
**
H-
*
*#
Y*
•K-
"i
-**
X-^-
0H
A
A
A
A
^
**-
•xlf
BZN
X*
+*
+
-x-
*
If
A
-f-
*
V*
XX
*
-K-
^•)f
**
0H
A
A
*x
A
*
X
XX-
XX
BRA
X*
X-
X
Y*
X
*"*•
0H
A
A
X
A
A
XX
MRS
X*
X
+
X
•+•
X
+
X
-f
X-
X
+
X
XX
Y*
X
X
XX
XX
0H
X
X-
A
XX
K\L
XT
X = significant^ 95%C.L.
XX = significant® 99%C.L.
A = undetected trend
+- = positively correlated
(slope >0 )
— = negatively correlated
(slope < 0)
-------
f
3. If the slope of X versus T is equal to zero, and that
of (j; versus T is less than zero, the element" s release
is significant and occurs at the same rate as AW
versus T.
4. If the slope of X.. versus T is less than zero, and
that of ty versus T is also less than zero, the
element's release from the coal is significant and
the element is released at a higher proportional rate
than the total volatile matter.
5. If the slope of (J> versus temperature is equal to zero
(indicating that no appreciable amounts of that element
have been released), the element is not volatile. If
it is less than zero, the element is volatile. The
behavior of simply mirrors that of \|» , but, more
H rl
importantly, it allows easy visual examination of the
data. Since (f> must range from 1.0 (complete retention)
to 0.0 (no retention), its value quickly indicates
whether the element is retained entirely in the char
or not. This is useful because the linear regression
analyses only detect consistent trends. If a retention
(jj) drops at low temperatures and then remains at a
constant value, no linear trend would be found;
nevertheless, it would be obvious that the element
had been released from the coal.
6. If no significant correlations are found for a given
element, it must be concluded that the scatter in the
139
-------
data is too large for any trend to be detected with the
statistical analyses used. A conservative statistical
criterion based on the two-tailed Student's t-test was
used to determine the significance level of the trends.
The statistical analyses must be used with caution,
however, as they only detect consistent trends. As indicated
above, if a retention ($,,) drops to low values at low tem-
peratures, no linear trend would be found. Another prob-
lem may be caused by the presence of clear outliers, which
may force rejection of clearly significant trends. Cases
where the linear regression analyses missed a drop in
have been marked as such in Table 5.3. Visual inspection
of the reduced data (shown in Appendix B.I.) showed that
rapid decreases in $.. for Hg, As, La, Cl, and Se were not
detected in a few cases. In general, however, the statis-
tical analyses proved to be accurate.
Reproducibility problems were encountered in all batch
experiments with Bottom Red Ash anthracite. The extreme
scatter in the data made it possible to detect only the
most obvious trends, and the precision problems due to the
low weight and elemental losses of the anthracite were
compounded by chemical analysis problems. Because of the
problems encountered with BRA, some of the following gene-
ralizations may not apply to this particular coal.
140
-------
It has been established that Sm, Cr, Th, Sc, Fe, and
Co are retained completely in the chars produced in the
batch pyrolysis experiments with nitrogen over a tempera-
ture range of 25 to 1200°C. As illustrations, the mass
fractions (X^ and fy*) of Fe and Sc are plotted versus
temperature in Figures 5.7 and 5.8. La, Ee, and As ex-
hibit intermediate volatility (<40% release); however,
the scatter in the data for La make this conclusion sus-
pect, as will be discussed later. Finally, S, Pb, I!g,
and Cl are highly volatile (>50% release). Mercury and
chlorine show losses greater than 70% at temperatures
below 700°C, and more than 75% of the lead is released at
temperatures above 1000°C. Figures 5.9 to 5.12 show mass
fractions (X and ^ ) of S, Pb, Hg, and Cl as functions of
temperature. The mass fractions of chlorine were found to
be below the neutron activation analysis detection limit
in some chars. The estimated upper limits for those mass
fractions are shown in Figure 5.12. The data for V and 5b
were too scattered for any conclusions to be reached.
The results found in this research agree to a sub-
stantial extent with the findings of Kuhn et al. (197B),
who analyzed the chars produced during the pyrolysis of
several coals heated in steps to 450 and 700°C in a nitro-
gen atmosphere (see Section 2.6.6). Kuhn's results for the
elements investigated in this study were: 1. Cr, Th, Sc,
141
-------
ou.u
20.0
10.0
15.0
10.0
15.0
ass Fractlon(^g/5
IV) _
P 01 0
o b b
2
oj 15.0
<
10.0
15.0
10,0
5.0
r\
I 1 1 1 1
H O
Q 2 °
o a a a a
o
0 °
o B a a D
0 ° 0
* § o a D
o o
o o
a a a
9
D §
BZN
O
n
O NB8
a
o
MRS
a
0
WKII
a
BRA
—
@
1 i
0.0
5.0
0.0
5.0
0 200 400 600 800 1000 1200 1400
Temperature (°C)
Figure 5.7 Iron Mass Fraction In Batch Chars
0.0
142
-------
3.5-
2.0-
1.0-
4.0-
3.0-
2A-
*— - «W
O»
^ 1.5-
| 1.0-
o
0
£
5CI-H
V)
o
2
o: 4.0-
3.0-
4n-
3.0-
2.0-
l.O-
i i I i i i i
O=^H 0 0 ° 0
BZN
O
D a a a D
a
0 o
o N88
0 0 D Q D 0
O
0 0 o
MRS
a ° a Q
O
o ° o
O WKII
0
Q
0 8 8
BRA
i
i i i i i i i
0.0
2.0
0.5
2.0
1 A
0 200 400 600 800 1000 1200 1400
Temperature (°C)
Figure 5.8 Scandium Mass Fraction in Batch Chars
143
-------
200 400 600 800 1000 1200 1400
z
o
o:
u.
CO
CO
1.0
0.5
0.0
1.0
0.5
0.0
1.5
1.0
0.5
0.0
3.0
2.0
1.0
0.6
0.5
0.4
O = xH+
— (-] = y -J-
H O
P g° o u
D D D D
0 8° o o o
D D D
o
a a o o o
a o
nan
D Q
a o o °
ODD
.0 eg @ g 8
o
_
BZN
D
n NB8
a
0
MRS
a
.
WKII
a
BRA
O
D
i
200 400 600 800 1000 1200 1400
Temperature (°C )
Figure 5.9 Sulfur Mass Fraction in Batch Chars
144
-------
c
o
o
o
tr
6.0
4.0
2.0
0.0
200 400 600 800 1000 1200 1400
15.0
10.
oO
5.0
0.0
10.0
5.0
0.0
10.0
5.0
OjO
10.0
5.0
0.0
0
o
a
°
a
80
a
o
a
a
n O
a a
§
e
a
Q P
a u
e R
8 S
8ZN
NB8
MRS
WKI 1
200 400 600 800 1000 1200 1400
Temperature (°C)
Figure 5.10 Lead Mass Fraction in Batch Chars
145
-------
1
c
J3
•»-
u
S
U-
w
in
O
Z
a:
<
O.IO
0.05
0.3
0.2
0.1
0.3
0.2
O.I
0.3
0.2
0. 1
0.3
0.2
O.I
o.o
0 ° 0 = XH+
O D=»H+
o a o BZN
a o ° „
° NB8
" 8a 8
H a i S
o
a MRS
° e8
H 988
P WKII
oo
Q D Q
H 8 © A
-o
BRA
§ §
. § , ,008,
0.0
0.0
0.0
0.0
200 400 600 800 1000 1200 1400
Temperature ( °C )
Figure 5.11 Mercury Mass Fraction in Batch Chars
0.0
146
-------
o
o
CO
-------
Fe, Co, and V were completely retained; 2. La, Sb, and
Gm showed moderate losses (<30%) in some coals, and
3. Cl, S, As, Pb, and Se showed significant losses (>30%).
The agreement between the results of the two studies
is gratifying considering the diversity of coals studied.
It appears that trace elements tend to exhibit the same
behavior in different coals. This study has extended the
results of Kuhn e_t al. to temperatures up to 1200°C, and
provides enough data for the development of a model des-
cribing the equilibrium release of volatile trace and
minor elements.
Examination of the elemental release data for the
elements found to be volatile suggests that there is a
critical temperature above which the elemental release be-
gins. Once this temperature is reached, the value of \^u
(mass of element in char/mass of element in feed coal) de-
creases until an asymptotic limit is reached. These ob-
servations suggest the following simple mathematical model
to describe the release of a given element from a coal
undergoing pyrolysis.
xc • TTC
148
-------
where Xc = mass fraction of element in feed coal
(JjH = mass of element in char per unit mass of
feed coal
TC -- critical temperature at which elemental
release begins
or in terms of the fractional retention $„ (mass of element
in char/mass of element in feed coal)
n = 1 + H(T - Tc)-{f(T-Tc) - 1} (5-3)
where
H(T - Tc) = unit step function
*„ = Vxc . (4-49)
An equation capable of correlating the data for all
elements with intermediate and high volatility is:
f(T-Tc) = 1 - Ane[l-e~b(T~TC)] (5-4)
v/here
Afl = asymptotic fractional elemental release
b = constant .
The asymptotic fractional elemental release (Afte) was esti-
mated by linear regression of the equation
AnH = - + Afie , AflH*0.67Ane (5-5)
where
d = constant
A« = Afi , T -> oo .
149
-------
Duhne (1977) has shown that estimates of asymptotic equili-
brium values for physical and chemical processes can be ob-
tained from linear regression of an equation of the form of
equation (5-5) . if the highest value of A% for each ele-
ment is assumed to be very close to A^e, the requirement
that the data lie within 33% of the equilibrium point
(relative to a total range between zero and equilibrium) is
usually satisfied if the highest three points for each ele-
ment and coal are used.
Least squares estimation of b and Tc in equation (5-4)
tends to yield values of Tp below room temperature for He
and Iig. The following equation was found to correlate the
data close to TC for all elements with data close to Tc,
and so provided a basis for estimating this parameter:
*II = al + a2£n(T) ' Ii<1'° <5-6>
where
al' a2 = constants
The constants a-^ and a2 were determined from least squares
regression of 4>. and £n(T) for experimental values of
4>,,<1.0. The critical temperature was then calculated by
setting (T ) =1.0. This projection of the decay of
H to the critical temperature is not influenced by the
value of the asymptotic retention (1-A^£).
This procedure was used to determine TC for I!g, Cl,
pb, and s. However, it could not be used to determine TC
150
-------
for As and 53e because of lack of data at 300°C. Because ef
the chemical similarity between those two elements and sul-
fur, the value of T determined for sulfur was also used
Vrf
for As and Se.
Marquardt's (1963) algorithm for nonlinear regression
was used to estimate the parameter b of equation (5-4).
Clear outliers (e.g., large negative values of Afyj) , mass
fractions below the detection limit (e.g., chlorine data
at high temperature), and the data for sulfur in the an-
thracite (BRA coal) (which was markedly different from that
for the lower rank coals) were not used in the regression.
The parameter b could have been obtained from simple linear
regression; Marquardt's algorithm was used only as a matter
of convenience since data plots and statistical analysis of
the regression could be obtained with a SAS program used
extensively in this study. This program (BMOD) is listed
in Appendix A.I; a listing of the data used and the statis-
tical analysis of the regression are in Appendix B.I.
The parameters of the elemental retention model
(equations (5-3) and (5-4)) are listed in Table 5.4. Com-
parisons between the model predictions and the experimental
data for As, S, Pb, and Hg are shown in Figures 5.13 to 5.16,
The model provides a good description of the experimental
data for five coals with a single set of parameters for
each element (except for sulfur in BRA). This result by
151
-------
Table 5.4 Equilibrium Elemental Release Model Parameters
Element
As
Pe
Pb
C
Cl
Hg
TC AQe
oc %
223 G6.3
223 55.4
499 100.0
223 77.1
234 100.0
110 97.3
b
8.83 x 10~4
48.21 x 10~4
26.57 x 10~4
20.08 x 10~4
39.42 x 10~4
25.81 x 10~4
152
-------
100
c
o
a>
cr
100
75
50
25
100
75
50
25
100
75
50
25
100
75
50
25
O
O
O
O
o
o
BZN
NB8
MRS
O
WKU
BRA
0 200 400 600 800 1000 1200 1400
Temperature (*C)
Figure 5.13 Arsenic Retention in Botch Chars
153
-------
c
CC
I 00
75
50
25
100
75
50
25
I 00
75
50
2 5
100
75
50
25
100
75
50
25
(O
o
o
o
0 = data from this study
X=Data from Suuberg et al
o
o o
o o o
BZN -
MRS -
WKII -
BRA -
200 400 600 800 1000 1200 1400
Temperature (°C)
Figure 5.14 Sulfur Retention in Batch Chars
154
-------
o
c
100
75
50
25
100
75
50
25
100
75
50
25
100
75
50
25,
100
75
50
25
0
1 o
o
o
o o
120107
O O
o
o
II
BZN -
NB8 -
MRS -
WKII -
BRA -
0 200 400 600 800 1000 1200 1400
Temperature (°C)
Figure 5.15 Lead Retention in Batch Chars
155
-------
•I
tr
i ZOO 4OO 600 800 1000 1200 I40O
Temperature (°C)
Figure 5.16 Mercury Retention in Batch Chars
156
-------
itself constitutes a significant finding, in that it indi-
cates that trace elements tend to behave in the same manner
during the pyrolysis of coals regardless of their rank, with
the exception of sulfur in anthracite. Therefore, the model
should provide excellent engineering estimates of the extent
of volatile trace and minor element release as a function of
temperature during the devolatilization stage of any gasi-
fication process.
Figure 5.14 shows the similarity between the data ob-
tained in this study and that obtained by Suuberg e_t al.
(1978) for Montana lignite during wire screen batch experi-
ments. The model developed in this study provides a fair
description of such data. The arsenic retention data ob-
tained in this study agree closely with the arsenic retentions
determined by Duck and Himus (1951) for four coals carbonized
at eight temperatures (ranging from 290 to 1050°C).
The sulfur release determined in this study agrees
with the findings of Kuhn e_t al. (1977) who reported that
most of the sulfur is lost between 300 and 400°C. They
report a 66% loss of sulfur at 700°C. Kuhn et: al. indi-
cated that most of the coals in their study showed similar
behavior; however, the results of this study indicate that
sulfur is evolved to a much lesser extent, at a given tem-
perature, in anthracite. The weight and sulfur loss data
from the transient experiments (shown in Figure 5.3) and
the equilibrium batch experiments (Figure 5.17) suggest
that the following equation holds for all ranks of coal:
157
-------
CO
co
O
c
tt>
E
-------
AS = AW* . (5-7)
It will be shown in later sections that this relation is
also valid for fast pyrolysis, and it might apply as well
to the steam-oxygen gasification of WKll coke. The reason
for such behavior is not clear since sulfur exists in at
least three primary forms in coal (pyrites, sulfates, and
organically bound sulfur). A possible explanation could
be inferred from the findings of Kuhn et al. (1977); they
report that the pyrite contained in coal is converted to
pyrrhotite and sulfur at 450°C or lower in nitrogen atmos-
phere. Their chemical analyses also indicate a greater
loss of sulfur from the pyrites than from organic sulfur
at low temperatures, whereas the reverse was found to be
true at high temperatures (>450°C). Furthermore, as dis-
cussed in Sections 2.2.5 and 2.2.7, pyrites react with car-
bon to form carbonyl sulfide and decompose releasing sulfur
in gaseous form.
The degree to which equation (5-7) correlates the
sulfur loss data of the equilibrium experiments is shown
in Figure 5.17. The relationship appears to hold fairly
well for all the coals studied, although better agreement
is obtained for the lower rank coals (BZN, NB8, and MRS).
In addition to examining the trends with temperature
shown by the mass fractions of individual elements, it is
useful to determine whether the rates of evolution of
159
-------
certain elements show high correlations. Element-to-element
correlations of X were tested for significance with com-
puter program BATCH (Appendix A.I), The calculated slopes
and coefficients of determination are shown in Appendix B.I.
Those found to be significant at the 95% and 99% confidence
levels using a table of correlation coefficients (Snedecor
and Cochran, 1967) are shown in Tables 5.5 to 5.7. It is
apparent that low volatility elements tend to be highly
correlated, while high volatility elements do not. This is
to be expected, since elements retained completely in the
chars are enriched at a rate inversely proportional to the
decrease in char weight, while very volatile elements tend
to be depleted at different rates.
The following low volatility elements: Sm, Th, Sc, La,
Fe, and Co are positively correlated in at least four of
the five coals studied. The only unexpected element in
this list is lanthanum. This result appears to indicate
that the previous classification of lanthanum as a medium
volatility element may have been due to scatter in the
data. As shown in Tables 5.5 to 5.7, volatile elements
tend to be negatively correlated with nonvolatile elements.
A possible application of such correlations is, therefore,
to estimate the volatility of elements whose mass fractions
cannot be determined by finding whether they are positively
or negatively correlated with readily measured low vola-
tility elements.
160
-------
Toble 5.5 Significant Element- Element Correlations- Batch Experiments
NB8 Coal
Element Sm Sb As Se V Cl La Th Cr Sc Fe Co Hg Pb S
Sm
Sb
As
Se
V
Cl
La
Th
Cr
-Sc
Fe
Co
Hg
Pb
S
\
-f-
**
*
*
••t-
*•
**
*
\
-t-
*
\
\
\
\
-r-
ft
*t
\
*
t
•ft
«•*
*
-4-
*
**
\
-f
*
*
**
*•*
*
•*•
I-t-
*
\
4-
**
-t-
*^«
»*
«• *
4-
*
4-
»*
-1-
\
4-
**•
*
~i
-t-
•4-
**
*
4-
**
**
\
4-
**
-»-
*
^
-t-
**
-*-
*
-h
*
**
-1-
4-
**
\
4-
~^
\
*
«
»*-
•ft
«^
*•»
*
\
*
*
^
\
WKII COAL
* = significant @ 95 % C. L.
** = significant @ 99% C. L.
+ = positively correlated
— = negatively correlated
161
-------
Table 5.6 Significant Element-Element Correlations-
Batch Experiments
Element
Sm
Sb
As
Se
V
Cl
La
Th
Cr
Sc
Fe
Co
Hg
Pb
BZN COAL
Sm Sb As S^ V Cl La Th Cr Sc Fe Co Hg Pb S
X
4-
X
4-
**
4-
X-
\
4-
XX-
X
X*
X-
4-
X-*
X--X
\
X-
•X--X-
-t-
XX
X-
\
-»-
x-x
+
*
+
X
\
\
-t-
x-x
+"
XX-
x-x
\
+
•X-
^f-X
-X-
-»-
•X
4-
X X
\
+
X
-h
X
\
-»-
X-
•X-
+
x-x-
-»-
X-
*
-t-
XX
\
•f
•X
-h
X*
-r-
XX
•f
XX
+
•X-
\
X
-h
X
•+•
X
-r-
X
-f-
X
-+-
X
\
x-x-
X
*
•X-
*
*•
\
X-
XX-
\
+
x^
+
X-
-r-
X^
4-
•X-
\
BRA COAL
-X- = significant at 95% C.L.
X* = significant at 99% C.L.
+• = positively correlated
— = negatively correlated
162
-------
Table 5.7 Significant Element - Element Correlations-
Batch Experiments
Element
Sm
Sb
As
Se
V
Cl
La
Th
Cr
Sc
Fe
Co
Hg
Pb
S
Element
\
X
-r-
XX
XX-
XX
-H
X-
X
X-
\
\
+•
4-
~*
\
+*
\
X
\
If
\
\
X^X
4-
XX
4-
X*
XX
\
4-
XX-
4-
X
X-
X X
\
\
x xx^
\
\
XXX N.
\
Sm Sb As Se V Cl La Th Cr Sc Fe Co Hg Pb S
MRS COAL
X- = Significant at 95% C,L.
XX = Significant at 99%C.L.
4- = positively correlated
- = negatively correlated
163
-------
6. ANALYSIS OF RESULTS FROM LAMINAR FLOW
REACTOR EXPERIMENTS
The purpose of these experiments was to study the de-
volatilization and elemental release of three coals (MRS,
NBb, and BZN) as functions of temperature and time. As
indicated previously, a laminar flow reactor was chosen to
carry out the experiments. Thus, rapid heating conditions
(103 - lo4 °C/sec) and small residence times (170 - 1500
msec) were attained.
Fifty-two runs were carried out. The first fourteen
runs were made with WKll and MRS , and were intended pri-
marily to shake down the reactor system. The results ob-
tained in these runs are not reported in this thesis.
Twenty-six runs were made witli MRS coal, at temperatures
ranging from 25 to 900°C. Nine runs were made with 1JB8
coal, five at bOO°C and four at 900°C. Finally, three runs
were made with LZN coal, all at 800°C.
A summary of run conditions is shown in Table 6.1 The
residence times calculated using Subroutine RESTIM in com-
puter program MODELS are shown in Appendix A.2. The iso-
thermal reaction time tj. (as used in equation (4-30) , Section
4.2.2) is defined here as tR minus three times the coal heat-
ing constant TT. This is done only for tne purpose of
allowing comparison of the results with the work of other re-
searchers (e.g. , Badzioch and Hawksley, 1970; Nsakala, 1976) .
164
-------
Table 6.1 Summary of LFR Run Conditions
en
en
Run No.
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
Coal
MRS
MRS
MRS
MRS
MRS
MRS
MRS
MRS
MRS
MRS
MRS
MRS
MRS
MRS
MRS
MRS
MRS
MRS
MRS
MRS
BZN
T
°C
294
395
398
398
402
600
600
598
597
799
800
800
800
799
799
800
801
601
870
602
803
fcR
msec
1523
482
714
976
1233
172
285
522
774
171
270
518
770
521
519
516
825
501
557
1017
517
fcl
msec
1416
380
612
874
1131
78
191
427
679
84
183
431
683
434
432
429
738
406
472
923
430
Coal Feed
Rate
g/min
0.45
0.42
0.47
0.54
0.37
0.79
1.44
1.10
0.51
6.00
0.43
0.32
0.43
0.48
0.36
0.37
0.36
0.32
0.32
14.50
1.08
Char
Recovery
%
44.50
68.00
54.50
—
52.00
74.00
75.00
68.67
63.50
83.00
69.17
68.00
45.00
61.00
73.50
69.00
26.00
71.00
-
-
-
CY3
Recovery
%
24.72
3.92
24.77
-
31.73
0.00
0.00
0.97
5.83
0.00 Bad
0.00
1.47
1.56
1.64
0.68
1.45
55.36
1.41
-
Bad
-
Comments
run-discarded
run-discarded
-------
Table 6.1 continued
en
en
Run No.
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
Coal
BZN
BZN
NB8
NB8
NB8
NB8
NB8
NB8
NB8
NB8
NB8
MRS
MRS
MRS
MRS
MRS
MRS
T
°C
804
808
801
802
805
803
802
900
900
900
900
900
900
900
900
25
25
fcR
msec
816
268
518
818
269
397
670
613
255
767
441
619
254
766
443
1428
488
t Coal Feed
Rate
msec g/min
727
181
431
731
182
310
583
527
172
684
358
534
170
683
360
1309
272
1.20
2.85
0.79
0.53
0.32
0.42
0.51
0.41
0.40
0.83
0.41
0.74
1.05
0.48
0.96
0.49
0.25
Char CY3
Recovery Recovery Comments
% %
51.00
74.67
_
54.50
77.33
81.33
64.00
59.33
73.00
60.50
62.50
52.00
76.00
50.00
57.50
66.67
92.00
0.00
0.00
_
0.00
0.00
0.00
0.00
0.00
0.00
0.83
0.00
0.00
0.00
0.00
0.00
25.00
2.90
-------
The sections that follow present the results obtained
in the laminar flow reactor runs. The ash tracer technique
is examined in detail; the results of the trace element
analyses are used to aid in the determination of the extent
of ash losses, and a procedure is developed to correct for
such losses. The devolatilization results are presented
and compared with the predictions of Kobayashi's (197t>)
parallel reactions model, a first-order model featuring
temperature-dependent asymptotic weight loss, and two ei.i-
pirical models with parameters fitted by least squares re-
gression.
Thirty trace and minor elements were analyzed in the
chars produced in three 800°C runs (2b, 27, and 29) with
MRS coal. Special care and precautions were taken in the
operation of the reactor system and the handling of the char
samples for these runs. A detailed analysis of the results
of these runs and a less detailed analysis of the results
of the elemental analyses from the other runs were carried
out. At least ten elemental analyses were performed for
every char produced with the LFR. Finally, statistical
trend analyses of elemental release were carried out, and
an attempt was made to model the rate and extent of ele-
mental release as a function of time and temperature.
167
-------
6.1 Weight Loss Estimation Errors
The ideal way to conduct laminar flow experiments
would be to feed a measured amount of coal, collect all of
the char produced, and determine by difference how much of
each analyte of interest was evolved. The difficulty in
collecting 100% of the char has led to the use of a non-
volatile material as a tracer. Ash is almost invariably
used for this purpose.
Weight loss on dry ash-free basis in laminar flow
reactor experiments is usually estimated with the equation
AW
Aj
II
1 -
H
(4-58)
where Ap = moisture-free ash fraction of feed coal
A! = moisture-free ash fraction of char.
The results obtained using this formula are shown in Figure
6.1 for MRS coal. NB8 and BZN exhibit similar behavior.
As the data of Figure 6.1 show, low temperatures
and low residence times consistently lead to negative
values of AW,.. Since negative u.a.f. weight losses
are a physical impossibility, it is clear that the ash
*
tracer method leads to underestimation of AW and therefore
of AW (m.f. weight loss) and AW+ (a.r. weight loss). The
168
-------
30
*< 20
£
M
M
O
-j 10
JC
111
o n
V
-10
i i i i i i i
U Q 900°C
^ O 870 °C
• 800 °C
Q O 6OO°C
0 X 400°C
A 300°C
•
*
^
o
o
— cT<5— $> ;
X *
x .. "
X
1 1 1 1 II
0 400 800 1200 1600
Residence Time ( msec)
Figure 6.1 D. A. F. Weight Loss by Ash Tracer Method-MRS Coal
169
-------
first two quantities are much more sensitive to this effect
and to errors in the ash analyses as large moisture losses
ttt-.l to make all values of AW* (T, r.R) positive, hiding the
ash-introduced error. Values of %AW^, %Aw|, and %AW* de-
termined by the ash tracer method are calculated with com-
puter program LFRS1 (Appendix A.2) and tabulated in Appendix
D.2.
The attainment of negative weight losses cannot be
attributed to error in the ash analyses. All analyses were
carried out in random order using blind samples, so that
errors in the analyses would have led to random variations
and not to the observed consistent pattern of decreasingly
negative d.a.f. weight losses with increasing time and
temperature.
It has been argued (Badzioch et al., 1968) that ash-
induced errors have a very large effect only at low decom-
position rates, i.e., at low temperatures and/or residence
times and high coal ranks. This can be shown mathematically
Ly taking the derivative of equation (4-58) with respect to
dAW
dA
t
i
Ac
.1 - Ac_
r 1 -
-(A*)2.
(fa-1)
170
-------
The rate of change of AWJ is higher for small values of A.'.,
c\ * '
so that underestimation of AVv\ at low values of A. is lar-
A li
ger than at high values of A?j for the same net amount of
ash loss.
However, the underestimation effect cannot be ignored
at high temperatures. The mass fractions of several ele-
ments in flRS feed coal (X*) and the chars (>-^,) from runs
25, 27, and 29 are shown in Figure o. 2. It is evident that
the chars are enriched in some of these elements, since Xj;
is greater than X*. However, for three of the elements
shown (P, As, and La), ^ (mass of element in char per uuit
ntass of feed coal estimated using ash as a tracer) values
are also larger than the feed values. Such behavior is
physically impossible as the following argument demonstrates,
The mass balance for coal pyrolysis is
wc = wn •*• wv
where W^ = feed coal weight
Wj, = char weight
W, = volatiles weight
The species mass balance is
xcwc = Vi, * xvv\ . (6-3)
If ash is used as a tracer,
AC
WH = VJC — {6'4)
171
-------
o
o
M
M
O
g
CO
.20
.15
.10
1 2
* • w
1.0
.80
360.0
310.0
260.0
1 O
1 • w
.80
.60,
6/"\
.u
5.0
4.0,
3.0
i i i
A
a
V - V +
x- XH
- O=YA*
: a=tH+ x
O
a
X
O
a
X
8
X
O
a
i i i i
i i i
X
9
X
O
a
X
0
0
X
8
X
0
a
i i
i
X"
9-
x"
8
x
8
X
x~
-
, 8
200 400
Residence Time (msec)
600
.05
.60
210.0
.40
3.0
800
Figure 6.2 Mass Fractions in LFR Chars, MRS 800°C Runs
172
-------
where AC = ash content of feed coal
Aj. = ash content of char.
These two equations combined with equations (4-43) and
(4-4B) yield
(fa-5)
where 4u = iKj estimated using ash as a tracer.
These relationships are valid for all bases, a.r., n.f., and
d.a.f., as long as all parameters are expressed on the same
basis.
Combining equations (6-2) and (6-4) and substituting
into equation (6-3),
_ XC - Xy(l - VA1;)
X - - . (fa-6)
Substituting into equation (6-5) ,
tyh = xc - xv(i - AC/AH) . (6-7)
Since
A,, > A (6-8)
h ~ C
it follows that
*A < XC . (6-9)
That is, the mass fractions of elements in the char, when
normalized to feed coal weight, must be less than or eaual
to the feed coal mass fractions of those elements. The
173
-------
condition that underlies this inequality is that there have
been no ash losses.
The contrary behavior of several elements (P, As, and
I-a) is shown in Figure 6.2. The values of 41* for the more
volatile elements, Kg and S, do meet the criterion set by
equation (6-9); however, this occurs at the conditions of
the runs shown only because the high volatilities of these
elements overcome the ash loss effect. The opposite result
is obtained for sulfur at lower extents of devolatilization
(i.e., lower T and/or tR), as illustrated in Figure 6.3.
Mercury is evolved too quickly under any conditions for the
effect to be observed. Figures 6.2 and 6.3 also depict
values of iy calculated by a corrected ash tracer method
which will be discussed later.
If only one element had shown this behavior, it could
have been attributed to sample contamination, or to analyti-
cal error. However, the result was obtained for several
elements whose concentrations were determined by four
different analytical techniques, so that the inadequacy of
the ash tracer method must be regarded as genuine. The re-
sults shown in Figure 6.2 correspond to MRS runs 25, 27,
and 29.
An interesting corollary of the above analysis is that
elements which are entirely, or at least very strongly,
associated with the ash should be lost from the coal in the
174
-------
c
o
I.Op
i
c
j
4
0.8 -
0.6
o.:
0.15-
Q
D
D a
O
n
O
O
O
)
O
175
rs.
'0.4
-j
r>
v
r't
'—^
i
0.05-
0.0
r 8
i
i
flit i 1
3 10 20
ill
30 40
M.F. Weight Loss (%&W*)
.3 Mass Fractions in LFR COO°C MRS
- Volatile Elements
175
-------
same proportion as the ash itself. Thus the mass fractions
of those elements normalized using equation (6-i) should
satisfy the equality of equation (6-9), i.e., yjA = X .
Sonic nonvolatile trace elements were found to show such
behavior in KRS coal. The results for four nonvolatile
elements are shown in Figure 6.4. The chars of MRS runs
25, 27, and 29 show enrichment in the mass fractions of
these elements (A*) as the residence time increases, while
the slopes of g;* versus tp for Sm and fin are not signifi-
cantly different from zero, indicating that ijj, = X, and
thus that these two elements are released from the coal at
the same rate as the ash. This finding suggests that sig-
nificant amounts of trace elements may be released from
coal as submicron ash particles during gasification. Vhe
statistical analyses discussed in later sections use this
information to determine the mode of elemental transport
from the coal particles to the gas stream.
A number of factors are potential sources of error in
values of AW calculated by the ash tracer method. Among
these factors are the following:
1. Measured ash contents of the feed coal and the char
may be in error either due to analysis errors or to
run-to-run variability.
2. The ash contents of coal and char particles vary with
particle size. The cyclones used to collect the char
176
-------
o
o
o
•?
3L
• i
E1
9
Jj
o»
Q|
a.
1
^
rz1
liu
.65
.55
1150
1050
950
850
100.0
90.0
80.0
70.0
2.25
2.05
1.85
1.65
1.45
111
X
0
Q
X=X|J
OsV+
V
0
O
-
X
o
D
.
X
-
O
D
riii
111
X
0
D
X
O
Q
X
O
D
X
O
,o
1 x
—
O"
a
*
•
o-
D
X
-
O.
a
o-
o
45
750
60.0
800
0 20O 40O 600
Residtnce Tim* (mstc)
Fiqur* 6.4 Mats Fractions in LFR 800* MRS Chars-
Nonvolatil* Eltmenlt
177
-------
impose a size classification, and thereby bias the
measured ash content of the product. If the cyclone
contents are mixed to provide the product char, in-
homogeneity and imperfect sampling constitutes another
potential source of error.
3. There is a difference between the experimentally
measured proximate ash content of an as-received coal
and the same coal which has been subjected to pre-
devolatilization.
4. Tars evolved in pyrolysis may condense in the collected
char, thereby decreasing the apparent weight loss.
5. The rate of heating of the coal and hence the de-
volatilization rate may be a function of the coal feed
and feeder gas rates, both of which influence mixing
and heat transfer near the feed inlet.
6. Some of the material that constitutes what is termed
"ash" in coal may in fact be lost during pyrolysis,
either by being converted to gaseous species or by
being expelled as particles too small to be collected
by the devices employed in the experimental system.
The sections that follow explore each of these effects,
quantifying them to the extent possible. The results of
these studies are then used to formulate in Section 6.2 a
weight loss calculation method that accounts in part for
possible ash losses during pyrolysis.
178
-------
6.1.1 Precision and Accuracy of Weight Loss Estimations
The magnitude of the d.a.f. weight losses cannot be
explained by the variability in the moisture and ash
analyses, as shown in the following discussion. The effects
*
that errors in chemical analyses have on values of %AW, ob-
tained using ash as the tracer may be estimated using the
following equation:
2
C
3 AW^\
+
id A/-* i
2 c-2
A+ +
C
3AW*\2 S2
1 A +
-(- j H
f3AW \
A
'3M-. '
2 Sf^ TAWV SM
C '^ + j
where
• th
9 Aw
A
3A+
(^
-1
A+
^ - MH ~ AH"
Ll - M^ - A*.
. Ac
A*
" i - *; - -; i
L(l - M? - A?)2.
II
(0-10)
S. = standard deviation of i parameter. From
equation (4-58),
(6-11)
3 AW
-f
II
A
i _
"
X
Ll -
(b-12)
9AWA
+
C
-Ac
A+
H
H H
/I ,5+ n + \ 2
(1 - MC - A )
(6-13)
179
-------
(6-14)
- Mi -
An intermediate temperature and residence time have
been chosen to estimate the theoretical S. * from the esti-
AWA
mated errors in the chemical analyses. For run No. 29
(coal = MRS, T = 799°C, tR = 519 msec, AWA = 0.08B4), the
variance components are estimated to be:
S2
1.089xlO~6 = S2.
A
H
S2 = 1.046xlO"5
M+
* 2
3 AW \
A\
3A
105.79
II
II
,* 2
'n'
3M.
3 AV:
* 2
70.91
1.39
1.05
180
-------
The variances of the average moisture and ash contents of
the feed coal were determined from five replicate analyses.
It is assumed that the moisture and ash analyses of the
chars have the same variances as those of the feed. There-
fore, the variances of the mean moisture and ash contents
2 2
were used for S 4. and S +. Equation (b-10) then yields
Vr A 2
?.,,* = 0.0148. However, this is not a aood estimate of
A\;A
the expected run-to-run variability, because the same
of feed coal was used for all the runs with a given coal
(MRS in this case) , and the average values of >:t and At were
used for the computation of AW in all the runs with that
coal. Consequently , S..+ and S7+ must be set equal to zero
in order to obtain the expected run-to-run variability c.ue
to analytical error. Equation (6-10) then yields Sf =
0.0093. The largest contributors to the error esti-
mates are the ash variance components, even though the
value of S + is an order of magnitude larger than the value
of S . If only the second variance component in equation
(t>-10) were used, S ^ = O.OObV. The conclusion is that
the error in AW* reflects almost entirely the error in the
A
ash determination.
In order to determine the experimental variance between
runs, four replicate runs were made. Hun numbers 2b, 2b,
29, and 30 were made at the same experimental conditions.
The average weight loss and standard deviation estimateu
from those four replicates are as follows:
181
-------
*
AU = 0.0904 , s = 0.0035.
>•• M:
A
'.!ii:.s, .lu" u::; i-r iinental variance of these four laminar ilow
t'xnor indents in sr.aller than the theoretical estimate.
icjv/evcr, not all experiments were conducted under conditions
.^s Cciri.-iully controlled as in these runs.
Porif arison of the theoretical and experimental stan-
*
uan; deviations with the ncnative values of AW, (see Figure
A
' .1) indicate that they are too large to have been caused by
..nalysis error or run-to-run variability. Furthermore, as
indicated 1. eforc, random errors in the proximate analyses
could not have caused the r.onotonic increases in negative
A\: valuos with increases in " and t,,.
A u
Variances in chemical analyses and between runs are
not large er.ounh to explain the magnitude nor the trend of
nt'
-------
a significant fraction of the ash exists as separate par-
ticles, as has been observed by T,i tt le johr. (1'Jot) .
The particle size distribution of MFC pulverized coal
is shown in Figure 6.5. The ordinates are the percentaoes
by weight of particles greater than and less than the par-
ticle size given in the abscissa. They were determined
from the weights retained in different L.S. Standard Sieves
during the size grading of the coal. The noisture and ash
contents of the different size fractions were also deter-
mined; the results are shown in Figure fa.6. It is evident
that the moisture-free ash content (A*) increases as the
coal particle size decreases. The d.a.f. sulfur contents
of the different size fractions (% S ) do not show a statis-
tically significant trend. This appears to indicate that
sulfur in MRS coal is associated primarily with the coal's
organic matter, in agreement with the findings of Ficne ct
al. (1978) shown in Table 3.2.
In an experiment suggested by the above results, two
runs were carried out in the LFR at room temperature. The
objective was to determine the moisture, ash and sulfur con-
tents of the coal fractions collected in the three LFR sys-
tem cyclones (depicted in Figure 3.2). The results are
shown in Table 6.2. The coal fraction collected in the
3/4 inch cyclone (CY-1) clearly has a reduced ash content
relative to the feed coal, while the fraction collected in
the 1/2 inch cyclone (CY-2) is enriched in ash. The material
183
-------
9
o
10
20
30
40
50
60
70
80
90
99.9
0
99.9
99
90
80
70
60
50
40 „
N
30 £
20
10
10
20
I
30 40 5060708090100
Av«rog« Partiel* Diam«t«r(
Figure 6.5 Rosin Rammler Plot of Montana Rosebud Coal
184
-------
12.0
11.0
< 10.0
oo
en
9.0
6.0
60X170
170X200
200X230
230X270
270X325
325X400
400 +
dp
170.0
82.5
69.0
58.0
49.0
41.50
36.50
20.76
16.73
16.24
16.67
15.25
12.72
13.05
1 T
%S*
1.2542
0.9793
.0899
.1115
.1225
.1215
.0835
O
I
I
I
30 40 50 60 70 80 90 100 110 120 130 140 ISO I6O 170 180
Avtrog* PoricU Oiom«t*r (jim)
Figut 6.6 Ath Conttnt of MRS Slzt Froctiont
-------
Table 6.2 Room Temperature LFR Runs with MRS Coal
Run No. t CY. No. Recovery %M
R e
%AW*
Vv
Feed
12.72 11.47
0.9929
0.0
51
co
CTl
1428
1
2
3
TOTAL
42.0
8.0
16.7
66.70
6.62
7.39
6.99
11.09
11.76
10.99
0.9612
0.7047
0.9753
-3.86
+ 2.80
-4.92
39.9
52
488
1
2
3
TOTAL
78.7
10.7
2.7
92.10
6.63 11.26 1.0184 -2.10
7.54 12.02 0.7115 5.18
not enough sample
14.11
-------
in CY-3 (another 1/2 inch cyclone), which traps the coal
particles that miss the collector, also shows a reduced ash
content. The results of the ash analyses appear to be con-
firmed by the sulfur analyses. There is a sharp reduction
in the sulfur content of the particles trapped in CY-2,
indicating that the particles trapped in this cyclone have
less organic matter.
Approximately a 6% a.r. weight loss (AV>+) is found, in
runs 51 and 52, due to the drying effect of the nitrogen gas
used (as evidenced by the reduction in noisture). However,
the ash tracer method yields AVI = 3% for the char collected
in CY-1. Furthermore, d.a.f. weight loss values calculated
by the ash tracer method (equation (4-58)) yield negative
values for CY-1 and positive values for CY-2 in runs 51 anu
52. Roth sets of values are erroneous, since roon tempera-
ture runs should not exhibit d.a.f. weight loss, much less
d.a.f. weight gain.
The source of errors is of course the ash loss of the
sample in CY-1 and the ash enrichment of the sample in C\-2.
The weighted average ash content of CY-1 and CY-2 (i.e.,
that which would be calculated if the two samples had been
mixed) is still much lower than the ir.f. ash content of the
feed (e.g., %A* = 11.20 in run 51). Therefore, it is clear
that a significant amount of ash is not collected by either
CY-1 or CY-2. The samples collected in CY-3 are depleted of
187
-------
ash in spite of the size of that cyclone (1/2 inch) quite
likely because of the reduction in the collection efficiency
due to the low flow rates of the gas stream passing through
it.
Table 6.3 shows a comparison between the composition
of CY-1 and CY-2 chars for several high temperature runs.
It is obvious that the chars collected in the two cyclones
are different. The lower sulfur contents of the chars in
CY-2 indicate that they are enriched in ash as was the case
in the room temperature runs. In addition, since the pro-
cedure followed in the sampling of the chars consisted of
combining the chars collected in the hoppers of CY-1 and
CY-2, incomplete mixing could have led to a large amount
of scatter in the trace element analyses. Fortunately, most
of the char was collected in CY-1 in the majority of runs.
To minimize this potential inhomogeneity problem, each
sample used for neutron activation analysis (NAA), atomic
absorption analysis (AAA), and moisture analyses weighed
0.5 g, which v/as typically 4% of the total amount of col-
lected char. Replicate sulfur analyses were carried out for
all chars, because each analysis required only a 0.1 g
sample.
Kobayashi (1976) suggested that the ash fraction of
coal used in the ash tracer method should be that of coal
particles which have experienced the same sampling process
188
-------
Table 6.3 Comparison of CY-1 to CY-2 A.R. Char Compositions
Run No.
38
39
40
41
42
43
44
45
46
47
48
49
50
Feed I
Coal CY-l
NB8
NB8
NB8 0.702
NB8
NB8
NB8
NB8
NB8 0.460
NB8
MRS
MRS
MRS
MRS
kM+ %A+
CY-2 CY-1 CY-2 CY-1
0.9459
0.9506
3.67 23.56 28.65 0.9152
0.9686
0.9707
0.8582
0.9952
5.37 38.20 28.05 0.8713
0.9784
0.7860
0.9330
0.7923
0.8947
+
CY-2
0.7145
0.6643
0.7448
0.7365
0.6963
0.7042
0.7333
0.7032
0.6929
0.6547
0.6309
0.6091
0.5877
-------
at room temperature as they would in the real experiment.
Therefore, the d.a.f. ash content of CY-1 in the room tem-
perature runs, 51 and 52 for MRS coal, was used. The char
collected in CY-2 was not taken into account because the
amount collected varied from run to run. Also, the amount
of char collected in CY-2 was small in most of the runs.
Uhen the CY-1 d.a.f. ash content was used in the calcu-
*
lations, the magnitudes of the negative AW values of MRS
chars (see Figure 6.1) were reduced by only about 10%.
Therefore, it is evident that there are other sources of
ash loss and/or bias of the data. No cold runs were made
for MB8 and BZN cools. It was assumed, as an initial es-
timate, that the same percentage of ash was lost as in the
runs with MRS, then these feed ashes were adjusted downward
until they met the criteria developed in Section 6.2 for
upper and lower bounds of weight loss estimates.
6.1.3 Chemical and Physical Ash Losses
The findings of Kobayashi (1976) show that ash losses
due to vaporization are not significant up to 1250°K.
However, Padia (1976) concluded, based on his studies of
ash in coal, that the ash produced from pyrolyzed chars
weighs 4.75 percent less than that produced from un-
pyrolyzed coal. Such behavior was said to be the result
of differences in CaSO4 formation during the ashing of
190
-------
coals preheated in an inert atmosphere and unpyrolyzed
coals. This phenomenon will be referred to as Padia'c
effect in this report. Kobayashi (1976) confirmed Fadia's
findings: his results also show approximately a 5% difference
in the ash produced from pyrolyzed and unpyrolyzed coals.
Padia's findings suggest that the ash-induced error
can be minimized by increasing the ash content of the chars
by about 5%. This correction, coupled with the use of the
reactor at room temperature, should provide better weight
loss estimates. This procedure was used in this study. Vhe
pertinent equations will be developed in Section 0.2.
t.1.4 Mechanical Ash Losses
As indicated in the previous section, ash losses due to
vaporization are not significant up to 1250°K. Another po-
tential source of ash loss is that during fast pyrolysis,
the volatile components of coal escape at very high veloci-
ties, thus carrying with them solid particles which may be
in their path. This hypothesis is supported ty calculations
made by Kobayashi (1976). From experimentally measured gas
flow rates through discs of coal and pressure drops at room
temperature (Karn et a_l., 1975), Kobayashi estimated coal
particle internal pressures of 100 atmospheres. This esti-
mate is not unreasonable considering the high volatile
matter flow rates expected during devolatilixation; it also
191
-------
agrees v.'ith Lewellen's (1975) prediction of a few hundred
atmospheres of internal pressure when a 70 micron particle
is iicated to 1000°C at 10 °C/sec. Moreover, microscopic
examinations of pyrolyzed coal particles usually reveal
"blow holes" on the surface of the char particles (Ilorton
and Snoot, 1975), indicating that volatiles (and probably
ash particles) are released from coals. Inorganic mineral
matter is known to be distributed in coal nore or less uni-
forialy as snail inclusions of variable composition of
approximately 2 urn mean size (Fadia, 1976). £uch small
particles would probably pass through the cyclones used in
this study. Unfortunately, this effect cannot be quanti-
fied.
6.1.5 Tar Condensation
Visual examination of the lucite cyclone hoppers during
the LFR experiments clearly indicated that yellow-brown
fumes of condensing tars deposited on the hopper walls, ihe
coal particles would be expected to act as condensation
nuclei for the tars. The extent of the error in weight loss
and eler.iental evolution measurements that results from this
phenomenon is unknown, but such an effect could be sub-
stantial for mercury and elements which are predominantly
associated with the organic fraction of the coal. This
effect could be reduced by lowering the cooling rate of the
192
-------
coal particles as they pass through the collector and the
cyclones. As in the case of mechanical ash losses, this
effect cannot be quantified.
0.1.6 Lffects of Coal and Gas Feed Rates* and
Residence Time Effect
As shown in Table 6.1, the coal feed rate varied sig-
nificantly in many runs. The evidence suggests that in-
creases in coal feed rate lead to increases in weight loss
ot a given temperature and residence time. This is very
likely due to the turbulence caused by the higher particle
loading at the feeder tip, leading to more rapid heating of
the gas surrounding the coal particles and consequent de-
crease in the value of the particle heating time constant.
This would not necessarily be true at very high coal loading
ratios (as shown by Reidelbach and Algerrcissen, 1976) because
of the increased enthalpy requirement.
An additional complicating factor is introduced by
changes in feeder gas velocity, Uf. The laminar flow re-
actor was operated with a coal feeder gas velocity in a
region where the coal particle heating time constant t
appears to Le highly sensitive to changes in u*. This effect
is shown by taking the partial derivative of equation (4-33)
with respect to u:
193
-------
9T,, -0.5348
II
U
exp
f
5.67238 -
4.0844T
10000
(6-15)
As uf becomes small, 3TH/3Uf becomes large. The value of
uf used in this research is lower than those used by all
other researchers in the field, a consequence of an attempt
to minimize the dispersion of the coal particles and there-
l.y to maximize the particle collection efficiency. Further-
more, the flow controller used in the feeder gas line was
oversized. This may have introduced variations in feeder
gas flow rate that may not have been detected.
Nevertheless, neither changes in coal feed rate nor
fluctuations in Uf could have caused the negative values of
*
AWA. Such changes can only perturb the extent of reaction
at a given T and tj,, i.e., cause scatter in the data. The
conclusion is that negative values of Av:. cannot be due to
experimental error, but must be attributed to ash losses.
Another reason for carrying out runs 51 and 52 was to
determine the effect of residence time on sample recovery.
This was done by using CY-3 as a device to estimate the
fraction of sample which missed the collector. The higher
the sample recovery in CY-3, the larger the amount of coal
that was not collected. The results shown in Table 6.2 in-
dicate that smaller residence times lead to better collection
efficiencies. Because of the design of the reactor collector
194
-------
base, total recovery of the char that misses the collector
is not possible. If it were, the gravimetrically deter-
mined weight loss calculated with the following equations:
- < - *« 3
1 - Mc - Ac
where
weight of char recovered in CY-1, CY-2, and CY-3
RH weight of feed coal
(6-17)
would equal zero in runs 51 and 52. The data in Table 6.2
show that this is not the case. However, it is seen that
the lower the char recovery in CY-3, the better does AW^
estimate Aw (which should be zero for runs 51 and 52).
Therefore, a negligible recovery in CY-3 would imply that
Aw* is a good estimate of Aw*. Table 6.1 shows that several
runs meet this criteria.
6.2 Calculational Procedure for Estimation
of Weight Loss
It has been shown that ash losses bias weight loss
estimates downward. In order to determine the behavior of
volatile components of coal, and of its elemental con-
stituents, it is imperative that the underestimation due
195
-------
to ash losses be corrected. The procedure that was used
for the estimation of d.a.f. weight losses (AW ), and
upper and lower bounds on AV.T is presented in this section.
Dry ash-free weight losses were estimated with the
following equation:
AW* = 1 - -£ (6-lb)
EH
where
*
E * d.a.f. ash fraction of particles collected
during cold runs (see Section 6.1.2)
*
E « d.a.f. ash fraction of char particles
corrected for Padia's effect
*
E is calculated from the equations
E,* - - ^ - ; (6-19)
1 - En - NH
where
£+ . 1'°475AH (6-20)
11 1 + 0.0475AJ
and +
* -
11 1 + 0.0475AJJ
E*j and wj are hypothetical quantities used only to estimate
d.a.f. weight loss. They are the values that the a.r. char's
ash and moisture would have if the yield of ash had not
196
-------
been reduced by the pyrolysis in inert gas (Padia's effect)
as discussed in Section 6.1.3.
*
It should be noted that the values of E~ used are
specific to the apparatus used in this study (e.g. cyclone
sizes, etc.) .
Actual moisture and ash contents of the feeds and
chars are used to carry out all other calculations in the
different bases. Once AVI is calculated using equation
(6-18) weight losses in a.r. and m.f . bases are calculated
using equation (4-47) rearranged to the form
(i- AW*)
wc *
and substituting into equations (4-43) and (4-45). Weight
loss values on the three bases are calculated by computer
program LFRSl listed in Appendix A. 2. The results are in
Appendix B.2.
This method only provides estimates of the true weight
losses, since it does not account for such phenomena as loss
of ash in submicron particles and condensation of tars on
the product char. An indication of the validity of the
estimate could be obtained if upper and lover bounds on the
value of AY; were known; an estimated value of &W* that olid
not fall within these bounds could then be rejected out of
hand .
197
-------
In fact, such bounds can be estimated from conside-
rations presented in Section 6.1. The uncorrected value
of AI-IA, v/hich does not account for ash losses, must be
lower than the true weight loss. (This is easily shown
Mathematically, but the fact that in extreme cases Aw is
A
negative provides a convincing heuristic demonstration.)
Cn the other hand, as was shown in Section 6.1.6, the
values of Aw determined gravimetrically from the weights
of the chars collected in all three cyclones (AVJ..) pro-
vides an upper bound on the pyrolysis weight loss, in that
some material is inevitably not collected in the cyclones.
The weight loss estimates and bounds calculated in
this manner are shown in Table 6.4. The bounds calculated
for MRS coal indicate that the value of L£ used for this
*
coal provided good estimates of AW . However, the upper
*
bounds for some NB8 and BZN runs (AWW) were slightly lower
than the Aw* estimates. Consequently, the EC values for
these coals were adjusted downward until the estimates of
AVJ* were between the bounds. This procedure is particularly
sound in the case of NB8 coal because its high ash content
allowed the calculation of good weight loss estimates by the
ash tracer method for those runs where the extent of decom-
position was high (e.g., run 45). in addition, most of the
char passed through the water-cooled collector, as evidenced
by the zero percent recoveries in CY-3 for the NB8 runs.
198
-------
Table 6.4 Weight Loss Estimates and Bounds
Run
No.
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
%AWJ
(Low
Bound) *
-7.19
-7.65
-6.27
-3.12
-2.58
-1.37
-1.41
-0.30
2.38
-2.32
-0.72
9.55
19.16
8.76
8.84
9.03
16.52
0.43
24.64
0.71
3.81
7.30
-1.60
5.06
23.95
-7.40
8.74
22.80
41.92
20.47
46.35
41.43
31.19
5.74
33.43
26.32
%AW*
0.64
0.21
1.48
4.41
4.90
6.03
5.99
7.02
9.50
5.14
6.63
16.15
25.05
15.41
15.49
15.66
22.61
7.62
30.14
7.95
9.89
13.16
4.82
3.02
22.31
0.00
6.77
21.13
40.67
18.76
45.20
40.17
36.21
12.61
38.29
31.70
%**;
(High
Bound)
•»
-
-
—
-
16.22
14.75
-
-
6.42
22.19
—
-
—
17.27
22.30
•
—
—
•
•
33.33
4.91
—
44.47
13.62
12.97
34.60
44.63
24.37
45.26
41.66
44.02
15.05
46.80
37.57
The actual lower bound must be greater than s«ro
199
-------
Dry ash-free weight losses calculated by the procedure
developed above are shown in Figures 6.7 to 6.9 for MRS,
NB8, and BZN respectively. The procedure used to estimate
AW yielded positive values for all runs except run 40 (for
which AV7 has been set equal to zero) while also indicating
low d.a.f. weight losses at low temperatures and/or resi-
dence times (e.g., MRS coal at 300°C). In addition, $+
values calculated using these estimates of AW approach the
criteria of equation (6-9) as shown in Figure 6.2 to 6.4.
Therefore, it is concluded that the procedure is basically
sound.
Quantitative comparison of the data obtained in this
study with data obtained by other researchers is quite
difficult to obtain, for several reasons. The data ob-
tained by each researcher reflects strongly the equipment
design and run parameters used (e.g., feeder, main, and
suction gas flow rates); residence time calculations and
estimations of particle time-temperature histories depend
on the model used and the assumptions made; and weight
loss estimations by the ash tracer method are dependent on
the efficiency of the devices used to collect the char.
Some researchers simply ignored negative d.a.f. weight losses
calculated by the ash tracer nethod (e.g., Uadzioch and
Ilawksley, 1968), others carried out all their experiments at
200
-------
<3
•»
o
u.'
40 -
400 800
1200 1600
TfiM
Ffgur« 6,7 D.A.F. Weight Lot* of MRS Cool
201
-------
e
<
cl
10 -
200 ~ 400 600
R«sid«A«* Tim* (m «tc)
Flgurt 6.8 D. A.F. Weight Lo«* of
NB8 Cool
202
-------
~ |O800*C
,A.F. W«igM Lo«t of
BZN Cool
too
203
-------
temperatures high enough for the ash loss effect to have
been overcome by the weight loss (e.g., Nsakala, 1976).
Nevertheless, a qualitative comparison is instructive.
The data of Nsakala for a Texas lignite is shown in Figure
6.10. Direct comparison of this figure and Figures 6.7 to
6.9 is not possible because the abscissa in Figure 6.10
corresponds to isothermal reaction times which are approxi-
mately 100 msec lower than the actual residence times. In
addition, Nsakala used a simplistic model, which assumed
a fully developed laminar profile in the reactor, to cal-
culate residence times. Since such is not the case in
reality, the actual residence times for those experiments
should be still larger. Furthermore, Nsakala operated his
reactor with a wall temperature approximately 100°C higher
than the reactor main gas temperature which was reported
to be 808°C in all his runs. Therefore, it is not clear
whether such data should be associated with 800 or 900°C
data in this study. Finally, differences in feeder gas
velocities should have led to different particle temperature*.
time histories which, in turn, would have led to different
weight losses for the same residence times. Nevertheless,
qualitative comparison of the sample data from Nsakala's
work and this study indicates that the same phenomena have
been observed in both studies.
204
-------
*
*
d
•08«C Nominal T
100 ISO
TUM (•••«)
6.10 D.A.P. «*lfM LOM *f T««M
From NMk«l« (!•?•}.
too
205
-------
6.3 Modeling of Coal Pyrolysis
The experimental weight loss data for MRS coal, des-
cribed previously, have been correlated with four models:
Badzioch and Hawksley's (1970) isothermal model, an iso-
thermal first-order model, Kobayashi's (1976) nonisothermal
first-order parallel reactions model, and a nonisothermal
first-order model.
Badzioch and Hawksley's model parameters have been curve-
fitted with the LFR data for MRS coal, using Marquardt's
(1963) nonlinear regression algorithm. The model equation
presented in Section 2.2.4 is
AW* - Q(VM*) (1-C) [l-exp{-At £exp(-B/T)]}] (2-2)
C - exp [-KX(T - K2)] (2-3)
t « tR - 3t (6-22)
where
VM* = ASTM proximate volatile natter of coal oh
° d.a.f. basis.
The value of VM* was determined to be 44.1% for the MRS coaJt
o
fraction used in the LFR experiments. Badzioch and llawksley
(1970) determined that C in their model was equal to 0.14
for all nonswelling coals. Nsakala (1976) used this value
successfully to model the pyrolysis of Montana lignite.
Since the free-swelling index of MRS coal is 0.0, C - 0.14
was used in the curvefit. The values of the other constants
206
-------
wore found to be Q = 1.77, A « 685,75 sec"1, and C »
7356.40°K. The statistical analysis of the regression is
shown in Appendix n.2; the f.AS program used to fit the tKito
(program BADZ) is listed in Appendix A.2. Figure 6.11
shows reasonable agreement between this model and experi-
mental weight loss values? the sum of squares of the error
(or residuals), SSE, is also shown in this figure.
The following first-order isothermal model was fitted
to the MRS-LFR data.
•t
* * *KtI
AV? » AU (1 - e ) (6-23)
CO
where
AVI* = B0 + 82T* (6-24)
K = Be"E/RT (*-25)
PO' P2• ™ constants
P = preexponential factor
r « activation energy
The model parameters were found to be Bo * -3.75, &2 *
4.47xiO~5 °K~2r B m 72.65 sec"1, and E » 8866 cal/mole.
A conparison of the model curvefit and the experimental
data for MRS coal is shown in Figure 6.11. The SAS pro*
gram MCDL6A used to fit the data is in Appendix A,2. The
-,*^T|
statistical analysis of the regression, included in
Appendix B.2, shows that the sun of 0c}iMAre« of the error
207
-------
*
$
-------
(SSE) for this model fit is .smaller than that of Badzioch
and Ilawksley's model data regression.
Among the nonisothermal pyrolysis models available in
the literature, Kobayashi's model was selected because of
its simplicity and proven ability to correlate with a
single set of parameters extensive amounts of devolatili-
zation data for various coals (among them Montana lignite),
This model was described in detail in Section 2.2.4. The
expressions for the overall d.a.f. fractional weight loss
are
t
t / (K! + K2)dt
AU* = / (O^K! + ct2K2)e dt (2-9)
o
-E-L/RT
where KI = Bj_e (2-5)
-E2/RT
K2 = E2e (2-6)
Values given by Kobayashi for the constants are a-, = 0.3,
5 i 7 i
a2 = 1.0, T>i = 2x10 sec ~*-t B2 = 1.3x10 sec x, E± =
25 kcal/mole, and E2 = 40 kcal/mole. These values are
used in subroutine PYROTII (Appendix A. 2) to calculate AV:
for all the coals (MRS, BZN, and NB8) by integrating
equation (2-9) numerically. Particle residence times and
time-temperature histories are calculated as indicated in
Sections 4.2.1 and 4.2.2. Values of AV/ predicted by this
209
-------
model for MRS coal are compared with experimental results
in Figure 6.12. Overall, agreement between the model's
2. rri°ri predictions and the experimental data is very
good. The sun of squares of the residuals is smaller than
those of the two isothermal models, indicating a better
fit. It must be stressed that this model's parameters have
not been fitted to the experimental data, as was done for
the first two models presented in this section.
A first-order nonisothermal model has been developed.
The rate equation (equation (2-1)) is expressed as:
= K(AV7* - AW*) (6-26)
CO
with boundary conditions
* *
f\v -*• AW^ as t •*• «>
AW* = 0 at t = 0
where
-E/RT
K = Be (6-25)
AV7* = f(T).
AN,, is the equilibrium d.a.f. weight loss at a given tempera-
ture. Since K and AW,,, are functions of particle temperature,
which in turn is a function of time, equation (6-26) is a
linear ordinary differential equation which is solved using
an integrating factor, yielding:
210
-------
I I I I
Koboyoshis (1976) Model
First Order
_ Non- Isothtrmol Model
i i i i I
10 20 90 40
Experimental O.A.F. Weight Loss (%&W*)
Figure 6.12 Comparison of Now- Isotfcennol Models' Prediction
211
-------
t
/ K dt
t o
. c ""- e at _ (s.27)
t
/ K dt
o
e
For isothermal conditions, this equation reduces to the
familiar form of equation (6-23).
Values of AVJ^, at each temperature can be obtained from
estimates of the asymptotic d.a.f. weight losses in the
LFR experiments. As discussed in Section 5, Duhne (1977)
has shown that estimates of asymptotic equilibrium values
for physical and chemical processes can be estimated from
linear regression of an equation of the form:
At;* = £_ + AW* , AW* > 0.67AW* (6-28)
tR °°
where
= constant
K. » AVv* as tR * «
If the highest value of AU* at each temperature is assumed
to be very close to At.'*, the requirement that the data lie
within 33% of the equilibrium point (relative to a total
range between zero and equilibrium) is usually satisfied
usinq the three highest AW points at each temperature.
Values of AW* estimated in this manner for MRP coal are
compared with batch equilibrium d.a.f. weight loss values
(AW*) in Figure 6.13. The LFR asymptotic values of Aw* are
212
-------
50
40
<3
5? 30
b.
•
<
•
O
10
x«
MRS cool
X
O
X
O
200
400 600
T«mp«ratir« (*C)
•00 IOOO
Flgur* 6.13 CoMpariMn of LFR Atymptotio O.A.P.
W«i«ht Lo«»«« «Hh BotehO.A.F.
LOMM
213
-------
consistently below the Latch values. The implication of
these findings is that high heating rates during fast
pyrolysis (3 x 103-104 oc/sec) lead to lower asymptotic »
weight losses than do the lower heating rates (5-45°C/sec)
that characterize batch pyrolysis.
These observations are consistent with findings by
other researchers. Kobayashi observed the same effect under
higher heating rate conditions (104-105 °C/sec) . lie found
that, during laminar flow experiments with small residence
times (5-220 msec), Montana lignite weight loss curves
leveled off at significantly lower values than the asymp-
totic values of weight loss at long residence times. lie
determined that for this lignite at 1240°C, about 30% of
its d.a.f. weight was lost in 100 msec, 45% in 1 sec, and
b7i in about 10 minutes. Lorton (1979) reviewed the fast
pyrolysis data of several researchers and noted that a
quasi-equilibrium is reached before the particles are
quenched. I!e indicated that, judged by the reaction
occurring during the experiments, the pyrolysis process
appears complete; however, additional devolatilization
occurs when the quenched char particles are reheated in a
batch reactor. The implication is that if the particles
had been maintained longer at the elevated temperature in
the primary reactor (wire screen or laminar flow), the
additional pyrolysis would have occurred there. In ad-
dition, the enhanced fast pyrolysis volatile yields reported
214
-------
in the literature usually include the weight loss during
fast pyrolysis and the additional proximate furnace vola-
tile matter loss. However, low rank coals have been found
to show relatively snail enhancement in volatile yields,
Therefore, finding that fast pyrolysis quasi-equilibriua
volatile yields are below slow pyrolysis equilibrium
values is reasonable.
Comparison of the fast pyrolysis data obtained in this
study for Montana Rosebud with that of Suuberg et al. (1978)
for Montana lignite is shown in Figure 6.14. The agreement
between the two sets of data obtained with two different
heating rates (10 °C/sec in Suuberg et aJL. *s experiments
and 3x10 3-104 °C/sec in this study) is entirely consistent
with the discussion presented by Kobayashi (1976) regarding
the effect of heating rates on volatile yields. He pointed
out that there should not be any such effect for heating
rates above the critical heating rate calculated with the
equation
[AT\
H *
I *r+
IT * ** <6-W)
where
E » first-order activation energy
Tw » temperature at t » -
B * preexponential factor
R » gas law constant
Z15
-------
d
w
1
so
*l 40
*
20
10
X« Su«b«r« at at'« (1978) data
O*thi« ftadyt data
MRS coal
x O
XX
'-* .,JV •; .'.;.-.
^ '.L.:.i'; 'fi, ' I
20O
400
tOO
too 1000
"^
6.14 Co*oar!s«« of Wlra-S«raana«4
Q««*i-ea»illftri«* O.A.P. W«lflM LM« data
216
-------
Kobayashi estimated that the critical heating rate of
Montana lignite is 336°C/sec. Both Suuberg's and this
study's heating rates are well above this value.
The values of AW calculated from the regression of
equation (6-28) have been curvefitted with a fourth order
polynomial
6o
where
temperature, °K.
The values of this equation's parajMters were found to bet
60 - -1.78, 0X - 1.14, 62 - -0.259, 63 - 2.48XKT2, &4 «
-8.23xl
-------
of E and B; however, the trial and error procedure was
found to be satisfactory.
The agreement between the model predictions and the
experimental data is shown in Figure 6.12. The sum of
squares of the residuals (SSE in the figures) for this
model is less than the values for all the other models
tested.
The preceding discussion suggests that none of the
fast pyrolysis models developed to date can be used in
real gasification systems where the heating rates may be
lower (e.g., because of particle size or bed depth) and
typical residence times may be in the order of minutes.
In order to provide a model which can be used to obtain
engineering estimates of the weight loss of coal during
the devolatilization stage in fluidized bed gasifiers , the
equilibrium batch d.a.f. weight loss data for each of the
five coals studied have been curvefitted with equation
(6-29). The nonisothermal first-order model (equation
(6-27)) will then yield asymptotic d.a.f. weight loss
values in agreement with the results of 20 minute resi-
dence time batch experiments. Dry ash-free weight losses
predicted by this method (using the sane Arrhenius para-
meters) are shown in Appendix B.2. The sum of squares of
the errors of this model fit for MRS coal to the fast
pyrolysis data is 421.0 which is larger than that obtained
with the LFR data; it is not unreasonable, however.
218
-------
In order to allow prediction of AH+ and AM?, the rate
of moisture evolution must be modeled. The following
equation is used as a rough estimate of the Moisture re-
lease rate:
(€-31)
M =
where
« ASTM moisture content of coal
AH « enthalpy of vaporisation of water plus
sensible heat for 600<>C steam (15943
cal/mole)
BM * empirical constant evaluated using MRS*
LFR data at 300^0 (1.652x10* sec-1).
Kobayashi's model and the first-order nonisothermal model
with batch Aw* values, coupled with the above moisture
evolution model, yield the predictions shown in Appendix
A. 2. The same appendix also contains the experimental data
calculated in a.r., m.f., and d.a.f. bases. This model pre-
dicts very rapid evolution of moisture, which is what is
observed experimentally. All of the runs except run 15
(the 300°C run) produced chars with loss than 21 moisture.
Such residual moisture is probably due to water conden-
sation in the collector and cyclones, and to laboratory
humidity absorption during sampling. This hypothesis is
supported by the random variations in char moisture at a
given T and tR.
219
-------
6.4 Analysis of Elemental Release Results
The results discussed in Section 5 show that several
elements axe evolved in significant quantities during batch
pyrolysis. It is of interest to determine how fast the
elemental release occurs. For this purpose, the chars pro-
duced in the laminar flow reactor (LFR) experiments were
analyzed for several major, minor, and trace elements in
addition to moisture, ash, and volatile matter. A listing
of the results of the analyses of the coals and chars is
shown in Appendix B.2.
The objectives of these experiments were to determine,
in a qualitative manner, the behavior of the elemental
components of coal during fast pyrolysis, and to develop
and test a kinetic model for the elemental release. Sam-
pling and analysis problems made determination of meaning-
ful kinetic parameters difficult (modifications that might
allow such determination will be discussed subsequently).
Therefore, simple statistical trend analysis was used to
determine whether different elements were retained com-
pletely in the chars, or whether they were evolved to a
measurable extent. Orcter-of-magnitude estimates of the
kinetic parameters for sulfur were obtained for a first-
order, single reaction, elemental release model.
Different groups of chars were analysed for different
species, for a variety of reasons. Excellent data on the
220
-------
release of the major elements and volatile matter during
fast pyrolysis is available (Kobayashi, 1976) for tempera-
tures above 800°C (see Figures 2.1 and 2.2). Therefore,
C, H, and ASTM proximate volatile matter were analyzed only
for low temperature runs. Because of its importance in
coal gasification, sulfur was analyzed in all chars. Most
trace and minor elements studied were analyzed only for
the higher temperature runs in order to diminish the effect
of tracer-introduced bias. Three 800°C runs with MRS coal
were analyzed for all the elements that the analytical
capabilities permitted. These runs have been used primarily
to find element-to-element correlations.
Values of H and H were calculated using equations
(4-48) and (4-49), and were also calculated using the ash
tracer method without any corrections. Equation (6-5) was
used for this purpose. To distinguish $ and $ values de-
termined using the best AW estimates from those determined
using the uncorrected ash tracer method, the latter are
subscripted with an "A" instead of an "H". The values of
XH' ^H» ^A» XH» ^H' ^A' ^H' and *A ^or a1^" elements and
coals were linearly regressed versus residence time and
temperature. The computer program (LFRSl) used for this
purpose is listed in Appendix A.2. The results of the data
analysis are tabulated in Appendix B.2.
The mathematical analyses of the time-temperature
histories of the coal particles and the pyrolysis modeling
221
-------
studies suggest that using linear regression to detect
trends with time and temperature may be an oversimplifi-
cation. However, in view of the scatter in the data and
the uncertainties in the residence time, particle tempera-
ture, and weight loss estimations, a more complex approach
is not justified. To allow for all these uncertainties, a
less stringent criterion than in the batch data analysis
was used to test for significance of the slopes with res-
pect to time and temperature, significance at the 80, 90,
and 98 percent confidence levels was determined using a
two-tailed t-test.
Tables 6.5 to 6.7 present the element-temperature and
element-time correlations. Caution must be exercised in
the interpretation of these tables. The uncertainties in
the sampling, analysis, and weight loss estimations may
have biased some results. In addition, the temperature and
residence time ranges over which the chars were analyzed
vary for different elements and coals. The reader is urged
to refer to Appendix B.2 for the specific T and tR values
which were used in the multiple linear regression of a
specific element. The conclusions shown in Tables 6.5 to
6.7 are valid only within the range of T and tR values of
the chars analyzed for each element.
The element-time correlations for the MRS coal runs
(25, 27, and 29) are shown in Table 6.8. As indicated be-
fore, special precautions were taken in the LPR operation.
222
-------
Table 6.5 Significant Correlations in MRS-LFR Chars
MRS Coal
ELEMENT
Vblatile Matter
C
H
S
Mn
Cu
Al
Ce
Se
Th
Cr
Sm
U
La
As
Sb(by NAA)
Element-Temperature
Correlations
xs
+
X-
XX
~X
-X
•+•
*
+
*
+
XX
Yt
XX
XX
XX
+
XV
•+•
e
+
X-
YH*
XX
XX
XX-
XX
•X
+
-X
+
X-
0A
XX
X*
XX
+
XX
0
+*
0H
XX
XX
XX-
XX
-X
-h
X
A
A
A
Element- Time
Correlations
x£
V V
+
©
+e
X
^®
-h
-X
x~x
%*
X-
X-
XX
V*
TH
©
X-
X-
x~x
0A
X-
X-
XX
0H
A
©
X-
•X-
A
XX
© = significant at 80% C.L.
X = significant at 90% C.L.
XX = significant at 98% C.L.
A = undetected trend
— = negatively correlated
-I- = positively correlated
223
-------
Tobla 6.5 ( Continued )
MRS Coal
ELEMENT
Br
No
K
Ti
Sc
EL-
Ru
Fe
Co
Zn
Cs
. P
Ha
=»b
V(byAA)
V(by MA A)
Sb(byAA)
Eltment- l»mp«rotur«
Corrt tot Ions
x£
e
"e
*
~i
4-
If
4-
**
Htf
WA
1
"*~ t •*" t" •*"
I
t
"e
"^
~^
$
^^
.
if
.
"^
!
^t-
*
if
e
0H
El«m*nt- Time
Correlations
XH*
® "&
t-
»
A
If
!
e
Vi/*
!
i
^
e
^j
$
if
Ntf
"""
0A
0H
's
i !
"e
^
if** #
e
e
£
Klf
0 3 significant of 60% C.I.
If = significant at 90% C.I.
** * significant at 98% C.L.
& * undtttcttd trtnd
— > ntgotlv«ly corrtlottd
+ * positively corr«lat«d
224
-------
Table 6.6 Significant Correlations in NB8-LFR Chars
NB8 Coal
ELEMENT
S
Se
Sm
La
As
Sb(by NAAJ
Hg
Pb
Fe
Co
Th
Sc
V(byNAA)
Element- Temperature
Correlations
Y*
AH
-r-
X
•»•
*
+
*•
+
*
Yl
-t-
*
•»-
*
+
*
-t-
X*
vs
-t-
®
H-
*
0
0A
-1-
^
-t-
^
-»-
^t-
H-
*•*
0H
A
A
A
A
+
•X-
A
A
+
e
Element- Time
Correlations
y*
AH
X-
+
**
vt
©
*
*
"i
**
YS
~e
~©
0A
e
•H-
*
©
X*
0H
A
A
A
~®
"©
= significant at 80% C.L.
= significant at 90% C.L.
= significant at 98% C.L.
A = undetected trend
- = negatively correlated
-I- = positively correlated
225
-------
Toble 6.7 Significant Correlations in BZN-LFR Chars
BZNCoal
ELEMENT
S
Se
Sm
Sb(by NAA)
Hg
Pb
Sc
Fe
Co
La
Th
As
V (by NAA)
Element- Ttn
Corrttat
XS
4-
®
•+-
X.
n
•»-
e
•4-
tt.*.
W
e
*
+
«..*
Vtrotura
on»
0A
e
-1-
4M(
0H
e
A
*
A
Z
-i-
K»
Etem«
Cor
X*H
e
•f-
*
-«-
*
H*
»
4-
*
"e
4-
**
nt- Tlmt
[••lotions
^
«
4-
*
4-
*4t
0A
»
4-
•M
e
4-
««
0H
•*
A
A
A
+•
*
A
A
•f
K «
© s significant at 80% C.L.
* = significant at 90% C.L.
** - significant at 98% C.L.
A = undetected trend
— = negatively correlated
+ = positively correlated
226
-------
Table 6,8 Significant Element-Time Correlations
at 800°C-MRS Runs 25,27, and 29
Element
C
H
S
Sm
U
La
As
Sb(byNAA)
Br
Na
K
Ti
Mn
Cu
V(by NAA)
Al
y*
AH
4-
*
*
4
©
+
•*
©
4-
*•*-
4
*
w*
TA
~©
**
*
+e
~<8
' •
VLlt
TH
**
*
•x-^-
-»-
©
•X-
-X-
©
-
0A
©
**
*
~®
©
.•
• .
0H
**•
*-
^K
A
©
A
A
*
-X-
""©
Element
Ce
Se
Th
Cr
Sc
Eu
Ru
Fe
Co
Zn
Cs
P
" ' • Pb
:Hgs
Sb(by AA)
V(by NAA)
Y*
AH
-h
X
*".:
+
*
+
X, «
w
4-
X^*
+
©
"
+
^
VA*
••
V*
TH
< i
r
0A
0H
A
A
A
© = significant at 80% C.L,
•X = significant at 90% C.L.
##• = significant at 98% C.L.
A = undetected trend
— = negatively correlated
+ = positively correlated
227
-------
char sampling/ and chemical analyses for these runs. In
addition/ these chars were analyzed for all the elements
which the analytical capabilities available allowed. There-
fore/ the results shown in Table 6.8 provide some additional
information which becomes apparent with the improved pre-
cision in the results.
The most important interpretations of the information
presented in Tables 6.5 to 6.8 are as follows:
1. If XH versus T has a slope significantly greater than
zero, indicating that the char is becoming progres-
sively enriched in the element as the temperature in-
creases/ and i|>u versus T also has a slope equal to or
greater than zero (the latter due to underestimation
of AW*)/ indicating that the absolute amount of element
in the char is remaining constant/ it may be inferred
that the element is not released from the particle in
the temperature range investigated.
2. If the slope of Xu versus T is greater than zero, and
f
that of ^u versus T is less than zero/ the element
n
release from the coal is significant/ but its frac-
tional release at a given temperature is less than that
of total volatile matter.
3. If the slope of X^ versus T is equal to zero, and that
of ty versus T is less than zero, the element release
H
is proportionally equal to the total volatile matter
release.
228
-------
4. if the slopes of XA and ^ versus T are less than zero,
H H
the element release is proportionally greater than the
total volatile matter release.
5,. If the slope of <|>H versus T is equal to zero (indi-
cating that no appreciable amounts of the element have
been released), the element is not volatile. If it is
less than zero, the element is released to an extent
that increases with increasing temperature. The be-
havior of 4>u simply mirrors that of ^u, but, more im-
n n
portantly, it allows easy visual examination of the
data. Since must range from 1.0 (complete retention)
to 0.0 (no retention), its value quickly indicates
whether the element is retained in the char or not.
This is useful because the linear regression analyses
only detect consistent trends. If a retention (4>H)
drops at low temperatures and then remains at a con-
stant value, no linear trend would be found; neverthe-
less, it would be obvious that the element had been
released from the coal.
6. Those elements for which the slope of X versus T is
f
positive, the slope of \\> versus T is zero (indicating
A
that the mass fraction normalized by the uncorrected
ash tracer method meets the equality criterion of
equation (6-15)), and the slope of ty^ versus T is
negative, are released to the same extent (with res-
pect to T) as the ash is lost from the particles.
229
-------
7. If the slopes of X and $ versus T are positive, and
t H A
the slope of ipH versus T is negative, the element is
released to a lesser extent (as a function of T) than
the ash.
8. If the slope of X^ versus T is positive and that of
<»A versus T is negative, the element is released to
an extent large enough to overcome the ash loss
effect.
9. The interpretations of the slopes of Xj, <*„, and $„
MM H
versus t_ are similar to those versus T. The rates
R
of elemental release are compared with AW versus t_
instead of AH versus T, however.
10. If no slopes are found to be significantly different
from zero for a given element, either the scatter in
the data is too large for any significant trends to
be detected, or the trends cannot be detected through
linear regression. Obvious undetected trends have
been marked as such in Tables 6.5 to 6.7.
Because of the large number of elements studied, an
element-by-element interpretation of the results shown in
Tables 6.5 to 6.7 is out of the question. However, since
the roost important information needed for the evaluation of
the fate of trace elements in coal gasification processes
concerns the release of volatile elements, only one non-
volatile element will be examined as an example. The bulk
230
-------
of the discussion that follows concerns only elements which
show significant release from the coals studied.
As was the case in the batch experiments, certain ele-
ments exhibit similar behavior in every coal. Furthermore,
the elements found to be released in significant quantities
during the batch experiments generally show the same be-
havior during LFR experiments. Nonvolatile elements are
retained in the chars: Xjj for those elements increases with
time and temperature, while fy remains at the feed value.
As an example, the behavior of iron is shown in Figure 6.15.
The slight increase in the <|> values in this figure suggest
H
that weight losses are still being underestimated slightly.
Other such elements include Co, Fe, Sc, Na, Ce, and V.
Other elements are found to be released in significant
quantities, as evidenced by the negative slope of ipjjj or H.
Such elements include C, H, S, Pb, Hg, As, Se, and La. Sm
and Mn appear to be released from some chars at the same
rate as the ash.
Temperature appears to be a more important factor
than residence time in determining the extent of elemental
release during fast pyrolysis of the three coals studied
over the temperature range (800 to 900°C for most elements)
and residence times (250 to 800 msec for most elements)
covered in this study. This is indicated by the larger
number of significant trends with respect to temperature
found in the multiple linear regressions.
231
-------
c*
"s.
a>
g
c
o
o
0
ul
#Svjot8Q^q
-
-
MRS
-
NB8
m
m
m
i iBZ^I
3.0
400
8(
1200
1600
R«sid«nce Time
Figure 6.15 Iron Mqss Fraction In Chars
232
-------
The data presented in Figure 6.16 complement the
findings of Kobayashi (1976) which were shown in Figures
2.1 and 2.2. Carbon, hydrogen, and ASTM volatile matter
are shown to be released from the coal particles in
measurable quantities during fast pyrolysis at low tem-
peratures. Figures 6.17 to 6.22 show the decrease in re-
tention (4>H) of La, As, Se, S, Pb, and ilg as functions of
T and tR. The solid lines in Figure 6.20 are model pre-
dictions which will be discussed later. The trends are
evident in all cases: $„ decreases as T and/or tR in-
crease. However, it is also evident that the scatter in
the data and the uncertainties in the manner in which the
weight loss data had to be estimated would make the deter-
mination of meaningful kinetic parameters quite difficult.
Another-problem, perhaps more important than random
data scatter, is the consistent pattern of retention values
greater than 100% at low temperatures and/or reaction times
(i.e., low extents of devolatilization), This problem is
severe in the case of the medium volatility elements (La,
Se, and As). It appears that the weight loss estimations
are still biased low, at least at low temperatures and/or
residence times. As indicated before, other evidence that
substantiates this observation is seen in Figure 6.15} the
normalized iron mass fractions l^> are consistently above
the feed mass fraction. They should have been equal to
zero, within experimental error.
233
-------
i <:u
100
80
60
1 20
r 100
£ 80
c
o
1 60
a> ' -
-------
100 •
aoL
60
40
1201
100
80
60
40
100
80
60
40
20
0
'o ' H
n
LJ
an
O ° o
o
-
MRS
D
o o o
a
-
N88
D a
a * eoo»c
O * 900°C
m
BZN
i i i i i i i
20
20
400
800
I ZOO
ieoo
Residence Tim«(iiiMc)
Figure 6.17 Lanthanum Retention in LFR Chart
235
-------
X
^SL
^^x
e
3
c
IOU
140
120
too
80
160
I4O
120
100
80
120
80
60
40
20
°<
ru — i 1 i 1 1 1
3 = 3D
D
So
^D MRS
O u
a a
a
O O
.
0
a
0
NB8
a
• ^
a
* = 30O»C
X s400°C
* *j s o^j^J C
i^i S ^14^^^^^*
- 3 * 3D °
BZN
i i i i i i l
6
6
) 400 80O 1200 1600
Residence Time (msec)
Figure 6.18 Arsenic Retention in LFR Chars
236
-------
o
I4O
120
100
80
60
100
80
60
40
20
too
80
60
40
9f)
Cw
a
t — .1 D i i i i •
—
° z D
°°0 1
\
0 ° |
J
/-> HRS
__ " Q
^ I08n
' ° 0°
-
o
o
-
N88
Q
.
8^ &f*4\Gf*
9 6QQ C n
» eoo*c D
O*9OO*C Q
2 « 2Q
BZN
i i i i I i i
40
0
400 800 1200
Rtsltfwic* TtMt
1000
Flgurt 6.19 S«l«niim Retention in UK
Char*
237
-------
-------
2
c
w
«
ac
too
8O
60
40
20
1 20
1 00
80
60
40
100
80
60
40
20
O
i i LJ i i i i i
A o a
nO °O O i
° - o , • .
-
-
MRS
o"D ° D
o o
o
-
-
NB8
D
a a
» «
" X«4OO*C
„ a«8oo»c
BZN
i i i i i i i
0
40O BOO I2OO
R««ld«nct Ti«« (MMC )
Flgurt 6.21 Hod fUtwtio* in LFR CMr«
I60O
239
-------
o
I
80
60
40
20
120
100
80
60
40
120
100
80
60
40
9rt
1 1 LJ 1 1 1 1
D
D
£
*BJ
&
0 °
-
(§1 D D
-
D
D
o o
O
-
a
D
*
m
D
£ = 6
a = 8
r 0=9
1 1 1 1 1 I
I
~
-
-
-
MRS
-
-
-
-
NB8
—
-
oo°c
oo°c
00«C -
BZN
20
400 800 1200
Residence Time (msec)
Figure 6.22 Mercury Retention in LFR Chars
1600
240
-------
Despite these problems, the trends shown by the re-
tention curves of the volatile trace elements suggest that
if the weight loss bias problem could be solved* the cal-
culation of kinetic parameters would be possible. To that
end, a first-order model is proposed for the evolution of
trace elements in coal during fast pyrolysis. The rate
equation is:
—- - «vi«*. - AQB) (6-32)
at •*
with boundary conditions
AOfl * AQ«» as t •» ••
A&JI - 0 at t - 0
where
-E/RT
K -Be (6-25)
Aft* is the equilibrium d.a.f . elemental release at a given
temperature. Since K and Afl. are functions of particle
temperature, which in turn is a function of time, equation
(6-32) is a linear ordinary differential equation which ie
solved using an integrating factor, yielding
f /
J KAQ^e °
K dt
dt
/ K dt
e o
241
-------
Values of Afl. at each temperature can be obtained using
the batch model in the form
-b(T - Tc)
AQ,, = H(T - Tc)» {Afie[l - e ]} (6-34)
Particle residence times and time-temperature histories are
calculated as indicated in sections 4.2.1 and 4.2.2.
The isothermal form of equation (6-33)
-Kt
AftH - Afl.d - e ) (6-35)
was used to determine model parameters for sulfur as a
rough estimating procedure. Linear regression of t versus
£n[l - Aftjj/Aftao] yielded -1/K as the slope; linear re-
gression of ln(K) versus 1/T yielded in(B) as the intercept
and -E/R as the slope. All the 800 and 900°C data for this
element (in the three coals) was used for the regressions
except for the AftH of run 31 which is an obvious outlier.
The value of the frequency factor was found to be B -
171.113 sec"1 with a value of the activation energy E *
2.502x10* caI/mole. No parameters could be obtained for
the other volatile elements because of the high degree of
scatter and/or bias of their AflH data.
Integration of equation (6-33) was carried out using
Simpson's rule with program LTEMP listed in Appendix A.2.
The model predictions (tabulated in Appendix B.2) are
242
-------
shown as the solid lines in Figure 6.20 (as $„» 1 - Afly) .
The model's predictions appear to be qualitatively correct.
These results suggest that given the proper Arrhenius
parameters the proposed model could predict the elemental
release data reasonably well. Regardless of the precision
of the kinetic parameters obtained , the model predictions
must converge to the batch model values CaQw) thus ensuring
the model's accuracy at long residence times (on the order
of minutes) .
In the absence of good kinetic parameters for the ele-
mental release model, the following findings may be used as
rules of thumb to predict rates of elemental release.
several volatile species have been found to be highly cor re*
lated with AW (d.a.f. weight loss); they are shown in
Table 6.9. ASTM volatile natter (VM), carbon, and hydrogen
were expected to be highly correlated sine* they sjoat be
evolved for the coal to lose weight. The other volatile
elements appear to be released at a rate proportional to
the d.a.f. weight loss. Therefore, the proportionality
constant (slope) may be used to predict the rate of el«
tal release through the equation
for
243
-------
Table 6.9 Significant D.A.F. Weight Loss-Element Correlations
Coal
Element/
taalyte
VM
C
; H
S
Se
t
AS
1
Sm
Hg
1
t
Pb
MRS
"Slope
(K.)
1.2
0.80
1.8
1.2
2.1
1.6
0.94
0.97
0.37
r2
0.93
0.89
0.95
0.87
0.41
0.44
0.49
0.24
0.05
•f
* *
•f
* *
•f
* *
•f
* *
+
* *
•f
* *
•f
* *
+
•
NB8
S"lope
(K.)
NA
NA
NA
1.1
1.1
0.89
0.72
1.2
0.66
r2
0.97
0.55
0.26
0.86
0.86
0.54
+
* *
•f
* *
•f
•
•f
* *
+
* *
•f
* *
BZN
slope
(KB)
NA
NA
NA
1.7
5.6
2.7
-0.90
1.7
2.6
r2
0.98
0.80
0.33
0.33
0.30
0.90
+
* *
+
•
+
*
-------
where
K « proportionality constant (slope).
s
This model presumes that an element is not released until
T reaches TC, and thereafter is released at a rate pro-
portional to the rate of d.a.f. coal weight loss (calcu-
lated with a suitable model or determined experimentally).
The proportionality constant was estimated from linear re-
gression of Aftjj versus AW* with program LFRS1 shown in
Appendix A.2. The results of the regressions for all ele-
ments and coals are listed in Appendix B.2. This model
should provide rough estimates for the release of S, Hg,
Pb, As, and Se as functions of temperature and residence
time, during the pyrolysis of pulverised coal. In the ab-
sence of better data, the model parameters should be use-
ful for use with low rank and medium rank coals.
The finding that K8 for sulfur is roughly equal to
one, determined during transient and equilibrium batch
pyrolysis, is confirmed by the values shown in Table 6.S.
Therefore, equation (5-7) is found to hold for all pyroly-
sis conditions.
As indicated previously, samples from MRS coal runs
at 800°c were analyzed for all the elements that the
analytical capabilities allowed. The mass fractions of
these elements were correlated with one another. The reason
for this data analysis (carried out with program If USA
listed in Appendix A.2) was to determine significant element*
245
-------
to-element correlations (XH to X ). The significant corre-
lations are shown in Table 6.10. These results are not as
consistent as were those of the batch runs; however, the
same general behavior is apparent. Low volatility elements
tend to be positively correlated with one another; high
volatility elements tend to be negatively correlated with
low volatility elements; and high volatility elements tend
not to be correlated with other high volatility elements.
246
-------
ELEMENT
Table 6.10 Significant Element-Element Correlations
c
H
S
S
-------
7. APPLICABILITY OF RESULTS TO A
PILOT PLANT GASIFIER
An air/oxygen-steam fluidized bed gasifier is being
operated at the Chemical Engineering Department at North
Carolina State University. This facility would have pro-
vided an excellent pilot plant scale test of the elemen-
tal release model. Unfortunately, throughout the duration
of this study, a chemical grade coke with very small vola-
tile matter content was used as the feed stock. The coke
was made from Western Kentucky No. 11 coal at a coking
temperature range of 1600 to 2000°F. Since this study
focused on the behavior of trace and minor elements during
the devolatilization stage of coal gasification, its appli-
cability to the elemental behavior of a predevolatilized
coal is limited. Nevertheless, some useful comparisons can
be made. In particular, the ash tracer technique can be
used to gain useful insights on the behavior of trace and
minor elements during the gasification process.
7.1 Plant Description
Descriptions of the pilot plant are contained in Ferrell
et al. (1977a, b). A schematic diagram of the gasifier and
the particulates, condensables, and solubles (PCS) removal
system is shown in Figure 7.1. Solid samples are obtained
from the coal feed hopper, char receiver, and cyclone;
248
-------
Rz PURGE
mnt
MONITOftCD VMIAIU*
T TEWEMTUKE
DT DIFFWfKTIAL PRfUME
f HJt>* MTf
L UVft
rigor* 7.1
HMlimt
&&lfia A» PARTJCUMTE, COOCtSABUS AM) SOU9LES (PCS) (STOVAL SYSTL1
SMOMIK PWOSS VMIA&E flWITORl* POIXTS
-------
liquid samples are obtained from the PCS tank; and gas
samples are obtained from a gas sampling port located
between the cyclone and the venturi scrubber.
7.2 Elemental Balances
Major, minor, and trace element material balances
have been made for Run GO-15, carried out on April 3, 1979.
The run consisted of the steam-oxygen gasification of
Western Kentucky No. 11 coke of 10x80 mesh size.
Nominal operating conditions for the reactor were
100 psig and 1800°F with a char feed rate of 25 Ib/hr, a
steam feed rate of 25 Ib/hr, and a bed height of 38 inches.
A carbon conversion of approximately 42% and a make gas
flow rate at the PCS system exit of 11.2 SCFM were ob-
tained. The principal operating variables of the run,
selected output variables, and major element material
balances are evaluated and logged with an existing data
logging and analysis program. The program output is shown
in Table 7.1.
All material balances correspond to the steady state
portion of the run. Key process variables are shown in
Figure 7.2. The data plotted were taken from a printed log
made during the run and are actual values of the variables
as computed and logged by a computer-based data acquisition
system.
250
-------
Table 7.1
Operating Variables for GO-15
DUN CO-IS
• NCSU BEPA«T«£»T OF CMEHICAL ENGINEERING •
• FLUIOIZED e't> COAL GASIFICATIO* REACTOR •
t «
fl/1/74, 5U5-6I05 PUN* a ON EXPTL PLAN
REACTOR SPECIFICATIONS
ro
in
JED SilHEifv? 2:8 in'. lo:i?2HnETEp!s)
- Estlf ATEO BEO-VOIOAOC-»-0;T9 :
SOLID FEED PROPERTIES
COAL
?tD EiPtupiriN FACTOR a j,9i
ST1"*TCD dfcAK KATC • O.T6 SCFH
SETTLE
-AVfcRAGL .
A-R MOISTURE
NTLC4Y a|| COAL CHAR. 15*8} "ESH
DNCFNSITT 8 U6*2 LB/FTlTl .l OEC.C
ZU.AO LO/HR AT J6S.J DCC.F
11.20 KG/MR AT |85.2 DEC.C
-6.»Z-t.B/*P--*T 76.9 -PFG.F-
O.CO KC'HW AT 20.9 DEG.C
.67 L;/MR AT
E:< • $*03
"PURSE ~«ii. «"~9.76"LB/HH
4,0 DFG.F
6ll
-------
Table 7.1 continued
Output Variables And Material Balances For GO-15
BUI.
- is
ti/l/79, Sl!5-til«5 RUN «« ON
f{.m
VAMUPLF.S
OUTPUT
PRESSURE DROP. 6VEP. 20-l
PCS.tiAS FLO* HATE
-CYCLO»t7.6*S- F.LOi>-,«»m
SOLID .HOLDUP
• ?,fl
( 0.260 PSD
ile6 LH-«OLE/HS (11.20 SCFH
1.87 UK-rClE/miO"* BASIS)
"•£•67 I. h*flOtE/HH" ( I ft, 00 SCFl"
1.87 le-rCLf/[ CB en *f"iiti"cEO/.LH cni
tM" 1 cij's ^-"Lili ?^ '^^?£t3. ?9*ttrtAF?
HE»TINC VALUE or MAKE,.G»S n 30*8,5 BTU/L»
o-^-eftlii^ "(J/.Kl —
HEATING V«LUE or SHEET SAS
6117.7 BTU/L?
270.1 RTU/lrF
CONCISION =
SCLIO
B4L*NCt
COAL >ED
SPENT CKAR COLLECTED
-ClCLOuC t>US1 COf
COAL GASIFIED
LLtCIED-- i
-7
86
:UI
:StS
56.11 UF ftfO
JI2* UF FEED
14.61 OF FtED
..._.... „ CHiH BEHQvAL R*TE
KJ/X6 CMAft SATE FOR-MASS RALAhCt
l«.9 I.B/HH
C*S
PCS EXIT CAS AN^L'SIS
**Uflm BALANCES I FLO«3 IN UB/MR
KA3S • • .C
COAL 25.5 20.81
P.AStS _^°.l 0.0
SH » R I u , 0
US' 0.8
CASES 59,0
--„ EwiTtR 0.0
Tni*L OUTPUT 72.8
DIFFERENCE -—!.«* -«.SX
C.flO
2,78
0.5S
10.62
o'.t olo
0.01
lO'.fl
0.02
-S1^
'o.'c
1.4S .t.OX
0,26
{6,'U I
•0,12
Oigi
o|o
17.10
2.«t
- a —
0.600
o.o.
«.1S1
0.021
oio
-7.lt-
-------
rvi
en
• TT-201
ATT-202
• U-203
Selected Process V«H«b1ei
for (tun ft)-IS
14 M IfcOl l*:00
Figure 7.2 GO-15 Run Summary
17:00
11:00
II00
-------
Gas samples were obtained at each sampling time shown
in Figure 7.2. Water samples were obtained at the first
and the last sampling times. Solid samples were obtained
before and after the run. A special sampling device was
used to obtain feed coke, char, and cyclone fines samples
corresponding to the steady state portion of the run.
A summary of the solid sample analyses is shown in
Table 7.2. All the analyses shown in this table, except
that for sulfur, were carried out using ASTM standard pro-
cedures. Sulfur was analyzed using a Fisher Model 470
sulfur analyzer.
The results of minor and trace element analyses are
shown in Table 7.3.
Trace and minor element balances around the gasifier-
PCS system are calculated as follows:
Basis: 1.5 hours at steady state
- TEH + TEF + TEpCS + TE6 (7.1}
where TEC » weight of element entering in feed coke
TEU * weight of element leaving in spent char
H
TEF - weight of element leaving in cyclone fines
TEpcs« weight of element accumulated in PCS tank
during steady state
TEG « weight of element leaving PCS in the gas.
254
-------
Table 7.2 Suimary of Solid Sample Analyses
no
tn
Sieve and Moisture Analysis
Sample
Feed Coke
Spent Char
Cyclone Fines
lethod/Instrument
Moisture
% As Received
0.977 * 0.003
0.278 t 0.001
1.69 i 0.02
ASTM-D-3173
Weight Retained on Sieve No.
12
. 2.7
0.2
0
20 30 40 60 80 100 200 235 PAN
28.3 13.6 14.1 20.1 9.5 4.0 5.2 1.7 0.8
14.0 14.2 20.7 32.0 11.0 3.7 3.0 0.4 0.4
000 0.4 1.0 3.6 33.2 36.8 24.6
U. S. Standard Sieves, Mechanical Sieve Shaker
Proximate Analysis
j ample
:eed Coke
Spent Char
Cyclone Fines
lethod/Instrument
% Moisture..
0.995 ± 0:005 -
0.234 i 0.001
2.00 + 0.02
ASTM-D-3173
* Ash
13.30 t 0.02
17.71 ± 0.07
14.33 t 0.14
ASTM-D-3174
S Volatile Matter
1.62 t 0.07
1.62 t 0.04
1.64 t 0.04
ASTM-D-3175
X Fixed Carbon
84.09
80.44
82.03
Ultimate Analysis
Sample
Feed Coke
Spent Char
Cyclone Fines
Method/ Instrument
% Carbon
81.87 t 0.93
78.93 t 0.10
79.76 i 0.05
ASTM-D-3178
% Hydrogen.
1.59 4 0.11
0.392 + 0.10
0.436 + 0.02
ASTM-D-3178
% Nitrogen
1.03 t 0.00
0.868 t 0.003
0.829 + 0.001
ASTM-0-3179
% Sulfur
2.54 + 0.005
2.62 ± 0.04
2.63 + 0.07
Fisher Model 470
% Oxygen
0.0
0.0
2.0
-------
Table 7.3 Suseaary of Trace Element Analyses
ro
Element
Food
Measured Trace Element Concentrations (ug/g) or (pg/ml)
Coke Spent Char Cyclone Fines NSS ww ss ww
AS*
Be
Cr
Hg
Ni
Pb
Sb«
V
SB
Ce
u
Se
Th
Cr
La
Sbn
Br
Se
Ru
Fe
Co
Eu
Asn
9.60 t 0.40
6.83
69.60
0.11
32.30
11.90
0.78
44.00
2.03
63.57
2.23
<
2.39
67.57
29.29
0.06
0.60
0.01
0.80
1.46
0.01
0.60
0.20
1.49
0.24
0.28
1.26
1.48
0.64 t 0.05
<5
5.53 ± 0.16
17.84 ± 4.54
43083 ± 793
7.85 ± 0.21
0.29 ± 0.08
8.05
12.40 ± o.40
8.75 ± 0.05
75.00 ± 0.60
0.08 t 0.01
131.50 t 1.50
11.90 t 1.46
0.55 t 0.02
57.00 ± 1.00
2.58
36.14
3.10
<5
3.42
108.75
16.23
0.63
4.83
8.40
36.56
51693
12.60
0.38
8.50
16.80 * 0.40
7.59 t 0.05
72.80 t 0.60
0.07 t 0.01
120.20 t 2.00
48.00 ± 1.46
0.85 t 0.01
71.00 t 0.60
2.24
68.68
2.99
<5
2.52
61.13
36.09
0.63
<5
5.99
16.83
31656
8.25
0.31
8.78
0.019
0.00115
0.055
0.0005
0.06
0.065
0.092
0.098
0.0019
0.098
<0.05
0.072
<0.05
1.98
0.074
0.027
<0.5
0.0021
<0.5
7.33
<0.5
<0.1
<0.015
0.020
0.00118
0.066
0.0009
0.06
0.051
0.108
0.011
0.0018
0.142
<0.05
0.046
<0.05
1.79
0.089
0.018
<0.5
Q.0020
<0.5
7.24
<0.5
<0.1
<0.020
*atonic absorbtion analysis
nneutron activation analysis
-------
Only TEG was not measured in this run; it will be assumed
to be zero.
The following operating variable values are obtained
for Table 7.1 and run data:
Coke Feed Rate = 11.59 kg/hr
Char Removal Rate = 7.67 kg/hr
Cyclone Collection Rate = 0.38 kg/hr
Volume of water in PCS tank = 617 liters
These values are used with the trace element analyses to
calculate the terms in equation (7-1).
The results of the trace and minor element mass
balances are shown in Table 7.4.
7.3 Evaluation of the Data
It is evident from inspection of Table 7.1 that major
element balances can be made with a high degree of accuracy.
Such results are comparable to those published for other
coal gasification pilot plants (e.g., Gasior, 1978).
Examination of the data shown in Table 7.2 reveals that,
as expected, the percentages of moisture, carbon, hydrogen,
and nitrogen are smaller in the spent char, while the ash
content has increased significantly. The size distributions
of the spent char and, to a much greater extent, the cyclone
fines are shifted toward smaller particle sizes.
257
-------
ro
tn
00
Table 7.4 Trace Element Mass Balance
Basis: 1.5 hours (SS)
Element
mg
AS*
Be
Cr
Hg
Ni
Pb
Sb«
V
Sm
Ce
U
Se
Th
Cr
La
Sbn
riN
Feed
TEC
166.90
118.74
1210.00
1.91
561.54
206.88
13.56
764.94
35.29
1105.16
38.77
41.55
1174.70
509.21
11.13
Out
Char
TCH
142.66
100.67
862.88
0.92
1512.91
136.91
6.33
655.79
29.68
415.79
35.67
39.35
1251.17
186.73
7.25
Out
Pines
TEp
9.58
4.33
41.50
0.04
68.51
27.36
0.48
40.47
1.28
39.15
1.70
BELOW
1.44
34.84
20.57
.36
Acc.
PCS
TEPCS
0.62
0.02
6.79
0.25
0.00
-8.64
9.88
-53.70
-.06
27.16
0.00
DETECTION LIMIT
0.00
-117.28
9.26
-5.56
rout
152.86
105.02
911.17
1.21
1581.42
164.27
16.69
696.26
30.96
482.10
37.37
40.79
1286.01
216.56
7.61
Alg. %
Sum. Recovery
14.04
13.72
298.83
0.70
-1019.88
45.61
-3.13
68.68
4.33
623.06
1.40
0.76
-111.31
292.65
3.52
92
88
75
63
282
79
123
91
88
44
96
98
109
43
68
-------
Table 7.4 continued
Element
ng
Br
Se
Ru
F«
Co
ftu
M»
UN
Feed
TEC
Out
Char
"H
Out
Fines
TEy
Ace.
PCS
^PCS
rout
Alg.
Sum.
%
Recovery
BELOW DETECTION LIMIT
96.14
310. IS
748998
136.47
5.04
139.95
96.64
420.62
594728
144.96
4.37
97.79
3.41
9.59
18044
4.70
0.18
5.00
-.06
0.00
-55.56
0.00
0.00
3.09
100.05
430.21
612772
149.66
4.55
105.88
-3.91
-120.06
136226
-13.19
0.49
34.07
104
139
82
110
90
76
•analysed by AA
by
-------
It is of interest to determine whether the ash tracer
method can provide good estimates of carbon conversion in
this gasification system, or if, as in the LPR experiments,
ash losses bias the results low. The carbon conversion cal-
culated by the ash tracer method (using equations (4-50)
and (6-6)) is 27.7%. The true carbon conversion, calculated
from a carbon mass balance, is 41.9% as shown in Table 7.1.
The large discrepancy indicates that a significant amount of
ash was removed from the char and left the reactor with the
gas. The ash was not trapped in the cyclone, as shown by
the low ash content of the cyclone fines, which indicates
that the ash particles leaving the reactor must be relatively
small. It is thought that those small ash particles are
trapped in the PCS system. This is supported by the large
amount of total residue (typically 0.3 kg) found in the PCS
tank at the end of a run.
Based on the discussion presented in Section 6.1,2, the
cyclone fine.8 were expected to show a slight increase in ash
content. Since such is not the case, the relative weight
loss of the spent char particles can be compared to that of
the cyclone fines using ash as a tracer. Equation 4-52
yields %AW+ - 24.9 for the char and %An£ -7.2 for the fines.
This calculation appears to show that the cyclone fines had
a smaller residence time in the reactor, and therefore did
not react to a large extent.
260
-------
One of the most important findings of this research is
. .! '
that sulfur appears to be released at approximately the same
rate as volatile matter during pyrolysis. This also appears
to be the case during the steam-oxygen gasification of coke.
The data in Table 7.2 shows that the sulfur mass fraction in
the char and cyclone fines remain fairly close to the mass
fraction in the feed coke despite a 32% sulfur loss (calcu-
lated using equations (4-43), (4-48), and (4-49)). Since
the d.a.f. weight loss (calculated with equation (4-47)) is
31%, it appears that equation (5-7) may also be used to es-
timate the release of sulfur from coal during steam-oxygen
gasification.
The trace element analyses show that the spent char is
enriched in most trace elements analyzed. Elements depleted
in the spent char include Hg, Sb, Cef and La. Based on the
results of the trace and minor element mass balances, it can
be concluded that most of the elements analyzed are released
from the coke in significant quantities. Furthermore com-
parison of the elemental mass fractions with the normalized
mass fractions calculated by the ash tracer method (equation
(6-5)) indicate that for several elements (e.g., V, U, and
Eu), the mode of transport out of the coal particles appears
to be escape with the ash; volatile elements (e.g., Hg) are
released faster, while a few (e.g., Sc) are released slower
than the ash. The normalized mass fractions are shown in
261
-------
Table 7.5 (see Section 6.1 for the discussion that leads to
these conclusions). These results are generally consistent
with the LFR results discussed in Section 6.
The erratic results of the PCS water analyses have been
found to be due to sampling error due to settling or unevenly
dispersed ash in the PCS tank. Trace and minor element con-
centrations determined by HAA in well shaken waste water
samples (i.e., including their residue) are an order of mag-
nitude higher than concentrations found in waste water
samples where the residue was allowed to settle before the
analysis sample was pipetted. Another complication is due
to the small steady state sampling interval. The small con-
centrations of trace elements in the PCS tank cannot be
expected to increase dramatically in 1.5 hours when coke is
gasified. To overcome these problems in future studies,
separate analyses will be made of the filtered PCS water and
the residue which will be collected in filters placed between
the tank and the drain. A water sampling train has been in-
stalled between the cyclone and the venturi scrubber. Its
use will permit the obtention of undiluted steam condensate
and ash samples.
Finally, the discrepancies observed between the AA and
NAA analyses can probably be attributed to sample inhomo-
geneity and to differences in the sensitivities of the two
techniques for the different elements analyzed. A more com-
plete discussion of these problems is presented in Appendix C,
262
-------
Table 7.5 ty Values in Spent Char
A
Element
As 9.31
Be^ 6.57
Cra 56.3
Hg 0.06
Ni 98.8
Pb 8.9
Sb 0.41
V 43
Sm 1.94
Ce 27.14
U 2.33
ThK 2.57
CrD 81.57
Lah 12.19
SbD 0.47
Sc 6.31
Ru 36.56
Fe 38821
Co 9.46
Eu. 0.29
As 6.38
aaAalyzed by atomic absorption
banalyzed by neutron activation
263
-------
8. CONCLUSIONS
The objectives of this research project were to derive
data on equilibrium retentions and kinetics of the release
of trace and minor constituents of coal. The principal con-
clusions are summarized below. In addition, some con-
clusions are drawn on volatile yields and kinetics of de-
volatilization at low to medium temperatures* and on some
aspects of the experimental methodology explored in the
research.
8.1 Trace and Minor Element Release
In the chars produced in batch pyrolysis experiments
with nitrogen over a temperature range of 25 to 1200°C:
Sm, Cr, Th, Sc, Fe, and Co are retained completely; As* Ser
and perhaps La» exhibit intermediate volatility (<50%
release); S* Pb, Kg, and cl are highly volatile (>50% re-
lease) . Mercury and chlorine show losses greater than 70%
at temperatures below 700°C, and more than 75% of the lead
is released at temperatures above 1000°C. These results
apply to coals with ranks ranging from lignite to anthra-
cite.
Fe* Co* Sc* Ma* Ce, and V are completely retained in
the LFR chars. C* H, S* As* Se, Pb* La, and Hg are released
in significant quantities during fast pyrolysis with the
264
-------
release occurring within a few tenths of a second. In
addition, some elements (e.g., Sm and Mn) appear to be
released with the ash expelled from the coal particles.
The equilibrium release (or retention) of As, Se, S,
Cl, Pb, and Hg has been successfully modeled (see Section
5.2). The model employed has three parameters: TC, the
temperature at which the trace element release begins; Aiie,
the asymptotic elemental release at high temperatures; and
b, a fitted constant. This model was found to describe ele-
mental retentions in ligniticr subbituminous, bituminous, and
anthracitic coals for all elements studied except sulfur in
anthracite. The model should provide good engineering esti-
mates of the equilibrium extent of volatile trace and minor
element release as a function of temperature during the
devolatilization stage of any gasification process.
The kinetics of sulfur release from lignitic and sub-
bituminous coals have been modeled with fair results. The
bias and imprecision of the retentions of other volatile
trace elements made the determination of their kinetic para-
meters difficult. A first-order model coupled with the
equilibrium release (batch) model was used. The sulfur
modeling results suggest that improvements in the trace
and minor element data could yield the proper Arrhenius
parameters for the model, which could then be used to pre-
dict the elemental release as a function of time and tem-
perature.
265
-------
Sulfur was found to be released in direct proportion
to the dry ash-free coal weight loss, regardless of coal
rank and pyrolysis conditions (temperature, residence tine,
and heating rate), thereby providing a simple model for
sulfur release during pyrolysis.
As would be expected, the mass fractions of non-
volatile elements in the chars tend to be positively corre-
lated with one another over a wide temperature range. The
mass fractions of volatile elements tend to be negatively
correlated with those of nonvolatile elements but not with
one another, indicating that volatile elements are re-
leased in different proportions as the temperature in-
creases.
8.2 Volatile Yields and Kinetics
of Devolatilization
The volatile yields of five coals ranging in rank from
lignitic to anthracitic exhibit similar behavior during
slow heating <5-45°c/sec). All coals exhibit a dramatic in-
crease in weight loss between 400 and 600°C, and equili-
brium is reached much more rapidly at the higher tempera-
tures. The equilibrium volatile yields exhibit the expected
dependence on rank, with lignite showing the highest and
anthracite the lowest yields. The equilibrium weight loss
data show the characteristic devolatilization behavior of
266
-------
coals pyrolyzed in batch reactors. Moisture evolution
occurs at 100°C; devolatilization begins at about 350°C,
and most of the weight loss 'occurs between 400 and 750°C.
Bituminous and anthracitic coals exhibit higher
equilibrium volatile yields when pyrolyzed in shallow beds
than in deep beds. Lower rank coals do not show such an
effect. This conclusion is consistent with the findings
of Kobayashi (1976).
The fast pyrolysis devolatilization rates found in
this study agree reasonably well with those found in other
studies. Nonisothermal pyrolysis models correlate fast
pyrolysis weight loss data in laminar flow reactors sig-
nificantly better than isothermal models. Kobayashi's
(1976) two-parallel first-order reactions model predicts
pyrolysis weight losses reasonably well for temperatures
ranging from 300 to 900°C and residence times ranging from
t
150 to 1500 msec. However, the best correlation of the
data was obtained with a first-order model featuring tem-
perature dependent asymptotic weight loss.
8.3 Experimental Methodology
Laminar flow reactor data are subject to several
sources of error, including uncertainties in particle tem-
perature and residence time. The largest source of scatter
in the data is caused by variation in particle heating
267
-------
rates (due to changes in coal feed rates and/or feeder gas
flow rates). In addition, the char collection devices tend
to make the chars inhomogeneous. As discussed previously,
the weight loss of coal in the LPR should be determined
gravimetrically. The ash tracer technique leads to under-
estimation of weight losses and hence should not be used
to study the kinetics of trace and minor elements during
pyrolysis. Ash tracer losses can be corrected to some ex-
tent by using in the weight loss calculations the feed
coal ash content after it has been subjected to a room tem-
perature run and by increasing the apparent ash content
of the chars by about 5%. However, such a procedure is
not precise enough for the study of the kinetics of trace
and minor element release from coal.
268
-------
9. RECOMMENDATIONS
The results of this research indicate that the study
of the kinetics of trace and minor elements during coal
pyrolysis, and coal gasification in general, is feasible.
However, considerable refinement of the methods and tech-
niques used is necessary.
Specifically, it was found that the largest source of
scatter in the weight loss data was the variation in coal
feed rates. This problem can be avoided through more care-
ful operation of the feeder system and frequent recali-
bration of the feeder valve. It is also recommended that
higher main gas and suction flow rates be used to improve
the particle collection efficiency such that only a neg-
ligible fraction of the coal particles are lost. Finally,
a much smaller feeder tip should be used. The resulting
higher feeder gas velocities will reduce the particle
heating times and make the particle heating time constant
less sensitive to variations in feeder gas velocity.
The weight loss of coal pyrolyzed in the LFR should
be determined gravimetrically. The ash tracer technique
should not be used to study the kinetics of trace and minor
element release. In order to ensure that all the coal par-
ticles enter the water-cooled collector and thus ensure
complete char recovery, it is recommended that the main gas
and suction flow rates be increased and the diameter of the
269
-------
reaction tube be decreased. The first two changes should
also improve the collection efficiency of the cyclones.
The suction flow rate should be increased well above the
isokinetic rate; the added uncertainty in the residence
time estimates should be amply compensated by the improved
reliability in the weight loss estimates. The velocity
of the coal particles (and thus their residence time) could
be measured with a laser doppler anemometer; the added cost
and complexity of the apparatus would not be justified
until other more important improvements are made, however.
It is obvious that tars condensed on the surface of
the quenched particles. The amount of tar deposited quite
likely depends on the size of the particles, and is pro-
bably a significant source of sample inhomogeneity and of
bias in the devolatilization weight loss estimation. One
solution to this problem may be the use of air or boiling
water as the coolant in the collector instead of cold
water. An increase in reaction time would result, but the
added uncertainty in the reaction time due to the slower
particle quenching would probably not be as significant as
the uncertainty now present in the results due to tar con-
densation.
Coal and char samples must be made more homogeneous
before analysis. The chars produced in the batch and LFR
experiments showed the formation of fused lumps. It is
possible that the lumps may be enriched or depleted in
270
-------
some elements. In general, the LFR chars showed more
scatter than the batch reactor chars. In order to overcome
sample inhomogeneity problems, the samples should be ground
to a very fine powder and mixed thoroughly in a roller
mixer. One drawback of this procedure is that the likeli-
hood of sample contamination is greatly increased by the
handling of all samples for trace element analysis. Pro-
per care should render that problem minimal, however. Also,
the largest possible sample size should be used in order
to "average out" sample inhomogeneities , and several rep-
licate runs should be made at each set of conditions.
Finally, it is recommended that future work be focused
on the volatile elements determined in this study, parti-
cularly Hg and Pb. More data is needed at low tempera-
tures for the equilibrium release model; such data should
allow better estimation of critical temperatures. Future
work with the LFR should be carried out exclusively at
temperatures above 800°C. Trace/minor element release
below that temperature appears to be small and errors in
the weight loss determinations appear to be larger than at
higher temperatures. Future studies should focus on one
coal; North Barber No. 8 from New Mexico is recommended
because of its high volatiles content and the ease with
which it can be handled in the coal feeder. This study has
shown that results obtained with one coal are applicable to
most other coals.
271
-------
LITERATURE CITATIONS
Agreda, V.H., Coal Gasification Project internal Technical
Report No. 10., "Design and Operation of an LFR and
a Batch Pyrolysis Reactor*, Department of Chemical
Engineering, North Carolina State University, Raleigh,
NC, 1979.
Anderson, G.L., A.H. Hill, and O.K. Fleming, "Predictions
on the Disposition of Select Trace Constituents in
Coal Gasification Processes." Paper presented at
Environmental Aspects of Fuel Conversion Technology
Symposium, Hollywood, PL, April, 1979.
Anthony, D.B., Sc.D. Thesis, Massachusetts Institute of
Technology, Cambridge, MS, 1974.
Anthony, D.B., J.B. Howard, H.C. Hottel, and H.P. Meissner,
"Rapid Devolatilization of Pulverized Coal," Fifteenth
International Symposium on Combustion, p. 1313, The
Combustion Institute, Pittsburg, PA, 1975.
Anthony, D.B., J.B. Howard, H.C. Hottel, and H.P. Meissner,
"Rapid Devolatilization and Hydrogasification of
Bituminous Coal," Fuel, 55, 121, 1976.
Attari, A., A.H. Corcoran, and G.S. Gibson, "Transformation
of Sulfur Functional Groups during Pyrolysis of Coal,"
in Proceedings of 172nd National Meeting of the American
Chemical Society, Division of Fuel Chemistry, Vol. 21,
pp. 106-111, American Chemical Society, Washington, DC,
1976.
Attari, A., J. Pau, and M. Mensinger, Fate of Trace and
Minor Constituents of Coal During Gasification, U.S.
Environmental Protection Agency Publication EPA-600/
2-76-258, September, 1976.
Babu, S.P., (Ed.), Trace Elements in Fuel, Advances in
Chemistry Series 141, Washington, DC, 1975.
Badzioch, S. and P.G.W. Hawksley, "Kinetics of Thermal
Decomposition of Pulverized Coal Particles," Ind.
Eng. Chem. Process Design Develop., 9, 521, 1970.
Badzioch, S., R.B. Sainsbury, and P.G.w. Hawksley, BCURA
Member's Circular, No. 340 ,1968.
272
-------
Bennett, W.S., "WESCO Coal Gasification Plant: Navajo Con-
siderations, " Los Alamos Scientific Laboratory Report
No. LA-6247-MS, February, 1976.
Bird, R.B., W.E. Steward, and E.N. Lightfoot, Transport
Phenomena, John Wiley & Sons, New York, NY ,1960.
Chukhanov, Z.F., Izv. Akad. SSSR, Otd. Tekh, Nauk, 8,
7, 1954.
Dryden, I.G.C., "Chemical Constitution and Reactions of
Coal," in Chemistry of Coal Utilization, Supplemen-
tary Volume, H.H. Lowry, ed., John Wiley & Sons,
New York, NY, 1962.
Duck, N.W. and G.W. Himus, "On Arsenic in Coal and its
Mode of Occurrence," Fuel, 30, 367-71, 1951.
Duhne, C.R., "Calculating the Approach to Equilibrium,"
Chem. Eng., 84, No. 18, August 29, 1977.
Eddinger, R.T., L.D. Friedman, and E. Rau, "Devolatili-
zation of Coal in a Transport Reactor," Fuel, 45,
245, 1966.
Ferrell, J.K., R.M. Felder, R.W. Rousseau, and D.W. Alexander,
"A Coal Gasification-Gas Cleaning Test Facility," Paper
presented at the U.S. EPA Symposium Environmental Aspects
of Fuel Conversion Technology, III, Miami, FL, September,
1977a.
Ferrell, J.K., R.M. Felder, R.W. Rousseau, and L.W. Alexander,
"A Coal Gasification-Gas Cleaning Pilot Plant at North
Carolina State University," Paper presented at the
Miami International Conference on Alternative Energy
Sources, Miami, FL, December, 1977b.
Fiene, F.L., J.K. Kuhn, and H.J. Gluskoter, "Mineralogic
Affinities of Trace Elements in Coal," Presented at U.S.
EPA Symposium on Coal Cleaning to Achieve Energy and
Environmental Goals, in Hollywood, FL, September, 1978.
Forney, A.J., W.P. Haynes, S.J. Gasior, R.M. Karnosky, C.E.
Schmidt, and A.G. Sharkey, Trace Element and Major
Component Balances Around the Synthane PDU Gasifier,
U.S. Energy Research and Development Administration
Publication PERC/TPR-75/1, August 1975.
273
-------
Bennett, M.S., "WESCO Coal Gasification Plant: Navajo Con-
siderations,* Los Alamos Scientific Laboratory Report
No. LA-6247-MS, February, 1976.
Bird, R.B., W.E. Steward, and B.N. Lightfoot, Transport
Phenomena, John Wiley ft Sons, Hew York, MY .i960.
Chukhanov, Z.F., I*v. Akad. SSSR, Otd. Tekh, tank, 8,
7, 1954.
Dryden, Z.G.C., "Chemical Constitution and Reactions of
Coal," in Chemistry of Coal Utilisation, Supplemen-
tary volume, H.H. Lowry, ed., John Wiley ft Sons,
New York, NY, 1962.
Duck, N.w. and G.w. Himus, "On Arsenic in Coal and its
Mode of Occurrence," Fuel, 30, 3(7-71, 1951.
Dunne, C.R., "Calculating the Approach to Equilibrium,"
Chem. Eng., 84, No. IS, August 29, 1977.
Eddinger, R.T., L.D. Friedman, and E. Rau, "Devolatili-
zation of Coal in a Transport Reactor," Fuel, 45,
245, 1966.
Ferrell, J.K., R.M. Felder, R.W. Rousseau, and O.N.
•A Coal Gasification-Gas Cleaning Tost Facility*" Paper
presented at the U.S. BPA Symposium Environmental Aspects
of Fuel Conversion Technology, XXX, Miami, FL, September,
1977a.
Ferrell, J.K., R.M. Felder, R.w. Rousseau, and U.W. Alexander,
•A Coal Gasification-Gas Cleaning Pilot Plant at north
Carolina State University," Paper presented at the
Miami International Conference on Alternative Energy
Sources, Miami, FL, Parember, 1977b.
Fiene, F.L., J.K. Kuhn, and B.J. Gluskoter, "Minaralogic
Affinities of Trace Elements in Coal,* Presented at U.S.
EPA Symposium on Coal Cleaning to Achieve Energy and
Environmental Goals, in Bollywood, FL, September, 1971.
Forney, A.J., W.P. Haynes, S.J. Gasior, R.M. Earnosky, C.I
Schmidt, and A.G. Sharkey, Trace Element and Major
Component Balances Around the Synthane poo Gasifier,
U.S. Energy Research and Development Administration
Publication PERC/TPR-75/1, August 1975.
273
-------
Gasior, S.J., R.G. Lett, J.P. Strakey, and W.P. Haynes,
"Major, Minor and Trace Element Balances for the
Synthane PDU Gasifier: Illinois No. 6 Coal," Paper
presented at the 175th National ACS Meeting, Anaheim,
CA, 1978.
Given, P.H., "The Distribution of Hydrogen in Coals and
its Relation to Coal Structure," Fuel, 39, 147,
1960.
Gluskoter, H.J., R.R. Ruch, W.H. Miller, R.A. Cahill, G.E.
Dreher, and J.K. Kuhn, Trace Elements in Coal:
Occurrence and Distribution, U.S. Environmental Pro-
tection Agency Publication EPA-600/7-77-064, June,
1977.
Goldschmidt, V.M., "Rare Elements in Coal Ashes,"
Industrial and Engineering Chemistry, 27, No. 9, p.
1100-1102, 1935.
Goldstein, S., Modern Developments in Fluid Dynamics, 2,
p. 599, Dove Publishers, New York, NY, 1965. ~
Gould, R.F. (Ed. ), Fuel Gasification, Advances in Chemistry
Series, No. 69, Am. Chem. Soc., Washington, DC, 1967.
Henry, J.P., and B.M. Louks, "An Economic Study of Pipe-
line Gas Production from Coal," Chem. Tech., p. 237-
247, April, 1971.
Hirsh, P.B., "Conclusions from X-Ray Scattering Data on
Vitrain Coal," Proc. Conf, Science in the Use of Coal,
p. A29, inst. Fuel, London, England, 1958.
Horton, M.D., "Fast Pyrolysis," in L.D. Smoot and D.T. Pratt,
Eds., Pulverized-Coal Combustion and Gasification,
Plenum Press, New York, NY, 1979.
Horton, M.D. and L.D. Smoot, "Exploratory Studies of Flames
and Explosion Quenching," Interim Report No. 3, U.S.
Bureau of Mines, Contract No. HO122052, Chemical
Engineering Department, Brigham Young University, Prove,
UT, 1975.
Horton, N. and K.V. Aubrey, "The Distribution of Minor
Elements in Vitrain: Three Vitrains from the Barnsley
seam," London, Journal of the Society of Chemical
industries, 69, Suppl. No. 1, p. 541-548, (revised) ,
1950.
274
-------
Hottel, B.C. and J.B. Howard, "New Energy Technology,"
M.I.T. Press, Cambridge, MS, 1971.
Howard, J.B., "Mechanism of Ignition and combustion in
Flames of Pulverized Bituminous Coal," Ph.D. Thesis,
Pennsylvania State univeristy. University Park, PA, 1965.
Howard, J.B. and Essenhigh, R.H., "Pyrolysis of Coal Par-
ticles in Pulverized Fuel Flames," Ind. Eng. Chem.
Process Design Develop. 6, 74, 1967.
Jahnig, C.E., R.R. Bertrand, and E.M. Magee, "Control
Technology Research and Development Needs," (Gov.
Res. Lab., Exxon Research and Eng. Co., Linden, NJ),
U.S. Environ. Prot. Agency, Off. Res. Dev., Rep.
EPA 1976, EPA-600/2-76-149, Symp. Proc,: Environ.
Aspects Fuel Convers. Technol., II, 1975; PB-257
182, 259-65.
Jensen, G.A. and G.T. Austin, "Iron From Coal Minerals,7
Ind. Eng. Chem. Process Design Develop., 16, (1),
44, 1977.
Kaakinen, J.W., "Trace Elemental Behavior in Coal-Firec
Power Plants," Environ. Sci. Technol., 9, 862-869,
1975.
Karn, F.S., R.A. Friedel, and A.G. Sharkey, "Mechanism of
Gas Flow through Coal," Fuel, 54, 279, 1975.
Kaskan, W.E., "The Dependence of Flame Temperature on Mass
Burning Velocity," Sicth Symposium (International) on
Combustion, The Combustion Institute, p. 34, 1956.
Kimber, G.M. and M.D. Gray,""Measurements of Thermal De-
composition of Low and High Rank Non-Swelling Coals
at MUD Temperatures," BCURA Document, No. MHD 32,
January, 1976.
Klein, D.H., "Pathways of Thirty-Seven Trace Elements
Through Coal-Fired Power Plants," Environ. Sci.
Technol., 9, 973-979, 1975.
Kobayashi, H., "Rapid Decomposition Mechanism of Pulverized
Coal Particles," M.S. Thesis, Department of Aeronautics
and Astronautics, Massachusetts Institute of Technology,
Cambridge, MS, 1972.
275
-------
Kobayashi, H., Sc.D. Thesis, Department of Mechanical
Engineering, Massachusetts Institute of Technology,
Cambridge, MS, 1976.
Kuhn, J.K., D. Kidd, J. Thomas Jr., R. Cahill, D. Dickerson,
R. Shiley, C. Kruse, and N.F. Shimp. EPA Contract
No. 68022130, 1979.
Kuhn, J.K., D. Kidd, J. Thomas Jr., R. Cahill, D. Dickerson,
R. Shiley, C. Kruse, and N.F. Shimp, "Volatility of
Coal and its By-products," paper presented at the
Third Annual EPA Symposium, Environmental Aspects of
Fuel Conversion Technology, Hollywood, FL, September,
1977.
Langhaar, H.C., ASME Trans. Vol. 64, 1942.
Lewellen, P.C., "Product Decomposition Effects in Coal
Pyrolysis," M.S. Thesis, Chemical Engineering Depart-
ment, Massachusetts Institute of Technology, Cambridge,
MS, 1975.
LittleJohn, R.F., "Mineral Matter and Ash Distribution in
'As-Fired* Samples of Pulverized Fuels," J. Inst.
Fuel, 59, February 1966.
Loison, R. and Chauvin, F., "Pyrolyse Rapide Du Charbon,"
Chem. ind. (Paris), 91, 269, 1964 (Reviewed by
Badzioch, S., BCURA Monthly Bulletin, 31, 1967).
Loran, B.I. and J.B. O'Hara, "Specific Environmental Aspects
of Fisher-Tropsh Coal Conversion Technology," Third
Symposium on Environmental Aspects of Fuel Conversion
Technology, Hollywood, FL, September 15, 1977.
Lowry, H.H., ed. chemistry of Coal Utilization, Supplementary
Volume, John Wiley & Sons, New York, NY, 1963.
Lukesh, J.S., "Thermal Decomposition of Arseno-pyrite," Am.
Mineral, 25, 539-42, 1940.
Lyon, H.S., Trace Element Measurements at the Coal Fired
Steam Plant, CRC Press, Cleveland, OH, 1977.
Malte, P.C. and D.P. Rees, "Pulverized-Coal Reaction:
Pollutant Formation," in Pulverized-Coal Combustion
and Gasification, L.D. Smoot and D.T. Prat, Eds.,
Plenum Press, New York, NY, 1979.
276
-------
Marquardt, D.W., "An Algorithm for Least Squares Estimation
of Non-Linear Parameters," J. Soc. Ind. App. Math.,
11 (2), 431-441, 1963.
Massey, L.G., (Ed,), Coal Gasification, Advances in
Chemistry Series 131, Washington, D.C., 1974.
McAdams, W.H., Heat Transmission, Third Edition, pp. 202
et seq., McGraw-Hill Book Co., New York, NT, 1964.
McCabe, W.L. and J.C. Smith, Unit Operations of Chemical
Engineering, Second Edition, McGraw-Hill Book Co.,
New York, NY, 1967.
Menster, M., "Devolatilization of Coal by Rapid Heating,"
in, L.G. Massey (Ed.), Coal Gasification, Advances
in Chemistry Series 131, Washington, DC, 1974.
Morgans, W.T.A. and NiB. Terry* '•Measurements of the Static
and Dynamic Elastic Moduli of Coal," Fuel, 37, 201,
1958.
Nsakala, N., "Characteristics of Chars Produced by Pyrolysis
Following Rapid Heating of pulverized Coal," Ph.D.
Thesis, Pennsylvania State University, University park,
PA, 1976.
0'Gorman, J.W. and P.L. Walker, Jr., "Mineral Hatter Charac-
teristics of American Coals of Various Rank," Report
to the Office c-f Coal Research, United States Department
of Interior, July 1, 1969.
Padia, A.S., "The Behavior of Ash in Pulverized Coal Under
Simulated Combustion Conditions," Sc.D. Thesis,
Massachusetts Institute of Technology, Cambridge,
MS, 1976.
Page, G.C., "Fate of Pollutants in Industrial Gasifiers,"
paper presented at the Third Annual EPA Symposium,
"Environmental Aspects of Fuel Conversion Technology,
Hollywood, PL, September 1977.
Pai, S.I., Fluid Dynamics of Jets, pp. 120 147, D. Van
* Nostrant Cov, Inc., New York, NY , 1954.
Pohl, J.H. "Fate of Fuel Nitrogen," Sc.D, Thesis,
Massachusetts Institute of Technology, Cambridge,
277
-------
Pohl, J.H. and A.F. Sarofim, "Devolatilization and Oxi-
dation of Coal Nitrogen," in Sixteenth Symposium
(International on Combustion, pp. 491-501, The
Combustion Institute, Pittsburg, PA, 1977.
Reidelbach, H. and J. Algermissen, "Theoretical Studies
of Coal Pyrolysis in an Entrained Bed Flow Reactor,"
J. Soc. Auto. Eng. pp. 469-475, 1978.
Reidelbach, H. and M. Summerfield, "Kinetic Model for
Coal Pyrolysis Optimization,* Amer. Chem. Soc.,
Div. Fuel Chem., 20, 1, 161, 1975.
Robertson, C.A. in Sainsbury et al. (1966).
Sainsbury, R.B., P.C. Yellow, S. Badzioch, and P.G.W.
Hawksley, BCURA Member's Circular, No. 309, 1966.
Sass, A., "Garrett's Coal Pyrolysis Process," Chem.
Eng. Progress, 70, No. 1, 1974.
Sevenster, P.G., "Studies of Some Physical Properties
of South African Coal," J. South Afr. Chem. Inst.,
7, 41, 1954.
Shapatina, E., V.V. Kalyuzhnyi, and Z.F. Chukhanov,
"Tehcnological utilization of Fuel for Energy , 1-
Thermal Treatment of Fuels," 1960 (reviewed by
Badzioch, S., BCURA Monthly Bulletin, 25, 285,
1961).
Shiley, R., J. Kuhn, R. Cahill, D. Kidd, and P. Dickersonf
"Volatile Inorganic Elements From Coal Pyrolysis,•
in Neil F. Shimp, Characterization of Coal and Coal
Residue, EPA Contract No. 68022130, 1978.
Snedecor, G.W. and W.G. Cochran, Statistical Methods,
Sixth edition, Iowa State University Press, Ames,
IA, p. 557, 1967.
Soloman, P.R., The Evolution of Pollutants During the
Rapid Devolatilization of Coal, Report R 76-952588-2,
United Technologies Research Center, East Hartford,
CN, 1977.
Spackman, W., "The Nature of Coal and Coal Seams," in
Short Course in Coal Characteristics and Coal Con-
version Processes, Pennsylvania State University,
University Park, PA, May 19-23, 1975.
278
-------
Stickler, D.B., R.E. Gannon, and H. Kobayashi, "Rapid
Devolatilization Modeling of Coal," Technical
Meeting, Eastern Section of the Combustion Institute,
November, 1974.
Stinett, S.J., D.P. Harrison, and R.w. Pike, "Fuel Gasi-
fication Prediction of Sulfur Species by Free Energy
Minimization," Env. Sci. Technol. 8, 441, 1974.
Suuberg, M., W.A. Peters, and J.B. Howard, "Product Com-
position and Kinetics of Lignite Pyrolysis," Ind.
Eng. Chem. Process Design Develop., 17, 1, 1978.
Tran, D.Q., "Kinetic Modeling of Pyrolysis and Hydrogasi-
fication of Carbonaceous Material," Ph.D. Thesis,
Department of Mineral Engineering, University of
Wyoming, Laramie, WY, 1978.
van Krevelen, D.W. and J. Schuyer, Coal Science, Elsevier,
Amsterdam, Holland, 1957.
van Krevelen, D.W., Coal, Elsevier, Amsterdam, Holland,
1961.
Vestal, M.L. and W.H. Johnson, "Desulfurization Kinetics of
Ten Bituminous Coals," SRI Corporation, Report No.
SRIC 69-10, June 1969.
wen, C.Y. , C.T. Li, s.H. Tscheng, and W.S, O'Brien, "Com-
parison of Alternative Coal Gasification Processes,"
65th Annual AIChE Meeting, New York, NY, November
26-10, 1972.
Wiser, W.H., G.R. Hill, and N.J. Kertamus, "Kinetic study
of the Pyrolysis of High-Volatile Bituminous Coal,"
Ind. Eng. Chem. Process Design Develop*, 6, 133, 1967.
Yurovakii, A.Z., V.s. Kaminskii, and A.L. Rubinstein,
Khim. i Tekhnol. Tapliva i Masel, 7, 20-3, 1957.
Zubovic, P., "Geochemistry of Trace Elements in Coal,"
in, Ayer, F.A., compiles. Symposium Proceedings,
Environmental Aspects of Fuel Conversion Technology,
Us (December, 1975), Washington DC, Environmental
Protection Agency, Environmental Protection Tech-
nology Series EPA-600/2-76-149, 1976.
279
-------
APPENDICES
280
-------
APPENDICES
Appendix A (Data analysis computer programs) and Appendix B
(Data, calculated results, and statistical analysis of data) have
been omitted in this report. They are included In the Ph.D.
dissertation of Victor H. Agreda*. copies of which may be obtained
from University Microfilms, Inc.
*Agreda, V.H., "Devolatlllzatlon Kinetics and Elemental Release
in the Pyrolysls of Pulverized Coal." M.C. SUt* University, lift.
281
-------
APPENDIX C
Chemical Analyses
Proximate, ultimate, and trace/minor element analyses
were carried out on the coals and chars used arid produced
in this study. Every effort was made to ensure the accuracy
and precision of the analyses and to eliminate any bias in
the data due to instrument drift or analyst bias. All
samples were analyzed as blind samples, i.e., with a neutral
label that did not identify the source of the sample, and in
. . . - - - - f • ~- . •
random order. Furthermore, certified standards were run
concurrently with the samples. This was done for almost
every analyte. If the analysis of the standard did not
agree with the certified value, the entire lot of analyses
was discarded and new analyses made.
C.I Proximate Analysis
Moisture in the coal and char samples was determined by
establishing the loss in weight of 0.5 g of sample heated to
104-110°C for 1 hour in a moisture oven with bone air circu-
lation. ASTM-D-3173 method was followed.
The ash content was determined by weighing the residue
remaining after burning 0.5 g of coal at 950°C in a muffle
furnace. The same sample used for moisture analysis was
282
-------
used. ASTM-D-3174 method was followed. Certified standards
were run concurrently with the analysis samples.
The volatile matter of the samples was determined by
establishing the loss in weight resulting from heating 1.0 g
of sample for 6.0 minutes at 550 ± 20°C and 6.0 more minutes
at 950 ± 20°C in volatile matter furnaces. ASTM-D-3175
method for sparking coals was used. Certified standards
were run concurrently with the analysis samples.
C.2 Ultimate Analysis
The .determination of carbon and hydrogen was made by
burning 100 mg of sample in a combustion train and fixing
the products of combustion in an absorption train after
complete oxidation and purification from interfering sub-
stances. ASTM-D-3178 method for total carbon and hydrogen
was followed. This method gives the total percentages of
carbon and hydrogen in the coal as analyzed, and includes
the carbon in carbonates and the hydrogen in the moisture
and in the water of hydration of silicates. The results for
hydrogen were corrected, such that the hydrogen in the mois-
ture would not be included, using the equations %H * %H -
0.1119x(%M). Benzoic acid samples were used to test the
accuracy of the analyses.
Sulfur analyses were carried out with a Fisher Sulfur
Analyser Model 470. The analyser combusts the sulfur in
283
-------
the samples and detects the sulfur dioxide produced through
and amperometric technique. This analyzer is extremely
accurate and precise in the analysis of sulfur in coal and
coke. However, the analyzer does not detect sulfate sulfur
which occurs in snail quantities in most coals* Certified
coal standards were used to calibrate the analyzer and to
check its accuracy periodically.
C.3 Trace/Minor Element Analyses
The majority of the minor/trace element analyses were
done by neutron activation. Atomic absorption was used for
the analysis of a smaller number of elements, in both cases ,
NBS certified standards were run concurrently to ensure the
accuracy of the analysis*
C.3.1 Neutron Activation Analysis
Neutron activation analysis is an analytical technique
dependent on the measurement of the niadber and energy of
Y- and X-rays emitted by the radioactive isotopes produced
in the sample matrix by irradiation with thermal neutrons
from a nuclear reactor.
All neutron activation analyses reported in this thesis
were done by the Activation Analysis Laboratory, Department
of Nuclear Engineering, North Carolina State University.
284
-------
Typical parameters used ares 2 to 4.5 hours irradiation time
at l.SxlO3 n/cn2-sec. The decay was Monitored for 200 to
1200 seconds with the counting done on an Ortec 24% GeLi
and an Ortec 16 on LEPO coupled to a computerised NO6620
system.
The estimated instrumental error for each element
analyzed by this technique is shown in Table C.I. The com*
puter does a convergianoe, (l) of pipeting errors (usually
<0.2%), (2) weighing errors (usually <0.1%), and (3) counting
statistics errors on standards and unknowns (ranging from
0.1 to 100% depending on the element). These error esti-
mates do not include sampling and char inhomogeneity effects.
C.3.2 Atomic Absorption Analysis
These analyses were carried out using a Perkin Elmer
603 Atomic Absorption Spectrophotometer, aa BGA-2200 graphite
furnace, and a cold vapor mercury analysis system* Bamp
heating was used for the drying and charring steps* and nor-
mal, temperature controlled, or time controlled heating was
used during atomisation depending on the volatility of the
element. Deuterium background correction was used in all
cases except in the analysis of mercury. Klectrodeleas
discharge lamps were used in all eases except in the analysii
of Cd, V, Be, and Bi where hollow cathode lamps were used.
285
-------
Table C.I Percent Error (Instrumental) NAA in Coals
Percent
Element Error
Samarium i 1%
Uranium ± 6%
Lanthanum ±1%
Arsenic (<1 ppm) ±5-10%
Arsenic (2-10 ppm) ±2- 5%
Antimony ±2%
Bromine ±3%
Sodium t 2%
Potassium ±10%
Titanium ± 5%
Manganese ± 1%
Copper (<100 ppm) ±25%
Vanadium ± 3%
Aluminum ±1%
Mercury (<0.10 ppm) ±10%
Cerium * 5%
Selenium ± 5%
Thorium ±2%
Chromium * 5%
Scandium ±1%
286
-------
Table C.I continued
„, ^ Percent
Element
Europiun ± 2%
Rubidium ± 5%
iron ± 2%
Cobalt l 1%
Zinc (<100 ppn) ±20%
Cesium 1 5%
287
-------
The waste water samples were digested by evaporative
reflux with nitric acid for the analysis of all elements
of interest except mercury. Raw waste water samples were
analyzed directly for mercury. The solid samples were pre-
pared by two methods: oxygen bomb combustion, and low tem-
perature ashing followed by acid bomb digestion. The liquor
obtained from the oxygen bomb combustion was used for the
analysis of Hg, Pb, As, Sb, and Cd. A LFE-LTA-504 Low Tem-
perature Plasma Asher was used for the oxidation of the
samples intended for the analysis of Cr, B, Be, Ni, P, and
optionally Pb and Cd. The low temperature ashes were then
digested in teflon lined acid bombs. All acid liquors were
then diluted to volume. Typical sample weight to solution
volume ratios were 0.5 gram to 100 ml. Waste water samples
were typically concentrated from 150 to 50 ml.
Analysis for each element was basically the same.
standard linearity was established for the range the samples
fell in, if possible. Generally, if the standards showed
curvature, the samples were diluted, or a less sensitive set
of parameters was used. All AA and HGA parameters (shown
in Table C.2) were optimized to give the best signal to
noise ratio. Direct calibration methods were used for the
analysis of Hg, Ni, P, Sb, Be, Cr, and V. The method of
standard additions was used for the analysis of Pb and As.
Nickel complexation was used to allow high temperature
charring during the analysis of As. The method described
288
-------
Table C.2 Summary of Atomic Absorption Analysis Parameters
ro
;
Dry Temp-Time Char Temp-Time Ati
Element Ramp Time cnar Time
°C-Sec <>C-sec
Cr
V
Be
Pb
Ni
As
Hg
"AA se«
sec sec
125-40 1100-34
125-42 1700-32
125-42 1200-30
125-40 600-34
125-41 . 1000-30
125-30 1000-30
10 'IB
Cold Vapor Analysis
M to be off by one nanometer
om. Temp-Time
°C-Sec
2700-8
2800-8
2800-8
2300-8
2700-8
2700-8
Wavelength* slit
nm nm
357
318
234
217
232
253
253
.9
.4
.9
.0
.0
.5
.5
(4)
(3)
(4)
(4)
(3)
(4)
(4)
0.7
0.2
0.7
0.7
0.2
0.7
0.7
-------
by R.D. Ediger, A.R. Knott, G.E. Peterson and R.D. Beaty
("The Determination of Phosphorus by Atomic Absorbtion
Using the Graphite Furnace," Atomic Absorption Newsletter
Vol. 17, No. lf 28 , Jan-Feb, 1978) was followed for the
analysis of phosphorus. The correlation coefficients of
all the calibration and standard addition lines were
greater than 0.94.
The most serious problem encountered in these analyses
was interferences from the coal's ash matrix. This problem
was particularly acute in the analysis of Pb in MRS coal of
325x400 mesh size. Otherwise, the results of the analyses
were satisfactory.
C.3.3 Assessment of Trace Analyses
Comparisons of the behavior of different elements under
gasification conditions must be made with caution. Very
often the large scatter in the data is not due to sample
inhomogeneity but to the small concentration of the element
or to the low sensitivity of the analytical technique used.
It is well known that the sensitivity and detection limits
of NAA and AA vary widely for different elements. A dis-
cussion of those variations among the elements studied in
this research is beyond the scope of this thesis. The
reader is referred to standard texts and manuals in atomic
absorption and neutron activation for that purpose.
290
-------
The analytical techniques must be refined. The same
coal or char analyzed several times by NAA and by AA show
small but significant differences. However, all analyses
made by the same method always appear to be internally con-
sistent. For that reason, elements analyzed by two diffe-
rent methods are reported separately according to the
method used. However, the overall accuracy and precision
of the chemical analyses used in this research are good and
quite comparable with results reported by other researchers
(see Sections 2.2.5 and 2.2.6 for references).
291
-------
TECHNICAL REPORT DATA , „ ,
(Please nod Instruction* on the reverse before completing} __^_^_____— -
EPA-600/7-79-241
4. TITLE AND SUBTITLE
Devolatilization Kinetics and Elemental Release in the
Pyrolysis of Pulverized Coal
7. AUTHOR(S)
V. H. Agreda, R. M. Felder , and J. K. Ferrell
U. PERFORMING ORGANIZATION NAME AND ADDRESS
North Carolina State University
Department of Chemical Engineering
Raleigh, North Carolina 27650
72. SPONSORING AGENCY NAME AND ADDRESS
EPA, Office of Research and Development
Industrial Environmental Research Laboratory
Research Triangle Park, NC 27711
^RECIPIENT'S ACCESSION- NO.
5. REPORT DATE
November 1979
6. PERFORMING ORGANIZATION CODE
8. PERFORMING ORGANIZATION REPORT NO.
10. PROGRAM ELEMENT NO.
EHE623A
11. CONTRACT/GRANT NO.
Grant R804811
13. TYPE OF REPORT AND PERIOD COVERED
Final; 9/77 - 9/79
14. SPONSORING AGENCY CODE
EPA/600/13
is. SUPPLEMENTARY NOTES TjjRL-RTP project officer is N. Dean Smith, Mail Drop 61, 919/
541-2708.
re. ABSTRACT The reDOrt gives results of a study of the evolution of volatile matter and
trace elements from pulverized coal during pyrolysis in an inert atmosphere, using
batch and laminar flow furnace reactors. Five coals were used, ranging in rank from
lignite to anthracite. Ash losses significantly affected calculated extents of devola-
tilization at any pyrolysis temperature, making the commonly used ash tracer tech-
nique a potential source of error in all experimental pyrolysis studies. Estimated
weight losses can be corrected for £hls~effect. Data on transient and equilibrium
elemental release and volatile yields were obtained in a batch furnace reactor, under
slow heating rates, over a wide range of temperatures and residence times. Weight
losses of all coals increased significantly with temperature. Sm, Cr, Th, Sc, Fe,
and Co were retained completely in the chars; As and Se showed intermediate vola-
tility; and S, Pb, Hg, and Cl showed high volatility. An empirical mathematical
model correlates the equilibrium release of Hg, Pb, As, Cl, and Se, as a function
of temperature, for the five coals. The same model correlates S release data for
coals with rank up to bituminous. Devolatilization kinetics data were obtained in a
laminar flow reactor for two lignites and a subbituminous coal, with rapid heating.
low to intermediate temperatures, rapid quenching, and 150-1500 msec residence.
^^ KEY WORDS AND DOCUMENT ANALYSIS
i DESCRIPTORS
, — . —
Dilution Elements
oal Coal Gasification
ulverized Fuels
yrolysis
olatility
inetics
DISTRIBUTION STATEMENT
Release to Public
.IDENTIFIERS/OPEN ENDED TERMS
Pollution Control
Stationary Sources
Devolatilization
Elemental Release
19. SECURITY CLASS (This Report)
Unclassified
20. SECURITY CLASS (This page)
Unclassified
. COS ATI Field/Group
13B 14B
21D 13H
07D
20M
20K
21. NO. OF PAGES
304
22. PRICE
292
------- |