U.S. Environmental Protection Agency Industrial Environmental Research EPA-600/7-77-1 38
Office of Rese;ii<;h and Development Laboratory
Research Triangle Park. North Carolina 27711 December 1977
SUPPORTIVE STUDIES IN
FLUIDIZED-BED COMBUSTION
Interagency
Energy-Environment
Research and Development
Program Report
-------
RESEARCH REPORTING SERIES
Research reports of the Office of Research and Development, U.S.
Environmental Protection Agency, have been grouped into seven series.
These seven broad categories were established to facilitate further
development and application of environmental technology. Elimination
of traditional grouping was consciously planned to foster technology
transfer and a maximum interface in related fields. The seven 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
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-agehcy 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 systems. The goal of the Program
is to assure the rapid development of domestic energy supplies in an
environmentally—compatible manner by providing the necessary
environmental data and control technology. Investigations include
analyses 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 environmental issues.
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 recommen-
dation for use.
This document is available to the public through the National Technical
Information Service, Springfield, Virginia 22161.
-------
EPA-600/7-77-138
December 1977
SUPPORTIVE STUDIES
IN
FLUIDIZED-BED COMBUSTION
by
A. Jonke, G. Vogel, I. Johnson, S. Lee,
J. Lenc, A. Lescarret, J. Montagna, F. Nunes,
J. Shearer, R. Snyder, G. Smith, W. Swift,
F. Teats, C. Turner, and I. Wilson
Argonne National Laboratory
9700 South Cass Avenue
Argonne, Illinois 60439
EPA Interagency Agreement No. IAG-D5-E681
Program Element No. EHE623A
EPA Project Officer: Walter B. Steen
Industrial Environmental Research Laboratory
Office of Energy, Minerals, and Industry
Research Triangle Park, N.C. 27711
Prepared for
U.S. ENVIRONMENTAL PROTECTION AGENCY
Office of Research and Development
Washington, D.C. 20460
-------
TABLE OF CONTENTS
Page
ABSTRACT 1
SUMMARY 1
TASK A. REGENERATION PROCESS DEVELOPMENT 11
1. Experimental 13
a. Materials—Sorbents 13
b. Materials-Coals 13
c. PDU—Combustion System and Procedure . . 13
d. PDU—Regeneration System and Procedure 16
2. PDU Regeneration Rate Experiments with Tymochtee Dolomite. . 17
a. Effect of Solids Residence Time and Temperature
on Extent of CaO Regeneration 18
b. Regression Analysis of Regeneration Data 20
c. Effect of Solids Residence Time and Temperature
on S02 Concentration in the Off-Gas 22
d. Effect of System Pressure on Extent of CaO
Regeneration and Off-Gas S02 Concentration 25
e. Formation of CaS 27
3. PDU Regeneration Rate Experiments with Greer Limestone ... 27
a. Effects of Solids Residence Time and Temperature
on Extent of CaO Regeneration 27
b. Effect of Solids Residence Time and Temperature
on S02 Concentration in the Off-Gas 30
c. Effect of System Pressure during Sulfation on
the Reactivity of Greer Limestone 31
4. Characterization of Causes of Defluidization during
Regeneration 31
a. Experimental Design and Procedure 33
b. Effect of Operating Variables on Defluidization
Velocity 34
c. Analysis of Variance for Defluidization Velocity .... 35
d. Regression Analysis for Defluidization Velocity 35
5. Cyclic Sorbent Life Studies with Tymochtee Dolomite 37
a. Experimental Conditions 38
b. Sulfur Acceptance during Combustion 38
c. Sulfur Release during Regeneration 38
d. TGA Sulfation Experiments 38
e. Estimate of Sorbent Makeup Requirements to Meet
EPA Sulfur Emission Limit 43
iii
-------
TABLE OF CONTENTS (Contd.)
f. Porosity of Dolomite as a Function of Utilization
Cycle 49
g. Coal Ash Buildup during Utilization Cycles 49
h. Electron Microprobe Analysis of Tymochtee Dolomite
from Tenth Utilization Cycle 59
i. Carbonate Levels of Sorbent Samples 66
j. Attrition and Elutriation Losses during Regeneration
and Combustion 67
k. Amount of Sorbent Processed per Cycle 70
6. Cyclic Sorbent Life Study with Greer Limestone 71
a. Combustion Step Results 71
b. Cyclic and Total Calcium Utilization 72
c. Porosity of Limestone as a Function of Utilization
Cycles 74
d. Limestone Reactivity as a Function of Cyclic
Utilization 74
e. Regeneration Step Results 75
f. Coal Ash Buildup during Cyclic Utilization 76
g. Attrition and Elutriation of Limestone Particles
during Regeneration and Sulfation 79
h. Total Cyclic Limestone Inventory 82
7. Regeneration Process Scale-up and Flowsheet Determination. . 82
8. Regeneration System Modifications 86
TASK B. REGENERATION PROCESS ALTERNATIVES 88
TASK C. SYNTHETIC SORBENTS FOR S02 EMISSION CONTROL 88
1. Introduction 88
2. Attrition Resistance 88
3. Bauxite Support 89
4. Cost 93
5. Conclusions 96
TASK D. LIMESTONE CHARACTERIZATION 97
1. Limestone Properties Affecting S0£ Reactivity 97
2. Effects of Precalcination and Heat Treatment 101
iv
-------
TABLE OF CONTENTS (Contd.)
Page
3. Limestone Attrition 104
4. Conclusions 108
TASK E. TRACE ELEMENTS AND COMBUSTION EMISSION STUDIES Ill
-1. The Effect of Additives on the Calcination/Sulfation
of Limestone/Dolomite Ill
a. Mechanism of Enhancement by Sodium Chloride Ill
b. Evaluation of Seven Additives 117
c. Effect of NaCl Additive on Several Limestones 126
d. Porosities of Limestones 130
2. The Determination of Inorganic Constituents in the
Effluent Gas from Coal Combustion 139
a. Coal Combustion Experiments 140
b. Charcoal Combustion Experiments 148
c. Lignite Combustion Experiments 152
d. Relation of Coal Ash Content to Sodium Vaporization. . . 154
TASK F. FLUE-GAS CLEANING STUDIES 156
1. Evaluation of On-Line Light-Scattering Particle Analyzers. . 156
a. Principles of the SDL Particle Morphokinetometer .... 156
b. Procedure for Comparative Flue-Gas Particle
Measurements 159
c. Experimental Evaluation of the PM Particle Analyzer. . . 161
d. The Effect of the PM Particle Analyzer Calibration
on Particle Measurements 167
2. Particle Removal from Flue Gas 171
a. Granular-Bed Filter 172
b. Acoustic Agglomeration 180
c. High Efficiency Cyclones 184
ACKNOWLEDGMENTS 184
REFERENCES 185
APPENDIX A: Physical and Chemical Properties of Coals and Sorbents
used in Sorbent Regeneration Studies 189
-------
LIST OF FIGURES
No. Title Page
1. Conceptual Fluidized Bed Coal Combustion Power-Generating
Facility Having Sorbent-Regeneration and Sulfur-Recovery
Capabilities ........................... 12
2. Simplified Equipment Flowsheet of PDU Fluidized-Bed
Combustor and Associated Equipment ................ 14
3. Six-Inch-Diameter Pressurized, Fluidized-Bed Combustor ...... 15
4. Experimental Sorbent Regeneration System ............. 17
5. Regeneration of CaO in Tymochtee Dolomite as a Function of
Solids Residence Time ...................... 20
6. The Extent of CaO Regeneration for Tymochtee Dolomite as
a Function of Temperature and Residence Time as Represented
by the Model Equation, Equation 1 ................ 21
7. Predicted and Experimental S02 Concentration as a Function
of Solids Residence Time at Three Regeneration Temperatures ... 22
8. Predicted S02 Concentration in the Dry Off-Gas as a Function
of Solids Residence Time, Regeneration Temperature, and
System Pressure ......................... 24
9. Regeneration of CaO in Greer Limestone as a Function of
Solids Residence Time and Regeneration Temperature ........ 29
10. Experimental S02 Concentration for the Regeneration of Greer
Limestone as a Function of Solids Residence Time and Temper-
ature .............................. 30.
11. Qualitative Plot of Bed Temperature and Pressure Drop as a
Function of Fluidizing-Gas Velocity for a Def luidization
Experiment ............................ 33
12. Sulfur Retention in Bed and SOa Concentration in Flue Gas
as a Function of Cycle Number .................. 40
13. Conversion of CaO to CaSOi+ as a Function of Time and Sulfation
Cycle as Determined by TGA Sulfation Experiments ......... 42
14. Percent Calcium as CaSOi^ as a Function of Time and Sulfation
Cycle as Determined by TGA Sulfation Experiments ......... 42
15. CaO/S Ratio Required to Achieve 75% Sulfur Retention as a
Function of Cycle ........................ 44
vi
-------
LIST OF FIGURES (Contd.)
No. Title Page
16. CaO Utilization at 75% Sulfur Retention as a Function of
Sulfation Cycle 47
17. Recycle Mechanism to Approximate the Steady-State Distribution
of Sorbent for a Continuous Combustion-Regeneration Process ... 47
18. Calculated Makeup and Total CaO/S Ratios Required to Achieve
75% Sulfur Retention as a Function of the Makeup CaO to
Total CaO Ratio 48
19. Pore Distributions of Dolomite Samples from Cycles Two
and Ten 50
20. Coal Ash Buildup as a Function of Utilization Cycle 52
21. Coal Ash Buildup as a Function of Particle Diameter in
Tenth Cycle Regenerated Particles 52
22. Photomicrographs of Sulfated and Regenerated Tymochtee
Dolomite Particles from the First and Fifth Utilization
Cycles 53
23. Photomicrographs of Sulfated and Regenerated Tymochtee
Dolomite Particles from the Tenth Utilization Cycle 54
24. Photomicrograph of Cross Section of a Tymochtee Dolomite
Particle from the First Combustion Cycle 56
25. Photomicrograph of Cross Section of a Tymochtee Dolomite
Particle from the First Regeneration Cycle 56
26. Photomicrograph of a Cross Section of a Tymochtee Dolomite
Particle from the Tenth Combustion Cycle 57
27. Photomicrograph of Cross Section of a Tymochtee
Dolomite Particle from the Tenth Regeneration Cycle 57
28. Photomicrograph of Cross Section of a Tymochtee Dolomite
Particle from the Tenth Regeneration Cycle 58
29. Photomicrograph of Cross Section of an Unreacted Tymochtee
Dolomite Particle 58
30. Electron Microprobe Analysis of a Typical Regenerated
Dolomite Particle (P-l) from the Tenth Cycle 60
31. Electron Microprobe Analysis of a Typical Regenerated
Dolomite Particle (P-2) from the Tenth Cycle 61
vii
-------
LIST OF FIGURES (Contd.)
No. Title Page
32. Electron Microprobe Analysis of a Typical Regenerated
Dolomite Particle (P-3) from the Tenth Cycle 62
33. Electron Microprobe Analysis of a Typical Partially
Sulfated Dolomite Particle (PS-1) from the Tenth Cycle 63
34. Electron Microprobe Analysis of a Typical Partially
Sulfated Dolomite Particle (PS-2) from the Tenth Cycle 64
35. Electron Microprobe Analysis of a Typical Partially
Sulfated Dolomite Particle (PS-3) from the Tenth Cycle 65
36. Comparison of the Reactivity of Tenth Cycle Regenerated
Dolomite (-14 +30 mesh) to that of a Crushed (-30 +100 mesh)
Sample from the Same Experiment 66
37. Weight Fraction of Unsulfated Calcium as Calcium Carbonate
in Sulfated and Regenerated Dolomite Samples as a Function
of Utilization Cycle 67
38. Cyclic Calcium Utilization for Greer Limestone for 84%
Sulfur Retention 73
39. Pore Distributions of Limestone Samples from Cycles Two
and Ten 74
40. Cyclic CaO/S Molar Feed Ratio Required to Retain 84% of
the Sulfur 75
41. Conversion of Available Calcium to CaSO^ as a Function of
Time 76
42. Coal Ash Buildup as a Function of Greer Limestone Utilization
Cycle 79
43. Photomicrographs of Sulfated and Regenerated Greer Limestone
Particles 80
44. Process Flowsheet for a 10-MWe FBC Boiler with Sorbent
Regeneration, Using Tymochtee Dolomite 84
45. Schematic of Regeneration System with Sorbent Preheater 87
46. Particle Size Distribution of a-A^Os (heat-treated at 1100°C)
Before and After Attrition Test 89
47. Particle Size Distributions of Sulfated Dolomite from First
Cycle of Ten-Cycle Experiment Before and After Attrition. .... 91
viii
-------
LIST OF FIGURES (Contd.)
No. Title^ Page
48. Particle Size Distributions of Regenerated Dolomite,
Before and After Attrition Test 91
49. Porosity of Bauxite 92
50. Calcium Utilization of Synthetic Sorbents 92
51. Rate of SOs Capture by Sorbents 93
52. Calcium Utilization at 900°C for Ten Precalcined (at 900°C)
Limestones in a Thermogravimetric Analyzer 97
53. Cumulative Pore Volume as a Function of Pore Diameter
for Ten Limestones 100
54. Calcium Utilization as a Function of Surface Area of Pores
having Diameters Larger than MPD for Ten Limestones 101
55. Calcium Utilization as a Function of Various Pretreatments. . . . 102
56. Increased Energy Cost for Pretreatment Required to Reduce
the Environmental Impact of Mining and Disposal of Sorbents . . . 104
57. Laboratory-Scale Fluidized-Bed Apparatus for Attrition
Experiments 105
58. Attrition Rate of Calcined Tymochtee Dolomite as a Function
of Superficial Gas Velocity; L/D = 1.39 107
59. Attrition Rate of Calcined Tymochtee Dolomite as a Function
of Bed Depth 107
60. Material Loss from Beds of Precalcined Limestones, Super-
ficial Gas Velocity: 1.46 m/s; L/D: 1.38 108
61. Material Loss at 1.46 m/s Gas Velocity, Room Temperature 109
62. Effect of Impurity Concentration on Attrition 110
63. Sulfation of Partially Sulfated Limestone Particle 114
64. Sulfation of Crystalline Calcite 116
65. Enhanced Sulfation in 4% S02 at 850°C of Greer Limestone
by Pretreatment 120
IX
-------
LIST OF FIGURES (Contd.)
No. Title Page
66. Enhanced Sulfation in 0.4% S02 at 850°C of Greer Limestone
by Precalcination ........................ 120
67. Effect of NaCl Concentration on Sulfation of Greer
Limestone at 850°C in 0.3% S02 .................. 122
68. Enhancement of Greer Limestone Sulfation with Na^O^ at
850°C in 0.3% S02 ........................ 123
69. Enhancement of Greer Limestone Sulfation with Na2C03 at
850°C in 0.3% S02 ........................ 124
70. Enhancement of Greer Limestone Sulfation with CaCl2 at
850°C in 0.3% S02 ........................ 125
71. Enhancement of Greer Limestone Sulfation with KC1 at 850°C
in 0.3% S02 ........................... 126
72. Weight of SOs Captured as a Function of Time for Eleven
Untreated Limestones. ... ................... 127
73. Weight of SOs Captured as a Function of Time for Eleven
Limestones with 2 wt % NaCl ................... 128
74. Effect of NaCl on Sulfation of Limestones at 850°C in
0.3% S02 after 7 Hours ...................... 129
75. Porosimetry Curves for Eleven Limestones Calcined One Hour
at 850°C in 20% C02 ....................... 131
76. Porosimetry Curves for Eleven Limestones Plus 2% NaCl,
Calcined 1 Hour at 850°C in 20% C02 ............... 132
77. Porosimetry Curves for Greer Limestone plus NaCl Calcined
at 1 Hour at 850°C in 20% C02 .................. 133
78. Effect of NaCl on Sulfation of Limestones (Both Precalcined
and Simultaneous Calcination/Sulfation) at 850°C in 0.3% S02
After Seven Hours ........................ 134
79. Sulfation at 805°C of Limestones as a Function of Total
Inerts Content in Precalcined Stones Treated with 2% NaCl .... 135
80- Average Pore Diameter as a Function of Total Inerts Content
of Limestones .......................... 136
81. Sulfation at 850°C of Limestone as a Function of Average
Pore Diameter in Precalcined Stones Treated with 2% NaCl ..... 137
-------
LIST OF FIGURES (Contd.)
No. Title Page
82. Average Pore Diameter as a Function of Percent NaCl Added .... 139
83. Bed Temperature and Effluent Composition in a Typical
Continuous Two-Batch Coal Combustion Experiment 142
84. Bed Temperature and Effluent Composition in a Typical
Continuous Two-Batch Coal Combustion Experiment 143
85. Relationship of 02 Content of Inlet Combustion Mixture
to 02 Consumption and C0£ Formation in Combustion of Coal .... 144
86. Relationship between Ash Content of Coal and Sodium
Vaporized during Combustion at 900°C of Coal Impregnated
with 0.5% NaCl 155
87. Spectron Development Laboratory's PM Analyzer System for
Velocity and Particle Size Measurement 157
88. Visibility as a Function of Particle Size and Fringe Period
for Two Particle Shapes 159
89. Schematic of FBC System with Modified Flue-Gas System 160
90. Sampling System for Flue Gas Particles 161
91. Partial (1.5-23 pm) Cumulative Mass Distribution Obtained
On-Line with the Spectron PM Analyzer Compared with that
Obtained with a Coulter Counter (SGL-1) 163
92. Partial (5-74 pm) Cumulative Mass Distribution Obtained
On-Line with the Spectron PM Analyzer Compared with that
Obtained with a Coulter Counter (SGL-1) 163
93. Fractional Mass Distribution of All Elutriated Particles
from the ANL PDU Combustor During a Combustion Experiment
(SGL-1) 165
94. Cumulative Mass Distribution of All Elutriated Particles
from the ANL PDU Combustor During a Combustion Experiment
(SGL-1) 165
95. Comparison of the Partial (1.5-23.9 pm) Cumulative
Distribution Obtained On-Line with the Spectron PM Analyzer
with that Obtained by Coulter Counter Analysis of a Particle
Sample (SGL-2C) 166
96. Fractional Mass Distribution of Particles Contained in the
Flue Gas Between the Secondary Cyclone and the Filter (SGL-2C)
Obtained by Coulter Counter Analysis 167
xi
-------
LIST OF FIGURES (Contd.)
No. Title Page
97. Cumulative Mass Distribution of Particles in the Flue Gas
Between the Secondary Cyclone and Metal Filter (SGL-2C) ..... 168
98. Comparison of the Partial (5-73 ym) Cumulative Mass
Distribution Obtained On-Line with the Spectron PM Analyzer
with that Obtained with a Coulter Counter (SGL-1) ........ 169
99. Comparison of the Partial (1.5-23 urn) Cumulative Mass
Distribution Obtained On-Line with the Spectron PM Analyzer
with that Obtained with a Coulter Counter (SGL-1) ........ 170
100. Comparison of the Partial (1.5-23.8 urn) Cumulative Mass
Distribution Obtained On-Line with the Spectron PM Analyzer
to that Obtained by Coulter Counter Analysis of a Particle
Sample (SGL-2C) ......................... I71
101. Granular-Bed Filter Assembly
102. Pressure Drop Across a Fixed Bed of -6 +14 Mesh Tymochtee
Dolomite as a Function of Mass Flow Rate Through the Bed .... 176
103. Pressure Drop Across a Fixed Bed of -14 +30 Mesh Tymochtee
Dolomite as a Function of Mass Flow Rate Through the Bed .... 176
104. Test Arrangement for the Determination of Particulate
Loadings in the Effluent Gas from the Granular-Bed Filter. . . . 178
105. Modified Flue-Gas System for Granular-Bed Filter Tests ..... I80
106. Schematic of Resonant Manifold System .............. 183
xii
-------
LIST OF TABLES
No. Title
1. Experimental Conditions and Results for the Regeneration of
Sulfated Tymochtee Dolomite by the Incomplete Combustion of
Triangle Coal in a Fluidized Bed ................. 19
2. Effect of Regeneration Pressure on the Regeneration of
Tymochtee Dolomite and the SC>2 Concentration in the Off-Gas ... 26
3. Experimental Conditions and Results for the Regeneration of
Greer Limestone by the Incomplete Combustion of Sewickley
Coal in a Fluidized Bed ..................... 28
4. Operating Conditions and Flue-Gas Compositions during the
Sulfation of Greer Limestone with Sewickley Coal. Combustion
at Two Operating Pressures .................... 32
5. Def luidization during Regeneration — Experimental Conditions
and Results of the Full 23 Factorial Experiment ......... 34
6. Analysis of Variance for Def luidization Velocity Data from
the AGL Series of Experiments .................. 36
7. Calculated Def luidization Velocities (Vj, m/s) at Different
Bed Temperatures, Reducing Gas Concentrations, and Bed
Particle Sizes for the Regeneration Process ........... 37
8. Operating Conditions and Flue-Gas Compositions for Combustion
Step of Cyclic Experiments .................... 39
9. Experimental Conditions and Results for the "Regeneration
Step of the Ten Utilization Cycles with Tymochtee Dolomite. ... 41
10. Comparison of the Experimental Cyclic Sulfation Results
Obtained at ANL with those reported by Zielke et al .- ...... 45
11. Porosity (cm3/g) of Tymochtee Dolomite as a Function of
Utilization Cycle ........................ 50
12. Calculated Coal Ash Buildup during Sulfation and Regeneration
of Tymochtee Dolomite, Based on Enrichment of Silcion ...... 51
13. Attrition and Elutriation Losses for Tymochtee Dolomite
during Regeneration in the Cyclic Utilization Study ....... 68
14. Decrepitation and Entrainment Losses and Calcium Material
Balances for the Ten Combustion Experiments in the Cyclic
Sorbent Utilization Study .................... 69
15. Gross Amounts of Sorbent Processed and Calcium Balances for
Each Half-Cycle in the Cyclic Combustion/Regeneration Study ... 70
Xlll
-------
LIST OF TABLES (Contd.)
No. Title Page
16. Operating Conditions and Flue-Gas Compositions for Combustion
Steps of Cyclic Experiments with Greer Limestone and
Sewickley Coal 72
17. Experimental Conditions and Results for the Regeneration
Step of Ten Utilization Cycles with Greer Limestone 88
18. Calculated Ash Buildup during Sulfation and Regeneration
of Greer Limestone 78
19. Losses of Greer Limestone Caused by Attrition and Elutriation
during Sulfation and Regeneration Steps in the Cyclic
Utilization Study 81
20. Total Reacted Limestone Inventory as a Function of Cycle
and Stage 82
21. Effect of Makeup CaO/S Mole Feed Ratio for Tymochtee Dolomite
in Boiler on Regeneration System 85
22. Fluidized-Bed Attrition Experiments 90
23. Chemical Compositions of Limestones 98
24. Eutectics and Double Salts Formed in the System, NaCl-CaC03-
MgC03-CaSOu-MgSOi+ 112
25. Effect of Nad on Reactions of Calcite 118
26. Effect of Additives on Simultaneous Calcination-Sulfation
of Crystalline Calcite at 900°C 121
27. A Material Balance of Sodium from Combustion of Illinois
Herrin No. 6 Coal Impregnated with 0.5 wt % NaCl 147
28. A Material Balance of Potassium from Combustion of Illinois
Herrin No. 6 Coal Impregnated with 0.5% NaCl 148
29. Material Balance of Sodium from Combustion of Activated
Coconut Charcoal Impregnated with 0.5 wt % NaCl 150
30. Material Balance of Potassium from Combustion of Activated
Coconut Charcoal Impregnated with 0.5 wt % NaCl 151
31. Material Balance of Sodium from Combustion of Glen harold
Lignite, North Dakota 153
32. Selected Interference Fringe Spacings and the Corresponding
Measurable Spherical Particle Size Ranges 158
xiv
-------
LIST OF TABLES (Contd.)
No. Title Page
33. Experimental Conditions for a Combustion Experiment in the
Evaluation of the SDL Particle Morphokinetometer (PM) 162
34. Experimental Determinations of e and 4>s in the Ergun
Correlation for the Flow of Gas Through a Fixed Bed of
Tymochtee Dolomite 177
35. Measured Particulate Loadings in Effluent Gas from Granular-
Bed Filter when Passing a "Clean" Gas through the Filter 179
XV
-------
SUPPORTIVE STUDIES IN
FLUIDIZED-BED COMBUSTION
Annual Report
July 1976—June 1977
by
G. J. Vogel, I. Johnson, S. Lee, J. Lenc, A. S. Lescarret,
J. Montagna, F. Nunes, J. Shearer, R. Snyder, G. Smith,
W. Swift, F. G. Teats, C. Turner, I. Wilson, and A. A. Jonke
ABSTRACT
These studies support the development studies for atmospheric
and pressurized fluidized-bed coal combustion. Laboratory and
bench-scale studies aimed at providing needed information on
combustion optimization, regeneration process development, solid
waste disposal, synthetic S02~sorbent studies, emission control,
and other tasks are included. Characterization of a variety of
limestones and dolomites from various parts of the country for
suitability in fluidized-bed combustors is also included. Reduc-
tion in solid waste volumes to reduce the environmental impact of
the waste sulfated limestone is one of the major goals of this
program. These studies are designed to supply data essential for
the application of fluidized-bed combustion units to public utility
and industrial systems.
This report presents information on: 10-cycle combustion-
regeneration PDU experiments using Greer limestone and Tymochtee
dolomite, bed defluidization, flowsheet development, preparation
of synthetic S02~sorbents containing metal oxides, limestone charac-
terization, coal combustion reactions, the enhancement of the
sulfation of limestone by Nad, evaluation of on-line particle size
analyzers, and status of the flue-gas cleaning studies.
SUMMARY
Task A. Regeneration Process Development
The feasibility of a sorbent regeneration process will depend on (1)
the ability to regenerate the stones and to generate an S02~rich off-gas
which can be treated in a sulfur recovery process, (2) the reactivity of
the regenerated sorbent during subsequent sulfation (coal combustion) cycles,
and (3) the availability of a sorbent that will not decrepitate beyond
acceptable levels.
In the sorbent regeneration process being investigated, CaSO^ is
reductively decomposed in a fluidized bed at ^1100°C. The heat and the
reductants required are produced by incomplete combustion of coal in a
-------
fluidized bed of sulfated stone. A solid-gas reaction by which regeneration
occurs is:
CaSO^ + CO ->• CaO + C02 + S02 (1)
The ability of the reductive decomposition process to fulfill the above require-
ments of a sorbent regeneration process for FBC boilers is being investigated.
*
PDU Regeneration Rate Experiments with Tymochtee Dolomite. A series
of regeneration experiments using once-sulfated Tymochtee dolomite and
Triangle coal has now been completed, and the effects of regeneration temper-
ature, solids residence time, and regeneration pressure on (1) the regeneration
of CaO and (2) the SC>2 concentration in the dry off-gas have been evaluated.
These data are being used in a model for the one-step regeneration process
to find process conditions that are optimum from technical and economic points
of view.
In experiments made at bed temperatures of 1000, 1050, and 1100°C, in
which the solids residence time was varied from 35 min to 7 min, the extent
of CaO regeneration decreased with decreasing solids residence time at all
temperatures. Nevertheless, at 1100°C the extent of regeneration remained
quite high (^75%), even at the lowest investigated solids residence time.
At 1000°C, the S02 concentration in the dry off-gas was relatively
unaffected (remaining at ^2%) as the solids residence time was decreased from
37 min to 11 min. At 1050°C, the concentration of S02 in the dry off-gas
increased from 3.0% to 4.8% when the solids residence time was decreased
from 34 min to 12 min. At 1100°C, the S02 concentration increased to a
greater extent than at 1050°C as the solids residence time was decreased.
It was predicted that the S02 concentration at 1050°C and 1100°C would be
at a maximum for a solids residence time of ^2.5 min. At 1100°C, the
maximum predicted S02 concentration in the dry off-gas would be 11.5%.
Also, the experimental regeneration data were statistically analyzed
to obtain an equation for the extent of CaO regeneration as a function of (1)
sorbent residence time in the reactor and (2) regeneration temperature. This
equation was used in the mass and energy constrained regeneration process
model to predict expected S02 concentrations in the off-gas in and beyond the
experimentally investigated temperature range (i.e,, at temperatures up to
1200°C). At 1100°C the S02 concentration in the off-gas from a pressurized
(1000 kPa) regeneration process was estimated to be ^2%, and ^12% at atmo-
spheric pressure (101.3 kPa).
In still another series of experiments, the regeneration pressure was
decreased by ^25%, from ^150 kPa to 115 kPa. The extent of regeneration was
not affected by pressure; the dependence of extent of regeneration on solids
*
Process development unit.
-------
residence time was equivalent at the two pressures. Regeneration pressure
did, however, affect S02 concentration in the off-gas via the extent of
dilution. At the lower pressure, higher 862 concentrations, up to 10.4%,
were obtained.
Temperature, solids residence time, and pressure had no significant
effect on the CaS content of the bed, which was low in all experiments.
PDU Regeneration Rate Experiments with Greer Limestone. A series
of regeneration experiments was performed to obtain CaO regeneration
rate data for Greer limestone. The dependence of CaO regeneration
on regeneration temperature and solids residence time in the reactor was
determined. The regeneration rate of Greer limestone was found to be equal
to that of Tymochtee dolomite. The experimental results were used to obtain
a "best fit" model equation for the dependence of CaO regeneration on
temperature and solids residence time in the reactor. This relationship
will be used in a process computer model to optimize the design process
conditions for a pilot plant.
The effects of bed temperature, 1050 or 1100°C, and solids residence
time, ^20, 11, and 7 1/2 minutes, in the regenerator on S0£ concentrations
in the off-gas were determined using sulfated Greer limestone as the feed
material. At the longest residence time little effect of temperature was
found, but at the shortest residence time the S02 concentration was twice
as high at the higher temperature. To obtain the highest S02 concentrations,
a short residence time and a high bed temperature are required.
Two experiments were performed to test the effect of combustion system
pressure on the reactivity of Greer limestone, one at a system pressure of
308 kPa (3 atm) and the other at 610 kPa (6 atm) ; both were at a bed temper-
ature of 855°C. It was found that, although the extent of calcination was
much lower in the higher pressure experiment, the reactivi y of the stone was
approximately the same in both experiments.
Characterization of Def luidization Causes du ring Regeneratio . The
effects of experimental variables on the def luidization characteristics
(agglomeration) of the regenerator fluid bed are being investigated in the
PDU regenerator. Sulfated Greer limestone obtained from PER and Sewickley
coal are being used. A statistical experiment has been completed (a full
23 factorial experiment plus two replicate experiments) in which the effects
of bed temperature (1050 and 1100°C) , total reducing gas concentration in
the off-gas (2.5 and 5.0%), an-i feed sorbent particle size (-10 +30 mesh and
-14 +30 mesh) on the def luidization velocity (minimum velocity required to
prevent agglomeration) were determined. Regeneration temperature and total
reducing gas concentration in the off-gas were found to have the greatest
statistical influence on the defluidization velocity. A model equation was
obtained that can be used to predict the minimum operable f luidizing-gas
velocity at which an industrial regenerator can be operated without agglom-
eration of the fluid bed.
Cyclic Sorbent Life Studies with Tymochtee Dolomite. The '-ffe^t of
repeated utilization cycles on the performance of Tymochtee dolomite as an
-------
S02-acceptor was evaluated in a ten-cycle combustion-regeneration experiment.
Combustion in each cycle was performed at a 900°C bed temperature, 810 kPa
pressure, 1.5 CaO/S mole ratio (ratio of unsulfated calcium in sorbent to
sulfur in coal), ^17% excess combustion air, 0.9 m/s fluidizing-gas velocity,
and a 0.9 m bed height. Regeneration in each cycle was at a nominal system
pressure of 158 kPa, a bed temperature of 1100°C, and a fluidized-bed height
of ^46 cm. The required heat and reductants for regeneration were provided
by incomplete combustion of Triangle coal.
During sulfation, the sorbent exhibited a steady loss in reactivity
with increasing utilization cycle. The results of TGA sulfation experiments
on samples of regenerated sorbent from the ten regeneration half-cycle
experiments are presented. They show that both the rate of sulfation and
the extent of sulfation decreased with utilization cycle. The TGA results
indicate a leveling-off of the loss of reactivity after the eighth sulfation
cycle which was not apparent from sulfation results obtained in the PDU
cyclic study.
The extent of regeneration remained at acceptable levels CWO%) for all
ten cycles, and S02 concentrations in the dry off-gas measured ^8.5% up
through cycle 7. In the three final cycles, the SC>2 concentration upon
regeneration was <7% because of a lower sulfur content of the sulfated
sorbent fed to the regenerator.
Based on the results of the cyclic study, an analysis was made of the
effect of makeup rate (ratio of makeup CaO to total CaO entering combustor)
on the total CaO/S ratio required during continuous recycle operation for
a sulfur retention of 75%. For example, a makeup rate of 0.18 requires a
projected makeup CaO/S ratio of M).27 and a total CaO/S ratio of ^1.5 for a
sulfur retention of 75%. Decreasing the makeup rate to 0.1 reduces the
required makeup CaO/S ratio to ^0.2 but increases the total CaO/S required
to 2.0 because of the lower reactivity of the recycled CaO.
The porosity of regenerated dolomite decreased with cyclic use, as did
its capacity to act as an S02-acceptor during combustion. The loss of
porosity and loss of reactivity towards S02 could be due to either or both
of the following: (1) the buildup of an ash layer around the particles or
(2) high-temperature (1100°C) exposure during regeneration. Sintering, which
begins to occur at the regeneration temperature, decreases the reactivity
by decreasing the beneficial effect of porosity.
It was found that after ten utilization cycles, ^13 g of coal ash had
accumulated with the sorbent for every 100 g of starting virgin dolomite.
Photomicrographs of particle surface and cross-section features revealed
the formation of shells on the dolomite particles. Electron microprobe
analyses were performed on cross sections of sulfated and regenerated dolomite
samples from the tenth utilization cycle to determine the composition along
cross sections of the particles and the role of the constituents during
sulfation reactions. The analyses confirmed the existence of the ash shell.
It was found that the coal ash shell was enriched in iron and calcium
(relative to the silica concentrations in the pure coal ash). The sulfur
concentration profiles in the tenth-cycle sulfated particles suggested that
-------
diffusion through the ash shell is not the limiting factor during sulfation
and that the loss of reactivity was within the dolomite particles—perhaps
due to sintering and loss of local porosity.
The carbonate level of sulfated and regenerated samples is also given.
It further indicates the decrease in reactivity of the sorbent during com-
bustion with increasing utilization cycle. The extent of recarbonation
steadily decreases, as does the extent of sulfation.
Over ten cycles, sorbent losses during each regeneration cycle have aver-
aged a relatively low 2% of the material fed. The combined losses due to
attrition and/or elutriation have been found to be ^8% per complete cycle of
combustion and regeneration. This is the minimum expected makeup rate for
Tymochtee dolomite in an FBC process utilizing sorbent regeneration at the
operating conditions used in these tests. A still higher makeup rate may be
required to maintain sufficient reactivity in the recycled sorbent.
Cyclic Sorbent Life Study with Greer Limestone. A cyclic sorbent
utilization experiment was performed with Greer limestone, which is the
stone that will be used by Pope, Evans and Robbins (PER) in the Rivesville
pilot plant, and Sewickley coal. All ten combustion/regeneration cycles
have been completed. The regenerability of the sulfated limestone was not
affected by cyclic utilization and ranged from 49 to 71%; this appears to
be acceptable. The S02 concentrations in the regenerator off-gas ranged
from 8.6% in an early cycle to 6.1% in a later cycle.
The reactivity of the regenerated limestone with 862 decreased with
cyclic use. The CaO/S mole feed ratio required to maintain a 84% sulfur
retention increased from ^2.9 in the first cycle to ^8.2 in the tenth
cycle. TGA sulfation experiments confirmed the reactivity loss as a function
of cyclic use. The porosity of the regenerated stones decreased with cyclic
usage due to sintering in the high-temperature reducing environment of the
regenerator reactor. The loss of beneficial porosity is believed to be the
primary cause of reactivity losses.
The total and cyclic calcium utilization in the regenerated stones
decreased with cyclic usage. Calcium utilization decreased from ^30% in the
first cycle to ^9% in the tenth cycle. These results on reactivity and
utilization of Greer limestone as a function of cyclic usage will be used
to predict fresh limestone makeup rates and to make flowsheet calculations
optimizing FBC processes which include sorbent regeneration.
It has been found that after ten utilization cycles ^25 g of coal ash
had accumulated for every 100 g of starting virgin limestone. Photomicro-
graphs of utilized limestone particles revealed ash adherence but no ash
encapsulation.
The averaged combined loss of limestone per cycle due to attrition and
elutriation is ^10%—8% during the combustion steps and 2% during the
regeneration steps. This is the minimum fresh limestone makeup required for
a FBC process with sorbent regeneration. Higher makeup rates may be required
to maintain the reactivity of the stone in the fluid bed of the boiler.
-------
Regeneration Process Scale-up and Flowsheet Determination. A process
flowsheet for a FBC process with sorbent regeneration is presented in which
a fresh dolomite makeup CaO/S mole feed ratio of 0.2 is used. A 10-MWe FBC
boiler module is used as the basis and Tymochtee dolomite is the sorbent.
The calculations in the flowsheet are based on the performance of Tymochtee
dolomite as a function of utilization cycle, which was established in a
previously completed cyclic experiment. The effect of varying the fresh
sorbent makeup rate to the boiler on the regeneration system performance and
size was also evaluated. However, the results confirm that a final flowsheet
can be established only after economic evaluations are performed.
Regeneration System Moclif ications . Installation of a f luidized-bed
sorbent preheater for the regeneration system has been completed. The pre-
heater will be used to evaluate the effect of sulfated sorbent feed temperature
on the regeneration process.
Task B. Regeneration Process Alternatives
No report for this period.
Task C. Synthetic Sorbents for SOa Emission Control
Synthetic sorbents for 862 emission control in FBC have been under
development as replacements for limestones. It was hoped that these synthetic
sorbents would result in a significant reduction in the environmental impact
that would occur if large amounts of limestone for S02 concentration were
quarried and disposed of. In this final report on this topic, data are
presented which indicate that the attrition resistance of CaO-Al203 type
sorbent is superior to that of natural stones.
The results of a brief study of the use of unpurified bauxite as an
alumina source for the preparation of synthetic sorbents are reported. The
sorbents had about half the S02 capacity of synthetic sorbents prepared from
purified alumina.
The probable cost of synthetic sorbents developed in these studies has
been estimated. It is shown that the anticipated reduction in environmental
impact which might be achieved by using these sorbents is not great enough
to justify their additional cost.
Task D. Limestone Characterization
A study is under way of the characteristics of limestones which influence
their utility as SC>2 sorbents in fluidized-bed combustion systems. Results
are reported on the S02 capacity of ten different limestones as measured
using a TGA method. Conversions of CaO to CaSO^ varied from about 12% to 96%
for the ten stones. It was found that the calcium utilization can be corre-
lated with the internal surface area of the calcined limestone if the pore
area is measured only down to a minimum pore diameter. The minimum pore
-------
diameter is a linear function of the CaCOs content of the stone. Stones
having lower CaCOs contents can utilize smaller pores.
Precalcination and heat treatment were studied as a possible way of
enhancing the S02 capacity of limestones. The energy cost for a precalcination
process (in mills per kWh) has been estimated.
The attrition of limestones is being studied using a small room-temper-
ature fluidized-bed test rig. Results are reported on the effect of fluidi-
zation velocity, the L/D ratio of the bed, and the compositions of ten
different stones. It was found that the impurity content had a large effect
on the attrition characteristics of the stone. Stones with high impurity
contents had lower attrition rates.
Task E. Trace Elements and Combustion Emission Studies
The Effect of Additives on the Calcination/Sulfation of Limestone. The
objective of these studies is to investigate the application of chemical
additives for enhancing the SC>2 reactivity of limestone. These studies are
based on the discovery made by Pope, Evans and Robbins that addition of
common salt to a spent partially sulfated limestone bed rejuvenated its
ability to capture SC>2 from the hot flue gas. In present work the mechanism
of action is being investigated, other substances are being tested for
their enhancement activity, and the practical aspects of the application
to AFBC systems are being studied.
To study the mechanism of the action of NaCl on sulfation, large
crystals of pure calcite were exposed to a S02~air gas mixture at 900°C
both with and without treatment with NaCl. It was found that the NaCl
treatment markedly increased the extent of the calcination and the sulfation
reactions. Highly crystalline forms of both CaO and CaSO^ were formed as
a result of NaCl treatment. It was concluded that NaCl aids these reactions
by facilitating the crystallization process, thereby leading to a product
(CaO or CaSO^) composed of larger crystals. The system is therefore more
porous, and diffusion of the reactive gases into the limestone particle is
facilitated. It has been hypothesized that the NaCl acts by forming a
surficial liquid film which provides a ready path for recrystallization.
Results are reported on the effect of the concentrations of NaCl,
Na2SOit, Na2C03, CaCl2, and KC1 on the sulfation of Greer limestone. These
salts are shown to decrease in effectiveness in the order KC1 > NaCl £
CaCl2 > Na2C03 > Na2SCh:.. Results are reported on the change in S02 capacity
produced by a 2% NaCl addition to eleven different limestones or dolomites.
It was found that limestones with the lowest S02 reactivities in the natural
state were most affected by NaCl addition.
To gain an understanding of the reasons for the large differences in
the effect of NaCl on different stones, porosity measurements were made on
each stone after calcination, both with and without prior NaCl treatment.
It was found that the NaCl treatment increased the mean pore diameter. The
increase, found to depend on the amount of salt added, is specific for each
-------
stone. The magnitude of the effect of NaCl on the mean pore diameter depends
on the impurity content of the stone, the effect being greater the smaller
a stone's impurity content. For stones treated with 2% NaCl, the conversion
of CaO to CaSOif was inversely proportional to the average pore diameter.
The Determination of Inorganic Constituents in the Effluent Gas from
Coal Combustion. Some chemical elements carried by combustion gases are
known to cause severe metal corrosion (for example, to turbine blades). A
study is under way to quantitatively determine which elements and chemical
compounds are present in the hot combustion gas of coal that may be important
to metal corrosion. A laboratory-scale batch fixed-bed combustor has been
designed and constructed for these studies. In this combustor, a sample
of coal can be burned under controlled conditions and the combustion gases
hot filtered and then cold trapped.
The results are reported for a series of experiments in which the
vaporization of sodium and potassium was studied. The sodium content of
the fuel was increased by the addition of 0.5 wt % NaCl prior to the com-
bustion. Experiments were done using, in different experiments, a bituminous
coal, a lignite, and charcoal as fuel. These three fuels had high, medium,
and low ash contents. The fraction of the sodium in the fuel which was
vaporized varied from about 20% for the low-ash (2%) charcoal to 1% for the
high-ash (16%) bituminous coal.) Analysis of the ash residue indicated
that the sodium was found as a complex silicate. When the hot (800°C)
alumina filter was used, between 80 and 95% of the sodium vaporized from
the fuel was captured by the filter. It is believed that the silicious
binder of the filter reacted with the vaporized sodium compound (probably
NaCl) to form a silicate. In runs in which the hot filter was not used,
both NaCl and KC1 were detected by X-ray diffraction analyses of the residue
deposited on the cold trap, which suggests that sodium and potassium are
carried as their chlorides in the hot flue gas.
Task F. Flue-Gas Cleaning Studies
Evaluation of On-Line Light-Scattering Particle Analyzers. In a
pressurized FBC, a continuous on-line analyzer capable of instantly measuring
the efficiency of particulate-removal equipment would be useful since such
an analyzer could be incorporated into a flue-gas control and alarm system
that would prevent gas turbine damage from high particulate loadings. Two
on-line particle size analyzers, each using a laser light source, are being
evaluated—a Spectron Development Laboratory split laser beam particle morpho-
kinetometer and a Leeds and Northrup single laser beam particle analyzer.
The experiments with the Spectron Development Laboratory particle
morphokinetometer (PM) have been completed, and the instrument has been
returned to the manufacturer. Preliminary results are reported. Spectron's
PM analyzer measurements are compared with those obtained on steady-state
particle samples with an Anderson cascade impactor and with a Coulter counter.
The results obtained with the Anderson cascade impactor and Coulter counter
were in good agreement, whereas the results obtained with the Spectron PM
analyzer diverged from Anderson/Coulter results for the larger diameters.
-------
An empirical calibration curve has been developed for the Spectron PM
which is based on comparative measurements obtained with the Spectron PM
and those obtained by Coulter counter and cascade impactor analysis. The
empirical calibration improved the agreement of the different measuring
methods, especially in the 1.5-23 ym measuring range.
Delivery of the Leeds and Northrup (L&N) analyzer has been postponed.
because final tests by L&N revealed that the laser unit was not performing
satisfactorily. This analyzer is now being tested by Leeds and Northrup.
Particle Removal from Flue Gas. In pressurized fluidized-bed combustion,
the hot flue gas from the combustor must be expanded through a gas turbine.
To prevent erosion (and possibly corrosion) of the turbine blades by partic-
ulate matter entrained in the flue gas, the particulate loading must be
reduced to very low levels. A program has been initiated at ANL to test
and evaluate promising flue gas cleaning methods for application in the
off-gas system of the ANL, 6-in.-dia fluidized-bed combustor. The techniques
which have been identified for investigation are acoustic conditioning of
the flue gas, granular-bed filters using the limestone or dolomite sorbent
as the granular bed material, and a Donaldson TAN-JET cyclone.
A small granular bed filter has been fabricated to study the use of
either fresh limestone or sulfated sorbent as the filter medium. Using these
materials, the filter cake would not have to be removed by blowback. Instead
the entire bed would be replaced when it has become loaded. With sulfated
sorbent, the discharged bed including the cake would be discarded. With
fresh stone, the stone plus cake would be fed to the combustor bed where
agglomeration of the particles in the cake could occur. Or the cake could
be separated from the stone prior to adding the stone to the combustor.
Initially, testing of the granular bed filter was performed at ambient
conditions to determine the pressure drop characteristics of the sorbent
and the possible contribution of dust from the sorbent to the effluent gas
from the filter. The pressure drop data was correlated with the Ergun
equation for fluid flow through a packed bed, and the values for e, bed
void volume fraction, and s, particle sphericity factor, were determined.
When a clean gas was passed through the granular-bed filter, dust loadings
exiting from the filter ranged from 0.0016 to <0.0001 grain/scf (0.0037 to
<0.0002 g/m3).
Acoustic conditioning of the flue gas is a technique to enhance the
natural tendency of polydispersed particulates to impact upon each other.
Thereby, the mean size of the particulate matter is increased (and the
number of particles is decreased) . The process is designed to increase
the collection efficiency of downstream dust collectors.
A suggested procedure has been prepared by a consultant for developing a
resonant manifold system for the evaluation of acoustic conditioning in
the FBC system at ANL. The principal components of the proposed system are
(1) a pulse-jet sound generator, (2) a resonant manifold for "splitting"
the resultant acoustic power, and (3) an acoustic treatment section where
agglomeration of the aerosol occurs.
-------
10.
A high-efficiency Donaldson TAN-JET cyclone would provide a means of
evaluating how effective upstream acoustic agglomeration is in increasing the
collection of dust in high efficiency collection devices. It is planned,
therefore, to proceed with the design, procurement, and installation of a
Donaldson TAN-JET cyclone as a part of the flue-gas cleaning studies.
-------
11
TASK A. REGENERATION PROCESS DEVELOPMENT
In support of ERDA's national program for the development and implemen-
tation of FBC technology, the feasibility of regenerating and recycling the
S02-accepting sorbents from the FBC boilers is being investigated.
Natural calcium-based stones such as limestones and dolomites are
receiving primary consideration as sulfur-accepting sorbents (sulfur reacts
with calcium to form CaSO^) in FBC boilers. The primary reasons are their
acceptable reactivity and low costs and the bountiful supply throughout the
United States. Approximately one tonne of natural stone will be sulfated
for every four tonnes of coal (^3 wt % S) combusted. In a 1000-MW electric
power plant (70% capacity factor) , ^2000 tonnes of stone per day will be
sulfated. Using the stone only once would generate large amounts of sulfated
stone for disposal. Regeneration of the CaO in the stone and recycling is a
potential solution to the waste disposal problem.
A conceptual power-generating facility utilizing fluidized bed coal
combustion and sorbent regeneration systems is schematically represented in
Fig. 1. Steam is removed frora the boiler and is expanded in turbines to
generate power. The flue gas from the boiler meets EPA chemical emission re-
quirements. The partially sulfated sorbent from the boiler is transferred to
the regenerator, where it is regenerated, and then it is combined with fresh
sorbent (to compensate for losses due to attrition and changes in reactivity)
and is recycled to the boiler. The S02-rich off-gas from the regenerator is
treated in a sulfur recovery plant.
For sulfur recovery, a process using coal as reductant, such as the
Foster Wheeler RESOX process* is recommended. Since the volume of the off-
gas from the regenerator and sulfur recovery system is much less than that
from the boiler, it is proposed that the off-gas from the sulfur recovery
process be recycled to the boiler. In this manner, the gas emission stream
from the regenerator which could contain some trace elements due to volatil-
ization from the sorbent during regeneration (not yet investigated) will be
eliminated, and the tail gas from the sulfur recovery system will not have
to be processed.
In the sorbent regeneration process being investigated, CaSOi, is reduc-
tively decomposed in a fluidized bed at temperatures of ^1100°C. The heat
and the reductants required are produced by incomplete combustion of coal
in the fluidized bed of sulfated stone. Two solid-gas reactions by which
regeneration occurs are:
+ CO -> CaO + C02 + S02 (1)
+ H2 -»• CaO + H20 + S02 (2)
At lower temperatures and under more highly reducing conditions, the formation
of CaS is favored:
CaSOl| + 4CO -»• CaS + 4C02 (3)
+ 4H2 -»• CaS + 4H20 (4)
-------
12
ATMOSPHERE
FLUIDIZED BED
BOILER
S02-RICH
ELEMENTAL
SULFUR
FLUIDIZED BED
REGENERATOR
AIR
Fig. 1. Conceptual Fluidized-Bed, Coal Combustion, Power-
Generating Facility Having Sorbent-Regeneration
and Sulfur-Recovery Capabilities
An oxidizing zone which forms at the bottom of the fluidized bed, where the
fluidizing gas (0£ and N2) is introduced, minimizes the buildup of CaS.
The feasibility of a sorbent regeneration process will depend on (1)
the ability to regenerate the stones and to generate an S02-rich off-gas
which can be treated in a sulfur recovery process, (2) the reactivity of the
regenerated sorbent during subsequent sulfation (coal combustion) cycles,
and (3) the availability of a sorbent that will not decrepitate above
acceptable levels.
_In this section of this report, (1) regeneration rate results (process
development unit-scale) for Tymochtee dolomite and Greer limestone are pre-
sented in which the effects of key variables on regeneration of the sorbents
were evaluated; (2) the effects of process operating variables on the tendency
of the fluid bed in the regenerator reactor to defluidize and agglomerate have
also been evaluated; (3) for Tymochtee dolomite and for Greer limestone, the
effects of repeated utilization cycles on the reactivity and the resistance
to decrepitation were evaluated in ten combustion (sulfation)/regeneration
cycles which were performed without fresh sorbent makeup; and (4) some process
flowsheet calculations for a FBC process with sorbent regeneration are
presented and discussed.
-------
13
1. Experimental
a. Materials—Sorbents
Tymochtee dolomite which had been sulfated during coal combustion
experiments (using Arkwright coal) at ^900°C and 810 kPa was regenerated.
Also, some Greer limestone which had been sulfated by ANL and some by Pope,
Evans and Robbins (PER) at 340-900°C and atmospheric pressure during
combustion of Sewickley coal was regenerated.
The sulfated Tymochtee dolomite contained ^9 wt % S as CaS04 (no
MgSC>4 present), 26 wt % Ca, 9.5 wt % C02, and had a nominal size distribution
of -14 +30 mesh before regeneration. In its virgin state, its main constit-
uents were CaCOs (50 wt %) and MgCOs (39 wt %). The virgin Tymochtee
dolomite had been obtained from C. E. Duff and Sons, Huntsville, Ohio.
The virgin Greer limestone used for this study contained 41.2 wt %
CaO, 32 wt % C02, and 4.3 wt % Si, and its size was (nominally) -14 +30 mesh.
When sulfated, it contained 5-9 wt % S in the cyclic experiment and ^8 wt %
in the remaining experiments.
b. Materials—Coals^
The coal used in the sulfation of Tymochtee dolomite was a Pittsburgh
seam coal obtained from the Consolidation Coal Company Arkwright mine. As
received, the coal contained ^2.8 wt % sulfur, ^7.7 wt % ash, and ^2.9 wt %
moisture and had a heating value of 7,610 kcal/kg and an average particle
size of 320 urn. Triangle coal was combusted under reducing conditions during
the regeneration of Tymochtee dolomite. It is a bituminous high-volatile
coal (32.6% volatile matter, dry) and has a high ash fusion temperature
(1390°C, initial deformation under reducing conditions). As received, it
contains 73.5 wt % C, 9.4 wt % ash, and 0.98 wt % S.
The coal used in the sulfation of Greer limestone was a bituminous
coal, Sewickley. As received, it contains 4.3 wt % S, 12.7 wt % ash, and
1.1 wt % moisture and has a heating value of 7,220 kcal/kg; sizes were (nomi-
nally) -6 +14 mesh during ANL sulfation and -1/4 in. during PER sulfation.
This same coal was combusted under reducing conditions in the regeneration
step of each cycle. It does not have a high ash fusion temperature (initial
deformation is at ^1120°C under reducing conditions) .
Additional data on the above sorbents and coals are presented in
Appendix A.
c. PDU—Combustion System and Procedure
The experimental equipment and instrumentation of the PDU (process
development unit) at Argonne consist of a 6-in.-dia, fluidized-bed combustor
that can be operated at pressures up to 1014 kPa, a compressor to provide
fluidizing-combustion air, a preheater for the fluidizing-combustion air,
peripheral-sealed rotary feeders for metering solids into an air stream fed
into the combustor, two cyclone separators and two filters in series for
-------
14
solids removal from the flue gas, associated heating and cooling arrangements
and controls, and temperature- and pressure-sensing and display devices. A
simplified schematic flowsheet of the combustion equipment is presented in
Fig. 2.
Details of the PDU combustor are presented in Fig. 3. The reactor vessel
consists of a 15-cm-dia, Schedule 40 pipe (Type 316 SS) , approximately 3.4 m
(11 ft) long. The reactor is centrally contained inside a 2.7-m (9-ft)
section of 12-in.-dia, Schedule 10 pipe (Type 304 SS). A bubble-cap type
gas distributor is flanged to the bottom of the inner vessel. Fluidizing
air inlets, thermocouples for monitoring bed temperatures, solids feed line,
and solids removal lines are accommodated by the bubble cap gas distributor.
The coal and sorbent are fed in a common line which extends 2 in. above the
top surface of the gas distributor plate and is angled 20° from the vertical.
A constant bed height is maintained in the combustor by use of a 36-in.-high
standpipe. The 6-in.-dia pipe is alternately wrapped with resistance-type
heating elements and cooling coils onto which a layer of heat-conducting
copper and then an overlay of oxidation-resistant stainless steel have been
applied. Additional cooling capacity is provided by three internal, hairpin-
shaped coils that extend down from the flanged top of the combustor to within
12 in. of the top surface of the gas distributor. The coolant is water
entrained in air.
TO GAS ANALYSIS
SYSTEM
TEST FILTER
STAINLESS STEEL
PRESSURE
CONTROL
VALVE
VENTILATION
EXHAUST
PREMEATER
Fig. 2. Simplified Equipment Flowsheet of PDU Fluidized-
Bed Combustor and Associated Equipment, The
"additive feeder" is actually a "sorbent feeder."
-------
15
60 AND 89 In.FREEBOARD
THERMOCOUPLES
INTERNAL COOLING
COIL LEADS
PURGE GAS OUTLET
HEATER CONTROL
THERMOCOUPLES
SHELL PURGE
GAS INLET
EXTERNAL COOLING
COIL LEADS -__
36 OR 48 in.
SOLIDS OVERFLOW
6, 12,AND 44 in.
BED THERMOCOUPLE WELL
RUPTURE DISK
FLUE GAS TO
CYCLONE AND FILTERS
EXPANSION BELLOWS
RUPTURE DISK
12-in.JACKET
EXTERNAL COOLING COIL
(WRAPPED ON 6-in:DIA WALL)
CALROD TYPE HEATER
INTERNAL COOLING COILS
BUBBLE CAP DISTRIBUTOR
ELECTRICAL HEATER
LEADS
PLIBRICO FILLED VOLUME
-~^^=- SOLIDS FEED LINES
FLUIDIZING AIR
Fig. 3. Six-Inch-Diameter Pressurized, Fluidized-
Bed Combustor
Fluidizing-combustion air is supplied by a 75-hp, screw-type
compressor capable of delivering 100 cfm at 150 psig. The air can be heated
to approximately 540°C in a 6-in.-dia, 10-ft-tall preheater containing eight
2700W, clamshell-type heaters.
Coal and either dolomite or limestone are pneumatically fed from
hoppers to the combustor, using two 6-in.-dia rotary valve feeders. The
feeders and hoppers are mounted on platform-type scales.
-------
16
The flue gas (off-gas) is sampled continuously and is analyzed for
the components of primary importance. Nitrogen oxide and total NOX are
analyzed using a chemiluminescent analyzer; sulfur dioxide, methane, carbon
monoxide, and carbon dioxide determinations are made using infrared analyzers;
oxygen is monitored using a paramagnetic analyzer; and total hydrocarbons are
analyzed by flame ionization. Prior to and during each experiment, the res-
ponse of each analytical instrument is checked, using standard gas mixtures
of flue-gas constituents in nitrogen. Batch samples of flue gas can be taken
and analyzed for constituents of secondary importance.
The combustion system is equipped with a Hewlett-Packard 2010C data
acquisition system to monitor and record the temperature, pressure, gas flow,
and flue-gas concentration for subsequent data handling and analysis.
Although the experimental procedure was subject to minor variations,
it was basically as follows: a preweighed amount (^15 kg) of (1) partially
sulfated sorbent from a previous experiment, (2) fresh unsulfated sorbent,
or (3) regenerated sorbent was charged to the reactor to provide an initial
bed of material. The starting bed temperature was then raised to about
430°C by passing fluidizing air (preheated to 430-480°C) through the combustor
and simultaneously employing the resistance heaters on the combustor wall.
Once the bed temperature reached 430°C, the system was brought to the desired
operating pressure, and coal (entrained in a transport air stream) was in-
jected into the bed. To prevent carbon accumulation in the fluidized bed
during startup, coal was initially injected in small amounts intermittently
until a rapidly increasing temperature and a changing flue-gas composition
confirmed ignition and sustained combustion. Continuous injection of coal
was then initiated, and the bed temperature was raised to a selected combustion
temperature. The desired temperature was maintained by the use of external
and internal cooling coils.
Injection of the sulfur-accepting sorbent (virgin or regenerated)
was begun when the bed reached operating temperatures. The air, coal, and
sorbent feed rates were adjusted to give a specified mole ratio of calcium
in the sorbent to sulfur in the coal, a specified superficial gas velocity,
and a specified level of oxygen in the flue gas leaving the combustor.
Sulfated sorbent was removed from the combustor by means of a standpipe to
maintain a constant fluidized-bed level.
d. PDU—Regeneration System and Procedure
Figure 4 is a schematic diagram of the regeneration system used in
this work. The reactor ID is 10.8 cm (4.25 in.), and the height of the
fluidized bed (^46 cm) is regulated by an overflow pipe that is external to
the fluidized bed. The pressurized, fluidized-bed reactor is lined with a
4.8-cm-thick castable refractory. The coal and the sulfated sorbent are
metered separately (for independent control) to a common pneumatic transport
line which discharges into the fluidized bed above the gas distributor.
Other components of the experimental system are an electrically
heated pipe heat exchanger for preheating some of the fluidizing gas and
for preheating air (used in startup only) to 'v400°C and a solids-cleanup
system for the off-gas. Continuous analyses of pertinent constituents (S02,
-------
17
TO GAS
ANALYZERS
FILTER
SAMPLE
GAS CONDITIONER
COAL
HOPPER"
SULFATED- SORBENT
HOPPER
FILTER
PRESSURE
CONTROL
VALVE
E»HAUST
REGENERATOR
(10.8-cmlO)
Fig. 4. Experimental Sorbent Regeneration System
02, CO, H2 , CHi^, and NO) in the off-gas was performed. Solids transport air
constituted ^40% of the total fluidizing gas in the reactor. The remaining
fluidizing gas was a mixture of pure nitrogen and oxygen. Oxygen and nitrogen
were metered separately and mixed to produce the required oxygen environment
in the reactor. Thus the oxygen requirement at different experimental condi-
tions could be satisfied without changing the fluidizing gas velocity. Oxygen
concentrations in excess of that in air were used in the feed gas for most
of the reported regeneration experiments. Large amounts of heat (per unit
capacity) were required to compensate for (1) the heat losses in the relatively
small experimental system and (2) the heat load imposed by feeding cold
sulfated sorbent to the system. In a large-scale industrial regeneration
system, such heat requirements would be absent and oxygen enrichment of the
fluidizing air would not be needed.
2. PDU Regeneration Rate Experiments with Tymochtee Dolomite
[J. Montagna (Principal Investigator), G. Smith, C. Schoffstoll,
R. Mowry, and J. Stockbar]
A requirement of a sorbent regeneration process is that CaO (S02 acceptor)
be regenerated sufficiently while an S02~rich off-gas is generated which can
be treated in a sulfur recovery process. The dependence of (1) regeneration
of CaO and (2) S02 concentration in the dry off-gas on key variables (such as
regeneration temperature, solids residence time in the reactor, and system
pressure) has been determined for the regeneration of once-sulfated Tymochtee
dolomite sorbent. This data is being used in a model for the reductive decom-
position regeneration process to find the optimum process conditions from
technical and economic points of view.
-------
18
a. Effect of Solids Residence Time and Temperature on Extent of CaO
Regeneration
The experimental conditions and results for thirteen experiments
are given in Table 1. Data on six of these experiments were reported earlier
in ANL/ES-CEN-1016. The experiments were performed at three temperatures:
1000°C, 1050°C, and 1100°C. Solids residence times ranged from ~7 to ^35 min.
Regeneration of CaO was calculated from chemical analyses of the
steady-state products. It was based on the sulfur to calcium ratios in (1)
the sulfated dolomite feed and (2) the steady-state product after regener-
ation. These calculated regeneration values are compared in Table 1 with
values based on off-gas analyses. The latter values are the ratios of the
total sulfur released into the off-gas stream to the total sulfur contained
in the sulfated dolomite feed. The off-gas flow rate was calculated using
the feed gas rate and the model of the regeneration process (discussed in a
later section) in which the gas volumetric expansion in the reactor is pre-
dicted. The CaO regeneration values obtained by chemical analyses of the
regenerated products generally agree within analytical accuracy with the
values calculated from off-gas analyses. The extents of regeneration (based
on off-gas and regenerated solids analyses) are plotted in Fig. 5 as a
function of solids residence time for the three temperature levels. The
solids residence time was varied by changing the solids feed rate and not
by changing the reactor volume.
At 1000°C, as the solids residence time was varied from 37 min to
11 min, the extent of regeneration decreased drastically from 77% to 11%
(based on solids analyses). The S02 concentration in the dry off-gas
decreased from 2.5% to 1.4%. At this relatively low temperature, the rate
of regeneration of CaO is low, and therefore long solids residence times
are required to obtain acceptable (^50%) regeneration levels.
At 1050°C, decreasing the solids residence time from 34 min to 12
min decreased the extent of CaO regeneration from 90% to 56%. The S02 concen-
tration in the dry, off-gas increased from 3.0% to 4.8%. At 1100°C, the
highest temperature level investigated, decreasing the solids residence time
from 37 min to 7.0 min caused the extent of CaO regeneration to decrease
from 95% to 75% and the S02 concentration to increase from 2% to 8.7%. (The
S02 concentration for Exp CCS-1 for which the solids residence time was 7 min,
was adjusted from 6.5% to 8.7% to compensate for the dilution due to the
higher fluidizing-gas velocity.) Because the rate of regeneration of CaO
is high at 1100°C, the extent of regeneration remained relatively high when
the solids residence time was as low as 7 min.
An improved rate of CaSO^ decomposition at higher temperatures
(_>1100°C) for the reductive decomposition of gypsum has also been reported by
numerous workers, including Martin et al. who used carbon as the reductant
Wheelock et al.^ who used CO as the reductant. As expected, the oxygen
required in the feed gas increased with sulfated sorbent feed rate and with
total amount of reductive decomposition, which is represented by the S02
concentration in the off-gas.
-------
Table 1. Experimental Conditions and Results for the Regeneration of Sulfated Tymochtee
Dolomite by the Incomplete Combustion of Triangle Coal in a Fluidized Bed.
Nominal Fluidized-Bed Height: 46 cm Pressure: 153 kPa
Reactor ID: 10.8 cm
Coal: Triangle coal (0.98 wt % S) Ash fusion temp, under reducing
conditions, 1390°C (initial deformation)
Sorbent: (1) -14 +50 mesh, 9.0 wt % S (CS-6, -7, -8)
(2) -14 +50 mesh, 9.4 wt % S (CS-10, -11, -12)
8.5 wt % S (CC-13 through -18 and CCS-1)
Exp.
No.
CS-6
CS-8
CS-17
CS-7
CS-16
CS-10
CS-15
CS-14
CS-ll
CSS-13
CSS-18
CS-12
CCS-1
Bed
Temperature,
°C
1000
1000
1000
1050
1050
1050
1100
1100
1100
1100
1100
1100
1100
Fluidizing-
Gas Velocity,
m/s
0.98
0.92
1.04
0.92
1.0
0.98
1.02
1.05
1.07
1.09
1.07
1.16
1.43
62 Cone in
in Feed
Gas, %
18
22
24
21
26
29
22
28
. 33
33
33
36
27
Feed
Rate,
kg/hr
5.0
10.0
16.1
5.4
11.1
15.0
5.0
10.0
14.3
31.6
13.2
19.5
26.4
Solids
Residence
Time, min
37
18
11
34
16
12
37
18
13
13
14
9.4
7.0
Reducing Gas
Concentration
in Effluent, %
1.4
2.2
2.4
1.9
2.2
2.5
1.5
2.2
2.9
2.2
2.4
2.9
2.7
Measured SO
in Dry
Effluent
Gas, %
2.5
2.5
1.4
3.0
3.3
4.8
2
4.8
6.4
6.1
6.3
7.8
6.5C
2
CaO
Regeneration,
%a / %b
83/77
39/30
18/11
84/90
53/53
53/56
82/95
94/82
80/85
89/72
93/80
79/77
67/75
.Based on off-gas analysis.
Based on chemical analysis of dolomite samples.
The S02 concentration adjusted for a 1.08 m/s fluidizing gas velocity is ^8.7%.
-------
20
100
. 80
o
o
O
g 60
z
o
£ 40
UJ
z
UJ
uj 20
-------
21
The values calculated by the model equation, equation 1, compare
well with the experimental results. A correlation coefficient ranging from
0.96 to 0.99 was obtained for the experimental data and the results calculated
from the model equation. Equation 1 with the calculated coefficients was
found to be a good mathematical model of the dependence of the extent of CaO
regeneration on regeneration temperature and solids residence time in the
reactor for the investigated experimental range. This relationship for the
rate of CaO regeneration is being used in a mass and energy-constrained model
for the regeneration process to predict sorbent behavior within and outside
the investigated operating range.
The extent of CaO regeneration as a function of temperature and
sorbent residence time as calculated by equation 1 has been plotted in Fig. 6.
The plot has been extrapolated to 1200°C, which is beyond the experimentally
investigated temperature range of 1000-1100°C. On the basis of these pre-
dictions, it is expected that the extent of CaO regeneration would increase
considerably (by ^1/3) if the temperature were increased from 1100°C to 1150°C
at a solids residence time of 5 to 7.5 min. At yet higher temperatures, the
extent of regeneration increases more slowly.
Fig. 6. The Extent of CaO Regeneration for Tymochtee Dolomite
as a Function of Temperature and Residence Time as
Represented by the Model Equation, Equation 1
-------
c.
22
Effect of Solids Residence Time and Temperature on S02 Concentration
in the Off-Gas
A least-squares best-fit relationship was obtained from the experi-
mental regeneration data for the functional dependence of CaO regeneration
on solids residence time and regeneration temperature. It was used in a
mass and energy constrained model for the regeneration process (equation and
model are described in a following section) . The S02 concentration in the
dry off-gas was predicted with the process model for solids residence times
ranging from 40 min to ^2.5 rain. In the model, gas volumetric changes due
to combustion and decomposition reactions (which affect the extent of dilution)
are included in the off-gas composition predictions. The predictions were
made for the three investigated temperature levels, a pressure of 153 kPa,
and a fluidizing-gas velocity of 1.07 m/s. A sulfur content of 9.5 wt %
was assumed for dolomite. The predicted S02 concentrations in the dry off-
gas (the curves) are plotted in Fig. 7, together with the experimentally
obtained concentrations.
O A D EXPERIMENTAL
PREDICTED S02 CONCENTRATION
ADJUSTED S02 CONCENTRATION
10 15 20 25 30
SOLIDS RESIDENCE TIME, min
Fig. 7. Predicted and Experimental S02 Concentration as
a Function of Solids Residence Time at Three
Regeneration Temperatures (pressure: 153 kPa)
-------
23
At 1000°C, the S02 concentration in the dry off-gas was predicted
to continually decrease as the solids residence time decreased, in agreement
with the experimental data. At this temperature, the regeneration rate was
low. At 1050°C and higher temperatures, the maximum SC>2 concentrations were
predicted at the lowest permissible solids residence time of ^2.5 min, at
which point oxygen that would be required would be greater than the total
fluidizing gas.
At 1100°C much higher SC>2 concentrations were obtained. Based on
the experimental data obtained for the extent of regeneration, an S02 con-
centration in excess of 10% is predicted for a 5-min solids residence time
(realistic). The experimental and predicted SC>2 concentrations in the
dry off-gas for all three temperature levels were in good agreement. These
experimental results show that for temperatures ^1050°C, as the solids
residence time in the reactor decreased (i.e., at higher rates of sulfated
sorbent throughput) extent of regeneration is sacrificed but S02 concentration
in the dry off-gas is increased. The effect of introducing room temperature
sorbent in the reactor for very short solids residence times, <7 min, has
not yet been evaluated.
Using equation 1 for the functional dependence of extent of CaO
regeneration on temperature and solids residence time in the above-mentioned
model for the regeneration process, the S02 concentrations in the dry off-
gas at three different system pressures are predicted in Fig. 8 for the
experimental conditions given. These predictions, also, were extrapolated
beyond the experimentally investigated temperature range of 1000-1100°C. The
effect of pressure on SC>2 concentration is very great, as discussed below.
It is predicted that with a system pressure of 1000 kPa (10 atm), S02 concen-
trations no greater than 4% can be obtained, even with regeneration temper-
atures as high as 1200°C and solids residence times as low as 5 min. At a
pressure of 100 kPa (1 atm), S02 concentrations as high as 20% are predicted
at regeneration temperatures up to 1200°C. However, based on the experience
to date with the regeneration process, it is expected that regeneration
temperatures in excess of 1100°C will not be feasible because of the increased
tendency of the sulfated sorbent (a mixture of sorbent and residual coal ash)
to agglomerate at high temperatures.
High temperatures were found to increase both the extent of CaO
regeneration and the 862 concentration in the off-gas. This effect of
temperature agrees with previous results presented by Montagna et al.^ in
which methane was used as the fuel for the regeneration of dolomite by
reductive decomposition. Higher SC>2 concentrations in the off-gas at higher
regeneration temperatures have also been reported by Hoke et at. ^ for the
reductive decomposition of pure CaSOi, in a fluidized-bed batch reactor.
Although at higher temperatures (above 1100°C) , the regenerability of the
sorbent would be further improved, the probability of agglomerating the
sorbent and residual coal ash would be increased, as reported by Skopp et al.^
Based on the above results with Tymochtee dolomite, an industrial regeneration
process should be operated at 1100°C and a solids residence time of ^5-8 min.
-------
o
Fig. 8. Predicted S02 Concentration in the Dry Off-Gas
as a Function of Solids Residence Time, Regen-
eration Temperature, and System Pressure
(Experimental Conditions: Solid Feed Temp, °C,
815; Fluidizing-Gas Velocity, m/s, 1.07;
Fluidizing-Gas Feed Temp, °C, 650; Sulfur
Concentration in Sorbent, %, 9.5)
-------
25
d. Effect of System Pressure on Extent of CaO Regeneration and Off-
Gas S(>2 Concentration
Thermodynamically, from an equilibrium standpoint, when the system
pressure is lowered, the SC>2 concentration in the off-gas will increase. In
the fluidized-bed process, the SC>2 concentration in the off-gas is determined
by the rate of total gas flow through the reactor and the rate of CaO regen-
eration in the reactor. Because of high air (fluidizing gas and/or combustion
oxygen) and energy (sensible heats of gas and solids, and the heats for the
decomposition reactions) requirements of the process, equilibrium S0£ concen-
trations in the off-gas cannot be achieved practically. At lower pressures,
less fluidizing gas is required and thus the extent of dilution is reduced.
The effect of system pressure on the extent of regeneration of CaO
and S02 concentration in the dry off-gas has been evaluated in six experi-
ments in which sulfated Tymochtee dolomite from the second and sixth regen-
eration cycles of a ten-cycle experiment (discussed in a following section)
was regenerated. The experimental conditions and results are given in Table
2. In experiments CCS-2A, -2B, and -2C, the pressure was ^115 kPa or ^75%
of that in CCS-2. In the experiments at the lower pressure, decreasing the
solids residence time from 8.8 min to 5.3 min caused the extent of CaO regen-
eration to decrease from 80% to 60% (based on solids analysis). These
results on extent of regeneration are in agreement with those obtained at
the higher pressure of 153 kPa (see Fig. 5) and equivalent solids residence
times. Specifically, results for CCS-2B and CCS-2 agree well.
At the lower pressure, the measured S02 concentration in the dry
off-gas increased from 8.9% to 9.6% as the solids residence time was decreased
from 8.8 min to 5.3 min. These concentrations are higher than those obtained
in the higher pressure experiments (Table 1 and Exp. CCS-2). A larger differ-
ence, and a more accurate comparison of off-gas S02 concentrations from the
lower and the higher pressure experiments, would be obtained by adjusting
the concentrations for a common fluidizing-gas velocity of 1.07 m/s (velocity
used for predictions in Fig. 8). The adjusted S02 concentrations could then
be compared with the results in Fig. 8.
During the regeneration step of the sixth utilization cycle of the
cyclic experiments (reported in a following section of this report) , an addi-
tional experiment (CCS-6A) was performed at a reduced pressure, 125 kPa
instead of 153 kPa (CCS-6) . The experimental conditions and results for both
of these experiments are also presented in Table 2.- As a result of lowering
the system pressure by ^20% while maintaining the fluidizing-gas velocity
almost constant (^1.2 m/s), the S02 concentration in the dry off-gas increased
by ^20% from 8.7% to 10.4% (the highest S02 enrichment obtained to date). The
S02 concentration in the off-gas increased because of less dilution at the
reduced pressure.
Relatively high S02 concentrations (^8%) have also been reported by
Gordon et al.7 for the regeneration of limestone at atmospheric pressure.
Although the number of experiments in this evaluation was limited and the
pressure variation was small, the system pressure was found to have a negligible
effect on the rate of CaO regeneration because the S02 concentration in the
-------
Table 2. Effect of Regeneration Pressure on the Regeneration of Tymochtee
Dolomite and the S02 Concentration in the Off-Gas
Nominal fluidized-bed height: 46 cm
Reactor ID: 10.8 cm
Temperature: 1100°C
Reducing gas concentration in off-gas: 3.0-3.2%
Coal: Triangle coal (0.98 wt % S)
Sorbent: -14 +30 mesh, sulfated dolomite in the second
(10.7 wt % S) and sixth (9.3 wt % S) utilization cycles.
Fluidizing
Exp.
CCS-2A
CCS-2B
CCS-2C
CCS-2
CCS-6A
CCS-6
Pressure,
kPa
115
115
118
153
125
153
Gas Velocity,
m/
1.
1.
1.
1.
1.
1.
s
35
41
44
26
22
18
Solids
Residence
Time,
CaO
Regeneration,
m±n %
8
7
5
7
7
7
.8
.4
.3
.5
.5
.8
78
69
57
67
73
75
S02 Concentration
in Dr
%
8
9
9
8
10
8
y Off-Gas,
/ 7C
.9/11.2
.5/12.5
.6/12.9
.6/10.1
.4/11.9
.11 9.6
Minor Sulfur Compounds
in Off-Gas, %
H2S
0.03
0.04
-
0.02
0.03
0.03
COS
0.1
0.1
-
0.1
0.1
0.2
CS2
0.1
0.1
-
0.1
0.1
0.1
.Based on chemical analysis of dolomite samples.
^Measured S02 concentration.
"Adjusted S02 concentration based on a constant fluidizing-gas velocity of 1.07 m/s.
NJ
-------
27
off-gas was not near equilibrium. However, pressure has been found to affect
the S02 concentration in the off-gas via the extent of dilution.
e. Formation of CaS
The sulfated dolomite, before regeneration, contained negligible
amounts of CaS. The buildup of CaS during regeneration of Tymochtee dolomite
for all experiments, including those performed at 1000°C, was found to be
£0.1% wt % (S2~). The investigated ranges of temperature, solids residence
time, and system pressure had no significant effect on the buildup of CaS
because of the existence of an oxidizing zone at the bottom of the fluidized
bed, as discussed by Montagna et al.k The CaS formed in the reducing zone
is oxidized in the oxidizing zone to either CaO or CaS04.
CaS + 202 -* CaSOi, (5)
CaS + 3/2 02 •* CaO + S02 (6)
The beneficial effect of an oxidizing zone on minimizing the buildup of CaS
has also been observed by Hoke et al. 5 and Swift and Wheelock.8
3. PDU Regeneration Rate Experiments with Greer Limestone
[J. Montagna (Principal Investigator), F. F. Nunes, G. Smith,
R. Beaudry, and R. MowryJ
The dependence of (1) CaO regeneration and (2) S02 concentration in the
regenerator off-gas on key variables such as regeneration temperature and
solids residence time in the fluid bed reactor was studied for once-sulfated
Greer limestone to aid in optimizing the regeneration process conditions for
the limestone.
A series of regeneration experiments with Pope, Evans and Robbins (PER)
sulfated Greer limestone was performed earlier at regeneration temperatures
of 1050°C and 1100°C. Some of these results have been discussed and compared
to similar results for Tymochtee dolomite in ANL/ES-CEN-1016. The rest have
not been reported because of inconsistencies in the results which are believed
to be due to (1) inconsistency (limestone type) in the (PER) delivered sul-
fated limestone batches and (2) the uncertainty of the history (e.g., salt
added or not) during sulfation.
Consistent regeneration rate data has been obtained in six new regener-
ation experiments in which the same batch of Greer limestone (which contained
7.7 wt % sulfur) was regenerated. This limestone has been sulfated during
combustion of Sewickley coal in the ANL 6-in., PDU combustor (no salt added).
The experiments were performed at 1050°C and 1100°C; solids residence time
in the reactor ranged from 7 to 23 min.
a. Effects of Solids Residence Time and Temperature on Extent of CaO
Regeneration
The six new experiments were performed as a part of the regeneration
step of the first cycle in the cyclic experiment with Greer limestone (reported
in a following section) to determine the effects of solids residence time and
temperature on the extent of CaO regeneration. The experimental conditions
and results for these experiments are given in Table 3.
-------
Table 3. Experimental Conditions and Results for the Regeneration of Greer Limestone
by the Incomplete Combustion of Sewickley Coal in a Fluidized Bed
Nominal fluidized-bed height: ^46 cm
Reactor ID: 10.8 cm
Pressure: 129 kPa
Coal: Sewickley (4.3 wt % S); ash fusion temperature (initial deformation)
under reducing conditions: 1119°C
Sorbent: -14 +30 mesh sulfated limestone (7.7 wt % S)
Exp.
No.
RGL-1A
RGL-1B
RGL-1C
RGL-1D
RGL-1E
RGL-1F
Bed
Temperature,
°C
1050
1050
1050
1100
1100
1100
Fluidizing-
Feed
Gas Velocity, Rate,
m/s
1.23
1.21
1.21
1.23
1.23
1.29
kg/hr
8.2
15.4
26.3
9.1
15.9
25.9
Solids
Residence
Time, min
22.54
11.93
6.99
20.28
11.59
7.12
Reducing Gas
Concentration
in Effluent, %
3
2
3
3
3
2
.2
.9
.4
.2
.3
.9
CaO
Regener . ,
%a / %b
72.2/83.0
43.1/53.3
28.3/27.3
79.5/92.0
72.3/82.8
66.9/70.9
Major Sulfur Compounds
in
S02
3.3
3.7
4.3
3.9
6.0
8.4
Dry Off-Gas, %
H2S
0.09
0.09
0.05
0.2
0.1
0.06
COS
0.07
0.08
0.1
0.1
0.1
0.09
CS2
0.06
0.05
0.05
0.05
0.03
0.02
Based on flue-gas analysis.
Based on chemical analyses of limestone samples.
.
oo
-------
29
The extent of regeneration values are plotted in Fig. 9 as a
function of solids residence time for two temperatures, 1050°C and 1100°C.
At 1050°C, when the solids residence time was decreased from 22.5 min to
7.0 min, the extent of regeneration decreased from 83% to 27% (based on
solids analyses). At 1100°C, the extent of regeneration decreased from 92
to 71% when the solids residence time was decreased from 20 min to 7.1 min.
The regeneration rate is higher at 1100°C, and therefore the conversion ratio
of CaSOij to CaO was less affected by a decrease in reactor particle residence
time. With a residence time of 7 min, the extent of regeneration was still
considerable, ^70%.
A "best fit" equation has been obtained by regression analysis
for the experimental extent of CaO regeneration as a function of regeneration
temperature and solids residence time. A similar analysis performed earlier
on regeneration rate data for Tymochtee dolomite was reported above. The
equation for the Greer limestone regeneration rate is
In (1 - R) = A-T + B-T:
(1)
where
A x
B x 10^ =
R =
T =
}2 _
3 _
T
t
extent of CaO regeneration (R = 1 for complete regeneration)
solids residence time (reactor particle contact time)
-12.AT - 3.98 (2)
3.25T - 1.24 (3)
(t - 1050)/50
= regeneration temperature, °C
The values calculated by the model equation (Eq. 1) compare favorably with
the experimental results. A correlation coefficient of ^0.97 was obtained
for the experimental data and the results predicted from the model equation.
Therefore, Equation 1 is a good mathematical model of the dependence of CaO
regeneration in Greer limestone on temperature and solids residence time for
the investigated experimental range. These best fit results for Greer lime-
stone are compared with results for similar experiments with Tymochtee dolomite
100
80
5 60
20
11000C,
Open Symbols: Solids Analysis
Closed Symbols: Off-Gas Analysis
O 1100° c
—V- 1050° C
Greer Lim.
~ Tymochtee Dol.
(ANL/ES-CEN-1017)
1 I J I ' i i
Fig. 9. Regeneration of CaO in
Greer Limestone as a
Function of Solids
Residence Time and
Regeneration Tempera-
ture
10
15
20
25
30
Solids Residence Time, min
-------
30
in Fig. 9. The regeneration rates for these two sorbents compare very
favorably. This relationship for the rate of CaO regeneration will be used
in the model for the regeneration process to optimize the design process
conditions and to scale up sorbent regeneration systems.
b. Effect of Solids Residence Time and Temperature on S02 Concentration
in the Off-Gas
In the fluidized-bed regeneration process, the S02 concentration
in the off-gas is determined by (1) the feed rate of CaSOi, to the regenerator
reactor (solids residence time and sulfur content of sulfated sorbent), (2)
the extent of regeneration of CaO, and (3) the gas flow rate through the
reactor.
In this series of regeneration experiments, a single batch of
sulfated limestone (containing 7.7 wt % sulfur) was used, and the fluidizing-
gas velocity was varied from 1.21 to 1.29 m/s (a small variation). The gas
flow rate through the reactor was not affected greatly by the fluidizing gas
velocity in the experiments. Therefore, the variation of S02 concentration
in the dry flue gas was due to the mass rate of CaO regeneration.
The experimentally obtained S02 concentrations in the dry off-gas
are plotted in Fig. 10. At 1050°C, the S02 concentration increased from
3.3% to 4.3% as the solids residence time decreased from 22.5 min to 7.0
min (i.e., as the sulfated-sorbent feed rate increased). At 1100°C, the S02
concentration in the dry off-gas increased from 3.9 to 8.4% as the solids
residence time decreased from 20.3 min to 7.1 min. At the longest solids
residence time (>20 min), the S02 concentration was found to be only slightly
higher at the higher temperature. With this long reaction time, most of the
CaO was regenerated at both temperature levels and hence the S02 concentration
in the off-gas was dependent on the CaSOi+ feed rate. For the shorter reaction
time (7 min), the S02 concentration was much higher at the higher temperature
(1100°C) due to the higher rate of CaO regeneration.
10
-O- HOO'C
1050°C
Fig. 10. Experimental S02 Concen-
tration for the Regeneration
of Greer Limestone as a
Function of Solids Residence
Time and Temperature.
Pressure: 129 kPa;
fluidizing-gas velocity:
1.21-1.29 m/s; limestone
sulfur content: 7.7 wt %
10 IS 20
Solids Residence Time, min
25
30
-------
31
c. Effect of System Pressure during Sulfation on the Reactivity of
Greer Limestone
The effect of system pressure on the reactivity of Greer limestone
has been evaluated using the ANL 6-in.-dia combustor. One experiment (SGL-1,
the first cycle in the cyclic experiments discussed in a following section)
was performed at a system pressure of 308 kPa and a bed temperature of 855°C,
and the other (Greer-2C) at 610 kPa and 855°C. The results are given in
Table 4.
In the 308-kPa experiment, at a Ca/S mole feed ratio of 2.9, a
sulfur retention of ^84% was obtained (656 ppm 862 in the dry flue gas).
This is slightly higher than the 78% retention obtained at the same CaO/S
mole feed ratio by Pope, Evans and Robbins9 at 100 kPa (1 atm) and a higher
fluidizing gas velocity (>3 m/s). The ratio of the C02 partial pressure
(^16% CC>2 in the dry flue gas) in the flue gas to the equilibrium C02 pressure
at 855°C was 0.78. The ratio of moles of C02 to moles of CaO in the sulfated
limestone product was 0.03, indicating that the limestone was fully calcined
during this experiment.
At the Ca/S mole ratio of 3.2 used in the 610 kPa (6 atm) experi-
ment, the sulfur retention was 85.5% (575 ppm of S02 in the dry flue gas).
The ratio of C02 partial pressure in the flue gas to the equilibrium C02
pressure at 855°C was 1.5. Since the ratio of moles of C02 per mole CaO in
the sulfated limestone product was 0.49, considerable calcination did occur.
Although this extent of calcination was unexpected, the results are not
surprising because the high C02 concentration in the flue gas is present
only near the top of the fluid bed. Near the bottom of the fluidized bed,
the C02 concentration in the gas bed is negligible and therefore calcination
is expected to occur there. The CaO/S ratios required to achieve a constant
sulfur retention in the two ANL experiments listed in Table 4 are very
similar, which suggests that there was no noticeable effect of pressure on
the reactivity of limestone. On the basis of findings reported by Westing-
house10 on the effect of the ratio of C02 partial pressure to equilibrium
C02 pressure during calcination at atmospheric pressure, it was expected
that the reactivity in the lower-pressure experiment (SLG-1) would be higher.
However, as mentioned above, the conditions are not constant throughout the
fluidized bed, and hence the degree of calcination cannot be controlled. To
enhance the reactivity of the limestones, calcination should be performed in
a separate reactor where the optimum calcination conditions can be maintained.
4. Characterization of Causes of Defluidization during Regeneration
[J. Montagna (Principal Investigator), F. F. Nunes, G. Smith,
R. Beaudry, and R. Mowry]
In the development of a sorbent regeneration process, an understanding
of the process by which a fluid bed agglomerates is necessary so that agglom-
eration may be controlled. During sulfation of sorbent (combustion step),
some coal ash is retained in the fluidized bed and is removed with the
sulfated sorbent. It is believed that in agglomeration in the fluid bed
during regeneration, coal ash is a contributor to initial coalescing of
particles, which is followed by loss of fluidity in the fluidized bed and by
-------
Table 4.
Operating Conditions and Flue-Gas Compositions during the Sulfation
of Greer Limestone with Sewickley Coal. Combustion at Two Operating
Pressures
Combustor: ANL, 6-in. dia
Coal: Sewickley, -8 mesh, 4.3 wt % S
Sorbent: Greer limestone, -14 +30 mesh
Temperature: 855°C
Excess air: ^17%
Nominal bed height: 0.9 m
Nominal gas velocity: 1.0 m/s
Combustion
Exp.
SGL-1
Greer-2C
System
Pressure ,
kPa
308
610
Feed Rate,
kg/hr
Coal Sorbent
6.36 3.38
12.68 6.42
CaO/S
Mole
Ratio
2.9
3.2
Flue-Gas
Analysis
(avg. values)
S02, 02,
ppm %
656 3.0
575 3.0
Sulfur
Retention,
%b / %c
83.6/84.8
85.5/87.3
Calcium
Utilization,
%
29.6
31.7
Moles of C02/
Mole of CaO
in spent
limestone
0.03
0.49
N>
EPA standard for Sewickley coal is ^635 ppm.
Based on flue-gas analysis.
"Based on solids analysis.
-------
33
increases in local temperatures. At these high local temperatures (>1100°C),
the bed agglomerates; reactions occur that form calcium silicates and calcium-
aluminum silicates.
A statistical experiment designed to study defluidization characteristics
(agglomeration) has been carried out in the PDU regeneration system. Sulfated
Greer limestone (from Pope, Evans and Robbins) and Sewickley coal were used
in the investigation. The effects of (1) regeneration temperature, (2) parti-
cle size distribution of the sulfated sorbent, and (3) total reducing gas
concentration (CO, H2, and CH4) in the off-gas on the defluidization velocity
(minimum velocity required to prevent agglomeration) were determined.
a. Experimental Design and Procedure
The statistical experiment consisted of a full 23 factorial design.
The three variables and their corresponding design levels were: regeneration
temperature (1050°C and 1100°C), feed sorbent particle size (-10 +30 mesh
and -14 +30 mesh), and total reducing gas concentration in the off-gas (2.5%
and 5.0%). The effects of these variables on the minimum gas velocity re-
quired to prevent agglomeration (defluidization velocity, V^) were studied.
This velocity (Vjj) is greater than the conventional minimum fluidization
velocity of a similar nonsticky particle system. The effect of stickiness
of bed material on the tendency of a fluidized bed to defluidize and agglom-
erate has been previously described by Gluckman et al.
Each defluidization experiment was begun by starting with a
fluidizing-gas velocity of ^2.0 m/s. The other design conditions remained
constant. The velocity was decreased in increments of <_0.15 m/s every 30
min until the bed defluidized. Defluidization and agglomeration were indi-
cated by a decrease in the pressure drop through the fluidized bed and the
formation of a vertical temperature gradient in the fluid bed, as illustrated
in Fig. 11. The regenerated sorbent samples taken just prior to agglomeration
in each experiment were saved and are being sized.
1200
ij 1100
1000
T3 15 cm above gas distributor plate
T^ 30 cm above gas distributor plate
2.0
\
0.4
0.2
Fig. 11. Qualitative Plot of Bed
Temperature and Pressure
Drop as a Function of
Fluidizing-Gas Velocity
for a Defluidization
Experiment. Bed height:
46 cm.
Fluldlzing-Cas Velocity m/s
-------
34
b. Effect of Operating Variables on Defluidization Velocity
The experimental conditions and results for this series of experi-
ments are presented in Table 5. The lowest defluidization velocity (V^ =
0.88 m/s) was obtained in AGL-5, using the smaller feed particles (-14 +30
mesh), the lower temperature (1050°C), and the lower reducing gas concen-
tration (2.5%) in the off-gas. The highest defluidization velocity (Vd =
1.56) was obtained in AGL-4 with the larger feed particles (-10 +30 mesh),
the higher temperature (1100°C), and the higher reducing gas concentration
(5.0%) in the off-gas. Higher temperature and higher reducing gas concen-
tration probably cause the sulfated particles (containing 5-10% coal ash)
to become sticky so that a higher fluidizing gas velocity is required to
overcome the adhesive forces between particles and prevent defluidization.
There are two possible reasons for particle size affecting the defluidization
velocity: The minimum fluidization velocity of large particles is higher,
and more kinetic energy must be imparted to larger particles to overcome
their adhesive forces.
Table 5. Defluidization during Regeneration—Experimental
Conditions and Results of the Full 2:
Experiment
Factorial
Nominal fluidized-bed height: ^46 cm
Sorbent residence time: ^25 min
Reactor ID: 10.8 cm
Pressure: 129 kPa (4 psig)
Coal: Sewickley (4.3 wt % S), nominal
(-12 +100 mesh); ash fusion temperature
(initial deformation) under reducing
conditions: 1119°C
Sorbent: Regenerated Greer limestone
Variables
Exp.
AGL-1
AGL-1-1R
AGL-1-2R
AGL-2
AGL-3
AGL-4
AGL-5
AGL-6
AGL-7
AGL-8
Temp,
°C
1050
1050
1050
1100
1050
1100
1050
1100
1050
1100
Mean Feed
Sorbent
Size,
ym
1271b
1271
1271
1271
1271
1271
969. 2C
969.2
969.2
969.2
Total Reducing3
Gas Cone . in
Off-Gas,
%
2.5
2.5
2.5
2.5
5.0
5.0
2.5
2.5
5.0
5.0
Response
Def luidization
Velocity,
m/s
1.19
1.02
0.99
1.24
1.15
1.56
0.88
0.96
1.01
1.51
Combined concentrations of CO, H2, and
Nominal -10 +30 mesh.
Nominal -14 +30 mesh.
-------
35
A requirement of a sorbent regeneration process is that an off-gas
with a high SC>2 concentration (^10% 862) must be obtained to reduce the eco-
nomic burden of a sulfur recovery system. The detrimental effect of a high
fluidizing-gas velocity on the extent of SC>2 dilution has been discussed,
along with previous ANL regeneration results.12
c. Analysis of Variance for Defluidization Velocity
An analysis of variance for the defluidization velocity is presented
in Table 6. The regeneration temperature and the total reducing gas concen-
tration were found to significantly affect the defluidization velocity (V^)
with the F-test for significance at the a = 0.1 level (90% confidence level).
The effect of the feed particle size on the defluidization velocity was
found not to be significant at the a = 0.1 level; however, it was significant
at a = 0.15.
•
The validity of this analysis depends upon the absence of inter-
actions between the controlled variables. For a 23 factorial experimental
design, the model equation is expressed by:
X. .. =y+a. + g. +Y, + I. . + I., + I.. + I. ., + e. .,
ijk i j 'k ij ik jk ijk ijk
where X^-j^ is the observed response; p is the mean of all possible responses;
a-^, $j, and Y^ are the treatment effects for the three controlled variables;
Iij> I^k, I-jk> and I^jk are tne interaction effects between the controlled
variables; and e^ik is the error between the observed and the expected
response.
The analysis of variance for a factorial experiment design with
partial replication cannot estimate the separate effects of the interaction
terms lij, Iik> Ijk» anc* -"-iik* These interaction effects are included in
the term Eijk.- ^n addition to inflating the error mean square, the presence
of interactions can give misleading results in F-tests for significance.
Comparing the error mean square (0.01576) with the variance (0.004634) cal-
culated from replicate experiments AGL-1, AGL-1-1R, and AGL-1-2R gives an
indication of the extent to which interaction effects have inflated the error
mean square. Since both of these values are estimators for a2 (the true
variance of the response data), the difference tends to indicate the presence
of interactions between the controlled variables. Thus, the lack of signifi-
cance (at a = 0.1) for feed particle size on the defluidization velocity
(indicated above) is possibly in error. Experimental results (AGL-1 vs
-5, AGL-2 vs -6, etc.) do suggest an effect of feed particle size on the
defluidization velocity.
d. Regression Analysis for Defluidization Velocity
A best fit relation for defluidization velocity (Vj) as a function
of regeneration temperature, mean feed sorbent particle size, and total
reducing gas concentration in the effluent from the regeneration fluid-bed
reactor was obtained by using least-squares techniques
V = 4.05 - 3.61 x 10~3T - 2.62R + 2.54 x 10~3TR + 5.26 x lO^F (4)
-------
36
Table 6. Analysis of Variance for Defluidization Velocity
Data from the AGL Series of Experiments
Source of
Variation
Temperature
Feed Particle
Size
Reducing Gas
Error
Corrected Total
Degrees of
Freedom
1
1
1
_4
7
Sum of
Squares
0.17701
0.04961
0.15401
0.06305
0.44368
Mean
Square
0.17701
0.04961
0.15401
0.01576
Calculated ,
F a F-,
e 1— a
11.23 4.54
3.15 4.54C
9.77 4.54
a
.The term Fe = ratio of variable mean square to error mean square
cFor a = 0.1 (90% confidence level).
For a = 0.15, the value of F. is 3.15.
1-ct
where VV(j) at which
industrial (or Rivesville) sorbent regeneration process reactors can be
operated without defluidization and agglomeration of the fluid bed can be
predicted. Predicted defluidization velocities for a reductive decomposition
regeneration process are given in Table 7. (The predictions for a 1500-um
mean feed sorbent particle size are applicable to a regeneration process at
Rivesville.) The effect of predicted minimum operable fluidizing-gas velocity
on the maximum expected concentration of 862 in a regeneration process off-
gas and its effect on the economic burden of a sulfur recovery system will
also be evaluated and reported.
Additional agglomeration experiments will be performed to evaluate
the effects on defluidization velocity of (1) using different coals (including
one with a high ash fusion temperature), (2) using a higher regeneration
temperature (1150°C) , and (3) extensive buildup of an ash layer on the sorbent
particles (dolomite from the tenth utilization cycle).
-------
37
Table 7. Calculated Defluidization Velocities (Vd, m/s)
at Different Bed Temperatures, Reducing Gas
Concentrations, and Bed Particle Sizes for the
Regeneration Process
V,, Defluidization Velocity, m/s
Reducing Gas
Concentration, % 1050°C 1075°C 1100°C
Mass Mean Particle Diameter, 1500 urn
1.18 1.22
1.29 1.39
1.40 1.56
1.51 1.74
Mass Mean Particle Diameter, 1000 pm
2
3
4
5
1.14
1.19
1.24
1.28
2
3
4
5
0.88
0.93
0.97
1.02
0.92
1.03
1.14
1.25
0.95
1.13
1.30
1.47
o
Approximate mean particle size of feed limestone that will
be used at the Rivesville pilot plant.
5. Cyclic Sorbent Life Studies with Tymochtee Dolomite
[J. Montagna and W. Swift (Principal Investigators), F. F. Nunes,
G. Smith, G. Teats, H. Lautermilch, R. Mowry, S. Smith, C. Schoffstoll,
and J. Stockbar]
The feasibility of sorbent regeneration technology will depend on the
ability to recycle the sorbent a sufficient number of times (1) without loss
of its reactivity for either sulfation or regeneration and (2) without severe
decrepitation. Unless both of these requirements are met, the sorbent makeup
rate will be so high that in comparison with the fresh sorbent requirements
and spent sorbent waste disposal for a once-through FBC process, regeneration
may not be economically justified.
Two ten-cycle sorbent utilization experiments were performed, therefore,
with no fresh sorbent makeup to evaluate: the changes in reactivity (sulfur
acceptance during combustion), the changes in regenerability (sulfur release
during regeneration), the extent of decrepitation, and the extent of coal
ash buildup as a function of utilization cycle. This section presents the
results of the ten-cycle sorbent-life study performed with Arkwright coal
and Tymochtee dolomite. (The results of the ten-cycle sorbent life study
performed with Sewickley coal and Greer limestone at 308 kPa pressure during
the combustion steps are presented in the following section.)
Since the processing capacity of the ANL PDU regenerator is much greater
than that of the combustor (by almost a factor of ten), the sorbent could
not be continuously recycled between the reactors. Thus, the sulfation and
regeneration experiments were performed batchwise.
-------
38
a. Experimental Conditions
The combustion experiments in each utilization cycle were performed
at a 900°C bed temperature, 810 kPa pressure, 1.5 CaO/S mole ratio (ratio
of unsulfated calcium in sorbent to sulfur in coal), ^17% excess combustion
air, 0.9 m/s fluidizing-gas velocity, and a 0.9 m bed height.
The operating conditions during the regeneration step of each cycle
were a nominal system pressure of 158 kPa, a bed temperature of 1100°C, and
a fluidized-bed height of ^46 cm. The required heat and reductants for
regeneration were provided by incomplete combustion of Triangle coal.
b. Sulfur Acceptance during Combustion
Representative steady state operating conditions and flue-gas
composition data for the ten combustion cycle experiments are presented in
Table 8. The level of sulfur dioxide in the flue gas and the corresponding
sulfur retention based on the flue gas analysis for all ten combustion cycles
are graphically presented in Fig. 12.
Sulfur dioxide levels in the gas increased from ^300 ppm in cycle
1 to ^950 ppm in cycle 10. This represents a decrease in sulfur retention
from ^88% in cycle 1 to ^55% in cycle 10. Although there is some scatter
in the data, it appears that the reactivity of the sorbent for sulfur reten-
tion decreased linearly with combustion cycle over the 10-cycle experiment.
c. Sulfur Release during Regeneration
The experimental conditions and results for a representative seg-
ment of each regeneration step are given in Table 9.
In the ten regeneration experiments, solids residence times ranged
from 6.8 to 8.1 min. The extent of CaO regeneration based on solids analysis
varied from 67 to 80%, with no apparent loss in regenerability over the ten
utilization cycles.
The S02 concentration in the dry off-gas from the regenerator
varied from 8.8% to 6.1%. In the first cyclic regeneration experiment (CCS-1),
the SC>2 was diluted by gas used to obtain the fluidizing-gas velocity of
1.43 m/s (other experiments were done at <1.26 m/s). In the three final
cyclic regeneration experiments, the lower SC>2 concentrations in the regener-
ation reactor off-gas were a result of the lower sulfur concentration in the
sulfated sorbent (7.1-7.9 wt % S instead of ^10% S). Although the combustion
steps of these cyclic experiments were performed with a constant CaO/S mole
ratio of ^1.5 with no virgin sorbent makeup, the total sulfur content of the
sulfated sorbent decreased in each cycle due to lowered sulfation reactivity
of the sorbent.
d. TGA Sulfation Experiments
To further test the reactivity of the sorbent for sulfation, samples
of the sorbent from each regeneration half-cycle experiment were sulfated in
a TGA apparatus at 900°C, using a reactant gas containing 0.3% S02, 5% 02, and
-------
Table 8.
Operating Conditions and Flue-Gas Compositions for Combustion
Step of Cyclic Experiments
Combustor: ANL, 6-in. dia
Coal: Arkwright, -14 mesh, 2.8 wt % S
Sorbent: Cycle 1, Tymochtee dolomite,
-14 +30 mesh
Cycles 2-10, Regenerated
Tymochtee dolomite
Temperature: 900°C
Pressure: 810 kPa
Excess air: ^17%
Nominal bed height: 0.9 m
Nominal gas velocity: 0.9 m/s
Additive
Feed
Combustion Analysis,
Cycle wt
REG- Ca
1 20
2 30.7
3 29.7
4 29.6
5 29.0
6 28.3
7 28.8
8 28.7
9 27.0
10 25.8
Feed
Rate,
% kg/hr
S
_
4.7
4.6
6.0
5.2
4.2
3.1
3.2
3.0
3.4
Coal
14.6
13.3
13.5
13.3
13.3
13.2
13.4
12.6
13.0
13.3
Sorbent
4.1
2.8
2.9
3.2
3.4
3.0
2.6
2.6
3.1
3.0
CaO/S
Mole
Ratio3
1.6
1.5
1.5
1.5
1.6
1.5
1.4
1.5
1.6
1.4
Flue-Gas Analysis
(avg values)
S02,
ppm
290
400
490
450
600
600
680
810
770
950
NO,
ppm
200
120
130
120
100
105
97
94
104
94
CH^,
ppm
32
30
(d)
30
31
21
47
43
49
49
CO,
ppm
90
40
20
45
55
40
59
40
56
64
C02,
%
16.0
15.5
16.0
16.0
16.6
17.0
16.1
16.2
16.5
15.9
02,
%
3.4
3.1
3.2
3.2
3.3
3.0
2.7
2.8
3.1
2.9
Sulfur
Retention,
%B / %c
88/81
82/79
78/86
79/66
72/64
72/66
69/61
61/57
63/52
55/50
.Ratio of unsulfated calcium in dolomite feed to sulfur in coal.
Based on flue-gas analysis.
.Based on solids analysis.
Analyzer inoperative.
-------
E
CL
CD
1100
1000
900
700
-600
Q
^500
2 Concentration
in Flue Gas as a Function of Cycle Number
the balance N2. The results of the experiments are shown in Figs. 13 and 14.
Figure 13 gives the percent of the CaO in the regenerated samples converted
to CaSOij as a function of time (i.e., the rate of conversion) for all ten
regeneration half-cycles. The numbers of the curves (2 through 10) correspond
to the numbers of the corresponding combustion half-cycles, and curve 11
represents sulfation of regenerated material from cycle 10. With the
exception of the curve corresponding to the fourth combustion cycle, the
rate of conversion decreased from the second through the eleventh sulfation.
An interesting observation is that after the eighth sulfation cycle,
the loss in reactivity with succeeding sulfation cycles is quite small. This
indicates that the reactivity of the sorbent may level off in the higher
sulfation cycles. This potential leveling-off of reactivity was not detected
in the cyclic combustion-regeneration experiments performed in the PDU.
Figure 14 plots the utilization of the stones for each sulfation
cycle as a function of time for the TGA experiments—i.e., the extent of con-
version. The starting point for each sulfation experiment is determined by
the extent of regeneration during the preceding regeneration half cycle.
-------
Table 9. Experimental Conditions and Results for the Regeneration Step
of the Ten Utilization Cycles with Tymochtee Dolomite
Nominal fluidized-bed height: 46 cm
Reactor ID: 10.8 cm
Pressure: 153 kPa
Temperature: 1100°C
Coal: Triangle (0.98 wt % S), ash fusion temperature
(initial deformation under reducing conditions): 1390°C
Sorbent: -14 +30 mesh sulfated Tymochtee dolomite
Cycle Expt.
No. No.
1
2
3
4
5
6
7
8
9
10
CCS-1
CCS-2
CCS-3
CCS-4
CCS-5
CCS-6
CCS-7
CCS-8
CCS-9
CCS-10
Fluidiz-
ing Gas
Velocity,
m/s
1.43
1.26
1.22
1.17
1.17
1.18
1.16
1.18
1.09
1.24
Solids
Residence
Time,
min
7.0
7.5
7.2
7.8
7.4
7.8
7.3
8.1
7.3
6.8
Sulfur
02 Cone Reducing Gas Cone in
in Feed Concentration Sulfated
Gas, in Off -Gas, Sorbent,
26.7
37.9
36.7
36.5
36.1
41.8
38.2
35.9
36.4
38.0
2.8
3.0
3.4
2.9
3.0
2.6
2.9
3.0
3.0
3.0
9.0
10.7
10.3
9.9
9.5
9.3
8.5
7.8
7.9
7.1
CaO
Regeneration
%a/%5
73/71
67/67
63/76
67/69
69/75
66/75
69/77
64/80
53/67
63/68
Major Sulfur
Compounds in
, Dry Off-Gas, %
S02
6.5
8.6
8.4
8.1
8.8
8.7
8.2
6.3
6.1
6.7
H2S
0.04
0.02
0.07
0.04
c
0.03
0.07
0.06
0.1
0.05
COS
0.06
0.1
0.1
0.1
c
0.2
0.1
0.07
0.1
0.08
CS2
0.04
0.1
0.1
0.1
c
0.1
0.1
0.07
0.1
0.09
a
Based on off-gas analysis.
Based on chemical analysis of dolomite samples.
Analysis not performed.
-------
42
C?
CO
o
o
o
o
o
O
CO
tr.
O
o
SULFATION
SYMBOL CYCLE
4 5
TIME, hr
Fig. 13. Conversion of CaO to CaSOi^ as a Function of Time
and Sulfation Cycle as Determined by TGA Sulfation
Experiments. Temperature: 900°C; reaction gas:
0.3% S, 5.0% 02, balance nitrogen.
SULFATION
SYMBOL CYCLE
o
CO
o
o
CO
o
O
_J
-------
43
However, after sulfation for a period of several hours, the utilization of
the stone decreases with sulfation cycle except for the inconsistency in the
fourth-cycle sulfation data. Again, the attainable sorbent utilization appears
to stabilize and become fairly constant at about the eighth combustion cycle.
e. Estimate of Sorbent Makeup Requirements to Meet EPA Sulfur Emission^
Limit
Based on the results of the cyclic combustion/regeneration experi-
ments, an analysis was made to estimate the sorbent makeup rate which would
be required in a continuous recycle operation to meet the EPA sulfur emission
limit. The makeup rate is essentially determined by three factors: (1) the
level of sulfur retention required, (2) the sorbent loss of reactivity with
increasing number of utilization cycles, and (3) the sorbent recycle rate,
which establishes the total CaO/S mole ratio during the combustion cycle. In
this discussion, the term CaO/S mole ratio is used to emphasize that the
mole ratio refers to the ratio of the available (unsulfated) calcium in the
sorbent feed to the sulfur in the coal feed.
The analysis was based on the results obtained during the ten
combustion experiments (REC-series experiments) in the cyclic combustion-
regeneration study. Conditions assumed for the analysis were, therefore,
a 900°C bed temperature, 810 kPa pressure, and a 0.9 m/s fluidizing-gas
velocity. The analysis was also based on maintaining a sulfur retention of
75%, which is slightly above the ^70% required to meet the EPA sulfur emission
limit.
The two requirements for the analysis were: (1) to determine the
utilization ("activity") of the sorbent at 75% sulfur retention as a function
of sulfation cycle and (2) to derive an analytical expression for the age
distribution (that is, the cycle number distribution) of the reactor charge
at steady state as a function of makeup rate.
In order to establish the first requirement in the analysis, it was
first necessary to determine the CaO/S mole ratio required as a function of
utilization cycle to maintain a constant sulfur retention of 75%. During
two of the combustion experiments (REC-7 and REC-8) in the cyclic combustion-
regeneration study, the CaO/S ratio was intentionally increased sufficiently
above 1.5 near the end of each experiment to bring sulfur retention to ^75%.
With this additional data, it was possible to determine the required CaO/S
ratio as a function of cycle to maintain a constant sulfur retention of 75%.
These results are illustrated in Fig. 15.
*
From the correlation of VAR-series experimental results, the CaO/S
ratio which would be required to achieve ^75% retention during the first
combustion cycle was calculated to be 1.0. From the correlation of the
REC-series cyclic combustion experiments at a CaO/S ratio of 1.5, a sulfur
A series of PDU experiments to study the effects of experimental conditions
on SC>2 retention.
-------
44
3.0
2.5
cc.
to
10
2
UJ
>
2.0
o
o
0.5
FROM REC-SERIES
EXPERIMENTS
•FROM VAR-SERIES
CORRELATION
FROM REC-SERIES
CORRELATION ATCoO/S=l.5
SULFUR RETENTION,%
o-74
a-75
0-76
Fig. 15.
3456789
CYCLE
CaO/S Ratio Required to Achieve 75% Sulfur
Retention as a Function of Cycle
10
retention of 75% occurs during combustion cycle 4.2. It was then experi-
mentally determined that during combustion cycles 7 and 8, CaO/S ratios of
2.1 and 2.5 were required to achieve sulfur retentions of ^74 and ^76%,
respectively. The resulting curve indicates that the CaO/S ratio required
in the eighth cycle to maintain a constant sulfur retention of 75% was
approximately two and one-half times that in the first cycle.
Zielke et al. 13 performed a similar cyclic combustion-regeneration
study using Tymochtee dolomite at ^155 kPa. A comparison of the results of
the two studies is given in Table 10. The results are obviously in very good
agreement; CaO/S ratios reported by Zielke et al. to achieve ^80% sulfur
retention were slightly higher than the ANL values of CaO/S to achieve ^75%
retention.
For purposes of correlation, a fractional combustion cycle such as 4.2 can be
used to indicate a material which is less reactive than sorbent being sul-
fated for the fourth time and more reactive than sorbent being sulfated for
the fifth time.
-------
45
Table 10. Comparison of the Experimental Cyclic Sulfation
Results Obtained at ANL with Those Reported by
Zielke et a1.a Tymochtee dolomite used in both
studies.
Conditions
Combustion
Pressure, kPa
Temperature, °C
Excess Air, %
Gas Velocity, m/s
Sorbent Size, mesh
Solids Residence Time, hr
ANL
^810
900
17
0.91
-14 +30
^5
Zielke et al.
M.55
980
20
0.91
-14 +28
^1.1
Regeneration
Pressure, kPa
Temperature, °C
Gas Velocity, m/s
Solids Residence Time, min
153
1100
VL.3
155
1065
^0.6
108
Cycle No.
Results
b
ANL •
Sulfur Retention,
CaO/S Ratio
7
/o
75
0.93
75
1.1
75
1.3
75
1.5
75
1.7
75
1.
9
75
2.2
75
2.
5
Zielke et al.
Sulfur Retention, %
CaO/S Ratio
79.
0.
3
95
80.4
1.4
80.2
1.9
78.7
1.9
77.
2.
6
2
79.
2.
3
3
80.6
2.6
N.D.
N.D.
Reference 13.
^Values obtained from Fig. 16.
"No data.
By the use of the correlation of Fig. 15, it was possible to calcu-
late the utilization ("activity") of the sorbent at 75% sulfur retention as
a function of sulfation cycle using the equation
U - °'75
n (CaO/S)
n
(5)
where Un = CaO utilization at 75% S retention for nth sulfation
(CaO/S)n = CaO/S mole ratio required for 75% sulfur retention during
nth sulfation
-------
46
The resulting correlation of the sorbent utilization at 75% sulfur
retention with the sulfation cycle is graphically presented in Fig. 18.
Utilization for a given sulfation cycle can be estimated from the equation
for the straight line in Fig. 16,
U = 0.92e-°'14n (6)
n
where n is the sulfation cycle number.
The second requirement in the analysis was to develop an expression
for the age distribution of the sorbent feed (recycle plus makeup) so that
the fractional amount of the feed being sulfated for the nth time can be
estimated. The approach taken was adapted from a procedure developed by
Nagier11* and is illustrated in Fig. 17.
In Fig. 17,
w0 = constant CaO makeup charged to each cycle
wn = total CaO charged to the nth cycle, n = l,2,...n
a = constant fraction of total charge rejected after each
cycle (includes decrepitation losses, incomplete
regeneration, and sorbent drawdown)
Therefore, (1 - a) represents the fraction of the CaO charged to
each stage which is recycled (via the regeneration process) to the next
stage. It can be easily shown that as n -*•<», wn converges to
w = [1 + (1 - a) + (1 - ct)2 + .... + (1 - a)n~1]w
n o
= wQ/a (7)
where the fraction, g , of the total charge, wn, being sulfated for the
nth time is expressed as follows:
gn = a(l - a)""1, n = 1, 2, .. . ,~ (8)
An equilibrium CaO utilization, Ueq, can then be calculated for 75% sulfur
retention at steady state as
U = I U g (9)
eq n=l n n
From this result, the total CaO/S ratio at equilibrium can be calculated as
Total CaO/S = 0.75/U (10)
eq
and the makeup CaO/S ratio as
Makeup CaO/S Ratio . = (Total CaO/S)-a (11)
-------
1.0
0.8
o:
a: 0.6
in
o
-------
48
The results of the analysis for Un as given by Eq. 6 are presented
in Fig. 18. As an>example of using Fig. 18, if a (makeup CaO/total CaO)
is 0.18, a makeup CaO/S ratio of ^0.27 and a total CaO/S ratio of ^1.5 are
required for a sulfur retention of 75%. Decreasing a to 0.1 (10% makeup)
reduces the makeup CaO/S ratio to ^0.2 (a reduction of 25%) but increases
the total CaO/S required to 2.0 (an increase of 33%). In comparison to the
once-through CaO/S ratio of 'vl.O for 75% sulfur retention, the makeup of 0.2
for a cyclic process corresponds to an estimated savings in quantity of lime-
stone needed of ^80%.
-------
49
It should be emphasized, however, that the choice of a is not
arbitrary. The value of a will affect both the process flow sheet and the
system economics. For a thorough discussion of the effect of recycle rate,
the reader is referred to a following section, "Regeneration Process Scale-up
and Flowsheet Determination."
f. Porssity of Dolomite as a Function of Utilization Cycle
The porosity of -25 +30 mesh particles was measured by the mercury
penetration method. The pore distributions of samples from cycles 2 and 10
are given in Fig. 19. The cumulative pore volume for pores >^O.A pm and also
for pores >Q.Qk pm in sulfated and regenerated Tymochtee dolomite are given
in Table 11. It has been reported by Hartman and Coughlin15 that most sul-
fation takes place in larger pores (^0.4 pm) and that pores smaller than
0.4 pm are relatively easy to plug. During sulfation of CaO, the pores
shrink as a result of molecular volume changes.
The porosity of sulfated dolomite was relatively unaffected by
utilization cycle, although the sulfur content decreased from 'x-lO wt % to
^7 wt % (Fig. 19). However, the porosity of the regenerated dolomite con-
sistently decreased with utilization, as did the sulfur content in the regen-
erated stones. The difference in porosity of the sulfated and regenerated
samples decreased from ^0.15 cm /g (pores >_0.4 pm) after the first cycle
to ^0.07 cm3/g after the tenth cycle. The porosity of regenerated dolomite
decreased with cyclic use, and thus its effectiveness as an SC>2 acceptor
decreased.
The loss of porosity in the dolomite could be due to (1) buildup
of an ash layer around the particle or (2) high-temperature (1100°C) exposure
during regeneration. At the regeneration temperature used, sintering begins,
decreasing the reactivity by decreasing the beneficial porosity of the parti-
cles. This limits local diffusion and reaction of S(>2. The effect of
porosity and reactivity of high-temperature exposure as a function of time
is being evaluated for Tymochtee dolomite and will be reported.
g. Coal Ash Buildup during Utilization Cycles
The extent of coal ash buildup in the fluidized bed during coal
combustion is of particular importance in evaluating its effect on (1) the
SC>2-accepting capability of the sorbent in the subsequent combustion step
and (2) the ash-sorbent agglomerating tendency in the regenerator reactor.
The ash buildups during all ten sulfation and regeneration steps
have been calculated from wet chemical analyses (Si and Ca) of the sorbent
product samples and are given in Table 12. As a basis for calculation, 100 g
of virgin dolomite was used. The ash buildup was based on bulk silicon en-
richment. The concentrations of silicon and the calculated coal ash buildups
are plotted in Fig. 20. After ten utilization cycles with no fresh sorbent
makeup, it has been found that for every 100 g of starting virgin dolomite,
^13 g of coal ash accumulated in the sorbent. The silicon concentration
increased from 2.1% in the virgin dolomite to 6.1% in the regenerated sorbent
from the tenth cycle. The differences in the concentration of silicon
-------
50.
4.7%S CYCLE-2 (REGEN)
2.7% S CYCLE-IO (REGEN)
CYCLE-IO
7.1% S (SULFATED)
10.7%S CYCLE-2
(SULFATED)
1.0 O.I
PORE DIAMETER,
Pore Distributions of Dolomite Samples from
Cycles Two and Ten
Table 11. Porosity (cm /g) of Tymochtee Dolomite as
a Function of Utilization Cycle
Cycle No.
Sulfated
Regenerated
Pores _M).4 ym.
DPores >0.04 u
Change, A
1
2
4
6
8
10
0.120a/0.164
0.120 /0.212
0.120 70.140
0.156 70.164
0.144 /0.156
0.132 70.140
0.268a/0.340
0.288 70.308
0.252 70.276
0.244 70.258
0.238 70.260
0.204 70.220
0.148a/0.176b
0.168 70.096
0.132 70.136
0.088 70.094
0.094 70.104
0.072 70.080
in sulfated and regenerated samples are due to the weight loss of the sorbent
during regeneration (CaSOi, ->• CaO) . The calculations for ash buildups do not
identify in which of the two process steps (sulfation or regeneration) most
of the ash buildup occurred. However, it is believed that most of the ash
buildup occurred during the sulfation step because the feed rate of fresh
ash (in the coal) was ^1 kg for every 3 kg of regenerated sorbent feed. In
contrast, during regeneration, the feed rate of fresh ash (in the coal) was
'vl kg for every 70 kg of sulfated sorbent feed.
-------
51
Table 12. Calculated Coal Ash Buildup during Sulfation and
Regeneration of Tymochtee Dolomite, Based on
Enrichment of Silicon
Mass Basis: 100 g virgin Tvmochtee dolomite
(20.0 wt % Ca and 2.14 wt % Si)
Cycle
No.
1
1
2
2
3
3
4
4
5
5
6
6
7
7
8
8
9
9
10
10
Cycle
Step
sa
Ra
S
R
S
R
S
R
S
R
S
R
S
R
S
R
S
R
S
R
Si Cone. ,
%
2.68
3.64
2.88
3.98
3.19
4.35
3.43
4.51
3.70
4.60
3.68
5.38
4.19
5.39
4.72
6.41
4.69
6.05
5.42
6.13
wt %
0
0
2.4
4.0
3.3
7.7
4.1
7.0
6.0
7.4
5.9
11.5
8.2
11.2
11.0
17.0
11.0
15.8
14.7
16.3
Ash Buildup
g ash
100 g virgin dolomite
0
0
2.1
2.7
2.8
4.8
3.4
4.9
5.2
5.2
5.1
8.1
7.0
7.8
9.6
12.5
9.7
12.2
13.3
12.7
o
S = sulfation step, R = regeneration step.
The coal ash buildup was also evaluated as a function of particle
diameter in a regenerated dolomite sample from the tenth cycle. The results
are plotted in Fig. 21. It was found that ash accumulation increased with
the particle diameter of the sorbent. The (nominal) 1700 pm particles (+14
mesh) were made up of agglomerates of smaller particles which are each coated
with coal ash. Thus the larger (+14 mesh) particles had more ash-coated
surface than did the smaller particles.
Sulfated and regenerated dolomite particles from the first, fifth,
and tenth utilization cycles were examined for macrofeatures under a low-
magnification microscope. Photomicrographs of these samples are given in
Figs. 22 and 23. The photomicrographs reveal that even the once-sulfated
stones were beginning to be coated with what is believed to be coal ash.
Particles from the tenth cycle (Fig. 23) appear to be completely coated with
ash. The coating can be more readily seen in color photographs. The cause
of the ash blisters is uncertain; however, their presence is probably bene-
ficial in that they expose reactive CaO in the particles.
-------
52
16
* 0
o~ 6
• -gosh/IOOq VIRGIN DOLOMITE
COMBUSTION STEP OF CYCLE
o-gosh/IOOg UTILIZED DOLOMITE
COMBUSTION STEP OF CYCLE
• -g ash/100 g VIRGIN DOLOLOMITE
REGENERATION STEP OF CYCLE
D- g ash/IOOg UTILIZED DOLOMITE
REGENERATION STEP OF CYCLE
R
Pi
o
n
ii
11
i i
! \
A-COMBUSTION STEP OF CYCLE
.0-REGENERATION STEP OF CYCLE
2 4 6 8
TYMOCHTEE DOLOMITE UTILIZATION CYCLE
10
Fig. 20. Coal Ash Buildup as a Function of
Utilization Cycle
40
^o
°* 30
a."
1 20
to
s! 10
0
FRACTIONAL SAMPLES
TOTAL SAMPLES
Fig. 21. Coal Ash Buildup as a
Function of Particle
Diameter in Tenth Cycle
Regenerated Particles
0 500 1000 1500 2000
FRACTIONAL PARTICLE DIAMETER,
-------
53
a. Cycle One, Sulfated (12.5 X)
b. Cycle One, Regenerated (12.5 X)
c. Cycle Five, Sulfated (12.5 X)
d. Cycle Five, Regenerated (12.5 X)
Fig. 22. Photomicrographs of Sulfated and Regenerated
Tymochtee Dolomite Particles from the First
and Fifth Utilization Cycles
-------
54
Fig. 23. Photomicrographs of Sulfated and Regenerated
Tymochtee Dolomite Particles from the Tenth
Utilization Cycle
-------
55
A petrographic examination was also made of the unreacted dolomite
and of samples of the_ dolomite after the first and tenth sulfation and regen-
eration half-cycles. As was observed in the above examination of the macro-
features, progressive buildup of a vitreous crust surrounding most of the
particles over the ten cycles was observed. The spotty beginnings of crust
formation during the first cycle are shown in Figs. 24 and 25. Representative
encrusted particles from the tenth cycle are shown in Figs. 26 and 27.
Where the crust is well developed (as in Figs. 26 and 27), it is
red with some black areas when viewed in ordinary light. Under the microscope,
a polished section viewed in reflected light shows that the black parts have
higher reflectivity and are magnetite (FesO^) . The less reflective red parts
resemble a silicate or silicate glass. The red coloration is probably due to
finely dispersed hematite (Fe2C>3) and is free of magnetite. The contrast
between these two parts is indicated in Fig. 27.
X-ray diffraction analyses were made of the surfaces of particles
from the first and tenth cycles. On the surface of first-cycle regenerated
particles, the presence (very minor) of Ca(Alo . yFeo . 3)205 was detected. On
the surface of tenth-cycle regenerated particles, evidence for the presence
of this compound was more pronounced. The tenth-cycle sulfated particle
surface contained a high concentration of a-
Although the two predominant crystalline phases in the crust are
magnetite and MgO, it is very likely that they are dispersed phases in a
vitreous matrix (of unknown composition), as evidenced by the vesicular
character shown in Fig. 28. In this respect, this material is similar
to the fusion crust of stony meteorites. No silicon-based compound was
found by X-ray diffraction (glassy silicate compounds would not be detected) .
The iron and aluminum compounds on the surface of these particles
are probably present in adhering coal ash. Specifically, the iron content
of the crust is much greater than the iron content of the average dolomite
particle. The unreacted dolomite particle shown in Fig. 29 contains the
highest visible amount of iron oxide and iron sulfide of some 24 particles
randomly selected and polished.
It is reasonable to speculate that ash particles stick to the
surfaces of dolomite particles during combustion. A glassy crust may form
by fusion at the combustion temperatures in the combustor (there is evidence
of crust formation during the first combustion cycle, Fig. 25) and later when
the particles are regenerated at 1100°C.
The presence of coal ash shells on the dolomite particles did not
cause routine def luidization during regeneration, although it is thought that
in .the regeneration reactor (at 1100°C under reducing conditions) this shell
is soft. The absence of agglomeration in the fluid bed during regeneration
Analysis performed by L. H. Fuchs, Chemistry Division, Argonne National
Laboratory.
-------
56
Fig. 24. Photomicrograph of Cross Section of a Tymochtee
Dolomite Particle from the First Combustion
Cycle (X75)
Fig. 25. Photomicrograph of Cross Section of a Tymochtee
Dolomite Particle from the First Regeneration
Cycle (X75)
-------
57
Fig. 26. Photomicrograph of a Cross Section of a
Tymochtee Dolomite Particle from the Tenth
Combustion Cycle (X75)
Fig. 27. Photomicrograph of Cross Section of a Tymochtee
Dolomite Particle from the Tenth Regeneration
Cycle (X75). Arrows indicate magnetite con-
centrations in crust; bracket = magnetite-free
area.
-------
58
Fig. 28. Photomicrograph of Cross Section of a
Tymochtee Dolomite Particle from the
Tenth Regeneration Cycle. The arrows
point to vesicles (X150)
Fig. 29. Photomicrograph of Cross Section of an
Unreacted Tymochtee Dolomite Particle.
The arrow points to a pyrite inclusion
(X75)
-------
59
was probably due to the fluidizing velocity which was >1.0 m/sec and high
enough to maintain stable fluidization for the -14 mesh particles used in
these experiments. The beneficial effects of high fluidization velocities
in "sticky" beds have been described by Gluckman et aZ.11
Evidence that the crust may be an effective sealant against gas
diffusion is an observation pertaining to samples from the tenth regeneration
cycle which contain a few particles having no crust. (The crust was probably
broken off prior to reduction.) An X-ray pattern of the interiors of these
crust-free regenerated particles showed strong CaO and MgO and weak Ca(OH)2;
in contrast, a pattern of the interiors of encrusted particles in the same
sample showed strong MgO, medium CaSOi, , and only weak CaO.
h. Electron Microprobe Analysis of Tymochtee Dolomite from Tenth
Utilization Cycle
Electron microprobe analyses were performed on cross sections of
sulfated and regenerated dolomite samples from the tenth utilization cycle
to confirm the existence of the coal ash shell and to determine its compo-
sition and its role during sulfation reactions.
Steady state samples of dolomite particles were screened (-20 +25
mesh), mounted in epoxy, and machined to remove the equivalent of one-half
of the nominal diameter (^400 ym). A thin carbon layer was applied to the
machined surface by vapor deposition to enhance the conductivity of the
mounts. Apatite (38.94% CaO), MgO (60.31% Mg), Si02 (49.88% Si), A1203
(52.93% Al), and FeS2 (46.55% Fe and 53.45% S) were used as standards to
obtain a quantitative estimate of local component concentrations. The measured
local concentrations of constituents were probably biased on the low side
due to surface irregularities (absent from the standards) which scatter the
characteristic emitted X-rays.
The radial component concentration profiles for three typical
regenerated particles (P-l, P-2, P-3) after ten cycles are given in Figs. 30,
31, and 32. The particle sections, when viewed with an optical microscope,
revealed a well-developed crust. Particle 1 (P-l), which is shown in Fig.
30, has a shell which is thicker on the side where the scan was initiated.
The electron microprobe scan of P-l confirms the existence of the ash crust
on the particle. Peak concentrations of Si ^12 wt %, Fe ^25 wt %, and Al
'W wt % were found in the crust. The concentrations of these components in
Arkwright coal ash, which was used during the combustion step of the experi-
ments, are: 12 wt % Si, 14 wt % Fe, and 12 wt % Al. The above-measured
concentrations of relatively major components in the particle crust were not
in the same proportion as in the coal ash; Fe/Si was ^2 in the crust and ^0,7
in the coal ash. The coal ash which encapsulated the particle was strongly
enriched with iron. The sulfur concentration profile shows that this particle
(P-l) had not been completely regenerated and that very little sulfur was
present in the particle crust.
The electron probe scans for two additional tenth-cycle regenerated
dolomite particles (P-2 and P-3) are given in Figs. 31 and 32. Both of
these particles also were encrusted with ash, as the scans for Si, Fe, and Al
-------
6Q
0>
Ash
Particle
Fig. 30. Electron Microprobe Analysis of a Typical Regen-
erated Dolomite Particle (P-l) from the Tenth
Cycle
reveal. In these particles also, a relative elemental enrichment of the ash
crust with iron was found. The calcium concentration in the crusts is as
high or higher than that in the particle interiors, which suggest the
possibility of calcium enrichment in the ash crust by diffusion from the
particle interior. The calcium concentration in the Arkwright coal ash, 3.5
wt %, is much lower than in the crust. The sulfur concentration profiles
show that these particles have been nearly completely regenerated, with peak
concentrations of <0.5% S. The bulk concentration of sulfur in the dolomite
-------
61.
0.4-
0.3-
0.2-
15--
10-
5-
4-
58 3-
2 2-
-I- !
— *J
~l
1
x
-
Ach
\
-
•v
-
-
--
-
-
_
-
__
-
_
-
-
--
...
--
-
-
••^
--
^-
-
-
-
E
•
-
-
-^
L Pnrtirle »
-
-
¥
V
A«;h
Fig. 31. Electron Microprobe Analysis of a Typical Regen-
erated Dolomite Particle (P-2) from the Tenth
Cycle.
before regeneration was 7.1 wt % (determined by chemical analysis). The
presence of the coal ash shell apparently does not prevent sulfur in the
particles from escaping during regeneration. Possibly, cracks and voids
in the ash crust could be major routes for gas transport during regeneration.
Sulfated particles from the tenth cycle (sulfated ten times and
regenerated nine times) were also analyzed with the electron microprobe.
-------
a* 30-
- 20-1
S 10-2
T
a
3$ 60-
5 20-;
^^
0.6
0.4
—
8
9
X
"
-/
/
L
-
\
—
—
r-
-
-
\
\
Sk
_
V
-
^
-
'
—
^
-•
»=
-
H^
'
—
-"
._
'-
"
.-
^1
_
~
/
/
-
m
s
-
s
f
ft
1
\
Ash-*|
Fig. 32. Electron Microprobe Analysis of a Typical
Regenerated Dolomite Particle (P-3) from
the Tenth Cycle
The analyses for three typical sulfated particles are given in Figs. 33, 34,
and 35. In all three particles, the formation of an ash crust (high
concentrations of Si, Fe, and Al) is again verified. Enrichment of iron
(in relation to Si and Al) in the crust was again observed, particularly
in the second and third particles. As was alio found in the regenerated
-------
63
< 2-
- ~
J
/
-
/
•N
•-As!
—
\
\
*
-
^
-
— *
_
"
-r
—
^
-
,**
-
\
-
•^
fX-
>-
-
"
"•*
-
=•
-
-
-
-
-
Particle
-
-
»-•
-
~
-
s
—
^
-
J*
""
--
-
=*•
-
—
—
-=
-
-
-
•V
^>
—
7
-
f_
f>.
-
s
-
Ash-*
Fig. 33. Electron Microprobe Analysis of a Typical Partially
Sulfated Dolomite Particle (PS-1) from the Tenth
Cycle.
particles, the ash crust was enriched in calcium, reaching concentrations
equivalent to those in the particle interiors.
The sulfur concentration profiles in the first two particles (PS-1,
PS-2) revealed uniform sulfation of the particle interiors, with local
sulfur concentrations ranging from 7.5 to 10%. (The bulk sulfur content of
the sulfated dolomiteysamples-was 7.1 wt %}) 'The calcium in the ash crust
does not appear to have reacted with sulfur. It could be present as a
-------
o
o
tt)
—| Ash
Ash*|
Fig. 34. Electron Mlcroprobe Analysis of a Typical Partially
Sulfated Dolomite Particle (PS-2) from the Tenth
Cycle
silicate. The sulfur concentration profile in the third partially sulfated
particle also shows that the calcium in the ash shell is not in a reactive
form. The sulfur concentration below the ash crust is highest near the
crust and decreases with penetration towards the center of the particle.
If diffusion through the ash crust or sulfated shell controlled the sulfation
reaction, the calcium adjacent to the crust would be expected to be more
fully sulfated (^10 wt % S) in a partially reacted particle. Also, a
sharper radial sulfur concentration gradient would be expected at the reaction
front. The electron microprobe analysis of sulfur concentration in tenth
cycle partially sulfated dolomite suggests a resistance at the reaction front.
-------
65
30-
20-
d*
30
o 20
o IOH
. -/
-X
!
i
•/• -
J
/
f^
_
f
\
\
«•
s
vh
\
>
M
/•
X,
—
7
/^
*
W,
/
^
;
fa
>
•%
*s
PI
•f
Jf
s*
v
^
P_
\M
>
^
Ml
ft
tfA
^
•/
*/
!\
fj
*^
*>
^
\
^
\
\
fS
^/
X
s*
V
V
ta
^
y
*/•
/
/
\
v
•\
/'
\
\
1^
1
*-
\
Fig. 35. Electron Microprobe Analysis of a Typical Partially
Sulfated Particle (PS-3) from the Tenth Cycle
The loss of reactivity could be due to a loss of local microporosity caused
by sintering within the dolomite particles.
Electron microprobe analyses performed on once-sulfated Tymochtee
dolomite particles were reported previously (ANL/ES-CEN-1016). In those
results, it was observed that a sulfated shell had formed on the particles
-------
66
and that the shell moved toward the center of the particles with increasing
extent of sulfation. There were sharp sulfur concentration gradients between
the sulfated shell edges and the unsulfated particle interiors. The local
sulfur content in the sulfated shell (for the incompletely reacted particles)
was as high (^10 wt % S) as that in the completely reacted particles. The
above observations suggest that diffusion through the sulfated shell was the
controlling step during sulfation in the first cycle and that the reaction
of sulfur (862) occurred in a well-defined reaction front.
A sample of tenth-cycle regenerated dolomite particles was crushed
and screened. The crushed (-30 +100 mesh) particles were sulfated in a TGA,
and the results are compared in Fig. 36 with TGA reactivity results for
uncrushed tenth-cycle dolomite. No notable difference in reactivity was
found, further suggesting that formation of an ash shell on the particles
is not responsible for the loss of sulfation reactivity in the dolomite
during cyclic utilization.
i. Carbonate Levels of Sorbent Samples
The weight fraction of the unsulfated calcium present as calcium
carbonate was derived for the sorbent at each half-cycle in the utilization
study up to the end of the sixth full cycle. The results are shown in Fig.
37. Initially, all of the calcium was present as CaC03 in the virgin dolomite
feed to the first combustion cycle. After the first combustion half-cycle,
the unsulfated calcium was still predominantly (^68%) CaC03. Following the
first regeneration, however, only 4% of the unsulfated calcium was present
as CaC03. Although the carbonate fraction remained essentially constant
during succeeding regeneration cycles, the carbonate fraction steadily
70
60
8 30
to
O
o
u 40
o
to
u
"S 30
20
10
O 10th Cycle Regenerated
^ 10th Cycle Regenerated
Crushed (-30 +100 mesh)
5.75
Time, hr
Fig. 36. Comparison of the Reactivity of Tenth Cycle
Regenerated Dolomite (-14 +30 mesh) to that
of a Crushed (-30 +100 mesh) Sample from the
Same Experiment
-------
67
o
CO
OL
CO
.00
0.9
G 0.8
~ 0.7
o
=3 0.6
-------
68
In the third approach, the losses were predicted by evaluating the
percentage of the calcium in the sulfated dolomite fed to the regeneration
reactor that was elutriated and subsequently removed from the off-gas by
the cyclones and filter. (The calcium contribution of the coal ash from
the Triangle coal used for regeneration was found to be insignificant.) The
losses, based on calcium content of particulates removed from the off-gas
stream, averaged 2.0% (see Table 13). Because the off-gas particles were
-30 mesh and the feed sulfated dolomite was normally -14 +30 mesh, it can
be assumed that the off-gas particles were attrited fragments of the regen-
erated dolomite. No apparent trend in the extent of attrition was found
for the ten cycles.
Table 13. Attrition and Elutriation Losses for Tymoohtee
Dolomite during Regeneration in the Cyclic
Utilization Study
S.D. = Ca in sulfated dolomite (feed), kg/hr
O.P. = Ca in particulate collected from
off-gas, kg/hr
Loss for Steady-State
Experiment Segment,
Regeneration /n P \
Cycle CCS- I ^V I x 10°
/O.P.\
I5'0'/
1
2
3
4
5
6
7
8
9
10
1.9
1.7
3.0
1.2
3.5
3.7
2.0
2.6
0.9
1.3
Avg 2.2
The extent of sorbent losses by decrepitation and/or entrainment
during the ten combustion half-cycles was also determined, based on the
steady-state calcium material balances around the combustor, which accounted
for the calcium in both the dolomite and the coal. The assumption was made
that all calcium entering the combustor in the coal was entrained in the
flue gas as fly ash with no correction for ash buildup in the fluidized bed.
Based on the rates involved and the relatively low calcium concentration
of the ash as compared with the calcium concentration of dolomite, such a
correction would then be very minor. The sorbent loss by decrepitation and/
or entrainment was then calculated as follows:
-------
Sorbent Loss (%) =
6.9
Calcium in Entrained
Particulate Matter
Calcium in
Coal Feed
(12)
(Calcium in Sorbent Feed)
The results are presented in Table 14 along with the calcium material balances
for the ten combustion experiments.
Table 14. Decrepitation and Entrainment Losses and Calcium
Material Balances for the Ten Combustion Experi-
ments in the Cyclic Sorbent Utilization Study
Sulf ation
Cycle
REC-1
REC-2
REC-3
REC-4
REC-5
REC-6
REC-7
REC-8
REC-9
REC-10
Sorbent Loss by
Decrepitation and
Entrainment, %
16
4
5
3
3
6
4
7
6
4
Calcium Material
Balance, (In/Out) x 100
108
94
104
96
99
102
100
101
91
96
The 16% loss reported for experiment REC-1 is a revision of the
previously reported value of 20-25% (ANL/ES-CEN-1016). Although the first-
cycle loss was still quite large, losses during the remaining nine combustion
cycles were reasonably small, averaging about 5% per cycle. It is quite
likely that after the first combustion cycle, the resistance of the sorbent
to decrepitation is increased by (1) residual sulfate in the sorbent following
regeneration and (2) the presence of the vitreous crust on the particles.
It is also possible that the rapid calcination of MgCC>3 during the first
combustion cycle contributed to the large sorbent loss during the cycle. On
the basis of these results, the loss of sorbent reactivity may be more signi-
ficant in affecting cyclic performance adversely than is the loss of sorbent
by decrepitation.
Although the thermal cycling is more extreme and the reactions are
more rapid (decomposition) during regeneration than during sulfation, the
extent of attrition during regeneration was lower. The lower sorbent losses
during regeneration can be attributed to the very short solids residence
time (^7.5 min) in the reactor in each regeneration step as compared with
the much longer solids residence time (^5 hr) in the combustor reactor for
each sulfation step. The effect of introducing solids that were at room
temperature into a hot reactor environment in each half-cycle cannot be
estimated. In an industrial process, the solids would be cycled between the
combustor and the regenerator reactor at the temperature of the reactors,
-------
70
and hence the thermal shock would be lessened. It is believed that lower
sorbent losses would be obtained with a continuous sorbent cycling system.
The combined losses due to attrition and/or elutriation per cycle
(sulfation and regeneration) have been found to be ^8%. In an FBC process
utilizing sorbent regeneration, it is expected that the makeup rate for
Tymochtee dolomite would have to be at least this high because of attrition.
An even higher makeup rate may be required to maintain sufficient S02~sorption
reactivity in the fluidized bed of the combustor.
k. Amount of Sorbent Processed per Cycle
Table 15 presents the amounts of sorbent processed during each
phase of the cyclic utilization study along with calcium balances. It
should be emphasized that losses of materials between cycles included handling
losses (spills, etc.), sampling losses, and (in the case of combustion)
losses during startup.
Table 15. Gross Amounts of Sorbent Processed and Calcium
Balances for Each Half-Cycle in the Cyclic
Combustion/Regeneration Study
Gross wt
in,
Cycle kg
Combustion
1
2
3
4
5
6
7
8
9
10
Regeneration
1
2
3
4
5
6
7
8
9
10
620
268
282
146
144
115
88.5
80.3
76.6
66.7
359
328
181
169
143
108
97.5
85.3
75.3
71.7
Ca
in,
wt %
20.0
33.9
29.7
29.4
29.0
28.3
28.8
28.7
27.1
65.9
26.3
22.3
23.4
24.1
22.9
22.9
23.3
22.8
22.6
22.1
Ca Gross wt
in, out,
kg kg
124.0
90.9
83.8
42.8
41.7
32.5
25.5
23.0
20.7
17.3
94.4
73.5
42.5
40.7
32.8
24.8
22.7
19.4
17.0
15.8
359
328
285
169
143
108
97.5
85.3
75.3
71.7
268
282
146
144
115
88.5
80.3
76.6
66.7
62.1
Ca
out ,
wt %
26.3
22.4
23.4
24.1
22.9
22.9
23.3
22.8
22.6
22.1
33.9
29.7
29.4
29.0
28.3
28.8
28.7
27.1
25.9
25.6
Ca
out,
kg
94.4
73.5
66.7
40.7
32.8
24.8
22.7
19.4
17.0
15.8
90.9
83.8
42.8
41.7
32.5
25.5
23.0
20.7
17.3
15.9
Calcium
Balance ,
%
76
81
80
95
79
76
89
85
82
92
96
114
101
102
99
103
102
107
102
101
-------
71
6. Cyclic Sorbent Life Study with Greer Limestone
[j. Montagna (Principal Investigator), F. F. Nunes, G. Teats, R. Beaudry,
R. Mowry, S. Smith, and J. Stockbar]
A cyclic sorbent utilization experiment was performed with Greer lime-
stone and Sewickley coal (this coal is used in both the combustion and the
regeneration steps). This limestone and coal are being used by Pope, Evans
and Robbins (PER) in the atmospheric pilot plant at Rivesville.
Evaluated as a function of utilization cycle (with no sorbent makeup)
were (1) the changes in reactivity (sulfur acceptance during combustion),
(2) the changes in regenerability (sulfur release during regeneration), (3)
the extent of decrepitation, and (4) the extent of ash buildup. In addition,
a balance for the trace elements of interest will be made for the combined
combustion/regeneration system.
The processing capacity of the ANL PDU regenerator is much greater
than that of the combustor (by almost a factor of ten), and the sorbent
cannot be cycled continuously between these reactors. This limitation was
overcome by performing the sulfation and regeneration experiments batchwise
with no fresh sorbent make-up. The results (as a function of utilization
cycle) are being incorporated in the regeneration process model and will
be used to predict the performance of the sorbent system in a boiler with
continuous regeneration—specifically, to evaluate the effect of fresh
sorbent make-up rates on the overall process.
a. Combustion Step Results
The combustion steps of the experiments were performed at 308 kPa
(^3 atm), %855°C (VL570°F), a nominal fluidizing-gas velocity of 1.0 m/s
and a constant sulfur retention of ^84% by the sorbent. This corresponds
to a sulfur concentration of ^640 ppm in the dry flue gas, which is the EPA
emission limit for Sewickley coal (which contains 4.3 wt % S).
Data for the combustion step of the ten cycles are presented in
Table 16. Sulfur retention was maintained at ^84% by adjusting the regenerated
sorbent feed rate so that the S02 concentration in the off-gas was maintained
at ^640 ppm. Therefore, the flue-gas-based sulfur retentions reported in
Table 16 are indicative of the ability to maintain the experimental design
conditions. The other reported values of sulfur retention were based on
the steady-state ratio of the amount of sulfur retained by the limestone
to that released by the coal. The differences between these values and the
designed retention are indicative of experimental errors. The largest
deviations (^27%) occurred in cycles four and five.
b. Cyclic and Total Calcium Utilization
The cyclic calcium utilizations and the calcium present as CaSOi^
in the combustor feed and product streams are given in Table 16. The percent
calcium present as CaSOi, in the combustor feed stream is dependent on the
extent of regeneration and the extent of sulfation in previous cycles. The
percentage generally decreases with utilization cycle as the stone becomes
-------
Table 16.
Operating Conditions and Flue-Gas Compositions for Combustion Steps
of Cyclic Experiments with Greer Limestone and Sewickley Coal
Combustor: ANL, 6-in.-dia
Coal: Sewickley, -6 +100 mesh, 4.3 wt % S
Sorbent: Cycle 1, Greer limestone,
-14 +30 mesh
Cycles 2-10, regenerated
Greer limestone
Temperature: 855°C
Pressure: 308 kPa
Excess air: ^17%
Nominal bed height: 0.9 m
Nominal gas velocity: 1.0 m/s
Flue Gas
Analysis,
Combustion
Cycle
No.
1
2
3
4
5
6
7
8
9
10
Sorbent
Total Ca,
wt %
29.4
41.1
40.0
39.7
39.5
39.9
40.8
40.4
39.6
39.8
Feed Rate,
kg/hr
Coal Sorbent
6.27 3.41
6.73 3.14
6.73 5.09
6.59 5.14
6.36 4.95
6.93 6.02
6.46 6.24
6.14 6.36
6.14 5.80
6.36 7.61
CaO/S
Mole
Ratio3
2.87
3.13
4.75
4.67
4.88
5.54
6.64
7.05
7.35
8.21
avg
S02;
ppm
628
636
674
598
630
696
580
620
615
642
values
, 02,
ppm
3.2
3.0
3.0
3.1
2.9
3.0
3.0
3.3
2.9
2.8
Sulfur
Retention,
%b / %c
83.6/84.8
84.3/68
83.7/65.3
85.5/57.4
84.2/57.4
84.0/63.5
86.6/78
84.1/86.3
84.9/85.1
84.0/71.4
Calcium Present
as CaSO^
Combustor
Feed,
%
0
13
15
18
15
14
9
8
9
7
in the
Streams
Product ,
%
30
32
26
28
25
23
20
20
19
15
Calcium
Utilization,
%
30
22
14
12
12
11
12
12
12
9
Ratio of unsulfated calcium in dolomite feed to sulfur in coal.
,Based on flue-gas analysis.
.Based on solids analysis.
Utilization of available calcium during cycle. Utilization is defined as the percentage of the
available CaO that is converted to CaSO^ in the combustion step.
-------
73
less reactive. The cyclic utilization value, -i.e., the percent of CaO that
is sulfated in each step, is plotted in Fig. 38. Calcium as CaSOit in the
product decreased from ^30% in the first cycle to ^15% in the tenth cycle.
The cyclic calcium utilization, also plotted in Fig. 38, decreased from ^30%
in the first cycle to vL2% in the 10th cycle.
Hammond and Skopp also reported cyclic limestone reactivity data
for a series of batch sulfation/regeneration experiments. A sulfur retention
of 80% was maintained in their sulfation steps. Using particles having an
average diameter of ^930 ym (equivalent to the size of the Greer limestone
used in this study), they found that calcium utilization decreased from
to ^15% in seven cycles.
32
28
^-24
o
S 20
lie
1 12
-------
74
c. Porosity of Limestone as a Function of Utilization Cycle
The porosity of -25 +30 mesh particles was measured by the mercury
penetration method. The pore distributions of samples from cycles 2 and 10
are given in Fig. 39. It has been reported by Hartman and Coughlin15 that
most sulfation takes place in larger pores (>_0.4 urn) and that pores smaller
than 0.4 pm are relatively easy to plug. During sulfation of CaO, pores
shrink as a result of molecular volume changes.
The porosity of sulfated limestone was relatively unaffected by
utilization cycle (Fig. 39), although the sulfur content (or total calcium
utilization) decreased from ^8.9 wt % to ^4.1 wt %. The porosity of the
regenerated limestone decreased with utilization cycle. Simultaneously, the
sulfur content of the regenerated stones decreased. The difference in porosity
of the sulfated and regenerated stones decreased from ^0.084 cm /g (pores
>_0.4 urn) after the second cycle to ^0.055 cm3/g after the tenth cycle. Most
of the porosity loss was experienced in the first six cycles. The loss in
porosity decreased the reactivity of the limestone with SC>2 • This loss can
be attributed to high-temperature (1100°C) exposure in the reducing environ-
ment of the regenerator. Loss of beneficial porosity limits internal
particle diffusion and reaction with S02•
d. Limestone Reactivity as a Function of Cyclic Utilization
As the reactivity of the limestone decreased with cyclic use, the
molar feed rate of CaO/S required to achieve 84% sulfur retention with no
virgin limestone makeup increased from 2.9 to 8.2 in ten cycles (see Fig. 40).
These results and those on cyclic calcium utilization will be used to predict
fresh limestone makeup rates and flowsheets for FBC processes with sorbent
regeneration.
0.20
0.16
.
o o
0.08
0.04
0.00
T
3.7% S CYCLE 2 (REGEN.)
~2.0%S CYCLE 10 (REGEN.)
8.9% S CYCLE 2 (SULF)
SGL-IO
.1% S CYCLE 10 (SULF)
10 I O.I
PORE DIAMETER ,
0.01
Fig. 39. Pore Distributions of Limestone Samples from
Cycles Two and Ten
-------
75
o
o
CJ
i r
I
I
Fig. 40. Cyclic CaO/S Molar Feed
Ratio Required to Retain
84% of the Sulfur
0 2 4 6 8 10
SULFATION CYCLE
TGA sulfation experiments were performed on regenerated samples
from this cyclic experiment. They were performed at 855°C and atmospheric
pressure, using a simulated flue gas containing 0.3% SC>2, 3% 02, and the
balance N2. The results obtained are shown in Fig. 41. These results con-
firmed the PDU results in that the rate of sulfation decreased with cyclic
use of the limestone.
In the PDU sulfation steps of the cyclic experiments, as the
reactivity of the limestone decreased with cyclic use, the CaO/S mole feed
ratio was increased to maintain 84% sulfur retention. Increasing the CaO/S
feed ratio caused the limestone residence time in the combustor to decrease.
This, together with decreasing reactivity, lowered cyclic stone utilization.
The TGA results and the PDU combustor limestone residence times in the
corresponding cycles were used as a basis for very crude predictions of
calcium utilization. The results (i.e., TGA data points) are given in Fig.
38, together with data obtained in the 6-in.-dia PDU combustor. Good agree-
ment was obtained between the PDU and the TGA data.
e. Regeneration Step Results
were:
The operating conditions during the regeneration step of each cycle
a nominal system pressure of 129 kPa (^4 psig), a bed temperature of
-------
76
30
I I I I
o VIRGIN CALCINED LIMESTONE
a TWICE-REGENERATED LIMESTONE
0
SIX-TIMES-REGENER-
ATED LIMESTONE
TEN-TIMES-REGEN--
ERATED LIMESTONE
1 1 1
) 1 2 3
TIME.hr
i
4 J
Fig. 41. Conversion of Available
Calcium to CaSOi+ as a
Function of Time. Ob-
tained with TGA at 855°C
and atmospheric pressure
1100°C, a fluidizing-gas velocity of ^1.2 m/s, a total reducing gas concen-
tration of ^3.0% in the dry off-gas, and a fluidized-bed height of ^46 cm.
The residence time of the sorbent was *W min.
Results for the regeneration step of the ten cycles are given in
Table 17. The SC-2 concentration in the dry off-gas in the ten cycles ranged
from 6.1 to 8.6%. The extent of CaO regeneration ranged from 49 to 81% during
the ten cycles. The lowest extent of regeneration was obtained in the third
cycle, in which the solids residence time was low. After ten cycles, the
regenerability of the limestone remained acceptable.
f. Coal Ash Buildup during Cyclic Utilization
Sewickley coal, which contains 12.7% ash, was combusted in the
sulfation and regeneration steps of the ten-cycle experiment. The extent of
coal ash buildup has been calculated based on wet-chemical analysis (Si and
Ca) of samples from each half-cycle of the experiment. The basis for calcu-
lations was 100 g of virgin Greer limestone. Coal ash buildup was based on
silicon enrichment. The results are given in Table 18 and are plotted in Fig.
42.
It was found that every 100 g of starting virgin limestone accumulated
25 g of coal ash in ten cycles. The silicon concentration increased from
-------
Table 17.
Experimental Conditions and Results for the Regeneration
Step of Ten Utilization Cycles with Greer Limestone
Nominal f luidized-bed height: ^46 cm
Reactor ID: 10.8 cm
Pressure: 129 kPa
Temperature: 1100°C
Coal: Sewickley (4.3 wt % S) , ash fusion temperature
(initial deformation) under reducing conditions:
Sorbent : -14 +30 mesh sulfated limestone
Sulfur
Cone, in 02
Regeneration Sulfated in
Cycle
No.
1
2
3
4
5
6
7
8
9
10
.Based on
Sorbent,
7
8
8
7
7
6
5
6
5
5
off-gas
%
.7
.6
.1
.9
.5
.0
.5
.0
.6
.1
analysis
Cone Fluidizing—
Feed Gas
Gas, Velocity,
%
44.1
44.1
42.9
41.1
39.9
39.5
38.5
40.8
40.3
37.9
*
m/s
1.29
1.20
1.23
1.18
1.32
1.25
1.25
1.28
1.29
1.24
Reducing
Solids Gas Con-
Residence centration
Time,
min
7.1
6.8
6.3
7.0
7.0
7.2
6.8
6.5
6.2
7.0
— i —
in Dry
Off Gas, %
2.9
3.7
3.2
3.4
2.9
3.1
3.0
2.8
3.0
3.1
CaO
Major Sulfur Compounds
Regeneration, in
%a / %b
67/71
55/63
50/49
58/58
59/65
79/62
63/60
54/60
54/73
64/81
S02
8.4
8.6
7.5
8.2
7.3
8.1
6.6
6.3
6.2
6.1
Dry Off-Gas, %
H2S
0.06
0.1
0.2
0.05
0.06
0.08
0.06
0.04
0.08
0.06
COS
0.09
0.1
0.1
0.1
0.07
0.08
0.07
0.04
0.06
0.06
CS2
0.02
0.07
0.2
0.1
0.09
0.1
0.07
0.08
0.06
0.1
-------
78
Table 18. Calculated Ash Buildup during Sulfation and
Regeneration of Greer Limestone. Based on
the Enrichment of Silicon.
Mass Basis: 100 g virgin limestone (29.4
wt % Ca and 4.27 wt % Si)
Cycle
No.
1
1
2
2
3
3
4
4
5
5
6
6
7
7
8
8
9
9
10
10
Cycle
Step
sa
R^
S
R
S
R
S
R
S
R
S
R
S
R
S
R
S
R
S
R
Si Cone,
wt %
7.52
3.94
6.70
7.79
7.08
7.63
6.50
7.56
7.40
8.25
9.22
8.87
8.50
9.04
8.59
9.28
8.10
10.25
8.58
wt %
13.98
0
10.46
15.58
12.09
14.46
9.50
14.16
13.47
17.12
21.3
19.79
18.20
20.52
18.59
21.56
16.48
25.73
18.54
Ash Buildup
g Ash
100 g Virgin Sorbent
15.4
0
12.3
21.2
14.35
19.67
11.88
19.01
15.9
23.24
25.28
27.47
22.04
28.20
22.88
29.0
19.73
34.83
22.77
»q
,S = sulfation step.
R = regeneration step.
4.3 wt % in the virgin limestone to 8.6 wt % in the product of the tenth
sulfation step. In the Tymochtee dolomite cyclic experiment, 13 g of coal
ash were accumulated for every 100 g of starting virgin dolomite. Arkwright
coal, which was used in the combustion (sulfation) steps of that cyclic
experiment, contained considerably less ash, 7.7 wt %, than did the
Sewickley coal, 12.7%. In both cyclic experiments, most of the ash was
probably accumulated during the combustion step (where the sorbent is exposed
to much more coal) rather than in the regeneration step.
Sulfated and regenerated limestone particles from the first and
tenth utilization cycles were examined with a low magnification microscope
for macrofeatures. Photomicrographs of these samples are given in Fig. 43.
Limestone particles from the first cycle contain some adhering coal ash.
The regenerated particles appear to contain more ash, but this is probably
caused by the sharper color contrasts in the regenerated particles. (This is
more apparent in color photographs.) Also, the surface of the once-regenerated
-------
79
40
g30
CO
20
-K)
I I I I
A g ASH/IOOg VIRGIN SORBENT,
COMBUSTION
• g ASH/IOOg VIRGIN f
SORBENT, REGENERATION
A * *
A'^ ' '
A
g ASH/IOOg UTILIZED
SORBENT, COMBUSTION
o g ASH/IOOg UTILIZED ~
SORBENT, REGENERATION
COMBUSTION
o REGENERATION
1_ I I
I
468
UTILIZATION CYCLE
Fig. 42. Coal Ash Buildup as a
Function of Greer
Limestone Utilization
Cycle
10
limestone particles is glossier and whiter than the surface of the once-
sulfated particles. This is probably due to the sintering effect of the
regenerator environment (1100°C and reducing) on the silicon-rich Greer lime-
stone. Particles from the tenth utilization cycle sample appear to contain
more ash than do the first-cycle particles. However, the particles are not
all encapsulated with coal ash, as was the case for particles from the cyclic
dolomite '.experiments. Many of the tenth-cycle Greer limestone particles are
visually identical to first-cycle particles (unlike the results from the
cyclic dolomite experiments), which would indicate that the ash layer thickness
is not increasing and that much of the coal ash is present as individual
particles in the bulk utilized limestone. The results in Fig. 43 suggest
that the maxmimum ash buildup to be expected when using Greer limestone and
Sewickley coal is ^20 wt % in the utilized stone.
g. Attrition and Elutriation of Limestone Particles during Regeneration
and Sulfation
The fresh limestone makeup rate into an FBC boiler will depend on
the losses of limestone caused by attrition and subsequent elutriation of
particles from the boiler and the regenerator, and on the amount of sulfated
limestone drawn off to maintain the reactivity of the bed in the boiler. The
sorbent losses from attrition and elutriation of particles have been deter-
mined for the sulfation and regeneration steps and the data are given in Table
19. Losses were based on the ratio of sorbent-attributable calcium in off-
gas particles to calcium in the feed sorbent. The limestone losses caused by
-------
80
a. Cycle One, Sulfated Particles
b. Cycle One, Regenerated Particles
c. Cycle Ten, Sulfated Particles
d. Cycle Ten, Regenerated Particles
Fig. 43. Photomicrographs of Sulfated and Regenerated
Greer Limestone Particles (X24)
-------
81
Table 19. Losses of Greer Limestone Caused by Attrition
and Elutriation during Sulfation and Regener-
ation Steps in the Cyclic Utilization Study
A = Ca in feed limestone (sulfated or
regenerated), kg/hr
B = Ca in particles collected from off-gas,
kg/hr
Loss = 100 x -
A
Cycle No.
Limestone Loss, % during
Sulfation
Regeneration
1
2
3
4
5
6
7
8
9
10
20.0
12.0
9.2
7.4
8.6
8.6
4.3
3.8
2.6
4.9
2.9
0.6
-
1.3
_
2.9
2.0
1.6
2.4
1.5
Avg
8.2
1.9
attrition were ^2.0% during each regeneration step. During sulfation, the
losses were ^20% in the first cycle and steadily decreased to ^4% in the
final cycles. During the first sulfation cycle, the limestone calcined. The
sulfated product contained 0.9 wt % CQ2 in sulfated limestone; ^3% of the
available calcium was carbonate. No recarbonation occurred in the entire
cyclic experiment. The heavier attrition losses in the first sulfation
step can be attributed to calcination. In subsequent cycles, the resistance
of the particles to attrition increased because of (1) sulfated hardening
and (2) partial sintering which occurs at the regeneration temperature.
The losses during sulfation were slightly higher for the Greer
limestone cyclic experiment than for the Tymochtee dolomite experiment (Table
14). However, combustion operating conditions were different in these
experiments. The Greer limestone was fully calcined at the system pressure
of 308 kPa and a bed temperature of 855°C whereas the CaCO$ in the Tymochtee
dolomite was not at 810 kPa and 900°C.
The combined average losses caused by attrition and elutriation
per cycle were ^10%. Therefore, the fresh Greer limestone makeup rate is
expected to be at least ^10% to replenish losses. A higher makeup rate may
be required to maintain the S02~sorption reactivity in the fluidized bed of
the boiler.
-------
82
h. Total Cyclic Limestone Inventory
A total cyclic inventory history for the Greer limestone is contained
in Table 20. The total loss per cycle was ^12—^9.0% during sulfation and
^3.0% during regeneration. These losses, based on total inventory (^12%),
are higher than those reported in the preceding section (^10%) which were
based on steady-state mass balances. The total losses include all losses
during cyclic handling (spills, etc.).
Table 20. Total Reacted Limestone Inventory as a
Function of Cycle and Stage
Cycle
1
2
3
4
5
6
7
8
9
10
5:
s*
Stage
S*
Rb
S
R
S
R
S
R
S
R
S
R
S
R
S
R
S
R
S
R
sulfated.
regenerated.
Total
wt, Ib
468
374.0
393
328.0
325
280.0
280.1
204
245
203
204
192
186
164
166
143
149
132
136
124
Ca,
%
32.4
41.1
34.6
40.0
34.9
39.7
36.4
39.5
34.7
39.9
34.9
40.8
35.6
40.4
36.2
39.6
35.2
39.8
36.1
39.0
Ca,
Ib-moles
3.783
3.835
3.440
3.273
2.830
2.773
2.544
2.010
2.117
2.023
1.776
1.954
1.652
1.657
1.499
1.413
1.309
1.311
1.225
1.209
Loss,
%
1.4
10.3
4.9
13.5
2.01
8.3
21.0
5.3C
4.4
12.2
10.02C
15.5
3.0
9.5
5.7
7.4
0.2
6.6
1.3
Avg S 8.7
Avg R 3.4
"Negative losses due to experimental inaccuracy.
7. Regeneration Process Scale-up and Flowsheet Determination
(J. Montagna and G. Smith)
The development of the fluid-bed, reductive decomposition regeneration
process for sulfated limestones has been successful. A stage has been
reached at which the potential integration of sorbent regeneration with a
FBC boiler plant is being considered by ERDA. As the next step a 10-MWe
boiler, equivalent in size to a cell in the Rivesville pilot plant, is being
used as a basis for the calculations. The actual size and location of the
"next stage" regeneration facility has not been determined by ERDA.
-------
83
A process flowsheet for a FBC process with sorbent regeneration is
discussed. Tymochtee dolomite is the sorbent used. The performance of
Tymochtee dolomite as a function of cycle number was established in a ten-
cycle experiment; most of those results are reported in a preceding section.
Because the sulfation step in those experiments was performed at 8 atm
instead of 1 atm used in Rivesville, the obtained sulfation reactivity
function may be high for atmospheric boiler predictions. Those results on
reactivity, calcium utilization, ash buildup, and elutriation as a function
of cycle number have been incorporated into the ANL-developed regeneration
process model (ANL/ES-CEN-1016) .
Flow diagrams containing mass and energy flow streams have been obtained
for different process conditions. These calculations are intended to evaluate
the effect of makeup CaO/S feed rates (feed rate of virgin dolomite into the
system) to the boiler on the size of the regeneration system, on the SC>2 con-
centration in the regenerator off-gas, and on the fuel burden of the sorbent
regenerator on the boiler or power plant.
The following base conditions are assumed for the boiler: ^12.1 m
(130 ft2) gas distributor plate area, 3.05 m/s (10 ft/sec) fluidizing-gas
velocity, 3% excess oxygen in the flue gas, and combustion of 90.8 tonnes/
day (100 tons/day or T/D) of Sewickley coal, which contains 4.3 wt % sulfur
and has a heating value of 7,220 kcal/kg (13,000 Btru/lb) .
A process flowsheet for the above boiler conditions and a fresh sorbent
feed CaO/S ratio of 0.2 is given in Fig. 44. The combined (virgin plus
regenerated) dolomite CaO/S feed ratio is ^2.0. In the absence of regener-
ation, a CaO/S feed ratio of ^1.0 would be required for Tymochtee dolomite
based on reactivity data that was obtained at 8 atm.
The sulfated dolomite (^50 T/D) is assumed to be introduced into the
regenerator at 843°C (1550°F), the temperature in the fluid bed of the boiler.
Boiler flue gas and air are mixed to provide the required oxygen concentration
in the regenerator (17.2% 62), and a fluidizing-gas velocity sufficiently
high to prevent agglomeration of the fluidized bed of the reactor. The
fluidizing gas velocity of 1.37 m/s is ^12% above the velocity predicted
to be required to prevent agglomeration of sorbent having a mean size of
1500 pm (-1/8 in.) and regenerated at 1100°C with 2% reducing gas in the
regenerator off-gas. The fluidizing gas to the regenerator is assumed to be
heated to 843°C by recovering waste heat from the regenerator off-gas and
other process streams.
The coal consumption by the regenerator reactor with 843°C solids and
gas feed streams was estimated to be 2.9 T/D. This includes an equivalent
coal thermal credit of 0.55 T/D for the sensible heat which is carried by
the hot regenerated sorbent to the boiler. The S02 concentration in the
regenerator off-gas is predicted to be 9.5% (dry), and the gas distributor
area for the regenerator is predicted to be 0.66 m2 (7.1 ft2).
The makeup (fresh sorbent) feed rate in a regenerative system is
dependent on (1) the sorbent losses due to attrition and (2) losses in
sorbent reactivity with usage. The total feed rate of sorbent (fresh sorbent
-------
84
Atmosphere
Sulfur
1
r "*
Boiler -*^
(130 ff)
Fresh Stone Makeup
5.5 T/D
(CaO/S=0.2)
f
100 T/D 1
Coal
| 4.8 T/D
Boiler Flue Gas
1
9.5 X S02
43.9 T/D
1093°C
49.8 T/D
1 843°C
V
5.5 T/D
i
843°C
(17. 23! 0^)
850 T/D
Air
i
1 i
Air
<
_
Sulfur
Recovery
Regenerator
7.1 ftZ
_ — —
SO.-Lean
Off-Gas
|
2.9 T/D
^
Coal
18.0 T/D
Fig. 44. Process Flowsheet for a 10-MWe FBC Boiler
with Sorbent Regeneration, Using Tymochtee
Dolomite. Regeneration Conditions: T =
1100°C, Sorbent residence time = 7 min,
P = 1 atm, Extent of regeneration = 70%,
V = 1.37 m/s, Bed depth = 0.46 m (1.5 ft)
and regenerated sorbent) into the boiler is dependent on the combined
reactivity of the sorbent.
The effect of makeup (fresh sorbent) CaO/S mole feed ratio to the
boiler on the regeneration system was evaluated and is shown in Table 21.
Increasing the makeup CaO/S feed ratio from 0.16 (5% of total CaO/S feed)
to 0.28 (20% of total CaO/S feed) would cause (1) the mass rate of sulfated
stone that must be regenerated to decrease from 85 T/D to 32 T/D, (2) the
sorbent waste stream (combined elutriated and draw-off sorbent) to increase
from 4.5 T/D to 7.9 T/D, and (3) the size of the regeneration system to be
decreased by a factor of ^3. The coal required for the regeneration step
would decrease from 4.6 T/D to 2.6 T/D. (Boiler coal consumption is 100 T/D.)
The S02 concentration in the regenerator off-gas would increase from 6.6%
to 12% over the same range of CaO/S makeup ratios (0.16 to 0.28). Reducing
the power plant's fresh sorbent requirement (and its spent sorbent waste
stream) would increase the size of the regeneration and sulfur recovery
-------
Table 21.
Effect of Makeup CaO/S Mole Feed Ratio for Tymochte.e
Dolomite in Boiler ,on Regeneration System
FBC boiler)
Regeneration conditions:
T = 1100°C, Extent of
regeneration = 65%
P = 1 atm, Solids
residence time = 7 min
Boiler CaO/S
Mole Feed Ratio
Makeup
0.16
0.2
0.28
Total
3.2
2.0
1.4
Effluent Boiler
Mass Rate, T/D
Waste
Stream
4.5
5.5
7.9
Regenerator
Feed
85
50
32
Goal Required
for Hot Regen-
eration, T/D
4.6
2.9
2.6
Thermal Credit
for Regen.
Sorbent, T/D
1.0
0.55
0.32
Required 02
in Regen.
Feed Gas, Regenerator,
% Size, ft2
14.1
17.2
20.3
12
7.1
4.5
S02 Cone.
in Regen.
Off-Gas,
%
6.6
9.5
12.0
oo
Boiler coal consumption is 100 T/D.
Includes thermal credit for hot regenerated sorbent.
-------
86
system, would decrease the S02 concentration of the regenerator off-gas
(which would increase the cost of sulfur recovery), and would increase the
fuel burden of the regeneration step on the boiler.
The above technical relationships must be combined with economic consid-
erations in choosing a sorbent makeup rate. Economic evaluation is being
performed at ANL.
8. Regeneration System Modifications
(J. Montagna and F. F. Nunes)
Installation of a fluidized-bed sorbent preheater in the PDU sorbent
regeneration system has been completed. A new top flange for the regenerator
designed to accommodate the sorbent preheater has also been installed. A
schematic of the regeneration system with sorbent preheater is shown in Fig.
45.
In the absence of the preheater, the sensible heat required to raise
the temperature of the sulfated sorbent from room temperature to ^1100°C
(normal operating temperature of regenerator reactor) was a large part of
the total process thermal requirement. Since the sorbent would enter the
regenerator reactor at 'v870°C or higher in an industrial process, this is
unrealistic.
The fluidized-bed sorbent preheater will be used to evaluate the effect
of sorbent feed temperature on the regeneration process. Then the results
will be compared with predictions.
For startup, the sorbent preheater is electrically heated through the
metal shell (6-in. pipe), which is lined with a castable refractory (3-in.
ID) to permit high internal operating temperatures without exceeding the
design metal wall temperature of the reactor (538°C) . To provide the required
heat during operation, kerosene is combusted under oxidizing conditions in
the fluidized bed of the sorbent preheater.
The bed height in the sorbent preheater can be varied by allowing the
sorbent to exit at any one of four vertically oriented overflow locations.
This allows the sorbent residence time in the preheater to be varied while
the sorbent feed rate into the fluidized-bed regenerator is kept constant.
These parameters can now be used to establish the experimental conditions
for testing the filter in the ANL flue-gas system using the Ergun correlation.
The correlation should also prove helpful in translating results obtained in
the ANL flue-gas system at temperatures of 250-300°F and pressures of 8 atm
to higher temperature (^1600°F) conditions.
-------
87
Off-Gas
Analyzer
1-ln. Rupture Disc
0 ' * Vent
N2,Air
Rotary
Feeder
Transp.
Air
Fig. 45. Schematic of Regeneration System
with Sorbent Preheater
-------
88
TASK B. REGENERATION PROCESS ALTERNATIVES
No report for this period.
TASK C. SYNTHETIC SORBENTS FOR S02 EMISSION CONTROL
(R. B. Snyder, W. I. Wilson, and Irving Johnson)
1. Introduction
For f luidized-bed coal combustion, naturally occurring limestones and
dolomites are the principal calcium-bearing materials being considered for
the sorption of S02 . This is primarily due to their low cost and vast
reserves. However, these materials have some disadvantages. Attrition rates
may be excessively high, especially at high superficial gas velocities of
4-5 m/s. Also, during cyclic sulfation and regeneration, the reactivity of
the limestone sorbent with S02 will decrease. Finally, it may be determined
that regeneration of these sulfated sorbents is not economical, in which
case large quantities of limestones must be quarried and disposed of. Due
to these potential disadvantages, synthetic S02~sorbent materials were
investigated as an alternative to limestones.
It has been determined that CaO in a-A^Os is the most promising synthetic
S02 sorbent. 7 The preparation method and" reactivity of this synthetic
sorbent18 have been previously reported (ANL/ES/CEN-1016) .
The three subjects studied and reported below are: (1) attrition
resistance, (2) bauxite support material, and (3) cost.
2. Attrition Resistance
The attrition rate of a 10. 4% CaO in a-A!203 synthetic S02 sorbent was
compared with that of Tymochtee dolomite, half -calcined Tymochtee dolomite,
sulfated Tymochtee dolomite, and 1100°C heat-treated (H.T.) a-A!203 support
material for synthetic sorbents. All starting material for attrition tests
was in the size range, -14 +30 mesh. The test materials were screened before
and after each test to help determine the attrition mechanism. These attrition
tests were performed in a 5.08-cm-dia fluidized bed at room temperature. The
attrition rate was determined as the amount of material lost overhead in
10 hr. The attrition results are shown in Table 22. The synthetic sorbent
had one-fifth the attrition rate of sulfated Tymochtee dolomite, only 0.6%
being lost. The uncalcined and half-calcined dolomite had high attrition
rates — 23 and 47%, respectively. The a-Al203 support material had a 3%
material loss in 10 hr; this indicates that impregnation with CaO to form
calcium aluminates hardens the material, making it more attrition-resistant.
The results are shown in Figs. 46-48 for (1) granular a-A!203 that had
been heat-treated (H.T.) at 1100°C, (2) sulfated dolomite, and (3) regenerated
dolomite for which results are not shown in Table 22. The weight percent of
material is given for various particle diameters (1410 to 30 ym) .
-------
89
50
i r
•Original distribution
•Distribution after 10-hr attrition test
200 300 600
PARTICLE DIAMETER,
1000
2000
Fig. 46.
Particle Size Distribution of ct-Al203
(heat-treated at 1100°C) Before and
After Attrition Test
3. Bauxite Support
Due to the high cost of synthetic sorbent, CaO in a-Al^Os (see below),
bauxite, a less expensive material, was tested as a support material. The
bauxite used in these experiments was approximately 70% A1203-30% Si02 and was
obtained from Harbinson-Walker Refractories, Chicago. The bauxite was
heat-treated at 1100°C for 6 hr or at 1500°C for 8 hr in an attempt to
enlarge the pore diameters in the support and to stabilize grain growth.
The porosity curves for the bauxite heat-treated at the two temperatures
are shown in Fig. 49, where they are compared with bauxite dried at 110°C
for 2 hr. Heat-treating at 1100°C increased the total porosity and increased
the average pore diameter. However, heat-treating at 1500°C caused particle
shrinkage and a loss of porosity.
Two synthetic sorbents were prepared from the heat-treated bauxites
and were tested for reactivity. The conversion of CaO to CaSO^ for the 13.2%
CaO in 1100°C H.T. bauxite and for the 14.2% CaO in 1500°C H.T. bauxite are
shown in Fig. 50. The bauxite-supported sorbents are compared with a 14.8%
CaO in a-A!203 sorbent and with Greer limestone. The bauxite sorbents had
the poorest performance—only conversion of ^31% of the calcium to CaS04.
-------
90
Table 22. Fluidized-Bed Attrition Experiments
L/D = 1.38; Fluidizing Gas Velocity =
0.6 m/s
Run
1
Sample
Tymochtee Dolomite
Particles in
Bed after
10 hr, %
52.6
Percent
Loss3 in
10 hr
47
(half-calcined)
(-14 +30)
2 Sulfated Dolomite 96.7 3
3 Tymochtee Dolomite 77 23
(uncalcined)
4 1100°C H.T. granular 97.4 3
a-A!203 support
5 10.4% CaO in 99.4 0.6
granular a-Al203
support
o
The percent loss determined as grams of overhead material x
100/gram original material. All overhead material was smaller
than 70 mesh.
The 1100°C H.T. a-A'!203 had only a 3% materials loss in 10 hr of
attrition, and therefore in Fig. 46, only a slight change in particle
diameter distribution is seen. The amounts of original material between
18 and 16 mesh (1000 ym, 1.3%) and between 16 and 14 mesh (1190 ym, 0.4%)
were small. In contrast, the sulfated dolomite distribution was skewed
toward large particle diameters (Fig. 47). This material also had a 3%
material loss (Table 22). In this graph, one can see the slight shift in
distribution to smaller particles. Figures 46 and 47 tend to indicate
that the mechanism for attrition is abrasion (the wearing away of surface
material), in contrast to the breakup or splitting of particles due to
particle-particle or particle-wall collision.
Figure 48 confirms that abrasion is the mechanism of material loss.
This graph shows the particle size distributions for dolomite before and
after a 10-hr attrition test. Material lost from the bed, which was col-
lected on a filter, constituted 50% of the bed. This overhead material had
very small particle diameters (37-74 ym) . Very little overhead material
was in the particle diameter range from 200 to 300 ym. There was no bed
material left in the final bed with diameters below 400 ym. If there had
been particle splitting, one would expect to find not a bimodal distribution,
but a shift of all particles to smaller sizes. The bimodal distribution
found indicates that abrasion was occurring and that the particles attrited
had a diameter range of 30 to 80 ym. Figures 46 and 47 show no bimodal
distribution for the material after attrition since only a very small quantity
of material was lost.
-------
91
CO
LU
_J
o
»—
QC.
-------
92
DRIED IIO°C
2hr
5 0.5
PORE DIAMETER,
Fig. 49. Porosity of Bauxite
0.05
0.005
CJ>
en
100
80
60
40
20
GREER LIMESTONE
° 13.22% CaO IN BAUXITE (70% AI203 30% SiOo,
H.T. IIOO°C, 6hr)
a 14.8% CaO IN a-AI203
• 14.2% CaO IN BAUXITE
(H. T. I500°C, 8hr)
0
0
Fig. 50.
234567
TIME, hr
Calcium Utilization of Synthetic Sorbents
8
-------
93
In Fig. 51, the number of grams of 803 captured per kilogram of sorbent
material is shown as a function of time. In 4 hr, the bauxite-supported
sorbent captured only 50 g of 803—in comparison, Greer limestone captured
190 g, almost four times as much. The poor reactivity, compared with that
of CaO in cc-A^Os sorbent, is probably due to the low reactivity of calcium
silicates in the bauxite.
oo
cc
O
CO
ro
O
CO
300
250
200
150
100
50
0
— GREER
° 13.22% CaO IN BAUXITE (70% AI203 30% Si02,"
H.T. IIOO°C, 6hr)
D 14.8% CaO IN a-AI203
• 14.2% CaO IN BAUXITE
(H.T. !500°C,8hr)
0
345
TIME, hr
Fig. 51. Rate of 503 Capture by Sorbents
Because of the poor performance of sorbents prepared from bauxite and
the loss of porosity upon heat treatment at 1500°C, these sorbents are con-
cluded to be unacceptable.
4. Cpst_
The main incentive for developing synthetic sorbents is to reduce the
environmental impact of the S02-emission control systems for fluidized-bed
coal combustion systems. In this section, the cost of using synthetic
sorbents will be compared with the cost of a once-through system using a
natural limestone.
-------
94
Synthetic sorbents of the type developed in this study will be more
costly than natural limestones. No large-scale preparational method for
synthetic sorbents has been devised and tested, and the laboratory method
can be used as the basis for an estimation of the cost. If refined hydrated
alumina (A^C^'^O) is used as a source of the alumina, then to prepare one
ton of a 20% CaOA^C^ sorbent, about 0.94 ton of alumina would be needed
which at $118 per ton would cost $111. Two-tenths of one ton of chemical
lime at $25 per ton would add $5 to the cost. About 900 Ib of nitric acid
at $4.50 per 100 Ib would add another $40 to the material cost, yielding
a total material cost of $156 per ton.
Hydrated alumina must be heat treated to form porous alumina support,
treated with Ca(N03)£ solution, dried, and given a final heat treatment.
These steps are estimated to cost about $100 per ton. Thus, the final
synthetic sorbent would be about $250 per ton. Since additional development
of the process should lower the cost, a cost of about $200 per ton does not
seem to be unrealistic at the present time. This cost is about 20 times
that of a natural limestone, which is assumed to be about $10 per ton de-
livered to the FBC site and disposal after sulfation as landfill.
The total cost of SC>2 emission control may be computed using the
equation:
DC
Total cost per kWh(e) = -^ [F$ + (1 - F)G] (1)
u
where:
R = Ca/S (mole) ratio needed to achieve the EPA S02 emission level
S = moles of sulfur generated per kWh(e)
C = moles of calcium per kg of sorbent
F = fraction of new sorbent which must be fed per kWh(e)
$ = cost of new sorbent, mills per kg
G = regeneration cost, mills per kg
The G factor will depend on the size of the regeneration plant, which will
depend on the recycle rate and the size of the FBC. The numerical value
of G is not known. An estimate for a pressurized fluidized-bed combustion
system, made by Westinghouse in 1975, indicated that the regeneration
system would added about 3 mills per kWh(e) to the electrical power cost.
The cost for an atmospheric FBC would be expected to be less than this—
probably of the order of 1.5 to 2 mills per kWh(e).
To generate 1 kWh(e) of power requires about 0.8 Ib of coal per hour.
If this coal contains 4% sulfur and has a heating value of 12,200 Btu/lb,
about 82% of the sulfur needs to be removed from the flue gas to meet the
EPA standard of 0.6 Ib of sulfur emission per 106 Btu. Thus for our system,
0.8 x 0.04 = 0.032 Ib of sulfur will be produced for each kWh(e), or 0.45
mole per kWh(e), the value of S. If Greer limestone is used in an AFBC, a
Ca/S ratio, R, of 3.1 is needed if the FBC is operated at 816-832°C with a
3.8-4.7 m/s superficial gas velocity. Greer limestone is a high SC-2 reactivity
stone which has been used for many studies in large FBC test rigs. Greer
-------
9.5
limestone contains 80.4% CaC03 and hence the value of C is 8.04 moles of
calcium per kg. The cost for a once-through system as obtained from Eq. 1
will be about 1.9 mills/ per kWh(e). Note that F = 1 in this case (i.e.,
there is no recycle) and the cost of sorbent, $, is 11 mills per kg. The
latter sorbent cost includes quarrying, shipping and disposal.
The Ca/S (R) ratio needed for a 20% CaO-Al203 synthetic sorbent must
be estimated from a comparison of the TGA curves (Fig. 51) for the synthetic
sorbent with a similar curve for Greer limestone. When this is done, it is
estimated that an R value of 5 would be needed to achieve the 82% sulfur
removal required to meet the EPA standard. This high value reflects the
rather poor SC>2 capacity of the synthetic sorbent in comparison to a good
natural limestone. Since the sorb'ent contains only 20% CaO, the value of
C is 3.57 moles per kg.
Although the attrition rate measured for a synthetic sorbent would
indicate that over 100 cycles would be possible, it seems best to assume
a lower value. We have therefore assumed that 25 cycles can be achieved.
On this basis, F is 0.04 and the sorbent cost when computed using Eq. 1 is
5.6 mills per kWh(e) or about three times the cost of one-through sorbent.
To obtain the total cost of using the synthetic sorbent !~he regeneration
cost must also be added. As noted above, this will probably add another
1.5 to 2 mills per kWh(e) to the cost, leading to a synthetic sorbent cost
of about 7 mills per kWh(e) compared with about 2 mills/kWh(e) for once-
through natural stone with the assumptions used above.
If one assumes that the environmental impact for the two systems may
be taken to be proportional to the sum of the quantities of fresh sorbent
feed and waste. In the case of the once-through system, the sum is 7.2
tonne/Mwd(e); for the synthetic sorbent, the sum is 1.2 tonne/Mwd(e). Thus
a six-fold decrease in environmental impact would have an additional energy
cost of about 5 mills/kWh(e). It is our opinion that this decrease in
environmental impact is not sufficient to justify the additional energy
cost.
Examination of Eq. 1 shows where any future work on synthetic sorbents
should be concentrated. Two factors appear amenable to change, the S02~
reactivity as represented by the Ca/S ratio, R, and the sorbent cost, $.
It should be possible, by better control of the porosity and heat treatment,
to obtain a synthetic sorbent at least as good as the best natural limestone.
The cost of the synthetic sorbent is probably where the greatest improvement
needs to be made. Although our limited experiments using a less expensive
support material (the bauxite studies) did not yield high-reactivity sorbent,
the results point to the direction where future studies should be concentrated.
It is also clear from the results of this study that inexpensive starting
materials must be used for synthetic sorbents since limestone is one of the
least costly materials available.
5. Conclusions
The synthetic sorbent, CaO in ct-Al203, presently is the most promising
synthetic sorbent for use in fluidized-bed coal combustors to minimize S02
-------
96
emissions. 1^0 and Na20 in alumina have higher sulfation rates; however,
their products (sulfates) decompose in the temperature range of interest
for combustion and regeneration. These compounds contribute to corrosion of
metals at high temperatures. Calcium oxide is the least expensive of the
metal oxides tested. CaO sorbents capture more 862 per unit weight of sorbent
than do barium oxide or strontium oxide; therefore, calcium ox.i.de is the
metal oxide of choice. The choice of cx-Al203 as the support material was
based on its stability in the temperature region being considered. The calcium
aluminates also add to the mechanical strength of the sorbent, minimizing
attrition.
The sulfation rate of CaO in a-Al2C>3 sorbent is highly dependent on its
physical properties, particularly the pore size distribution. Both pore
diffusion and gas-solid diffusion control the rate of SC>2 capture by the
sorbent; therefore, large pores in the support (with diameter larger than
about 0.2 pm) are beneficial. Sorbents containing higher CaO concentrations
are less porous. However, the optimum CaO concentration depends on the
residence time required in the combustor and the CaO concentration effect
on sorbent strength, since the quantity of S02 captured per unit weight of
sorbents in a given time is independent of the sorbent"s CaO concentration.
Natural sorbents (dolomite or limestone) contain higher concentrations
of CaO than do the synthetic sorbents and therefore capture more S02 per
unit weight of material. Thus, more synthetic sorbent material may be
needed for each pass through the combustor (possibly twice as much). However,
with regeneration of synthetic sorbent, the overall consumption of synthetic
sorbent should be lower.
The estimated large increase in the cost of electricity to obtain the
moderate decrease in environmental impact that would be achieved by using
synthetic sorbents is considered to be unacceptable. Therefore, at this
time, synthetic sorbents are not believed to be a viable option for reducing
S02 emissions from a fluidized-bed coal combustor. A final topical report
on these studies has been prepared (ANL/CEN/FE-77-4).
-------
97
TASK D. LIMESTONE CHARACTERIZATION
(R. Snyder and I. Wilson)
A research program is under way to characterize limestone for fluidized-
bed coal conversion plants. That is, the reactivity of limestone with S02
under various environmental conditions is being determined and correlated
with limestone physical properties. Pretreatment (precalcination and heat
treatment) is being investigated and the mechanism of S02 capture is being
studied. Finally, the attrition rate of the various limestones in fluidized
beds was determined.
1• Limestone Properties Affecting S0? Reactivity
Ten limestones and dolomites were fully calcined at 900°C in 20% C02-
80% N2 in a thermogravimetric analyzer (TGA), after which they were immediately
reacted at 900°C with a 0.3% S02-5% 02-20% C02-balance N2 synthetic combustion
gas mixture in the same apparatus. The results are shown in Fig. 52.
100
+— +—1343
D : Q-1336
Q- 1359 AND
DOLDWHITE
Fig. 52. Calcium Utilization at 900°C for Ten Precalcined
(at 900°C) Limestones in a Thermogravimetric
Analyzer. Sulfation gas composition: 0.3% S02,
5% 02, 20% C02, and balance N2.
-------
98
Different stones vary greatly in (1) reactivity of the CaO with S02
and (2) calcium utilization. In 5 hr of sulfation, 95% of the calcium (as
CaO) in Tymochtee dolomite was converted to CaS04, whereas for dolowhite
and limestone 1359, only 15% of the calcium was utilized.
To be able to predict the rate of SC>2 capture or calcium utilization
of a given limestone, one must determine which limestone physical properties
affect S02 reactivity. Below (in Table 23), the chemical compositions of
the limestones and dolomites studied are given. The stones are listed in
order from highest to lowest calcium utilization after 5 hr of reaction.
Since MgC03 converts to MgO on calcination, helping to make the material
porous, one can also attempt to correlate MgC03 content with calcium utiliza-
tion. There is a trend of greater calcium utilization with higher MgC03
concentration except for dolowhite, which has a high MgC03 concentration,
44.4% but has a low calcium utilization. Additional limestones with high
magnesium contents will be tested to determine if dolowhite is an anomaly.
Table 23. Chemical Compositions of Limestones (listed in
order of highest to lowest calcium utilization
after 5-hr reaction time)
Composition, wt %
Tymochtee dolomite
Limestone 1337
Limestone 1351
Limestone 1360
Greer limestone
Limestone 2203
Limestone 1343
Limestone 1336
Limestone 1359
Dolowhite
Ca
51
53
61
81
80
96
89
92
95
55
C03
.5
.4
.2
.6
.4
.0
.8
.6
.3
.2
MgC03
43.0
45.4
28.7
11.6
3.5
3.6
2.2
5.3
1.3
44.4
Si02
3.
0.
3.
1.
10.
0.
4.
1.
0.
0.
6
7
2
9
3
2
0
3
8
2
A1203
1.
0.
0.
0.
3.
0.
1.
0.
0.
0.
5
08
5
2
2
01
0
4
3
01
Fe203
0.4
0.07
5.6
0.9
1.2
0.2
0.7
0.2
0.1
0.09
Na?0
0
0
0
0
0
0
0
0
0
0
.07
.08
.13
.10
.23
.04
.1
.1
.03
.02
H20
0
0.3
0.67
3.7
1.17
0
2.2
0.1
2.2
0.1
High silica content also may affect calcium utilization, possibly
decreasing calcium utilization due to the formation of stable calcium sili-
cates. However, Greer limestone does not seem to be strongly affected by
its unusually high Si02 content of 10.3%. High sodium content would be
expected to increase reactivity; however, no trend was observable from Table
23 data. Apparently, calcium utilization can not be predicted from chemical
composition alone.
Unpublished data obtained at ANL indicate that calcium utilization is
more likely to correlate with the surface area of CaO in "sufficiently large
pores" than with chemical composition. Therefore, porosity measurements were
-------
99
performed on all of the limestones to determine what is a "sufficiently large
pore." From a porosity measurement, one also obtains the cumulative volume
of pores which have a maximum diameter between 100 urn and a specified
smaller size. From this information, the internal surface area can be
obtained as a function of decreasing pore diameter.
The environmental history of a limestone sample affects its porosity
curve. The porosity curves were determined on natural limestone — 18 to 20
mesh material which had been calcined at 900°C for 15 min in a 20% C02-80%
N2 atmosphere. Changing of any of the above conditions would produce
different results which would most likely prevent meaningful interpretation.
The reasons for selecting the conditions used are as follows: Different
sizes of particles have different reaction rates and different porosities.
To minimize the effect of particle size distribution, a narrow size range
was chosen — the starting material was virgin limestone which had been sieved
to 18-20 mesh. Porosity determinations were made on calcined limestones
prepared from the same lot as were the calcined 18-20 mesh limestones (sul-
fated on the TGA) , and the porosities of these two materials were compared.
The calcining conditions prior to porosity measurements must exactly
reproduce the calcining conditions used on the TGA because (as is well
known) the rate of calcination is affected by temperature and C02 concentra-
tion. The rate of calcination affects the pore size distribution and the
crystal structure of the limestone and hence its reactivity. Since the
sulfation reactions (for which results are given in Fig. 52) included
calcination at 900°C in 20% CC>2 , these same conditions were chosen for
calcining the limestones prior to porosity measurements. Porosity measure-
ments were made immediately after a sample was calcined since CaO is an
excellent desiccant and reacts with water to form calcium hydroxide, causing
a loss in total porosity.
Figure 53 shows the porosity curves for the ten calcined limestones.
Limestone 1360 had the greatest total porosity (0.75 cm3/g) , and it had more
internal pore volume contributed by large pores (pore diameter, >0.2 urn)
than did the other stones. Limestone 1336 had the least total porosity
(0.31 cm3/g). However, total porosity does not correlate with reactivity
nor with calcium utilization. Most of these limestones are uni-modal, having
only one major size cluster of pores. For example, the diameters of a
majority of limestone 1337 pores range from 0.035 to 0.07 ym. Dolowhite
has a low reactivity and its pore cluster occurs at pore diameters that
are smaller (0.025-0.045 pm) than for the other stones. This implies that
these small pores plug or close off quickly during sulfation due to the
formation of
It is concluded that there is a minimum size pore whose surface area
of CaO is reactive with S02 — that is, there is a minimum effective pore
diameter. Below this size, pores will close off prematurely during sulfation
since CaSOi, has a larger molar volume than does CaO. The cumulative surface
area was determined for all pores with diameters larger than each of several
specified minimum diameters. Correlation of calcium utilization with this
cumulative surface area was then attempted. A correlation was not observed
-------
100.
0.65
0.60
0.55
0.50
0.45
0.40
0.35
0.30
0.25
0.20
0.15
0.10
0.05
I - TYMOCHTEE
2 -1337
3-1351
4-1360
5-GREER
6-2203
7-1343
8-1336
9-1359
IO-DOLOWHITE
100 50
20 10
2 1.0 0.5 0.2 O.I
PORE DIAMETER, Mm
0.05 0.02 001 0.005
Fig. 53. Cumulative Pore Volume as a Function of
Pore Diameter for Ten Limestones
when a specified minimum pore diameter was used for each limestone. It is
highly unlikely that all limestones have the same minimum effective pore
diameter. In fact, one would expect that the greater the calcium content
of the limestone, the larger the pores would have to be in order not to
prematurely plug with CaSOi,.
The inert materials (Si02, Al203, Fe203) probably do not affect plugging
behavior since they are not part of the calcium-magnesium crystal structure.
Therefore, a minimum pore diameter, MPD, was determined based on the Mg/Ca
ratio in the limestone.
MPD =
174
5400
%MgC03
(1)
CaC03
MgC03
+ 1000
where
MPD = minimum pore diameter, pm
% MgC03 = wt % MgC03 in virgin limestone
% CaC03 = wt % CaC03 in virgin limestone
The MPD was assumed to be linear with respect to the limestone's Mg/Ca
ratio. It was also assumed that for a pure limestone, Mg/Ca = 0, the MPD is
-------
101
0.174 pm, which is its value at 1000 psi during porosimeter measurements.
Borgwardt and Harvey20 have indicated that all calcined limestones with
pores larger than 0.2 pm have high reactivity with S02.
Since dolowhite is nonreactive, it was assumed that the MPD for dolowhite
is 0.05 urn, which reduces the effective surface area of dolowhite for
reactivity to approximately 6500 cm2/g.
By use of the above equation, the calcium utilization of various lime-
stones was plotted against the surface area of pores having a diameter
larger than the MPD. These results are shown in Fig. 54. This correlation
applies reasonably well to all stones.
2. Effects of Precalcination and Heat Treatment
In atmospheric fluidized-bed coal combustion, large quantities of
limestone sorbent (Ca/S ratio of 4/1 to 6/1) may be required so that the
flue gas will meet EPA S02-emission standards. Pretreatment of limestones
to enhance their reactivity and their calcium utilization may reduce
limestone requirements. Therefore, the effect of calcination-heat treat-
ment on calcium utilization of Greer limestone has been studied and a
preliminary economic-environmental impact assessment made.
Greer limestone was precalcined at 900°C in a 20% C02~80% N2 gas stream,
then heat-treated at 900°C for 0, 2, 6, and 22.2 hr. The pretreated Greer
limestones were then sulfated at 900°C on a TGA, using 0.3% S02-5% 02 in
N2.
100
I- TYMOCHTEE
2-1337
3-1351
4-1360
5-GREER
6-2203
7-1343
8-1336
9-1359
10-OOLOWHITE
60 80 100
SURFACE AREA I03cm2/g
Fig. 54. Calcium Utilization as a Function of Surface
Area of Pores having Diameters Larger than
MPD for Ten Limestones
-------
102
In Fig. 55, the percent conversion of CaO to CaSOi4 is given as a
function of sulfation time for the various pretreated limestones. Their CaO
conversions are compared with that of Greer limestone which was simultaneously
calcined and sulfated with 0.3% S02-5% 02-20% C02 in N2 gas. The Greer
limestone which was simultaneously calcined and sulfated had the poorest
conversion—only 28% of the CaO was utilized to capture S02. This result
is in excellent agreement with the data obtained on Greer limestone by Pope,
Evans and Robbins in a 9 ft2 atmospheric combustor. They reported calcium
utilizations of 25 to 28 percent with a limestone residence time of approxi-
mately 4 hr.
The Greer limestone tested in the TGA (Fig. 55) calcined completely in
approximately 10 min. This was one-half the calcination time for the lime-
stone which had been simultaneously calcined and sulfated. The lower calcin-
ation rate during simultaneous calcination-sulfation probably produces a
limestone having smaller pores and thereby results in less calcium utilization.
Precalcination increased the calcium utilization from 28% to 43% for a
5-hr sulfation time. Heat treating the limestone for 2 hr after a 5-min
precalcination further increased the calcium utilization from 43% to 48% (a
residence time of 15 hr gave a 51% calcium utilization). Heat treating for
6 hr and 22.2 hr gave calcium utilizations of 52 and 55% in 5 hr. The pre-
calcination and heat treatments definitely increased the sulfation rate and
calcium utilization. However, after heat treating for 6 hr, additional heat
treatment gave a minimal change in sorbent performance.
o
O
O
o
O
O
C/}
80
70
60
50
40
30
20
10
0
O-PRECALCINED GREER LIMESTONE
_ D-PRECALCINED, 2hr H.T. GREER LIMESTONE
I-PRECALCINED, 6hr H.T. GREER LIMESTONE
• -PRECALCINED, 22.2 hr H.T. GREER LIMESTONE
- •-SIMULTANEOUS CALCINATION-SULFATION
4-PER 857°C, MONTHLY _ __., IP—.J-
- REPORT NO. 39
0
I I I I I I I I I
I I
I I
23456
SULFATION TIME, hr
Fig. 55. Calcium Utilization as a Function of Various
Pretreatments. Sulfation at 900°C with 0.3%
S02-5% 02 in N2
-------
103
To apply the results shown in Fig. 55 to a practical fluidized-bed
combustion system, a preliminary economic analysis was performed to determine
the cost of a precalcination-heat treatment step. The economic analysis was
made for a 600-MW atmospheric fluidized-bed coal combustor.
The pretreatment costs were based on the costs quoted by the Kennedy
Van Saun Corporation.22 A kiln (calciner and heat treating unit) was
estimated to cost $5 million/(1000 tons of calcined product per day). A
limestone residence time of 178 min was assumed> This unit requires two
operators per shift and also requires 5,000,000 Btu of energy per ton of
product (calcined, heat-treated limestone). From this information, a first
order economic analysis was performed to determine the cost of a pretreater
for various limestone residence times and for various quantities of limestone
per day.
Five costs were included in the analysis: capital, installation,
operating, maintenance, and fuel costs. The capital cost was estimated to
be a function of the limestone residence time and lime production rate to
the 0.6 power. The installation cost was assumed to be twice the capital
cost. On the basis of two men per shift, the operating cost was estimated
to be 300,000 dollars per year. Maintenance cost and fuel cost are directly
a function of the lime production rate. Limestone requirements were estimated
to be 1720 tons/day for a plant using a Ca/S ratio of 4. For heat treatment
times longer than 2 hr, the capital cost (installed) is the largest cost;
the next largest is fuel cost.
From the above information, the increased energy cost can be estimated.
If a 2-hr residence time and a Ca/S ratio of 4 are assumed, the increased
energy cost of precalcination and heat treatment would be 0.85 mill/kWh.
That is, it would cost 0.85 mill/kWh to decrease the environmental impact
by 42 percent (Fig. 56).
Figure 56 illustrates the energy cost for reducing the environmental
impact of Greer limestone. The environmental impact of Greer limestone with
no precalcination treatment was arbitrarily set at 1.0. This can be con-
verted to a given quantity of limestone that must be mined and disposed of.
As shown in Fig. 56, the environmental impact can be reduced by increasing
the energy cost (by means of pretreatment and thus greater calcium utili-
zation). However, decreasing the limestone requirements further would
require a large increase in energy cost. In fact, however much is spent,
the environmental impact can not be reduced more than 50% by heat treating.
The above analysis is only for Greer limestone which has been tested at
900°C. Also, the cost analysis for pretreatment kilns is only a first
approximation, and the effect of pretreatment on attrition has not yet been
determined. Nevertheless, this is the type of cost us environmental impact
information needed for assessment of the viability of pretreatment.
-------
104
co
2. Co/S =
3. Ca/S = 6
0
0.4
0.5 0.6 0.7 0.8 0.9
ENVIRONMENTAL IMPACT, ARBITRARY UNITS
Fig. 56. Increased Energy Cost for Pretreatment
Required to Reduce the Environmental
Impact of Mining and Disposal of
Sorbents
3. Limestone Attrition
The effects of fluidization velocity, the L/D ratio of the bed, and
stone composition on limestone attrition rates are being studied in a room
temperature (cold) fluidized bed. Attempts will be made to correlate the
results with limestone attrition rates in fluidized-bed coal combustors.
Also, the mechanism of attrition is being studied. In addition to the small
room-temperature fluidized-bed test rig used in this work, a small 2-in.-dia
high-temperature (850°C) fluidized-bed test unit is being constructed to
determine the effects of calcination and continuous sulfation on the attrition
rates of various limestones.
High attrition rates are undesirable since they may decrease SC>2
retention, require more or larger equipment to capture the higher dust loadings
(in order to meet EPA emission standards), and necessitate an increase in
the limestone fresh feed rate. In the case of pressurized fluidized-bed
combustion, the dust loading must be low to meet turbine requirements.
The "cold" fluidized bed apparatus is schematically depicted in Fig. 57.
The column has a 78.74-cm height which can be increased or decreased by
adding or removing sections. In the present assembled apparatus, the bottom
section is 30.48 cm high and has a 5.08-cm diameter. The top section
-------
10.5
" 3LI
FILTER CYCLONE
48.26cm
\
763-cm
ID
30.48 cm
POROUS-^ V
METAL PLATE *
5.08-cm
ID
•ROTAMETER
AIRORN,
INPUT
'PRESSURE
GAUGE
PRESSURE
GAUGE
Fig. 57. Laboratory-Scale Fluidized-Bed Apparatus
for Attrition Experiments
(removable) is 48.26 cm high ana has a 7.63-cm diameter. The top portion
of the bed is a disengagement section, which functions to decrease the
velocity of the fluidizing gas and particles, minimizing particle entrain-
ment.
At the bottom of the bed, there is a porous metal plate that functions
as a gas distributor. A pressure regulator controls the fluidizing gas flow.
A cyclone and a filter are located downstream from the bed to collect the
overhead particles produced from particle-particle and particle-wall
collisions (abrasion) during fluidization.
The sorbent is loaded into the apparatus after sections of the bed have
been separated. The fluidizing gas fed into the bed is house air or nitrogen.
The percent of the bed material that is elutriated overhead is considered
the material loss due to attrition. During each 10-hr attrition test,
the bed material is periodically weighed to determine the quantity of material
that has been lost overhead.
-------
106
Many parameters affect the attrition rate of limestones (fluidization
velocity, bed depth, tube arrangement, particle size, limestone composition,
calcination rate, extent of sulfation, temperature). Although there is a
number of variables involved in determining attrition rates, Tymochtee
dolomite, fully calcined, -14 +30 mesh, was initially studied as a function
of only three variables: fluidization velocity, bed depth, and tube arrange-
ment .
The attrition rates for calcined Tymochtee dolomite (no tubes in the
bed) are shown in Figs. 58 for a L/D of 1.38 and fluidization velocities
of 0.88, 1.19, 1.46, and 2.13 m/s. As can be seen in the figure, as the
superficial gas velocity increases, the attrition rate increases. At all
gas velocities, the attrition rate is high initially, then decreases to a
steady state rate in approximately 2-4 hr. The steady state attrition rate
is approximately proportional to the superficial velocity squared; which is
the theoretical attrition dependency on velocity. The results for L/D ratios
of 0.28 and 0.55 are for well-mixed fluidized beds. For a L/D of 1.38, the
fluidized bed is in the transition region between well-mixed and slugging;
for a L/D of 2.2, the operates entirely in a slugging mode. The effect of
bed depth on attrition is shown in Fig. 59 for a gas velocity of 1.46 m/s.
In deeper beds, where there apparently are more particle-particle collisions
per kg of limestone, attrition rates are higher. At 2.13 m/s, the attrition
rate is independent of L/D since an entrained bed develops (for -14 +30 mesh
particles), causing very high attrition rates.
The introduction of cooling coils (simulated by using copper tubing)
was expected to decrease the limestone particle velocity, "quieting" the
bed and thereby decreasing limestone attrition rates. There were six layers
of horizontal tubes, six tubes at each level, and a 1.2-cm spacing between
adjacent levels. The first layer of tubes was 1.2 cm above the gas distri-
butor plate. At the lower velocities of 1.19 and 1.46 m/s, the attrition
rates were decreased by a factor of 2.8 and 2, respectively. At 2.13 m/s,
an entrained bed develops, and thus the attrition rate is the same in the
presence and absence of simulated cooling coils.
The attrition rates of ten limestones (precalcined) were tested at room
temperature, a superficial gas velocity of 1.46 m/s, and a L/D of 1.38
(expanded). Figure 60 shows the total material loss in 10 hr. The limestones
are in the order of decreasing reactivity with S02 from left to right, as
determined in a TGA. No relationship between S02 capacity and attrition
is seen.
Figure 61 shows the amount of material lost overhead as a function
of time. The attrition rate is high for the first 1/2 hr, then decreases
rapidly. Of the ten limestones tested, limestone 1337 had the highest
attrition rate—a loss of 55% of the bed material. Greer was the strongest,
losing only 4.5%. The overhead material was finely powdered (smaller than
70 mesh), indicating that the material lost from the bed was due to attrition,
not elutriation.
The wide variation in attrition resistance can be correlated with the
impurity (Si, Fe, Al) concentrations of the limestones. In Fig. 62, the
-------
GAS VELOCITY, m/s
468
TIME, hr
Fig. 58. Attrition Rate of Calcined Tymochtee
Dolomite as a Function of buperficial
Gas Velocity; L/D =1.38
0.55
1.38
2.20
468
TIME, hr
Fig. 59. Attrition Rate of Calcined Tymochtee
Dolomite as a Function of Bed Depth.
Superficial gas velocity, 1.46 m/s
-------
10.8
DU
O
50
CO
8 40
I 30
LU
\—
^ 20
o in
— IU
• •
i —
i—
2 can be correlated with
composition (MgO content); however, more limestones samples must be tested
to determine how accurate this correlation is.
The calcium utilization of limestones can also be correlated with the
"accessible" internal surface area of the stone. Precalcination of limestones
may increase the calcium utilization of the stone; however if residence
times in a preheater are long, the capital cost of pretreatment equipment
may become too high for the procedure to be considered.
-------
109
60
50 -
-40 -
CO
CO
O
a:
LU
£
30 -
20
10
0
0
Fig. 61.
1337
1360
2203
1336
DOLOWHITE
TYMOCHTEE
1351
1343
'D
GREER
4 6
TIME, hr
8
10
Material Loss at 1.46 m/s Gas Velocity, Room
Temperature.
Attrition rates of limestones can be correlated with the concentrations
of impurities (Al, Si, Fe). High impurity concentration levels increase
attrition resistance.
-------
110
60
j= 50
o
S 40
CO
CO
o
CO
30
5 20
10
0
1337
2203
9
»I360
1336
DOLOWHITE
»TYMOCHTEE
»I359
"1343
0 2 4 6 8 10 12 14 16
IMPURITIES (S\, Fe, Al), %
Fig. 62. Effect of Impurity Concentration on Attrition
-------
.Ill
TASK E. TRACE ELEMENTS AND COMBUSTION EMISSION STUDIES
1. The Effect of Additives on the Calcination/Sulfation of Limestone/Dolomite
(J. Shearer and C. Turner)
The use of additives in a fluidized-bed coal combustion system to increase
the S02~sorption capabilities of limestone and dolomite has received consid-
erable attention. In particular, the addition of small amounts of NaCl to
enhance the S02 absorption characteristics of limestones has been used with
success by Pope, Evans and Robbins21 in atmospheric pressure fluidized-bed
combustors.
In order to fully understand the role that the salt mixture plays, a
clear picture of the reaction mechanism must be obtained. An extensive search
of the literature (during the report period) has revealed a multiplicity of
uses of NaCl (and of other salts) as a catalyst or mineralizer in related
areas.223'29
Apparently, no detailed theoretical study of the effect of catalysts on
the calcination/sulfation system has been performed. Several researchers
have looked at the individual systems of calcination and sulfation separately
under laboratory conditions and have concluded that the sorption capacity is
diffusion-controlled for the most part, depending greatly on the physical
characteristics of the stone itself. Therefore, any effects of mineralizers
must be directly related to diffusion and/or rearrangement in the limestone/
dolomite matrix.
Pope, Evans, and Robbins21 suggested that the addition of sodium chloride
leads to a physical disruption of the pore structure of the limestone by the
sodium ions, so that compounds form having sufficiently different lattice
constants to strain the system during replacement. This concept was put forth
on the basis of gross effects observed in a large fluidized-bed coal combustor
with limestone additive and NaCl catalyst.
A review of various compound melting points and eutectic points in the
system NaCl-CaC03-MgC03-CaSOi<-MgSOi1 , as shown in Table 24, suggests a mechanism
involving low-temperature melts and solid solutions for the enhancement of
calcination and sulfation of limestone and dolomite.
DTA studies of dolomite have shown a depression of temperature of initial
decomposition in the presence of NaCl223' similar to the effects on limestone.
a. Mechanism of Enhancement by Sodium Chloride
It is proposed that a liquid phase exists upon the dissolution of carbon-
ate and/or sulfate into NaCl when these components are in contact at temper-
atures above 750°C. The enhancement of interaction is directly attributable
to the presence of these molten salt films on particle surfaces.
The accelerating effect of NaCl on partially sulfated stone is due to
the appearance of the above-mentioned liquid phase at high temperatures. The
-------
112
Table 24. Eutectics and Double Salts Formed in the System,
NaCl-CaC03-MgC03-CaS04-MgS04
Species
NaCl
CaC03
Decomposition
(d) or m.p. ,
°C Eutectic
801
d.825 2NaCl - CaC03
m.p. ,
°C
690
Double Salt
Decomposition
(d) or m.p. ,
°c
Na2Ca(C03)2 770 d.817
MgC03
MgS04
Na2C03
CaCl2
MgCJL2
d.350
1450
d.1124
850
844
770
712
4NaCl -
721
624
- Na2C03 634
4NaCl - NaoSOu 638
Mg3Ca(C03)b d.<350
Na2Ca(SOi4)2 915
Mg2Ca(SOi,),, 1201
Na2Mg3(SOil)i4 700
Na2(S02,C03) 612
CaCl2-CaC03
d.700
liquid phase greatly increases the area of contact of the reactants between
the salt ions, and the calcium and carbonate ions may weaken the ionic
strength of the crystal lattice—that is, the mobility of ions in the crystal
lattice increases and some of the ions on the surface of the crystal separate
and dissolve in the salt films.
Some of the salt ions may enter the crystal, causing the mobility of the
ions in the crystal to be increased by the formation of a solid solution or
a compound. When calcium ions or oxygen ions move in the salt layer, the
mobilities of these ions may depend on the viscosity of the molten salt. The
presence of carbon dioxide from the decomposition of the stone creates a
highly mobile liquid with low viscosity, increasing the diffusion rate of
migrating ions. As the diffusion rate increases, the rate of interaction of
the components to form the new crystallization phase, CaSO^, increases.
Possibly, the formation of the new crystalline phase at high temperatures is
preceded by removal of individual ions and molecules from the solid CaC03
lattice, which leads to the appearance of molecularly porous substances that
form pseudomorphs of the original crystals and effectively lower the surface
energy barrier to recrystallization. Ongoing crystallization of CaO to its
normal lattice structure initially increases the specific surface with respect
to S02 capture, allowing more complete sulfation to occur. The appearance of
these molecularly porous structures has a considerable influence on the capture
of S03 introduced into the system and on further penetration of the crystal
by the melt, leading to considerable contact of the components.
-------
113
Figure 63 is a sketch of a portion of a partially sulfated limestone
particle undergoing sulfation in the presence of NaCl. The NaCl vapor dis-
solves 863 gas and small amounts of CaSO^ on the surface of the particle,
forming a liquid melt [CaSO^ + 803 + NaCl]L which penetrates the shell of
previously formed CaSOii (Zone I) along grain boundaries and imperfections.
At the same time, the CaSO^ dissolves and recrystallizes as the liquid moves
into the interior until it contacts unreacted CaCO^ (Zone II). At the reaction
interface, the NaCl-CaSOi+-S03 melt provides a medium for the dissolution of
small amounts of CaCOs, incorporating it into a liquid phase [CaCOj]-^. The
[CaCOslL readily dissociates into Ca2 and C032~ with the subsequent release
of C02 gas and the simultaneous formation of CaO and/or CaSOij from interaction
with the dissolved 803. The CC>2-saturated liquid is highly mobile and diffuses
inward. As the C02 escapes outward, the liquid dissolves more CaCO^, pre-
cipitates CaSOi^, and continues to penetrate the unreacted portion of the
limestone lattice until the reaction is brought to a halt when the amount of
CaCOs is too small to form a liquid with the NaCl, and diffusion barriers
arise.
The effect of the additive is to enhance crystal growth by providing
nucleation centers while simultaneously separating and dissolving the lattice
structure of the original limestone. The ions of the incipient CaSO^ and
CaO crystal surfaces also undergo separation, redissolution, and recrystal-
lization via the salt film, leading to a more open porous structure with
little diffusional resistance. Diffusional barriers due to blockage of pore
structures are removed by this continual dissolution and recrystallization
in the NaCl-containing liquid phase. Rapid diffusion of ions and gases
through a mobile liquid accounts for the effective lowering of the temperature
of reaction at which these processes occur.
As part of this investigation of the mechanism of the catalytic effect
of mineralizers and its relevance to coal combustion, a small laboratory
reactor was built for controlled-atmosphere and -temperature experiments.
The current experiments are concerned mainly with common salt, NaCl, and
calcite (CaCOs) i-n order to provide experimental verification of the proposed
mechanism of interaction.
Large (0.6-cm) pseudocrystal rhombs of calcite spar were chosen for
this study of sulfation and calcination because of (1) their low reactivity
in these reactions, (2) their high purity, and (3) their extremely low
porosity, which results in a layered effect in the reaction products. The
reaction proceeds only from the crystal surface inward, not throughout the
entire sample as in natural limestones and dolomites.
A horizontal tube furnace capable of achieving 1100°C was set up with
a gas-mixing system for providing appropriate combinations of air, C02, and
S02.
Preliminary experiments were performed in an air medium with a small
flow of S02 at 900°C. The gas mixture passed for short time periods over
calcite samples, which were in quartz boats in the furnace. In most runs,
NaCl vapor was evaporated from a boat filled with fused NaCl in the upstream
portion of the system. In a few runs performed with NaCl directly deposited
-------
114
S02
Particle Surface
CaS04
Advancing Interface via
Grain Boundaries
Zone L CaS04 product layer
CaSO^s) + NaCI(g) + $03(91 — ICaS04 + 503 + NaCIIL
ICaS04lL^ CaS04
-------
115
on the crystal surface, most of the Nad was lost by evaporation from the
crystal surface before the length of time required for reaction elapsed, and
only the crystal surface was affected. Blank runs with no salt present
provided samples for comparison.
Figure 64 illustrates the dramatic results obtained upon salt addition.
Photos I and II show calcite crystals exposed to 862 in air at 900°C for 15
min. There is a very thin reaction rim of CaSOi4 crystals, with some CaO
microcrystals on the samples; the major part is unreacted CaCOs . The reaction
proceeds along existing fractures and crystal defects at the surface.
Photos III and IV show calcite crystals exposed at the identical condi-
tions with NaCl vapor present. It can readily be seen that the amount of
reacted material is much greater in both cases than for the no-salt cases
and is more variable in extent. There is a highly crystalline layer of CaSO^
on the outside over a thicker layer of CaO which surrounds the residual
CaCOs- An X-ray study shows the outer layer to be highly crystalline CaSOi.
with minor amounts of NaCl present, indicating that NaCl penetrated the
sample at least to this depth.
That there is a gradation of sulfate in the CaO layer was indicated by
some preliminary microprobe scans intended to determine the sodium distri-
bution in the reaction layers. Due to problems with maintaining a polished
surface, the results were ambiguous. The CaO is very finely crystalline,
with little coherence, and is highly reactive with water vapor. In many
cases, the outer layer of sulfate crystals had seoprpfed from the oxide as
a thin coherent shell, making it difficult to mount the entire specimen.
From the relative thicknesses of the product layers, it can be seen that
NaCl increases the extent of calcination more rapidly than it increases the
extent of sulfation.
Several runs were performed without salt present and with C02 levels
such that no calcination occurred during the experiments. Salt vapor was
then added to the system, and calcination proceeded despite the high C02
levels in the furnace. It is hoped that by suitable adjustment of C02 levels,
a much greater extent of sulfation than calcination can be achieved . Previous
work31 has shown the importance of the calcining conditions when calcined
stones are reacted with S02•
In the sulfation experiment illustrated by the photographs of Fig. 64,
the no-salt samples show a very thin layer of calcium sulfate and calcium
oxide, with the sulfate predominating. The effectiveness of NaCl in promoting
calcination and producing an initially porous product shows clearly in the
very thick layers of CaO in the samples with additive. The sulfate layer in
these samples is revealed under the microscope as consisting of fairly large
crystals growing out of the CaO layer. Literature descriptions of alkali
effects commonly refer to enhanced crystallization of the product phases and
increased porosity.25'29 It has been found that the reaction of S02 with
limestone is greatly influenced by the reduction in porosity caused by the
sulfation reaction.32
-------
116
n
CRYSTALLINE CALCITE SULFATED AT 900°C IN
AIR FOR 15 MINUTES
nr
CRYSTALLINE CALCITE SULFATED AT 900°C IN
AIR IN PRESENCE OF NaCI FOR 15 MINUTES
Fig. 64. Sulfation of Crystalline Calcite
-------
117
The microscopically observed penetration of NaCl along grain boundaries
and fractures, with reaction occurring at crystal defects and pore surfaces,
supports the mechanism that transient surficial melts form which increase
the reactivity of the dissolved phases and lower the resistance to diffusion
of gases. Scanning electron microscope photographs are being taken to deter-
mine if indeed the Nad penetrated the reaction layers and to indicate the
final disposition of the salt.
Preliminary calcination experiments were done, with no S02 present, on
calcite spar samples to note any qualitiative changes as a result of the
presence of NaCl. The sample exposed to air and C02 without salt was
partially calcined for 30 min. The product was white and brittle and was
finely crystalline under the microscope. When NaCl was added in similar
experiments, the product was more completely calcined and had a more porous
and crumbly texture. The outer surface, however, had begun to sinter into
a rigid shell. This product was discolored, with brownish tones.
Table 25 summarizes the observations made during a series of diagnostic
experiments performed in a horizontal furnace, which led to the design of
thermogravimetric experiments to provide quantitative results. From the
table, it can be seen that under simultaneous calcination and sulfation
conditions, there is a marked decrease in the total amount of reacted
material during a given period of time as compared with the amount of reac-
tion under simple calcination conditions. The diffusional barrier of incoming
S02 slows dov/n the calcination of the CaCQj. Addition of the mineralizer,
NaCl, increases both the extent of calcination and the extent of sulfation
by improving the porosity of the calcined stone. When simple calcination
is performed under a high enough C02 pressure that calcination does not
occur readily, the addition of NaCl vapor to this same system effectively
overcomes the diffusional barriers, and calcination proceeds rapidly.
When a precalcined stone is reacted with SC>2, the amount of sulfation
both with and without salt addition is increased in comparison to simultaneous
calcination-sulfation. The addition of NaCl, however, increases the amount
of sulfation and of interpenetration of CaO by the crystallizing CaSO^, with
a fluid phase following defects and grain boundaries within the crystal.
This differs considerably from the sharply defined reaction front observed
with no salt present.
b. Evaluation of Seven Additives
Compounds other than NaCl that have been found to be effective mineral-
izers23"29 have been studied to evalunte their potential usage in coal com-
bustion, as reported here. Among the compounds reported in the literature
are NaOH, Na2C03, KC1, CaCl2, MgCl2, Na2SOi4, Na3POi4, and kaolin and other
clays. The noncorrosive aspects of some of these compounds may make them
more attractive than NaCl if all other considerations are favorable.
Most workers agree that the reaction mechanisms of both calcination and
sulfation have rate-controlling diffusion-limited steps so that any mineral-
izing effects in calcination may have similar results in sulfation. Many
-------
Table 25. Effect of NaCl on Reactions of Calcite
Experiment Additive
NaCl Introduction
Technique
Reaction
Conditions'
Comments
NaCl-la
NaCl-lb
NaCl-2a
NaCl-2b
NaCl-3b
NaCl-3B
NaCl-4a
NaCl-4b
NaCl-5a
NaCl-5b
None
NaCl
None
NaCl
None
NaCl
None
NaCl
None
NaCl
Vapor
Vapor
Vapor
Vapor
Vapor
Calcined at 900°C
in air, 15 min
Calcined at 900°C
in air, 15 min
Calcined at 900°C
in high C02, 15 min
Calcined at 900°C
in high C02, 15 min
Simultaneous 900°C
calcinat ion/sulfat ion,
15 min
Simultaneous 900°C
calcination/sulfat ion,
15 min
Precalcined at 900°C
30 min, then exposed
to S02 for 15 min
Precalcined at 900°C
30 min, then exposed
to SC-2 for 15 min
Partially sulfated stone
further reacted, 15 min
Partially sulfated stone
further reacted, 15 min
Reaction almost complete
Complete reaction, salt gave
a more porous product
No reaction
Almost complete reaction,
porous product
Surficial reaction, white
coating with thicker layer
of CaO
Some CaC03 remains in center,
a thick layer of CaO, outer
thick layer of crystalline
Thin layer of CaS04 , flaky
Thick layer of crystalline
intergrown with CaO
No appreciable change
Thicker shell developed having
a more crystalline character
oo
Sulfation atmosphere was 20 vol % S02 in air,
-------
119
salts have been shown to be effective in the calcination of limestone and
are used routinely to promote the formation of reactive limes. Table 26
summarizes the results of qualitative experiments in which NaCl and several
other additives were used. The only salts comparable to NaCl (in terms of
magnitude of effect ) are CaCl2 and MgCl2; NaOH, Na2CC>3 , and Na2SOi, all
behave similarly; the first two salts sulfated readily during reaction and
then effectively behave as Na2SOi, in the system. One set of experiments
(vapor salt introduction mode) was done using open boats of salt introduced
into the furnace to supply a vapor to interact with the calcite. If a porous
limestone instead of calcite spar crystals were used, the salts could be
directly deposited within the stone and thus >:ould perhaps avoid the diffi-
culties associated with subsequent sulfation of the additive.
The volatility of the Na2S04 is low compared to that of NaCl and CaCl2 •
Its melting point is fairly high compared to the reaction temperature,
whereas NaCl and CaCl2 are easily melted below 900°C. Sulfation with Na2S04
and the other related salts (NaOH and Na2C03) occurred only when the salts
were placed directly on the calcite crystals. These effects support the
view that a fluid phase is present during the reaction. The presence of
low-melting eutectics of salt and matrix appears to open up the system to
further sulfation. In porous stones, these salts may indeed have much
greater effects due to more intimate surface contact of the salt with the
stone.
These qualitative results indicated the desirability of pursuing quanti-
tative experiments on a thermogravimetric apparatus to measure weight changes
during sulfation and to deduce absolute values for sulfur capture when
mineralizers are present. The effects on sulfur absorption of varying the
concentration of the additives are reported below.
Figure 65 shows the effect of precalcination of Greer limestone (in the
tube furnace) on sulfur retention and the enhancement by NaCl. The NaCl was
introduced by immersion of the limestone in an aqueous solution. After
the limestone was soaked in a 20% NaCl solution, it was dried at 150°C,
leaving approximately 1% NaCl by weight in the stone. The sample was then
calcined in the furnace assembly at 900°C in an atmosphere of 20% C02 in N2 ,
reweighed to verify the completion of calcination, and then exposed to a
gas mixture containing 4% S02. Despite the loss by evaporation of most of
the salt, the effects upon sulfation can be seen, with the stones' capacity
for S02 increased by approximately 50%. The effect of simultaneous calcin-
ation/sulfation can also be seen — a substantially lower reactivity — although
within the time per j.od , the total amount of conversion eventually reaches
the same level as does the precalcined stones at this concentration of S02.
At lower concentrations of S02 (Fig. 66), as in a flue gas, the effect is
much more apparent because the uncalcined stone does not reach the same
level of sulfation as the precalcined stone in any reasonable time interval.
Figure 66 shows experimental data points from the thermogravimetric analyzer
with 0.4% S02.
The initial use of the horizontal tube furnace assembly instead of a
thermogravimetric analyzer was prompted by the possibility of extensive
corrosion when volatile alkali salts are used. Figure 67 gives the results
-------
120
o
CO
o
o
O
o
o
o
O
CO
o:
oPRECALCINED WITH 0.88% NoCI
• PRECALCINED GREER
nGREER LIMESTONE SIMULTANEOUS CALCINATION/SULFATION
Fig. 65. Enhanced Sulfation in U% SC>2 at 850°C of
Greer Limestone by Pretreatment
50
c?
-------
Table 26. Effect of Additives on Simultaneous Calcination-Sulfation
of Crystalline Calcite at 900°C
Expt.
CS
CS
CS
CS
CS
CS
CS
CS
CS
CS
CS
CS
CS
1
2a
2b
3a
3b
Aa
4b
5a
5b
6a
6b
7a
7b
Additive
None
NaCl
NaCl
CaCl2
CaCl2
Na2C03
Na2C03
NaOH
NaOH
Na2SOi4
Na2SOi4
MgCl2
MgCl2
Salt
Introduction
Mode
None
Vapor
Solidb
Vapor
Solid
Vapor
Solid
Vapor
Solid
Vapor
Solid
Vapor
Solid
Reaction
Time,
min
30
30
30
30
30
30
30
30
30
30
30
30
30
% S02
in Air
25
25
25
25
25
25
25
25
25
25
25
25
25
Extent of
Reaction
Minor
Ma j or
Major
Major
Major
Minor
Major
Minor
Major
Minor
Maj or
Major
Maj or
Comments
Surficial white deposit on CaSOi^
plus CaO
Highly crystalline CaSOi^ over a
thick layer of CaO
As in CS 2a, over entire crystal
Similar to NaCl though less CaO
formed
Similar to NaCl
Some surficial reaction
Reaction localized with small
effects elsewhere
Very little (NaOH appears to boil
away and sulfate)
Very localized (boils away)
Small amount of surface reaction
Entire crystal surface reacted
Similar to CaCl2, though less
extensive reaction
Similar to CaCl2
Additive in a boat upstream.
""Additive on calcite surface as slurry.
-------
122
c
ID
UJ
0
CO
O
(J
O
J—
O
0
u.
0
z
Q
55
o:
UJ
i
0
O
^
o
i-
IUU
90
or*
ou
70
60
50
40
30
20
10
0
f
i---i — i i - •!•••• » i » i >;
TF TGA
^ 0 GREER LIMESTONE TREATED WITH NaCI SOLUTION
0 D UNTREATED GREER LIMESTONE
.
y y A
ss ^^
---O~~ ' // ^ "
x^A A
fl
-
A^
.
-cy °
/
/
7
i i i i // i // i n
\ f\ * s\ n o i f ort *' K. f\ * \r\ r\ ' oo
.0
CONCENTRATION OF NaCI, wt %
Fig. 67.
Effect of NaCI Concentration on Sulfation of
Greer Limestone at 850°C in 0.3% SO.-/. A
comparison of TGA with horizontal tube
furnace (TF) .
of two series of experiments at 850°C using Greer limestone with NaCI the
mineralizing additive; these series were performed on the TGA and the tube
furnace (TF) under simultaneous calcination/sultation conditions. After
six hours, the two methods are in excellent agreement for a given concentra-
tion of sodium chloride. Some samples were prepared by immersion in a near-
boiling brine of NaCI and then drying in an oven at'100°C. The sodium
chloride was thereby uniformly distributed throughout the sample—within
pores and on particle surfaces. Samples used were all sieved to 18-20 mesh
size. The maximum concentration of salt introduced in this way was 2% by
weight.
High concentrations of NaCI were incorporated into samples of Greer
limestone by evaporating a slurry of water and salt or by a dry mixing of
finely ground salt with the Greer. Both of these methods yield salt particles
unattached to the limestone; hence, to avoid corrosion of the TGA, no experi-
ments with high-NaCl limestones were carried out in the TGA. The plot of
the percentage conversion of CaO to CaSOi^ in Greer limestone versus the
concentration of NaCI admixed with the stone in Fig. 67 shows that the data
for both the TGA runs and the horizontal tube furnace runs are in excellent
agreement. The lower value for the 2% run in the horizontal tube furnace
compared with the 1.8% run in the TGA reflects the different methods of prepa-
ration; for the tube furnace, the method was slurry evaporation wherein some of
the free salt (most of which had deposited on the surface of the particles
-------
123
during evaporation) evaporates without contributing to the reactivity of the
stone. These higher concentration runs thus represent a lower value of salt
concentration than was initially introduced. (Samples prepared by simple
mixing of finely ground salt with the stone gave even lower values of con-
version. )
As can be seen from the curve (Fig. 67), at low concentrations of salt,
doubling the amount of NaCl added doubles the percent conversion. Conversion
levels off near 2% NaCl, and there is very little increase in conversion
with further addition of salt.
By adding NaCl to the original Greer limestone (which itself has a
sodium content equivalent to 0.16 wt % NaCl), the amount of conversion is
substantially increased (Fig. 67). The untreated stone levels off after
6 hours near 20% conversion; in the runs with the highest concentrations of
NaC.1 (from 2% up to ^20%), conversion levels off near 60-65%, a factor of
three greater. The nonlinearity of the relation between salt content and
amount of conversion illustrated in Fig. 67 suggests that at these high
concentrations of salt, fusion of the sample and/or blockage of porosity
may hinder further sulfation.
The effect of a series of salts (Na2f)0i4, Na2C03, CaCl2, and KC1) on Greer
limestone reactivity is presented in Fig. 68, 69, 70, and 71, respectively.
In each graph, the natural stone without additive is represented by a dashed
curve.
80
. 70
O
«» 60
o
o
50
k-G-21 3.15 wt % No2S04
A-G-16 1.95 wt % Na2S04
O -G-24A 0.93 wt % Na2S04
D-G-260.60wt % Na2S04
O-G-I GREER
Fig. 68. Enhancement of Greer Limestone Sulfation
with Na2S04 at 850°C in 0.3% S02
-------
124
Figure 68 contains plots of the percent conversion versus time for
Greer limestone doped with Na2SOit. For this salt, the increase in SO?
absorption is not as great as with NaCl. Comparison at 2% mineralizer con-
centration shows a 35% conversion for Na2SOit and a 60% conversion for NaCl
(Fig. 67). On the basis of formula weights, the salts have approximately
equivalent concentrations of sodium. Increasing the concentration of Na2SOi4
does not increase the conversion percentage as much as increasing the NaCl
concentration does. In fact, when corrections are made for the amount of
sulfate possibly exchanged with CaO, the increases in conversion versus
concentration of Na2SOtt are barely noticeable. Results of analysis of the
reaction products appear to agree with this conclusion; further checks will
be made on all of the experiments performed.
Figure 69 is a plot of the percent CaO converted to CaSOi^ for Greer
limestone containing Na2C03 additive. Soaking the stone in a saturated
Na2C03 solution (thereby adding approximately 3% salt) increases the con-
version to sulfate from 20% to 50%, i.e., by more than a factor of two.
Larger amounts of salt will be added in future experiments to obtain a
complete picture of the relationship between conversion and concentration
of salt.
2.85 wt % Na2C03
A -G-32 1.36 wt % Na2C03
O -G-IO 0.62 wt % Na2C03
32 wt % Na2C03
O -G-l GREER
O
0
Fig. 69. Enhancement of Greer Limestone Sulfation
with Na2C03 at 850°C in 0.3% S02
-------
1.25
Figure 70 shows conversion versus time for Greer limestone doped with
CaCl2 solutions. Conversion increases significantly at low concentrations
of CaCl2 but appears to level off rapidly above ^1% CaCl2, as indicated by
the 50% conversion level reached; conversion with 5.5% CaCl2 was only slightly
higher than with 0.6% CaCl2- These experiments indicate that sodium ion
is not the only effective mineralizer in these systems. Chloride ion has
an effect on limestone sulfation that is as large as the effect of sodium.
This may partly explain the effectiveness of NaCl being greater than that
of the other sodium salts.
The enhancement due to KC1 is shown in Fig. 71. The percent conversion
of Greer limestone, 20%, increased to over 50% with ^2.5% KC1 in the stone.
The effectiveness of this salt is close to that of NaCl.
The weight percentages of salts prese-.t in the samples were compared,
and the individual additives were arranged in decreasing order of effective-
ness: KC.1 > NaCl £ CaCl2 > Na2C03 > Na2S04.
k-G-20 5.51 wt % CaCI2
A -G-I2 0.60 wt % CaCI2
o -G-I4 0.19 wt % CaCI2
D-G-15 <0.l wt % CaCI2
O-G-I GREER
- — O
Fig. 70. Enhancement of Greer Limestone Sulfation
with CaCl2 at 850°C in 0.3% S02
-------
126
80
'^ 70
o
CO
o 60
»- 50
A
O
D
G-27 WITH 3.57% KCI
G-28 WITH 2.40% KCI
G-29 WITH I . 30% KCI
G-30 WITH 0.70% KCI
G-31 WITH 0.60% KCI
O G-l GREER
Fig. 71. Enhancement of Greer Limestone Sulfation
with KCI at 850°C in 0.3% S02
c. Effect of NaCl Additive on Several Limestones
The greater part of this work so far has dealt with the reaction of
Greer limestone with sulfur dioxide and several additive salts, namely, NaCl,
Na2SOi,, Na2C03, CaCl2, and KCI. The results of tests on a series of lime-
stones having various compositions and morphologies are next discussed.
Eleven limestones and dolomites have been reacted for seven hours in
a horizontal tube furnace assembly at 850°C with a 0.3% S02-5% 02-20% C02-
balance N2 synthetic flue gas mixture. Figure 72 shows SOs captured/kg of
sorbent for the untreated stones.
There is great variation in both the rate of reactivity of the stones
with S02 and the calcium utilization among the different samples. Initial
reaction rates differ considerably but do not necessarily correlate with
total calcium utilization for the sulfation period. Comparison of limestone
1337 with Tymochtee dolomite shows that initial reaction rates are very
similar; changes . • -ur after 2 hr, 1337 having a steeper slope; later,
Tymochtee picks up S02 faster than 1337 and has a steeper slope. After 7 hr
of reaction time, the percentage of calcium (as CaO) in the stones converted
to CaSOn ranges from 10% for 1359 up to 90% for Tymochtee. The plot i*
corrected for the amount of actual CaO present and available for reaction.
These data may be compared with results on limestone characterization
by Snyder with several reservations. Snyder's work was performed at 900°C
-------
127
LJ
m
o:
o
400
350 -
300 -
250 -
200 -
ro
O
CO
D - TYMOCHTEE
0-1337
O-I35I
V-1360
A-2203
+ -I343
k-GREER
X -DOLOWHITE
D-1336
O-CALCITE SPAR
O-I359
Fig. 72. Weight of 803 Captured as a Function of
Time for Eleven Untreated Limestones.
using precalcined stones; the data reported here was collected at 850°C
under conditions of simultaneous calcination/sulfation. The relative order
of reactivity for the more reactive stones is the same for the two series.
However, at 850°C there is considerable rearrangement of the low to medium
reactive stones in comparison with their order at 900°C; this illustrates
the large effects of reaction conditions. There is no comprehensive
correlation between percent calcium utilization and composi'_ion. For
example, Dolowhite, which has about the same composition as Tymochtee, is
particularly unreactive. Porosity determinations are being made on stones
calcined at 850°C to attempt to verify a correlation of porosity-composition-
reactivity suggested by Snyder.
-------
128
Figure 73 is a plot of the same series of limestones reacted under
conditions similar to those used in earlier runs (Fig. 72) except that 2%
by weight of pure sodium chloride was added. The sodium chloride was added
by evaporating a water slurry of the stones and salt. This method was
used since it yielded a more intimate mixture than does simple mixing of
dry salt and stones. Preliminary experiments have shown that the loss of
salt by evaporation is greater when salt has been added as a powder. Such
evaporation leads to difficulties in evaluating the actual amount of salt
present during reaction. Slurry evaporation deposits most of the salt on
the surfaces of the limestone particles, where it interacts rapidly with
the stone on heating.
The reaction curves in Fig. 73 clearly show changes in the rate of
reaction for each stone with 2 wt % Nad present, as compared with the case
with no salt present (Fig. 72). For most of the samples, initial reaction
rates are higher than in the absence of salt, exceptions being the highly
reactive stones 1337 and Tymochtee whose initial rates are lowered by salt
addition. Noticeable also is the tendency of the reactivity of several
stones to continue at a high level during the entire 7-hr period with no
sign of leveling off. Dolowhite has a significant slope—even after seven
hours (Fig. 73)—whereas with no salt present this stone levels off at a
much earlier time (Fig. 72).
400
350 -
300 -
250 -
CE
Si
200 -
V- 1360
0 - 1337
- GREER
O- 1351
Q - TYMOCHTEE
A- 2203
O - CALCITE SPAR
D- 1336
+ - 1343
O- 1359
X - DOLOWHITE
Fig. 73. Weight of SOs Captured as a Function of Time
for Eleven Limestones with 2 wt % NaCl
-------
129
For a clearer representation of the effect of NaCl on sulfation of
limestone, Fig. 74 is presented as a bar graph to illustrate the results.
It is arranged in order of increasing sulfation reactivity of the pure stones.
The hatched bars represent the amounts of sulfation (in percent CaO convert-
ed to CaSOi,) occurring after seven hours in the absence of salt; the open
bars represent the additional conversion of CaO to CaSOi, due to the presence
of 2 wt % Nad.
In the case of calcite spar, the incomplete calcination of the raw stone
in the absence of salt interferes with the measured reactivity. When salt
is present, calcination is rapid and complete. If the entire calcium content
of the raw calcite is considered available, rather than the partially cal-
cined stone, conversion would amount to only 5% of the total stone.
CALCITE
SPAR
(INCOMPLETE
CALCINATION)
1359
(GROVE)
1336
DOLOWHITE
GREER
1343
2203
1360
1351
1337
(NEGATIVE
EFFECT)
TYMOCHTEE
(NEGATIVE
EFFECT)
RAW STONE
RAW STONE
PLUS 2% NaCl
BY WEIGHT
I
I
I
I I
I
I
I
I
0 10 20 30 40 50 60 70 80 90 100
CONVERSION OF CaO TO CaS04.%
Fig. 74. Effect of NaCl on Sulfation of Limestones at 850°C
in 0.3% S02 after 7 Hours. All stones except cal-
cite spar were totally calcined.
-------
130
From this graphic representation, it is readily apparent that the
limestones with low reactivity show a greater effect of NaCl than do the
rest of the samples. The highly reactive stones, 1337 and Tymochtee, are
actually hindered by the presence of the salt. If, as has been suggested,
the effect of the mineralizer is to change the crystallization characteristics
of the calcining and sulfating stone with subsequent changes in the porosity,
one might expect that any changes in stones normally undergoing nearly
complete sulfation would be detrimental. Preliminary porosity measurements
suggest that drastic changes in the pore size distribution occur during
calcination when salt is present. Porosity tests have been completed for
the entire series of limestones in both the calcined and the sulfated
condition with and without NaCl present and are reported below.
d. Porosities of Limestone^
The porosity curves of accumulated pore volume vs pore diameter for
samples calcined in 20% COz at 850°C for 1 hr are presented in Fig. 75. It
can be readily seen that each stone has its own pore distribution and pre-
dominant pore size. Conditions of calcination being constant, what is
reflected here are the differences in composition, morphology, and impurity
level of the stones. Limestone 1360 decrepitates badly and the curve measures
interparticle space for the powder produced, as well as intraparticle
porosity. The eleven stones represent a wide range of limestones having
various amounts of extraneous constituents. Enhancement of reactivity with
S02 appears to depend a great deal on the formation of larger pores which
can react more completely with large molecules such as sulfur trioxide.
These curves represent the baseline calcines with no salt addition, i.e.,
with only the small amount present naturally in the stone. The C02 level
during calcination must be specified since the reactivity of a lime can be
greatly affected by precalcination at high C02 levels.
Two weight percent sodium chloride was added to the stones, and calcin-
ation was carried out in a 20% CC-2 at 850°C for 1 hr. The resulting porosities
of the eleven stones are shown in Fig. 76. The effect of the sodium chloride
is to shift the curves to a larger average pore diameter. Each stone responds
in a unique way. The different porosities appears to reflect the different
compositions of the stones, but more analysis of the results is needed
before the exact relations are defined. The effects are dramatic in every
case, with the largest changes occurring in the more calcitic limestones.
Correlations with S02 conversion of CaO to CaSO^ are not readily discernible
and await further analysis of the data.
For the present, one example from the eleven will serve to illustrate
the effects of salt on limestone porosity. Greer limestone has been chosen
as an example since there is a great deal of data on Greer's reactivity
with S02. Figure 77 is a family of porosity curves for Greer limestone
illustrating the effect of varying the concentration of NaCl added to the
stone before calcination. The average pore diameter shifts to a higher
size for every incremental addition of salt. The shape of the curves remains
the same, with total porosity gradually rising, peaking at 1% NaCl, and
falling with further addition. At concentrations higher than 1% NaCl, a
-------
131
0.70
0.60 h
PORE DIAMETER,
10 I O.I
1350
-+GREER
TYMOCHTEE
1351
DOLOWHITE
1360
*-* 1343
CALCITE SPAR
2203
10 100 1,000 10,000
ABSOLUTE PRESSURE, psi
Fig. 75. Porosimetry Curves for Eleven Limestones
Calcined One Hour at 850°C in 20% C02
great deal of the added salt vaporizes and has no contact with the stone;
thus, the effect levels off rapidly. Data not reported here shows the
effect of increased exposure time at the calcining conditions for Greer
with 1% NaCl—only a slow growth of pores over a 6-hr timespan at 850°C.
All of these porosimetry curves represent calcination without sulfation.
For simultaneous calcination/sulfation, the resulting porosity is expected
to differ, especially in the presence of NaCl. The salt particles act
as centers of nucleation, causing rapid recrystallization and ionic diffusion.
Surficial melts lower the energy barrier for ions to move from one structural
lattice position to another, speeding the coalescence of grains within the
crystallizing CaO and CaSOi, and simultaneously maintaining an open-pore
structure which counteracts the loss of surface area due to this increase
in particle size. Further measurements are being made on other stones to
attempt to define the effects of mineralizers on the porosity of the stones
and how that affects the activity of the lime with respect to S02 capture.
-------
132
0.70
0.60 h
PORE DIAMETER , ^m
100 20 10 21 0.2 O.I 0.02 0.01
ro
o
II I I I I—I 1 [Mill I I
1360
1351
2203
~ TYMOCHTEE
- ~ DOLOWHITE
_*-* GREER
-* 1343
CALCITE SPAR
Fig. 76.
10 100 1,000 10,000
ABSOLUTE PRESSURE, psi
Porosimetry Curves for Eleven Limestones
Plus 2% NaCl, Calcined 1 Hour at 850°C
in 20% C02
In order to relate the porosity measurements to reactivity, samples of
the limestones were precalcined under identical conditions and then reacted
with sulfur dioxide, thereby avoiding effects due to simultaneous calcination/
sulfation. These sulfation runs were performed at 850°C in 0.3% S02, 5% 02,
20% C02, and the balance N2 with and without 2 wt % NaCl added.
The percent conversions to sulfate are presented in Fig. 78, along
with calcination/sulfation data refined from earlier work. The bars repre-
sent the various modes of reaction with S02 . The clear bars ("raw stone")
are the percent conversions of CaO to CaSO^, measured for simultaneous
calcination/sulfation. Similarly, the shaded bars represent the total con-
versions under simultaneous calcination/sulfation with 2 wt % NaCl added.
The widely spaced hatched bars are the conversions to sulfate for the
precalcined raw stone, while the closely spaced hatched bars represent the
-------
0.60
0.50
0.40
f 0.30
o
0.20
0.10
0
133
PORE DIAMETER, ^m
100 10 I O.I 0.01
[Mill I—I 1 [II I I I I—I 1 1 I I II I I—I 1 1 II I I I I I I | I i I I
— GREER PLUS 10% NaCI
D—O
U*-* GREER PLUS 5% NaCI
GREER PLUS 2% NaCI
GREER PLUS 1.0% NaCI
GREER PLUS 0.6% NaCI
GREER PLUS 0.5% NaCI
GREER
CALCINED Ihr AT 850°C
IN 20% CO.
10 100 1,000 10,000
ABSOLUTE PRESSURE, psi
Fig. 77. Porosimetry Curves for Greer Limestone
plus NaCI Calcined 1 Hour at 850°C in
20% C02
conversions to sulfate for precalcined stone that had been treated with 2 wt
% NaCI. The graph is arranged in the order of increasing reactivity of the
raw precalcined limestone, from "calcite spar" tu Tymochtee. With a few
exceptions, this order is also the order of increasing inerts content ("inerts"
are constituents that do not react with S02, including MgO) of the calcined
stone. There is a fair correlation between percent conversion and total
inert content for untreated stones which is improved if one considers that
stones 1360 and 2203 undergo considerable decrepitation and dusting when
calcined; their anomalously high reactivities are due to the increased surface
areas presented by the powders. The glaring exception to the general trend
is the conversion of dolowhite, which is a highly crystalline pure dolomitic
material and has very small pores. Its apparent unreactivitv must be due
to its initial low-porosity—perhaps constrained by its highly crystalline
state.
-------
u>
.p-
Fig. 78. Effect of Nad on Sulfation of Limestones (Both Precalcined and
Simultaneous Calcination/Sulfation) at 850°C in 0.3% S02 After
Seven Hours.
-------
135
The reactivity of pr-ecalcined limestones containing 2 wt % NaCl again
shows the above relationship between inerts content and reactivity. Figure
79 is a plot of sulfur conversion versus tota] inerts content of the calcined
stones treated with NaCl. Here, the correlation is excellent, including that
for previously anomalous dolowhite, which here has a reactivity comparable to
that of the other dolomitic stones studied. Apparently, the salt treatment
opened up the structure to an extent, overcoming the restraints induced by
the original crystallinity of this dolomite.
The porosity curves measured for these limestones have been used to
obtain an "average" pore diameter by dividing the pore volume measured on the
porosimeter by the calculated surface area. This gives one-half radius or,
after multiplying by four, an "average" pore diameter for all pores that
are larger than =0.02 urn. Smaller pores were not included since they play
no important part in sulfur reactivity in the time period considered.
o CALCITIC
• DOLOMITIC
10 20 30 40 50
TOTAL INERTS CONTENT OF
CALCINED STONES, %
Fig. 79. Sulfation at 850°C of Limestones as a Function
jf Total Inerts Content in Precalcined Stones
Treated with 2% NaCl.
-------
136
Figure 80 is a plot of "average" pore diameter for precalcined stones
(both raw stone and stones that had been treated with 2% NaCl) as a function
of total inerts content of the calcines. There appears to be a strong
correlation for the salt-treated stones that includes both the calcitic and
the dolomitic limestones.
Of the untreated stones, the calcitic limestones form a straight line
trend of increasing pore diameter with increasing inerts content. The
dolomitic stones are in a separate cluster. Their high MgC03 content un-
doubtedly plays a major role in determining initial porosity and pore size
distribution. The effect of NaCl on the pore diameter can be seen to increase
as the inerts content of the stones decreases. Pure calcitic stones are
affected to a much greater extent by the given amount of salt than are the
impure limestones and highly magnesian dolomites.
CALCITIC o CALCINED WITH 2% NaCl
DOLOMITIC • CALCINED WITH 2% NaCl -
CALCITIC a CALCINED RAW
DOLOMITIC • CALCINED RAW
'0 10 20 30 40 50
TOTAL INERTS IN CALCINED STONES, %
Fig. 80. Average Pore Diameter as a Function of
Total Inerts Content of Limestones
-------
137
The correlation between percent conversion of CaO to CaSOij and "average"
pore diameter for the salt-treated precalcined stones is shown in Fig. 81.
The larger the pore diameter, the lower the conversion of the salt-treated
stones. In the case of untreated precalcined stones (no graph presented),
no trend is clear (i.e., the points scatter), but the relationship appears
to be the inverse of that for treated stones—stones with large pores appear
to have greater reactivity. This apparent contradiction can be resolved if
pore size distribution is considered along with the known fact that pores
smaller than a certain size do not react appreciably with SC>2 (as discussed
below). The trend of increasing reactivity in the series of untreated
precalcined stones with larger pores reflects the fact that most of the pores
in these stones are too small to contribute to S02 reactivity. As a result,
some stones with pores of a more favorable size distribution react more with
S02- When NaCl is added, most of the pores in all stones now are larger than
this lower limit, and now the controlling variable is surface area, which
decreases dramatically as pore diameter increases.
0 0.2 0.4 0.6 0.8
D, AVERAGE PORE DIAMETER,
Fig. 81. Sulfation at 850°C of Limestone as a Function
of Average Pore Diameter in Precalcined Stones
Treated with 2% NaCl
-------
138
In the case of the dolomitic stones, their inherent porosity is already
at a near maximum with respect to sulfur capture due to the additional
porosity contributed by MgCC>3 decarbonation; the porosity of dolowhite is
anomalous since its pores are extremely small. When salt is added to dolomitic
stones, the effect is that surface area is lost by growth of large pores at
the expense of small pores; however, dolowhite's porosity is thereby shifted
to a more favorable pore size distribution.
The selection of 2% as the NaCl concentration was arrived at on the
following basis: (1) its favorable effect on one limestone (Greer) and (2)
the fact that this concentration was used by Pope, Evans and Robbins to
improve sulfur capture in their early atmopsheric fluidized-bed combustor.
It is expected that by varying the amount of salt, an optimum reactivity
for each stone should be achievable. Experiments are being done to study
this.
Figure 82 illustrates how varying of the concentration of added NaCl
affects the "average" pore diameter for a pure calcite spar and an impure
(up to 24% inerts) limestone (Greer). As can be seen, the effect of NaCl
addition on pore diameter is greater in the pure stone although the percent
conversion of CaO to CaSOt, with 2% NaCl addition is only 19% for calcite as
compared with 50% for Greer (Fig. 78). Hence, reactivity is not a function
of pore diameter alone. Surface area and direct effects of salt on the
sulfation reaction must be considered.
It has been reported in the literature that the only property of lime-
stone that reliably correlates with reactivity is the sodium content. This
relationship holds true for this series of stone in a very loose way. The
initial sodium contents of most of the samples are so low that comparisons
would be unreliable. However, an overall trend exists, confirming the
reported correlation. A larger number of stones must be considered before
any definite relation can be observed .
One further point supports the concept that the effect of NaCl is
related to the inerts content of the stones: The analyses of the products
after sulfation for various limestones treated with 2% NaCl show a general
increase in sodium content with an increase in the total inerts content of the
calcined stone. This suggests that the inert materials are interfering with
and tying up some of the effectiveness of the sodium. For the few dolomitic
stones, no apparent differences due to the inert content appear. The
formation of silicates, aluminates, etc, could effectively reduce the
concentration of sodium available for interaction with the calcium and
could hinder the recrystallization and growth of pores caused by salt.
Thus, we find residual sodium in the impure limestones and dolomites, and
we find that essentially no sodium is retained in the pure stones after 7
hr of sulfation. This point has consequences beyond this question of
mechanistic effect. This loss of sodium in these experiments plays a role
in determining the disposition of sodium in a fluidized-bed combustor and
may be an important consideration in relation to corrosion, which appears
to be related to alkali concentrations in the flue gas.
-------
139
Fig. 82.
6 8
NaCI, wt %
Average Pore Diameter as a Function of
Percent NaCI Added
In the above discussion, no attempt is made to evaluate those effects
of the salt that may contribute to sulfation (such as continuous changes
in porosity during reaction and surficial melts contributing to recrystal-
lization and decomposition). As further analysis of the data is completed
and scanning electron microscope photographs and elemental scans are
correlated, the details of the mechanism will be formulated clearly.
2. The Determination of Inor^ggnic Constituents in the Effluent Gas
from Coal Combustion
(S. Lee)
Some chemical elements in combustion gases are known to cause severe
metal corrosion. The objectives of this study are to determine quantitatively
which elements are present in the hot combustion gas of coal, in either
volatile or particulate form, and to differentiate between volatile and
particulate species. It is desirable to identify which compounds are present
as particulate species and their amounts and to determine the amounts of
compounds present as condensable species.
-------
140
Assembly of the laboratory-scale fixed-bed batch combustor system to
be used for this study has been completed. A schematic diagram of this com-
bustor system was presented and described in the preceding annual report
(ANL/ES-CEN-1016). Several shakedown coal-combustion experiments have
demonstrated satisfactory performance of the combustor system. During this
period of reporting, systematic studies have been started to investigate the
transport of alkali metals during the combustion of coal. The findings from
these studies are discussed and presented in this report.
a. Coal Combustion Experiments
In initial coal combustion experiments, the problem was encountered
that tar and/or soot condensed on the surface of the cold trap. This was
found to be due to insufficient oxygen in the combustion air during the early
stage of the experiment and was eliminated by introducing afterburning air
downstream from the combustion section. The afterburning air was added to
facilitate combustion of volatile matter evolved during the early stage of
an experiment. With the combustion of coal successfully controlled, a standard
experimental procedure was established and is described in the following:
After the alumina filter and the cold trap are installed in the
combustor, the filtration section of the combustor is preheated to the
desired temperature (800°C was used in all experiments), using external
furnaces. When the temperature of the filtration section reaches a constant
value, a sample is placed in the reactor and the entire system is flushed
with nitrogen gas for a few minutes. The effluent gas analyzers are cali-
brated with calibration gases at this time. Prior to heatup of the combustion
section, afterburning air is introduced into the combustor at a flow rate
of 5.0 L/min. The coal sample is then heated indirectly by induction heating
(applied to the combustor pipe at the combustion section of the combustor).
When, during heatup of the coal sample, C02 gas is detected in the effluent
by the effluent gas analyzer, the inlet gas mixture of 02 and N2 is intro-
duced into the combustion section to burn the coal. During initial feeding
of the inlet gas mixture (65% 02 in N£ at a flow rate of 6.4 L/min), volatile
matter evolved during heatup of coal is completely combusted. This high-02-
content gas mixture also burns coal at a high rate to heat the bed to the
combustion temperatures rapidly. When the desired coal 'bed temperature is
reached, the 02 content of the inlet gas mixture is gradually reduced to the
desired value, and the afterburning air stream is shut off. The combustion
temperature of the coal bed is controlled at the desired level by regulating
both the 02 content of the inlet gas mixture and the induction heating energy
input.
The success of the standard experimental procedures in eliminating
the condensation of undesirable tar and/or soot on the cold trap was shown
by a two-part test run designed to burn two batches of coal, one immediately
after the other, under the same experimental conditions. This test run was
also designed to test the capability of obtaining reproducible experimental
conditions for the two batches by controlling the operational characteristics.
This is a necessary capability because two or more batches of coal have to
be burned under the same experimental conditions in order to collect enough
condensate on the cold trap for analysis.
-------
141
In each part of this test run, 50 g of -20 +40 mesh Herrin No. 6
coal from Montgomery County, Illinois, was burned. Neither the hot filter
nor the cold trap was removed from the combustor during loading of the
second batch of coal. The analyses of effluent gas compositions and the
coal bed temperature for the entire course of the experiment are shown in
Fig. 83 (batch 1) and Fig. 84 (batch 2). The experimental controls used for
the two batches differed: a preheated inlet gas mixture was used in the
batch 2 experiment but not in the batch 1 experiment, and the afterburning
air was shut off earlier in the batch 2 experiment than in the batch 1 experi-
ment. As shown in both figures, the first peak in the €62 concentration
curve was a result of the sudden increase in flow rate caused by introduction
of the inlet gas mixture. The second sharp peak in the C02 curve (which ran
over-scale on the graph) and the corresponding sharp drop in the 02 curve
shows that the volatile matter evolved during this stage burned rapidly. No
carbon monoxide was detected, indicating that volatile matter was combusted
completely.
This rapid combustion of volatile matter also resulted in a rapid
rise of the coal bed temperature. The bed temperature was controlled stably
at 855°C (on this average) for both batches. The effluent gas compositions
for both batches were also observed to be nearly identical, indicating that
the rate of combustion of coal was the same for both batches. This test
run has demonstrated the capability of controlling the burning of coal under
a single set of experimental conditions in this combustor system. The cold
trap was stably controlled at 150°C in this test run. It was dry and com-
pletely free of tar and/or soot.
On the basis of the rate of C02 formation, the rate of burning of
coal at the experimental conditions has been calculated and found to be 10
to 15 g/hr. This rate decreased with a decrease in the particle size of
the coal used. A possible explanation for this is that finer coal particles
form a more packed bed that offers more resistance to the diffusion of oxygen
into the bed. When the oxygen content in the inlet combustion mixture of
oxygen and nitrogen was plotted against the oxygen consumption and also against
the formation of carbon dioxide, straight lines were obtained (Fig. 85).
This linear relationship indicates that burning of coal in this batch system
is controlled by the rate of diffusion of oxygen into the bed.
The cold trap is made of 304 stainless steel. At the end of each
coal combustion experiment, bluish condensates were always collected on the
surface of the cold trap. They were identified as iron sulfates by X-ray
diffraction.* It was .shown in subsequent '.experiments that the bluish conden-
sates were products formed by the attack of sulfuric acid on the stainless steel
cold trap. To solve this acidic corrosion problem, the cold trap was electro-
plated with a very thin layer of rhodium metal (about 0.00005 in.). No thicker
layer was possible because, without special treatment, stress cracking would
occur in an electroplated layer thicker than 0.00005 in. The choice of
rhodium (instead of platinum) is based on its inertness (as compared with
platinum) in corrosive environments, lower price, hardness (three times
*
Done by B. Tani.
-------
142
900
800
700
- 600
cc
UJ
Q_
500
400
300
200
iooi
0
5 25
2 20
CO
I l5
o
o
CO
10
0
DURING THIS PERIOD, ANALYZERS WERE NOT STABLE
DUE TO GRADUAL CHANGE IN INLET GAS MIXTURE FLOW
02
. ll I I I I
A—*—A—A—A_.
A—A—
0 10 20 30 40 50 60 120 180 240 300 360 420
TIME, min
Fig. 83. Bed Temperature and Effluent Composition
in a Typical Continuous Two-Batch Coal
Combustion Experiment. Batch No. 1.
-------
143
900
800
700
_>
r600h
UJ
500
400 i
300
200
100
0
r
I- I
1
o 25
o
.DURING THIS PERIOD, ANALYZERS WERE NOT STABLE
/DUE TO GRADUAL CHANGE IN INLET GAS MIXTURE FLOW
,02
/C02
***-A_A_A_A£A_A_A_A_A_A_A_A_A-A-A-A
10 20 30 40 50 60 120 180 240 300 360 420
TIME, min
Fig. 84. Bed Temperature and Effluent Composition
in a Typical Continuous Two-Batch Coal
Combustion Experiment. Batch No. 2.
-------
144
to
o
15
x
o
10
LU
O
O
CVJ
O
15 20 25
Go CONC. IN INLET COMBUSTION MIXTURE
OF OXYGEN AND NITROGEN,%
Fig. 85. Relationship of 02 Content of Inlet Combustion
Mixture to Oa Consumption and COz Formation in
Combustion of Coal
harder than platinum at room temperature), and its technical availability
for electroplating. After several runs of experiments using this electro-
plated cold trap, however, the rhodium metal was found to peel off gradually
from the surface of the cold trap, resulting in the acidic attack on the
surface of the cold trap again. This acidic corrosion problem was finally
solved by coating the surface of the cold trap with a layer of porcelain
enamel (about 0.01 cm).
In a series of systematic studies completed in this reporting
period, the transport of alkali metals from the combustion of high-chlorine
coal was examined. Study of high-chlorine coal is of interest because of past
findings obtained from the operation of boiler furnaces, that is, the chlorine
content of coal is closely related to its fouling and corrosive effects on
the fire-side of a boiler furnace. Data has been obtained showing that the
rate of deposition in boiler furnaces does not become significant until the
chlorine content exceeds about 0.3%. Because naturally occurring high-
chlorine coal was not available in this laboratory, a simulated high-chlorine
coal was used in this study. The simulated high-chlorine coal was synthesized
by impregnating coal with NaCl, using waster solution. The choice of NaCl
instead of other chlorine compounds to increase the chlorine content of the
coal to the desired concentration is based on a general agreement that chlorine
in coal is largely present as a chloride, especially as a sodium chloride.
The NaCl used was 99.5% purity A.C.S. reagent.
-------
145
The first series of experiments (SL-1 series) was carried out to
burn 0.5% by wt NaCl-impregnated Herrin No. 6 seam coal from Montgomery
County, Illinois. It is a high-volatile bituminous B coal containing 3.68 %
S, 0.135% Na, 0.195% K, and about 0.1% Cl34 on a dry basis. The proximate
analysis on a dry basis of this coal is 38.22% volatile matter, 45.13% fixed
carbon, and 16.54% ash. In this series of experiments, five batches of
-20 +40 mesh coal were burned at the same experimental conditions—850°C
atmospheric pressure, and a flow of 8% 02 in N2 mixture at a flow rate of
6.0 L/min. The cold trap and the A1203 filter were not removed until all
five batches of coal had been burned. A total of 215 g of coal was combusted
in this series of experiments.
To obtain quantitative information on sodium transport during
combustion, the cold trap was carefully rinsed with distilled water after
this series of experiments to collect all deposits on the surface for analysis.
The A1203 filter was leached, first with distilled water, and then with 5%
HC1 solution at a gentle boiling temperature for two hours. Both leaching
solutions were collected for analysis. (Before it was used in the combustor,
this A1203 filter had been new and had been first treated as described
above for 15 hr to remove both water- and acid-soluble salts, and then it
had been heated in a muffle furnace at 900°C in an air flow to completely
remove the absorbed HC1.) The combustion residues left in the combustion
boat after each experiment were also collected for analysis. All analyses
were done by atomic absorption.
The material balance for sodium element collected on the A1203
filter, collected on the cold trap, and retained in the ash bed has been
computed. Results show that the total sodium in the "output" is about 23%
higher than that in "Input." This wide spread may be compared with +5%,
the analytical error obtained when analyzing sodium by the atomic absorption
method. Since this analytical error can not account for the discrepancy
in the material balance, possible sources of error verified later are:
(1) nonhomogeneity of the coal sample and (2) contamination of the combustion
ash residue by the mortar and by both "Joy" and "Comet" detergents used for
cleaning purposes during the experiments. In contrast, the total amount of
sodium collected from the cold trap and the A1203 filter shows only 0.2%
of the sodium vaporized from the coal bed.
To study whether some trace elements in coal vaporized at the
experimental conditions (850°C and atmospheric pressure) and to obtain the
chemical compositions of the A^OS filter, (1) condensates collected on the
cold trap and (2) A1203 filter powder scraped from the filter element before
and after the SL-1 series of experiments were analyzed, using emission
spectrometry .^ The analytical results show that aluminum and silicon are
the major constituents of all filter scrapings. The source of other elements
present at a trace level is the chemical binder used in the manufacture of
the filter. Within the analytical accuracy and the possible inhomogeneity
*
.Done by R. Bane.
Work done by J. Paris.
-------
146
of the filter, the results show that except for possible volatilization of
manganese and lead, the composition of the filter generally remains the
same upon exposure to the flue gases. Several elements were also found
in trace amounts in the condensates collected on the cold trap. Since all
elements found in the condensates were also found in the A1203 filter
scrapings, these elements might have volatilized from the A1203 filter,
instead of the coal bed. Further study is required to verify this.
The combustion temperature is known to be an important factor
affecting the evolution of alkali metals during coal combustion. To investi-
gate this effect, the same 0.5% NaCl-impregnated Illinois Herrin No. 6 coal
as was used in the SL-1 series was combusted at 900°C and atmospheric
pressure. Two series (SL-2 and SL-3 series) of experiments were carried
out at these experimental conditions. The only difference between SL-2
and SL-3 series was that a porcelain enamel-coated cold trap was used in
the SL-3 series, whereas an uncoated stainless steel cold trap was used in
the SL-2 series. In each series of experiments, five batches (a total of
237 g in SL-2 and 228 g in SL-3) of coal were burned in a flow of 8% 02 in
N2. The flow rate was 6.0 L/min. The A1203 filter, the condensates collected
on the cold trap, and the ash residues left in the combustion boat were
treated the same as were those in the SL-1 series.
At the end of the SL-3 series of experiments, the entire cold trap
was completely free of bluish products which, in previous work of this study,
had been formed by the attack of acids on the uncoated stainless steel cold
trap. The cold trap was observed to be intact. These observations indicate
that the porcelain enamel coating was effective in preventing acidic attack
on the stainless steel under the experimental conditions. The surface of
the cold trap finger appeared to be substantially clean, and no condensate
and/or ash could be collected mechanically; however, scattered liquid drops
were observed on the surface. They showed both strong acidic reactions when
tested with pH paper and the presence of SO?" and Cl~ when tested with
solutions of BaCl2 and AgNOs, respectively. These results indicate that the
liquid condensates on the cold trap finger very likely were H2SOi+ and HC1.
The material balances for sddium and potassium for the SL-2 and the
SL-3 series of experiments are shown in Tables 27 and 28. As noted in these
tables, fairly good agreement is observed for both series of experiments,
indicating the reproducibility of the experiments. However, the totals of
sodium and potassium in the "Output" are again higher than those in the
"Input," especially for potassium. As discussed in a previous paragraph,
it is believed that nonhomogeneity of the coarse coal samples used in these
experiments and contamination of the ash residues by the mortar and by both
"Joy" and "Comet" detergents used for cleaning purposes account significantly
for the errors observed in the tables. However, because of the large dis-
crepancy in the material balance for potassium, there may be unknown sources
of potassium contamination.
Tables 27 and 28 indicate that about 1% of the sodium and 2% of
the potassium evolve from the coal bed and that most of the sodium and
potassium evolved is captured by the Al20s filter. Upon comparison with 0.2%
Na evolution observed in the SL-1 series of experiments, these results show
-------
147
Table 27. A Material Balance of Sodium from Combustion of
Illinois Herrin No. 6 Goal Impregnated with
0.5 wt % NaCl. Combustion was at 900°C and
atmospheric pressure.
a
mg
SL-2 Series SL-3 Series
In Coal 312.5 312.5
Added as NaCl 492.0 492.0
Total 804.5 804.5
Output
Collected on Cold Trap 0.3 0.6
Captured by A1203 Filter
(a) Water Leaching Solution 4.7 1.7
(b) Acid Leaching Solution 3.6 3.3
8.6 5.6
Left in the Combustion Residue
(a) Water Leaching Solution
(b) Acid Dissolution
Total
92.4
740.9
833.3
841.9
100.7
742.5
843.2
848.8
o
Obtained by an atomic absorption method; estimated precision
is +5%. Analyses done by Ralph Bane.
only a slight effect of combustion temperature on the evolution of alkali
metals from the combustion of Herrin No. 6 coal from Montgomery County,
Illinois. Tables 27 and 28 also show that both sodium and potassium are
essentially retained in the ash bed, especially in compound forms that are
not soluble in water. Results similar to these were also reported by the
Combustion Power Co. in California when a mixture of 221 samples of Illinois
No. 6 coal was combusted intheir CPU-400 fluidized-bed combustor at 870°C.35
The reactions that cause NaCl to be retained in the ash may be
very complicated. Possible chemical reactions are those in which NaCl,
silica, and metal oxides (such as CaO, A1203, or Fe203> react to form end
products with high melting points. Examples of these products are devitrite
(Na20-3CaO'6Si02), Acmite (Na20'Fe203-4Si02), and sodium aluminum silicates
«2Si02). Some of the reactions involved are:
2 NaCl + S + 1 1/2 02 + H20 -> Na2S04 + 2 HC1
Na2SOit + 3 Si02 -+ Na20-3Si02 + S02 + 1/2 02
2 NaCl + 3 Si02 + H20 ->• Na20-3Si02 + 2 HC1
-------
148
Table 28. A Material Balance of Potassium from Combustion
of Illinois Herrin No. 6 Coal Impregnated with
0.5% NaCl. Combustion was at 900°C and atmo-
spheric pressure.
a
nig
SL-2 Series SL-3 Series
In Coal 475.0 475.0
Total 475.0 475.0
Output
Collected on Cold Trap
Captured by A1203 Filter
(a) Water Leaching Solution
(b) Acid Leaching Solution
Left in the Combustion Residue
(a) Water Leaching Solution
(b) Acid Dissolution
0.1
1.3
2.3
3.7
4.7
577.9
582.6
Total 586.3
0.4
1.9
7.6
9.9
5.0
584.3
589.3
599.2
0
Obtained by an atomic absorption method; estimated precision is
+5%. Analyses done by Ralph Bane.
Na20«3Si02 + 3Si02 + 3 CaO -> Na20«3CaO6Si02
Na20-3Si02 + Fe203 + Si02 -> Na20-Fe203«4Si02
Na20-3Si02 + A1203 -> Na20-Al203-2Si02 + Si02
2 NaCl + H20 + Al203-2Si02 -*• Na20-Al203-2Si02 + 2 HC1
Mineral matter in coal is the source of silica and metal oxides for all
these reactions.
b. Charcoal Combustion Experiments
Sodium chloride has a significant vapor pressure at both 850°C
(1.55 mm Hg36) and 900°C (3.39 mm Hg36) . Significant amounts of NaCl are
expected to vaporize if silica and metal oxides are not available or are
available in inadequate quantity to fix NaCl by the reactions shown above.
This appeared to be the case when a mixture of NaCl and coconut charcoal
was combusted. A mixture of 20 g of 8 to 12 mesh activated charcoal and
1 g NaCl has been burned under the same experimental conditions as those of
the SL-1 series. At the end of the experiment, substantial amounts of NaCl
(which was identified by X-ray diffraction) were collected on the cold trap,
indicating that NaCl had vaporized at the experimental conditions and was
condensed on the cold trap. Since the charcoal has low sulfur (0.4%) and
-------
149
ash (1.3%) contents for retaining the NaCl in the bed, NaCl vaporized, as
would be expected .on the basis of its vapor pressure.
In the experiment with activated charcoal, the combustor pipe for
the combustion and filtration sections was maintained at 800°C throughout
the experiment. The purpose was to keep NaCl vapor from condensing on the
pipe before it reached the cold trap. At the end of this experiment, all
stainless steel sheaths of the thermocouples exposed to hot flue gases were
observed to be severely corroded, and substantial amounts of black scale
had spalled off the combustion and filtration sections. These phenomena had
not been observed when charcoal alone or coal was burned under similar
experimental conditions. Apparently, the severe corrosion observed in this
experiment resulted from attack of the NaCl vapor on the metal surfaces of
the thermocouple sheaths and the combustor pipe. The corrosive property
of NaCl vapor has also been reported in the technical literature.37"39
Alexander"" proposed that NaCl destroys the normally protective chromium-
iron spinel oxide layer by the reaction:
24 NaCl + 10 Cr203 + 9 02 + 12 Na2Cr04 + 8 Cr2Cl3
In a fluidized-bed coal combustion system, the addition of a small
amount of NaCl to the bed has been shown to improve the S02 absorption
characteristics of limestones. However, the experimental results reported
above show that the use of NaCl may also dangerously increase the potential
of the flue gases to corrode the metal surfaces of components located down-
stream from the combustion system.
As shown in the preceding paragraphs, the samples in the SL-2 and
SL-3 series of experiments were believed to be contaminated in the mortar
used for grinding (i.e., from the commercial detergents used for cleaning).
To investigate this contamination, another set of experiments has been
carried out using activated coconut charcoal in place of coal. This set
of experiments was also designed to study quantitatively the performance of
the cold trap and the characteristics of the A12C>3 filter with respect to
its ability to retain alkali metal compounds.
Two experiments were performed: SL-4 and SL-5. For both experi-
ments, the charcoal was impregnated with 0.5 wt % NaCl, using a water
solution, and then combusted at 900°C and 3-psig pressure using an inlet
combustion gas of 30% 02 in N2 at a flow rate of 3.2 L/min. A section of
glass wool was packed inside the flue gas exhaust line right at the end
cap of the combustor. This was used as a second filter trap, called a down-
stream filter, to collect any alkali metal compounds not captured by the
upstream A1203 filter and cold trap. The cold trap was air-cooled in this
set of experiments; the temperature at the tip of the cold trap was controlled
at about 200°C. The Al20s filter was used in the SL-5 experiment but not
the SL-4.
The material balances for sodium for both experiments are given in
Table 29. Within the limits of analytical and experimental errors, very
good material balances were obtained for both experiments. No commercial
detergents were used, and the ash residues were not ground in the mortar;
-------
150
Table 29. Material Balance of Sodium from Combustion of Activated
Coconut Charcoal Impregnated with 0.5 wt % Nad.
Combustion was at 900°C and atmospheric pressure.
Q
Sodium, mg
SL-4 SL-5
(1) in Charcoal 13 13
(2) Added as NaCl 39 39
Total 52 52
Output
(3) Collected on Cold Trap 5 (10%) 2 (4%)
(4) Captured by A1203 Filter
(5)
(6)
(a) Water Leaching Solution
(b) Acid Leaching Solution
Retained in Combustion Residue
(a) Water Leaching Solution
(b) Acid Dissolution
Collected by Downstream Filter
-
-
~5
26
14
40
6
Total 51
(10%)
(77%)
(11%)
1
6
9
30
13
43
1
53
(17%)
(83%)
(2%)
o
Obtained by an atomic absorption method; estimated precision is +5%.
therefore, these results appear to support the findings that the samples in
both the SL-2 and the SL-3 series of experiments were contaminated by
detergents from the mortar and pestle.
It may be seen that the quantities of sodium retained in the
combustion residue in compound form that dissolved only in acid solution
(item 5b in the table) are about the same as those in the charcoal (item 1);
therefore, the sodium vaporized during the experiments (items 3+4+6) was
from the NaCl added (item 2). In the SL-4 experiment, in which no A1203
filter was used, about 21% of the NaCl (items 3+4+6) was vaporized, and
the cold trap captured slightly less than 50% of the NaCl vaporized.
When the A1203 filter was placed upstream from the cold trap
(experiment SL-5), 74% of the NaCl vaporized was captured by the A1203 filter
[item 4/(item 3 + item 4 + item 6)]; the sodium captured by the filter was
essentially in compound forms not soluble in water, but soluble in 5% HC1
solution. The A1203 filter was maintained at about 800°C during this
experiment. These results indicate that the Al20o filter reacts with the
NaCl to (probably) form sodium alumina silicates, most of which are known
to be insoluble in water but soluble in acid. The small amount of sodium
compounds soluble in water (item 4a) is possibly in the form of simple sodium
silicates or sodium aluminate (NaA102).
-------
151
Results of X-ray diffraction analysis of the condensates collected
on the cold trap finger (from SL-4) showed that a significant amount of KC1,
in addition to NaCl, was present. In order to determine the source of the
KC1 and also to look into the chemical behavior of KC1 during the combustion
of charcoal, all samples collected from these two experiments were also
analyzed for potassium. The material balances for potassium in both SL-4
and SL-5 are given in Table 30.
Table 30 shows significantly different potassium contents in two
different samples of the activated coconut charcoal. They were obtained
from analyses of two 20-g charcoal samples collected from the same bottle
of sample. The great variation observed in potassium concentrations in the
charcoal samples indicates the nonhomogeneity of the charcoal (-8 +12 mesh)
used in these experiments.
Table 30. Material Balance of Potassium from Combustion of
Activated Coconut Charcoal Impregnated with 0.5
wt % NaCl. Combustion was at 900°C and atmo-
spheric pressure.
o
Potassium, mg
SL-4
SL-5
(1) In charcoal (one analysis
for each of two 20-g
samples)
130;171
130;171
Output
(2) Collected on Cold Trap
(3) Captured by A1203 Filter
(a) Water Leaching Solution
(b) Acid Leaching Solution
(4) Retained in Ash Residue
(a) Water Leaching Solution
(b) Acid Dissolution of Ash
13 (12%)
c
c
2 (2%)
75
II
88 (78%)
(5) Collected by Downstream Filter 11 (10%)
Total 112
13 (12%)
77
"93 (85%)
_A (1%)
109
Obtained by an atomic absorption method; estimated precision
is +5%. Analysis by Ralph Bane.
A separate 20 g sample of charcoal was combusted in each
experiment.
No A1203 filter was used.
-------
152
To obtain a more homogeneous sample, the particle size must be
reduced. The activated coconut charcoal is a product of MCB (Metheson
Coleman & Bell Manufacturing Chemists), Norwood, Ohio. The information
provided by MCB indicates that the potassium is an inherent element in the
charcoal. The exact compound form in which the potassium exists in the
charcoal is not known; however, X-ray diffraction results showed that
potassium was present as chloride vapor in the combustion gas.
As for the material balances of sodium (Table 29), the repro-
ducibility obtained for the material balances for potassium was good (Table
30). Table 30 also shows a possible loss of potassium in the combustor
system, which needs to be further investigated. Despite this, it is
noticeable that the material balances of potassium are very similar to
those of sodium insofar as the distribution of potassium in the various
"Output" categories is concerned. This similarity shows that the chemical
behavior of KC1 during combustion is similar to that of NaCl. The potassium
captured by the A1203 filter was essentially in compound forms not soluble
in distilled water but soluble in 5% HC1 solution. This indicates that the
A1203 filter material reacts with KC1 similarly to the way it reacts with
NaCl, probably forming potassium alumina silicates.
As mentioned above, the A1203 filter shows high reactivity with
the KC1 and NaCl vapors under the experimental conditions. This suggests
that solid material with a chemical composition similar to that of the A^Os
filter would be a good solid sorbent for removing alkali compounds from hot
combustion gas of coal. However, in this study, the material removed in
this Al£03 filter consists not only of particulates but also of most of the
NaCl vapor (and probably other alkali compounds too); as a consequence, with
the Al20s filter in place, the quantity of condensation collected by the
cold trap is hardly enough for analysis. A filter that captures only partic-
ulates from flue gas must be fabricated from another type of filter material.
c. Lignite Combustion Experiments
Experimental results so far obtained for the combustion of Illinois
Herrin No. 6 coal (Tables 27 and 28) and of activated coconut charcoal (Tables
29 and 30) have shown that vaporization of alkali metals during combustion
is related to the ash content of the coal. In order to investigate this
relationship and the transport behavior of the alkali metals from the com-
bustion of low-rank coal, this study has been extended to the burning of
lignite from the Glen-harold seam of North Dakota. The proximate analysis
on a dry basis of this lignite is 38.93% volatile matter, 52.97% fixed
carbon, and 8.10% ash. Lignite with a particle size range of -20 +40 mesh
was used in this study. It was dried at 110°C in a nitrogen flow before
being combusted.
Three series of experiments were completed. In all of these
experiments, lignite was combusted with air at a flow rate of 3.5 L/min.
Each series of experiments consisted of five batch combustion runs, and in
each run 50 g of dry lignite was burned. In two of the three series of
experiments (SL-6 and SL-8), lignite was impregnated with 0.5% by weight
NaCl, using water solution to increase the chlorine content of the lignite
-------
153
and thereby simulating high-chlorine coal. To allow comparison, plain lignite
was burned in the third series of experiments (series SL-9). The material
balances for sodium for these experiments are tabulated in Table 31.
Table 31. Material Balance of Sodium from Combustion of
Glen-harold Lignite, North Dakota
Lignite was combusted in air at atmospheric
pressure.
Experimental Series
Combustion Temp, °C
Amount of Coal Combusted, g
Amount of NaCl added, g
Input per Series of Experiments
(1) In Lignite
(2) Added as NaCl
SL-8
850
250
1.25
SL-6
900
250
1.25
Sodium, mgc
SL-9
900
250
None
1670 1670 1670
490 490 —
Total: 2160 2610 1670
Output per Series of Experiments
(3) Left in Ash Residue
(4)
(5)
(6)
(7)
(a) Water Leaching Solution
(b) Acid Dissolution of Ash
Collected on Cold Trap
Captured by Downstream Filter
Total:
Lost
Total Vaporized
[ (4)+(5)+(6) ]
1180
730
1910
57
30
1997
163
250
(11.5%)
1020
850
1870
71
26
1967
193
290
(13.5%)
910
690
1600
N.A.b
N.A.
70
(4.2%
aObtained by an atomic absorption method; estimated precision is +5%;
.done by Ralph Bane.
Not determined due to tar and soot condensation.
In series SL-6 and SL-8, combustion of lignite was complete, and
substantial amounts of condensates were collected on the cold trap finger
and in the downstream filter; however, in Series SL-9, in which lignite had
not been impregnated with NaCl solution, significant amounts of tar and
soot were condensed on the cold trap finger. For this reason, no condensates
could be collected for analysis, as shown in Table 31. Condensation of
tar and soot was due to incomplete combustion of volatile matter, which
rapidly evolved during the early stage of combustion. Since all three
series of experiments were conducted under the same operational controls
during combustion, the observed tar and soot condensation in the SL-9 series
indicates the effect of NaCl on the combustion characteristics of lignite.
-------
154
NaCl appears to suppress the evolution of volatile matter from lignite
during the early stage of combustion—possibly by blocking of the pores
in lignite by NaCl.
As shown in Table 31, when untreated lignite was combusted at
900°C (series SL-9), sodium essentially remained in the ash residue, mostly
in compound forms that are soluble in distilled water. Sodium has been
shown to be present in lignite primarily as salts of humic acids,1*0 to be
uniformly distributed within lignite,1*1 and not to be readily removed by
washing with distilled water. Apparently, the combustion results for
lignite show that most of the sodium in lignite was converted to compound
forms that can be readily leached out with distilled water after combustion.
Among those compounds are (possibly) sodium sulfates, chlorides, simple
silicates, and aluminates.
Also shown in Table 31 is the vaporization of 11.5 wt % and 13.5
wt % of the sodium, respectively, when lignite impregnated with a 0.5 wt %
NaCl was combusted at 850°C (SL-8) and 900°C (SL-6). Comparison with the
results for series SL-9 shows that sodium vaporized essentially from the
NaCl added.
The condensates, which were primarily collected on the hemispheric
surface of the cold trap finger, were gray-white. X-ray diffraction analysis
of the condensate indicated that the sodium was oresently mostly as NaCl and
to a smaller extent as Na2SC%. The relative quantities of NaCl and Na2SOi4
are being determined by wet chemistry methods and will be reported.
d. Relation of Coal Ash Content to Sodium Vaporization
Figure 86 is a plot, as a function of the percent ash in each
coal, of the percent sodium vaporized from the combustion of 0.5% by wt NaCl-
impregnated Herrin No. 6 coal, lignite, and coconut charcoal at 900°C. It
can be seen that a fairly good linear relationship exists between the ash
content of coal and the quantity of sodium vaporized during combustion. The
greater the amount of mineral matter in coal, the more the NaCl is tied up
in the ash and, therefore, the less NaCl is vaporized. The amount and form
of mineral matter vary widely from one coal to another. Mineral matter in
Illinois Herrin No. 6 coal consists of more than 50 wt % clay minerals and
about 15 wt % S102.1*3 On the other hand, lignite is estimated to contain
20 to 40 wt % clay and SiC^-**1 It is believed that clay minerals are
responsible for tieing up the sodium in the ash. This needs to be further
investigated.
Work continues on study of the transport of alkali metals from
the combustion of other coals of different ranks. The effect of operating
variables on the transport of alkali metals and the effectiveness of some
clay minerals on the retention of alkali metals will be quantitatively
evaluated. Among the operating variables are combustion temperatures, ratio
of coal to additive (limestone or dolomite), and type of additive.
-------
155
o HERRIN NO. 6 COAL
GLEN-HAROLD LIGNITE
ACTIVATED COCONUT
CHARCOAL
0 2
Fig. 86
8 10 12
ASH, %
Relationship between Ash Content of Coal
and Sodium Vaporized during Combustion
at 900°C of Coal Impregnated with 0.5%
NaCl.
-------
156
TASK F. FLUE-GAS CLEANING STUDIES
1. Evaluation of On-Line Light-Scattering Particle Analyzers
(J. Montagna, G. Smith, G. Teats, H. Lautermilch, and S. Smith)
In the development of pressurized fluidized-bed combustion systems,
a continuous on-line particle analyzer for the flue gas would be useful
(1) in measuring the efficiency of particulate-removing devices (cyclones
and filters), (2) for evaluating particulate gas loading characteristics
to establish gas turbine performance with different particulate loadings,
and (3) to protect turbines in an industrial-size plant. In a pressurized
FBC system, the flue gas will be at ^900°C and vLOOO kPa (vLO atm) between
the boiler and the turbine. If no on-line particle analyzer is available,
routine batch sampling of the hot gas (using inertial impactors, for example)
will be necessary. Batch sampling from a pressurized hot flue gas environ-
ment is difficult; another disadvantage is the long time lag between
sampling and analysis of samples.
Two on-line light-scattering particle analyzers are to be evaluated in
the ANL fluidized-bed combustion system under ERDA sponsorship. Both instru-
ments use a laser light source. The experiments with the Spectron Development
Laboratory (SDL) split laser beam particle morphokinetometer (PM) have been
completed, and the instrument has been returned to the manufacturer. Prelim-
inary results are reported comparing (1) measurements with the PM analyzer,
(2) Coulter counter measurements of steady state samples, and (3) cascade
impactor analyses. A meeting of representatives of ANL and SDL was held on
January 27, 1977, to discuss the results. Final analysis of the results is
being performed.
The Leeds and Northrup single laser beam analyzer was due to arrive
at ANL in February 1977; however, because of final adjustments being made
by Leeds and Northrup, its shipment has been delayed.
a. Principles of the SDL Particle Morphokinetometer
The particle morphokinetometer (PM) measures the sizes of particles
and their velocities by measuring the light scattered from each particle
as it crosses an interference pattern generated by the intersection of two
laser beams. A schematic of a typical PM instrument is given in Fig. 87.
The laser beams are directed radially into the flue-gas duct of the ANL PDU
combustion system through specially designed windows. The region of measure-
ment, called the PM probe volume or sample space, is at the center of the
duct where the two coherent beams intersect and generate the interference
pattern. The light scattered from the sample space is detected and changed
to an electric signal of the type shown in Fig. 87. The light may be detected
in either a forward or backward observation mode. Forward-scattered light
is being detected in the ANL evaluation.
The parameter measured to determine particle velocity is the time
period (see Fig. 87) of the scattered light signal.
U)
-------
157
INTERFERENCE FRINGES
l/e< MODULATION CONTOUR
2 ENLARGED VIEW
OF REGION OF
CROSS-FOCUS POINT
BACKSCATTER
OBSERVATION
MODE
FORWARD SCATTER
OBSERVATION MODE
PMT
TIME
Fie 87 Spectron Development Laboratory's PM Analyzer
System for Velocity and Particle Size Measurement
where
v = particle velocity . .
6 = distance between fringes in the sample space (fringe period)
T = time period of the scattered light
As the size of the scattering particle increases relative to 6 (which is
controlled by the intersection angle of the laser beams), the illumination
of the particle becomes less uniform, and it is averaged over the cross-
sectional area of the particle. Nonuniform illumination of the particle
results in a reduction'in the contrast or visibility of the scattered light
signal. Visibility, V, is defined as:
V =
I - I .
max mm
I + I .
max mm
(2)
where I is the maximum value of intensity in a period of the s""ered
Sght fr1- a particle, and Imin is the next successive mini«™-
shown that V can be equivalently expressed as the ra 10 of the AC
divided by the DC amplitude of the scattered signal (see Fig. 87).
-------
158
By imposing some geometric limitations on the particles (D is the
particle diameter) that will be measured in terms of the radius of the laser
beams, b
D < 0.2 b (3)
and on the distance between fringes, 6
6 < 0.2 b (4)
a simplified analytic expression of visibility, V, has been obtained for
spherical particles.
V * 2[J1(TTD/6)]/(Trd/6) (5)
where Ji is a first order Bessel function of the first kind.
The visibility, V, for a cylindrical particle has been obtained
as:
V ^ sin (irD /6)/(irD 6) (6)
*
where D is the length of the major axis of the cylindrical particle.
Equations 5 and 6 are plotted in Fig. 88 to illustrate the features of the
visibility in particle size measurement. From this plot, it is apparent
that particle size cannot be unambiguously determined unless V >^ 0.15 for
spherical particles and V ^0.23 for cylinders. This imposes an upper bound
on the size of particles than can be measured for a given fringe period.
The interference fringe periods and their respective detectable spherical
particle size ranges that will be used in this evaluation are given in
Table 32. A more detailed description of the principles of a Spectron PM
analyzer is available in the literature.****
Table 32. Selected Interference Fringe Spacings and
the Corresponding Measurable Spherical
Particle Size Ranges
Fringe
Period,
ym
71.4
22.3
2.94
Min. Particle,
Diameter,
pm
4.9
1.5
0.2
Max. Particle
Diameter,
ym
74
23
3.1
-------
159
i.o
0.9
0.8
0.7
» 0.6
0.5
" 0.4
0.3
0.2
0.1
0.0
0.0
Major Axis
Of Fringe
r
0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.6 1.8 2.0
Fig. 88. Visibility as a Function of Particle Size
and Fringe Period for Two Particle Shapes
b. Procedure for Comparative Flue-Gas Particle Measurements
The flue gas system in the ANL fluidized-bed combustion system
(PDU system) has been modified for these evaluations, as shown in Fig. 89.
Windows for particle analyzers have installed in two locations; one pair
is upstream from the primary cyclone. The other windows are near the system's
outlet, and there is the capability of routing the flue gas past these
windows, either upstream or downstream from the metal filters. With this
arrangement, the coarse entrained particles from the combustor, the smaller
particles escaping the two cyclones, and the smallest particles leaving the
metal filters (representative of particles that might enter turbines) were
sized. Downstream from each window location, sampling ports have been
installed that allow particle size analysis of representative samples with
cascade impactors. Also, steady state particle samples were obtained from
the cyclones and test filter.
-------
160
AIR j
PREHEATER
PA
-I-
CYCLONES FILTERS
COAL
SORBENT
V
AIR-*
V
SPO SI S2
COMBUSTOR
i
S4S5;
V-Cr0-
S3
-TEST
FILTER
OSP
PA
PA-WINDOWS FOR PARTICLE
ANALYZERS
SP-SAMPLE PORT FOR SAMPLING
WITH SAMPLE CYCLONE AND
CASCADE IMPACTOR
Fig. 89. Schematic of FBC System with Modified Flue-Gas System
The particle size measurements were obtained (1) with the on-line
Spectron PM Analyzer, (2) using an Anderson impactor, and (3) from steady
state samples obtained from the cyclones and/or the test filter. The steady
state samples were analyzed by sieve anlaysis and a Coulter counter. Here-
after, samples analyzed by this method are referred to as Coulter counter
samples. It was assumed that all observed particles were spheres of equal
density (apparent particle density, 1 g/cm3, for Anderson impactor) and
that the particles observed with the on-line instrument (PM) were identical
to those that were mechanically removed from the system and later analyzed.
The validity of these assumptions will be discussed in subsequent reports as
more results are compared. Since the density is assumed to be constant for
all particle diameters, the fractional volume and mass distributions are
equivalent.
Cascade impactors are the devices used most often for obtaining
airborne particle size distributions from process or ambient air in the size
range, 0.3-30 ym. In this study, an Anderson cascade impactor was used to
obtain combustor flue gas particle size distribution data for flue gas
samples. Each stage of the impactor consists of equidiameter orifices
followed by a target plate for collecting the particles. Smaller orifices
are used in successive stages, and thus the size of the particles collected
in successive stages becomes smaller. The particle size distributions are
calculated from experimental data by relating the mass collected on each
stage to the corresponding stage diameter. The impactor designs (including
the design of the Anderson impactor) are based on the theoretical development
of Ranz and Wong.1*5
-------
161
The sampling system in the FBC system for the cascade impactor is
illustrated in Fig. 90. The particle-laden flue gas flows by the optical
windows (located in the line downstream from the combustor or cyclones or
metal filter) where the particles are sized with a light-scattering particle
analyzer (Spectron PM). Next, the off-gas is expanded to reduce the velocity
to that required for isokinetic sampling by the cascade impactor. The tip
of the sampling probe (0.78-cm ID) is machined to enhance aerodynamic
stability near the probe entrance. The sample line is electrically heated
to maintain the temperature of the gas sample above its water dew point. The
cascade impactor is contained in a heated pressure shell to permit sampling
from the pressurized (3 r.o 8 atm) combustion system. After passage of the
gas sample through the impactor, its volumetric flow rate is measured with
a rotameter, after which the gas is depressurized. The gas velocities are
based on the measured gas flows and temperatures.
c. Experimental Evaluation of the PM Particle Analyzer
Some results of particle size measurements obtained with the PM
analyzer for several combustion experiments have been compared with size
distributions obtained for steady state particle samples. The measurements
with the laser instrument were made in the PDU combustion system's off-gas
duct between the combustor and the primary cyclone, where all particles
leaving the combustor were observed. Also, measurements between the secondary
cyclone and the metal filters are reported.
TC - THERMOCOUPLES
PI -PRESSURE INDICATOR
PA -ON-LINE ANALYZER
HL - HEATED LINE
TO FLUE-GAS
SYSTEM
FROM FLUE-GAS
SYSTEM
SAMPLING
PROBE (3/8 in.)
HEAT EXCHANGER
CYCLONE
IMPACTOR
—PRESSURE SHELL
Fig. 90. Sampling System for Flue Gas Particles
-------
162
The conditions for the first combustion experiment (SGL-1) for
this evaluation are given in Table 33. In this experiment, the distribution
of particles leaving the combustor was measured. This experiment was per-
formed in two one-day segments. Two fringe periods were used on Spectron's
PM analyzer in the reported experiment, 71.4 ym and 22.3 vim, on separate day
segments of this experiment. With a fringe period of 71.4 pm, the measurable
spherical particle diameter range is ^5-74 urn, and with a fringe period of
22.3 ym, the range is ^1.5-23 ym. In this experiment, Sewickley coal was
combusted in a fluidized bed of Greer limestone at 805°C, and the measured
particles consisted of limestone fragments, coal ash, and unburned coal.
Table 33. Experimental Conditions for a Combustion Experiment in the
Evaluation of the SDL Particle Morphokinetometer (PM)
Location of PM Windows: Between PDU combustor and
cyclones, SGL-1.
Between second cyclone and
filter, SGL-2C.
Sorbent: Greer limestone Coal: Sewickley
System Pressure, kPa: 308 (3 atm)
Fluidizing Gas Velocity, m/s: 1.0
Conditions Near Probe
at Sampling Duct
Conditions at
PM Windows
Exp
SGL-1
SGL-2C
Combustor
Temp,
°C
850
855
Gas
Velocity,
m/s
5.2
11.76
Temp,
°C
350
123
Gas
Velocity
Vfg.
m/s
-
3.26
Temp,
°C
-
110
Ratio of Duct
Velocity to
Probe Gas
Velocity,
vf /vs
rg s
-
0.99
In the first comparison, only the mass distributions inside each
measurable particle size range (1.5-23 ym and 5-74 pm) of the Spectron PM
analyzer were compared with the corresponding distributions obtained with
the Coulter counter. Comparisons of the partial cumulative mass distributions
by the two methods for the small (1.5-23 urn) and the large (5-74 pm) PM size
ranges are given in Figs. 91 and 92, respectively. The mass mean particle
diameter obtained was 8.5 jam with the Coulter counter and 20 urn with the PM
analyzer for the small particle size range. For the largp particle size
range, the mass means of the partial distribution were found to be 26 pm
with the Coulter counter and 70 pm with the PM analyzer. From these early
results, it is apparent that the difference between the two measurements is
greater for larger particles. Some reasons for the difference are given
below:
a. The mass loading downstream from the combustor was ^14 grains/
scfm (20 grains/acf), which is quite high. Thus, the chance
that there would be more than one particle in the sample
space of the PM analyzer was high. Over 98% of the signals
-------
163
99
98
95
90
- 80
llJ
§ 70
-I
O 60
~ 50
S 10
S 30
UJ
I"
I '°
" 5
2
I
05
0.2
O.I
V COULTER COUNTER
O SPECTRON PM
10
PARTICLE DIAMETER ,
30
Fig. 91. Partial (1.5-23 pm) Cumulative Mass
Distribution Obtained On-Line with
the Spectron PM Analyzer Compared
with that Obtained with a Coulter
Counter (SGL-1)
99
98
90
#
_- 80
1 7°
o 60
t 50
to
(A 40
<1
z 30
S 30
2
I
0.5
0.2
O.I
^7 COULTER COUNTER
O SPECTRON PM
PARTICLE DIAMETER,
Fig. 92. Partial (5-74 urn) Cumulative Mass
Distribution Obtained On-Line with
the Spectron PM Analyzer Compared
with that Obtained with a Coulter
Counter (SGL-1)
-------
164
were rejected by the PM analyzer because of particle
coincidence interference. (Signal rejection rates between
97 and 90% are considered acceptable by Spectron.)
b. The assumption that the measured particles are spherical
would bias the PM measurements towards large diameters if
the particles are actually nonspherical; the bias would
become more pronounced for large measured particles (see
Fig. 89).
c. The fragile particles might have broken up as the particle
samples were collected in the cyclones and test filter and
as they were prepared for Coulter counter analysis (dis-
persed in an electrolyte). This effect could impose a
bias towards small diameters in the Coulter counter measure-
ments.
These possible sources of the discrepancies in the two particle measurements
will be discussed in a subsequent report upon completion of this evaluation.
The total particle distribution leaving the combustor, in combustion
experiment SGL-1, .was characterized. The fractional mass distribution of
all particles between 1 ym and 1000 ym was obtained by combining Coulter
counter measurements and sieve analyses. They are presented in Fig. 93. The
elutriated-particle distribution is trimodal, with peaks at 750 ym (elutri-
ated sorbent), 70 ym, and 6.4 vim. A similar distribution was obtained by
combining the measurements made with the Spectron PM analyzer with sieve
analysis; it also is presented in Fig. 93. This distribution is also trimodal,
but the mass contributions of the small-diameter fractions are much smaller
because of the possible measurement biases discussed above. The cumulative
mass distributions obtained from these fractional distributions are compared
in Fig. 94. The mass mean obtained with the Coulter counter and sieve
analysis was 54 ym, and the one obtained with the PM analyzer and sieve
analysis was 58 ym.
On-line particle analyzers are intended to be used downstream from
FBC particle removal devices to monitor particle distributions and loading
in the flue gas entering gas turbines. The results obtained in this first
experiment, in which there were large (^14 grains/scf) particle loadings
(upstream from particle-removal devices) , are not representative of those
at a turbine inlet. It is encouraging that the characteristics of the
fractional distributions obtained by the two different methods were the same
and that the discrepancy between the two measurements (PM US Coulter counter)
became smaller for smaller particles (<20 ym). The smaller particles <10 ym
and >1 ym are expected to erode turbine blades.
The conditions for combustion experiment SGL-2C in which the sizes
of particles between the secondary cyclone and metal filters of the PDU
combustion system were measured are given in Table 33. In this experiment,
Sewickley coal was combusted in a fluidized bed of Greer limestone at 855°C,
and the measured particles consisted of limestone fragments, coal ash, and
unburned coal. Particles with diameters between 1.5 and 23.8 ym were sized
with the Spectron PM.
-------
165
COULTER COUNTER
O SPECTRON PM
Q SCREEN ANALYSIS
Fig. 93. Fractional Mass
Distribution of All
Elutriated Particles
from the ANL PDU
Combustor During a
Combustion Experi-
ment (SGL-1)
100
1000
PARTICLE DIAMETER,
99
98
95
90
- 80
70
Z
3
^50
2 40
Z 30
UJ
52°
_j
z I0
5 5
Z
I
0.5
0.2
O.I
V COULTER COUNTER
O SPECTRON PM
O SCREEN ANALYSIS
10
100
1000
PARTICLE DIAMETER,
Fig. 94 . Cumulative Mass Distribution of All Elutriated
Particles from the ANL PDU Combustor During a
Combustion Experiment (SGL-1)
-------
166
In the first comparison, only that part of the mass distribution
within the measurable size range of the Spectron PM analyzer was compared
with the corresponding distribution obtained with the Coulter counter.
Particles outside the Spectron PM range were ignored. The resulting partial
cumulative mass distributions for particles contained in the flue gas exiting
from the secondary cyclone are given in Fig. 95. The mass means of these
distributions are 3.5 ym (Coulter counter) and 17 ym (Spectron PM) . In the
distribution obtained with the Spectron PM, the particle mass population in-
creases sharply above 15 ym. Since at this point, the flue gas has passed
through two cyclones, most particles larger than 10 ym should have been removed.
Hence, the distribution obtained by Coulter counter analysis appears to be more
representative. The total (0.5 ym-100 ym) fractional mass distribution
obtained with the Coulter counter of the solids in the flue gas downstream
from the secondary cyclone is given in Fig. 96. The largest mass fraction
was found to be between 3.0 and 3.8 ym (3.38 urn, geometric mean), which
contained 22.5 wt % of the total (0.2 grains/scf) particulate mass loading.
The total cumulative mass distributions obtained for two consecutive
measurements of suspended particles from SGL-2 with the Anderson cascade
impactor are given in Fig. 97. Aerodynamic mass mean diameters of 2.7 ym
and 3.0 ym were obtained, with the assumption that the apparent density of
all particles was 1.0 g/cm3. (The measured apparent density of sulfated
dolomite is ^1.9 g/cm3, and literature values for devolatilized coal and coal
ash are ^0.6-0.9 g/cm3.) The sampling conditions for the Anderson impactor
samples are given in Table 33. The ratio of duct velocity to inlet sample
probe velocity was 0.99, indicative of isokinetic conditions.
Fig. 95. Comparison of the Partial
(1.5-23.9 ym) Cumulative
Distribution Obtained
On-Line with the Spectron
PM Analyzer with that
Obtained by Coulter
Counter Analysis of a
Particle Sample (SGL-2C)
-------
167
Fig. 96. Fraction Mass Distribution
of Particles Contained in
the Flue Gas Between the
Secondary Cyclone and the
Filter (SGL-2C) Obtained
by Coulter Counter Analysis
P;irl lr It- I)iami-1 «.• r . urn
The total cumulative mass distribution obtained with the Coulter
counter is also given in Fig. 97 . The mass mean diameter was found to be
3.5 urn, which compares favorably with the distribution means obtained with
the Anderson cascade impactor.
From the above results, it appears that the Spectron PM analyzer
with originally supplied calibration is biased towards large diameters.
d. The Effect of the PM Particle Analyzer Calibration on Particle
Measurements
A discussion by ANL and SDL representatives disclosed two possible
effects that would lead to bias in the PM measurements:
a. Large particles could block the response of the electronics
to small particles at relatively high particulate. loadings.
b. The original calibration for spherical particles, which was
obtained with low number densities of mists and aerosols
by microscopic measurements, is greatly influenced by particle
shape and orientation in the sample space.
Because of the difficulties in evaluating these and other effects, an empiri-
cal correlation of the experimental data was used to obtain a best fit for
the calibration curve. In this analysis, it was assumed that the Coulter
counter and inertial impaction measurements were correct. The particle
distribution measurements obtained with the Coulter counter and cascade
impactor were expressed as histograms with intervals equivalent to the PM
analyzer increments. The diameter-dependent factor necessary to force the
fractional contributions of the histogram intervals from the PM analyzer into
agreement with those from the Coulter counter and impactor measurements was
-------
168
99.9
99
98
95
90
uj 80
5
3 70
§ 60
v> 50
< 40
u 30
I 2°
_j
i 10
O.I
0.5
V COULTER COUNTER ANALYSIS
O ANDERSON IMPACTOR
5 10
PARTICLE DIAMETER,
50
100
Fig. 97. Cumulative Mass Distribution of Particles in
the Flue Gas Between the Secondary Cyclone
and the Metal Filter (SGL-2C)
obtained. The ratios of the fractional contributions obtained from the
Coulter counter and impactor measurements to those obtained from the PM
analyzer for a considerable number of measurements were correlated with the
reduced diameter of the interval, D/6. (D is the particle diameter and 6 is
the fringe period of the laser probe volume.) By use of least squares
techniques, the following correlation was obtained:
(7)
In (K) = -4.91 ln(|) - 2.5 [In (|
where K is th,e ratio of the expected mass concentration in the diameter
interval to the measured number of particles within the PM interval. This
correlation gives a fair fit (a correlation coefficient of ^0.88) for the
relationship between K and the reduced diameter, D/6.
The previously discussed experiments (SGL-1 and SGL-2C, see Table
33) were reanalyzed using this empirical calibration.
The measurements from these experiments were adjusted using the
calibration function (Eq. 7). The resulting truncated distributions (actual
particle distributions were wider than the range of the analyzer) are compared
-------
169
with the corresponding truncated distributions obtained by Coulter analysis
in Figs. 98 and 99. These figures also contain the previously reported
distributions which were based on Spectron's originally supplied calibration
function,
K
1/D
(8)
The mass means of the truncated particle size distribution were calculated
to be 26 um (using the Coulter counter measurement) and 47 pm (with the
PM analyzer) for the large PM particle size range. (A 70-um mass mean
was previously obtained with the PM analyzer using the Spectron supplied
calibration.) For the small PM particle size range, the Coulter counter
analysis produced a mass mean of 8.5 pm; that from the PM analyzer was
5.2 um.
99
95
90
80
60
_ A Spectron PM (ANL Calibration)
Spectron PM (Supplied Calibration)
O Coulter Counter
20
10
1
0.5
0.2L-L
/
I I I
10
Particle Diameter, yra
100
Fig. 98. Comparison of the Partial (5-73 pm) Cumulative
Mass Distribution Obtained On-Line with the
Spectron PM Analyzer with that Obtained with a
Coulter Counter (SGL-1)
-------
170
99.9
99.8
99.5
99f-
98
95
90
80
70
60
50
a-«
„• *°
01
£ 30 h
u
•H 20
a
.H
3
J 10-
5 -
2
1
0.5 -
0.2
0.1
A
O
Spectron PM (ANL Calibration
Spectron PM (Supplied Calibration)
Coulter Counter
5 10
Particle Diameter, gra
20
30
Fig. 99.
Comparison of the Partial (1.5-23 pro) Cumulative
Mass Distribution Obtained On-Line with the
Spectron PM Analyzer with that Obtained with a
Coulter Counter (SGL-1)
The results of experiment SGL-2C, in which the particle measurements
were performed on material upstream from the final filters, are presented in
Fig. 100. A fringe period of 23.0 pm was set in the PM analyzer, which
corresponds to the approximate size range 1.5-23.8 pm. The mass mean
particle diameter obtained by the PM analyzer was 8.8 pm and that by the
Coulter counter was 3.5 um.
Although these adjusted analyses continue to show some deviation
between the different measuring techniques, the agreement between the measure-
ments has improved considerably. Further analysis of the remaining experiments
is being performed which will help determine the usefulness of the experi-
mentally obtained calibration. Included in the experiments to be reported
-------
171
99.9
99.5
99
98
o
or
o
CO
CO
90
70
50
30
10
o
O.I
I III
A SPECTRON PM
(ANL CALIBRATION)
SPECTRON PM
(SUPPLIED CALIBRATION)
o COULTER COUNTER
I
2 5 10
PARTICLE DIAMETER,
50 100
Fig. 100. Comparison of the Partial (1.5-23.8
Cumulative Mass Distribution Obtained On-Line
with the Spectron PM Analyzer to that Obtained
by Coulter Counter Analysis of a Particle
Sample (SGL-2C)
are measurements made on suspended virgin limestone particles which were
generated by fluidizing a bed of virgin Greer limestone in the 6-in.-dia
combustor (no coal was combusted). These experiments were performed to
evaluate the effect of different particle densities and chemical compositions
on the measurements of suspended particles in the off-gas from combustion
experiments.
2. Particle Removal From Flue Gas
(W. Swift, G. Teats, S. Smith7 and H. Lautermilch)
In pressurized fluidized-bed combustion, the hot flue gas from the
-------
172
combustor must be expanded through a gas turbine. To prevent erosion of
the turbine blades by particulate matter entrained in the flue gas, the
particulate loading must be reduced to acceptably low levels. What consti-
tutes "acceptably low levels" of particulate loading for a gas entering a
gas turbine is, however, still not fully understood. Estimates of acceptable
loadings range from 0.05 to 0.0005 g/m3 (0.02 to 0.0002 grain/scf). If it
is assumed that the loading in the flue gas as it leaves the combustor is
^50 g/m3 the particle removal efficiency required to meet the estimated
allowable loadings for the turbine would be between 99.9 and 99.999%.
The critical factor in defining an acceptable loading for gas turbines
is the particulate size distribution. Obviously, the finer the particles
entering the gas turbine, the higher the permissible mass loading. Westing-
house, for example, has estimated an acceptable loading of 0.002 grain/scf
(^0.005 g/m3) provided that ^80% wt % of the particulate is smaller than
10 ym in diameter and that ^40 wt % is smaller than 2 urn. If 100 wt % of
the particulate is smaller than 10 ym and ^40 wt % is smaller than 2 ym,
Westinghouse estimates the tolerable particulate loading to be ^0.03 grain/scf
(MJ.07 g/m3).1*6
Existing devices readily adaptable to high-temperature, high-pressure
particulate removal (e.g., conventional cyclones) are not very efficient.
in removing particulate matter with diameters smaller than ^10 ym. Achieving
the "acceptable loading" necessary for PFBC requires, therefore, the develop-
ment of highly efficient methods for removing from flue gas the particulate
solids having diameters between 2 and 10 ym.
Prior to our undertaking an experimental program on removing small
(<10 ym) particles from a flue gas, a literature survey was made of the
existing technology. The history of particulate agglomeration and separation
at high temperatures and pressures was examined, with particular emphasis
on the unique features of the direct-cycle application of fluidized-bed
combustion. The basic long-range mechanisms of aerosol separation were
examined, and the effect of high temperature and high pressure upon usable
collection techniques was assessed. Primary emphasis was placed on those
avenues that are not currently attracting widespread research. This survey
is being published as a separate report.
An experimental program is under way at ANL to test and evaluate promising
flue gas cleaning methods in the off-gas system of the 6-in.-dia fluidized-
bed combustor. Two techniques which have been identified for investigation
are acoustic agglomeration and granular-bed filtration. A third approach,
the use of high-efficiency, controlled-vortex cyclones, is also being
considered; installation and testing are planned. The high-efficiency
cyclone, in addition to being independently evaluated, would be used in
measuring the effectiveness of (upstream) acoustic flue gas conditioning for
increasing the removal of fine particulate matter from the flue gas.
a. Granular-Bed Filter
Granular-bed filters currently under development can be generally
classified by the condition of the granular bed during filtration as fixed-bed
-------
173
moving-bed, or fluidized-bed collectors. The concept of collection to be
investigated at ANL is the use of fresh or sulfated limestone or dolomite
as granular-bed material in a fixed-bed collector with periodic bed replace-
ment. Use of sorbent from the combustion process as the collection medium
has the advantage of eliminating the need to "backflush" the filter to
remove the ash material that has been trapped during the forward filtration
cycle. As is done with fixed granular beds employing backflush cleaning,
several granular bed modules would be operated in parallel. Periodically,
each module would be taken off line, and sorbent plus trapped particulate
matter would be replaced with fresh bed material. The sorbent containing
the trapped particulate matter could then be transferred either to the
combustor (if fresh sorbent had been used as the filter material) or to the
regenerator or disposed of (if sulfated sorbent was used as the filter
material). This scheme would eliminate the need for inventorying and dis-
posing of an additional solid material if some other material was used in
the granular-bed filter.
The test filter itself is illustrated in Fig. 101. The dirty flue
gas enters through the centrally located pipe in the top flange of the filter
housing. Inside the housing, the gas passes'downward through the granular
filter chamber suspended from the top flange of the filter housing. The
filter chamber is circular in cross section and has an inside diameter which
can be changed to be either 3 or 6 in. The fixed, horizontal granular-bed
filter at the lower end of the filter chamber is supported by a wire mesh
screen over a perforated plate. After the gas passes through the granular
bed, the gas, now cleaned, exits through a port in the side of the granular
bed filter housing. Pressure taps are located .just above the top surface
of the granular bed and below the support plate to measure the pressure drop
across the bed. A thermocouple just below the bed support plate will monitor
the gas temperature during each test.
Tymochtee dolomite has been selected for initial testing. By
appropriate selection of bed cross section and bed depth during a series
of runs, the particle size of the granular bed sorbent, the superficial
velocity of the gas being filtered, and the AP of the clean bed at the start
of filtration can each be varied while the other two variables are constant.
The principal response variables of interest are the overall gas cleaning
efficiency and the efficiency of cleaning as a function of flue gas partic-
ulate size.
Initially, testing of the granular-bed filter was carried out at
ambient conditions to determine under what conditions testing of the filter
in the flue gas system of the ANL combustor would be warranted. Specifically,
tests were run:
1. To determine a relationship for the pressure drop of gas
flowing through the sorbent in a fixed bed as a function
of the mean particle diameter of the sorbent, the super-
ficial gas velocity, and the bed depth. The tests were
made with the bed material screened to -6 +14 or -14 +30
U.S. mesh.
-------
174
FILTER HOUSING
q*" GAS OUT
Fig. 101.
Granular-Bed
Filter Assembly
3"OR 6" FILTER
ASSEMBLY
THERMOCOUPLE
2. To determine any contribution of dust from the sorbent
to the dust loading in the effluent gas from the filter.
These tests were made to determine what treatments (such
as washing) would be required if the natural dustiness
of the as-screened sorbent material was found to contrir-
bute to the dust loading in the effluent gas from the
filter. These tests included a comparison of unsulfated
and sulfated sorbent.
It is well known that the pressure drop through a granular bed
is proportional (1) to the gas velocity at low flow rates and (2) to the
product of the gas density and the square of the velocity at high flow rates
APR
= aU + bpU
(1)
where AP/L is the pressure drop per unit length of bed, U is the superficial
gas velocity, p is the gas density, gc is the gravitational constant, and
a and b are constants which are functions of the granular bed geometry and
gas viscosity. Dividing by the gas velocity gives a linear expression:
UL
= a + bG
(2)
-------
175
where G equals pU, the mass velocity through the bed. According to the
correlation of Ergun, ** the factors a and b can be expressed by:
a = 150 r (3)
(4)
where e equals the fractional void-volume in the bed, g is the gas viscosity
and Dp is the diameter of a sphere having the same specific surface area,
Sv, as a bed particle. If d is defined as the diameter of a sphere having
the same volume as that of a bed particle, Dp can be expressed by
D = <)> d (5)
P s p
where 4>s represents the sphericity of the particle and is the ratio of the
surface area of a sphere having the same volume as a bed particle to the
surface area of the bed particle. Thus, <|>s is less than or equal to unity.
For nonspherical particles of a wide size distribution, a mean
equivalent particle diameter, d , can be approximated from the following
relation:
d
1 Pi
where X^ equals the weight percent of particles with particle diameter dp
Thus the Ergun correlation for fluid flow through packed columns can be
expressed as:
= 150 -JL_ + 1>75 ^JL __ (7)
The characterization tests were made by measuring the pressure
drop across the granular-bed filter as a function of mass flow through the
bed at several bed levels and with two different particle size distributions
(-6 +14 and -14 +30 U.S. mesh) of Tymochtee dolomite. For each experiment,
the quantity APgc/UL was plotted as a function of the mass flow G. Values
of a and b were then determined by a linear regression analysis of the data.
Using a value for dp calculated from screen analysis data for the two particle
size distributions, the expressions for a and b (Eqs. 3 and 4) were then
solved simultaneously for the parameters E and $s.
Figures 102 and 103 are pressure drop curves obtained for -6 +14
and -14 +30 U.S. mesh Tymochtee dolomite as a function of mass flow and taken
at different bed levels. Variations in the curves for a given particle size
distribution at different bed levels would be expected, due to minor random
-------
30
28
26
24
22
20
I
tn
18
V
e
u
et 16
^U 14
Cu
<
12
10
Curve
Mass Mean Particle Diameter: 2150 \im
I I I I I I
10
G x "10 ,
cm s
Fig. 102. Pressure Drop Across a Fixed Bed of -6
+14 Mesh Tymochtee Dolomite as a Function
of Mass Flow Rate Through the Bed
16
15
14 -
13 —
12 —
11 —
10 -
9 -
Curve Symbol Bed Height, cm
I I
D
Mass Mean Particle Diameter: 740 urn
10
7 -9-1
G X 10% g cm - s
Fig. 103. Pressure Drop Across a Fixed Bed of
-14 +30 Mesh Tymochtee Dolomite as a
Function of Mass Flow Rate Through
the Bed
-------
177
variations in e, the fractional void volume of the bed, for each experiment.
The pressure drop curves are very sensitive to this parameter (see Eq. 7).
The results of the linear regression analysis to determine the
parameters, e and s, are presented in Table 34. For both particulate size
distributions, there was considerable variation in the calculated values for
both the e and s parameters. The variations in <$>s are measures of the
experimental error in making the measurements.
Average values of e and s for the -6 +14 mesh dolomite were 0.36
and 0.78, respectively. For the -14 +30 mesh dolomite, average values of e
and fys were 0.34 and 0.78, respectively. The calculated bed voidages agree
fairly well with the. measured bed voidage of 0.38 for both particle size
distributions based on apparent particle density (2.4 cm ) and bulk density
(1.5 g cm~3) measurements. The calculated value for <}>s of 0.78 also agrees
favorably with the value of 0.8 reported by Saxena,l+y who usea the Ergun
correlation to correlate the incipient fluidization velocity of Tymochtee
dolomite.
Table.34. Experimental Determinations of e and $s in
the Ergun Correlation for the Flow of Gas
Through a Fixed Bed of Tymochtee Dolomite
Test
No.
1
2
3
4
5
6
U.S.
Mesh
-6
-6
-6
-6
-14
-14
+14
+14
+14
+14
+30
+30
dp,
ym
2150
2150
2150
2150
740
740
Bed
Height,
cm
5.1
9.5
10.2
15.2
10.2
15.2
Intercept, a.
g cm"3 s"1
8
7
10
10
92
103
.23
.62
.8
.1
.8
Slope, b,
cm
153
124
152
183
Average
598
497
Average
e
0.
0.
0.
0.
0.
0.
0.
0.
34
38
38
33
36
31
36
34
*
0.
0.
0.
0.
0.
0.
0.
0.
s
88
72
63
87
78
93
64
78
A second series of tests was performed at ambient conditions with
essentially clean gas fed to the filter to measure the dust loading in the
gas leaving the filter. The tests were performed using three different
granular-bed materials, two bed heights, two particle size ranges, and two
superficial gas velocities through the bed.
The test arrangement is illustrated in Fig. 104. The filter was
charged with the material to be tested and then inserted in the filter housing
in the test arrangement. Flow of compressed air to the filter and sampling
of the effluent gas from the filter for the determination of entrained
-------
178
r-FLOW
\ CONTROL
VALVE
PRESSURE
GAUGE
FLOW
METER LJ
I
COMPRESSED
AIR
VENT
1
SAMPLE
PROBE
SAMPLE
COLLECTION
FILTER
VENT
'/SAMPLE RATE
^CONTROL VALVE
GRANULAR
BED FILTER
SAMPLE RATE
FLOW METER
^PRESSURE
GAUGE
Fig. 104. Test Arrangement for the Determination of Particulate
Loadings in the Effluent Gas from the Granular-Bed
Filter
particulate matter were then initiated at time zero. At the conclusion of
the dust sampling period, the sample collection filter was weighed and the
average dust loading over the sampling period was determined.
The results of the tests are presented in Table 35. The particulate
loadings in all tests were quite low, less than 0.004 g m~3 (<0.002 grain/
scf). In tests 1 and 9, in which sampling was done during a 4-hr period,
the particulate loadings were <0.0002 and 0.0007 g m~3, respectively.
Test 1, made with a 5.1-cm bed of -6 +14 mesh Tymochtee dolomite
and a gas velocity of 0.61 m/s, served as a baseline experiment. In sub-
sequent tests (tests 2 and 9), changes in the baseline test conditions were
made to check the possible effect on the particulate loading in the filter
effluent of (1) sampling time interval, (2) bed height, (3) bed particle size,
(4) gas velocity, and (5) bed material (sulfated and unsulfated). Evan at
the unreasonably high bed velocity of 2.4 m/s in tests 5 and 6, the loadings
were quite low (0.0037 and 0.0021 g/m3, respectively).
Thus, these tests indicate the feasibility of using sorbent materials
in fixed granular-bed filters with no necessity of washing or air classifi-
cation to control fines in the screened bed material.
The granular-bed test filter has been installed in the flue gas
system of the ANL PDU combustor. Testing is scheduled to begin shortly.
A schematic diagram of the granular bed test system is given in
Fig. 105. The particulate-laden flue gas from the secondary cyclone of the
FBC system will be directed through the granular-bed test loop once the
system is operating at steady state conditions. Initially, the dirty gas
in the test loop will bypass the granular bed filter and the porous metal
cartridge filters for sampling in the cascade impactor sampling system
-------
Table 35. Measured Particulate Loadings in Effluent
Filter when Passing a "Clean" Gas through
Gas from Granular-Bed
the Filter
Test
1
2-A^
2-B
2-C {
2-Dj
3
4
5
6
7
8
9
Bed
Material
Tymochtee
>• Tymochtee
Tymochtee
Tymochtee
Tymochtee
Tymochtee
1337 Dolomite
Sulfated
1337 Dolomite
1337 Dolomite
Normal
U.S.
Mesh
-6 +14
-6 +14
-6 +14
-14 +30
-14 +30
-6 +14
-14 -+30
-14 +30
-14 f30
Tesr. Conditions
Time
Interval,
min
0-240
{0-30
30-60
60-90
90-120
0-30
0-30
0-30
0-30
0-30
0-30
0-240
Gas Bed
Vel, Height, AP,
m/s cm kPa
0.61 5.1 2.24
0.61 5.1 0.50
0.61 10.2 1.00
0.61 5.1 4.18
2.44 5.1 18.7
2.44 5.1 5.23
0.61 5.1 1.24
0.61 6.4 5.85
0.61 5.1 1.24
Particulate
Loading
g m grains/scf
<0.0002
0.0002
0.0023
0.0007
0.0037
0.0009
0.0021
0.0037
0.0021
0.0016
0.0018
0.0007
0.0001
0.0001
0.0010
0.0003
0.0003
0.0004
0.0009
0.0016
0.0009
0.0007
0.0008
0.0003
Variable
Tested
Baseline Test
Loadings During 30'
min Intervals Over
2-hr Sampling
Period
Increase Bed Ht
Decrease Bed Part.
Size
Increase Gas
Velocity
Increase Gas
Velocity
Different Bed
Material
Sulfated Bed
Material
Increase Sample
VO
Time
-------
180
FLUE GAS FROM
SECONDARY CYCLONE
•ceo—
EXPERIMENTAL
GRANULAR BED
FILTER
TID—
D8&-
POROUS
METAL
CARTRIDGE
FILTERS
FILTER
BYPASS LINE
-----DAPI
-OSP
SYSTEM PRESSURE
CONTROL VALVE
t
pOTAMETER FOR
B SAMPLE FLOW
YRATE
J MEASUREMENT
SP-SAMPLE PORT FOR CASCADE IMPACTOR
API-AP INDICATOR
Tl-TEMPERATURE INDICATOR
Fig. 105. Modified Flue Gas System for Granular-Bed
Filter Tests
located downstream from the filter. Thereby, the particle size distribution
and the mass loading of the inlet gas to the granular bed filter can be
determined. To assess the efficiency of the filter as a function of partic-
ulate size and test conditions in the filter, the dust-laden gas will then
be directed through the granular bed filter, with the clean effluent gas
from the filter again sampled in the cascade impactor sampling system.
Variables considered for study include particle size of the granular
bed sorbent, face velocity of the gas being filtered (acfm of gas being
cleaned per sq ft of granular bed surface area), and bed starting AP (as
determined by bed height). Response variables include efficiency of cleaning
as a function of particulate size and the energy loss per cubic foot of gas
cleaned.
b. Acoustic Agglomeration
The general objective of this agglomerating technique is to enhance
the natural tendency of polydispersed particulates to impact upon each other.
Thus, the use of acoustics in controlling fine particle emissions is a process
whereby the mean size of the effluent particles is significantly increased
(and correspondingly their number is decreased) by exposure to finite ampli-
tude acoustic fields. As described here, sonic agglomeration is a conditioning
process designed to increase the collection efficiency of downstream dust
collectors.
As a result of visits made to laboratories currently investigating
acoustical agglomeration of aerosols, Dr. David S. Scott, Chairman of the
Department of Mechanical Engineering at the University of Toronto, Canada,
has been retained as a consultant for testing and evaluating the use of acoustic
-------
181
agglomeration to condition the flue gas (from the ANL 6-in.-dia combustor)
to increase particulate removal efficiency. Professor Scott has been
investigating the fundamentals of acoustic-aerosol interactions for several
years and has been a proponent of acoustic conditioning to help control
industrial aerosol emissions.5
Discussions with Dr. Scott have led to the following understanding
of acoustic-aerosol interactions and its proposed application to PFBC.
(1) The use of nondimensional groups in the description of finite-
amplitude sound propagation through aerosols. Acoustic-aerosol interactions
can be studied from two viewpoints. The first consideration is the effect
of the particulate matter upon the sound propagating through the aerosol, and
the second is the effect of the acoustic field on the particulate matter.
In the latter case, the effect that is of interest is an increase in the
aerosol agglomeration rate resulting from an increase in particle-particle
collision frequency. In different situations, it is possible for either
the sound or the particulate matter to dominate the interaction.
The important parameters in characterizing the aerosol are the
momentum relaxation time and the mass loading ratio. The momentum relaxation
time is essentially a fundamental description of the "type" of aerosol; the
relaxation time is directly proportional to the particle density and the
particle diameter squared and is inversely proportional to the gas viscosity.
The mass loading ratio can be considered the "how much" parameter in the
characterization of the aerosol.
The corresponding "type" and "how much" parameters which characterize
the acoustic field are its frequency and its acoustic Mach number (essentially
the amplitude of the acoustic wave form), respectively.
In a comparison of the "what type" parameters for the aerosol and
the acoustic field, the following observations can be intuitively made. If
the particle relaxation time is very short with respect to the acoustic cycle
time, the particles will behave as elements of the fluid. There will be
little attenuation of the acoustic field, and particle-particle collisions
may not be enhanced because no significant differential motion is induced in
the particulate matter. If the particle relaxation time is long with respect
to the acoustic cycle time, the particles will essentially remain stationary,
with no enhancement of particle-particle collisions. Thus, for enhanced
coagulation of the entrained particulate matter, the optimum coupling of
particle relaxation time (for particles in the size range of interest) and
the reciprocal of the acoustic frequency occurs when both are of approximately
the same order of magnitude.
Coupling of the "how much" parameters is equally important. If the
amount of aerosol is large relative to the amount of sound, the sound is
rapidly attenuated and there is insufficient energy for enhanced coagulation.
If the amount of aerosol is very small relative to the amount of sound,
enhanced coagulation becomes less likely due to the large interparticle
distances in the aerosol.
-------
182
(2) Applicability of acoustic agglomeration to the removal of
fine particulate (2 to 10 ym) from high-temperature high^-pressure flue gases.
The conditions, mass loading, and particle-size distributions expected in the
flue gases from pressurized, fluidized-bed combustors have been discussed
with Dr. Scott. The relatively high loadings expected from a PFBC, coupled
with a relatively low mass contribution from particulate matter below 2 ym,
appear favorable to the use of acoustical conditioning of the flue gas. In
work done at the Ontario Research Foundation, Mississauga, Canada, and
reported by Scott,50 a very fine aerosol (^85 wt % below 15 ym) of ZnO dust
was exposed for 2.5 s to either a 160-dB or a 165-dB acoustic field. With
exposure to 160-dB sound, the mass mean diameter of the ZnO increased to
^4 urn from ^1 ym. At 165 dB, the mean diameter increased to ^6 ym. When a
cyclone having a poor efficiency was intentionally used downstream, the
collection efficiency increased from ^30-40% for untreated aerosol to as
high as 80% for acoustically treated aerosol at the highest dust loading
investigated.
(3) Capital cost and power requirements. The limited commercial
application of finite-amplitude sound for agglomerating aerosols has resulted
primarily from the consideration of high capital cost and high specific power
requirements. Scott, however, cites (1) the potential cost advantages of
progressive saw-tooth waves for acoustic agglomeration (as compared with
standing-wave fields conventionally used) and (2) the advantages of pulse-
jet sound generation that indicate economic feasibility in comparison with
other gas cleaning systems.5
Dr. Scott has submitted to ANL a suggested procedure for the devel-
opment of a resonant manifold system for the evaluation of acoustic dust
conditioning (ADC) in the FBC system at ANL. The principal components of the
proposed system are (1) pulse-jet sound generation, (2) a resonant manifold
for "splitting" the resultant acoustic power, and (3) an acoustic treatment
section.
There appear to be several promising features of pulse jet acoustic
dust conditioning. For instance, the heat of combustion of the pulse jet
fuel simply adds to the overall process heat. Furthermore, due to the elevated
pressures of PFBC, the pulse jet will run at higher power levels than if it
were at ambient conditions and should in principle, therefore, be a more
effective sound generator.
The pulse jet will be designed to operate with a frequency of
400-500 Hz. Although flexibility in the choice of the ultimate design is
being retained, a flap-valve unit utilizing gasoline as the fuel now appears
most feasible.
A problem in the use of the pulse jet is that of scaling. A pulse
jet, which is a quarter wave-length device, cannot be scaled up or down without
changing the acoustic characteristics. A resonant manifold system is re-
quired, therefore, to split the resultant pulse-jet acoustic power and off-gas
into a waste stream and one or two process streams.
-------
183
A tentative schematic design for the resonant manifold system is
given in Fig. 106. The manifold will be designed as a "length resonator "
In this configuration, the ends of the diameter will become pressure antinodes
tor standing waves set up within the resonant manifold. The acoustic field
will be drawn from the flat end walls.
The inlet to the vent will be a distance A/8 (X equals acoustic
wave length) from the end wall, to minimize leakage of sound through this
orifice. However, the depth to which the vent is immersed in the resonant
manifold will be adjustable to control the amount of acoustic energy wasted
and thereby control the sound levels transmitted to the process sound ducts'
(see Fig. 106) which will carry the sound to the acoustic treatment section.
The acoustic treatment sections will be interchangeable sections
of pipe of various configurations which can be installed in the PDU flue
gas system between the combustor and the primary dust collector It is
in these sections that the sound and aerosol will interact and agglomeration
or the aerosol will occur.
The design and development of the pulse jet, the resonant manifold
and the acoustic treatment sections will be performed by Dr. Scott at the
University of Toronto. As soon as the components have been fabricated and
tested for acoustic performance, they will be transported to ANL for instal-
lation and testing in the ANL flue gas system. It is anticipated that delivery
ot the pulse jet-resonant manifold system to ANL will be in the first quarter
of 1978.
PULSE-
JET
PROCESS
SOUND DUCTS
Fig. 106. Schematic of Resonant Manifold System
Notes: (a) estimate A/2 = 75 cm
(b) estimate 4d = 30 cm
(c) all dimensions involving A
to be adjustable
-------
184
c. High Efficiency Cyclones
The third method of particulate control being considered for
investigation is the TAN-JET high-efficiency cyclone. Similar high-efficiency
cyclones will also be installed with granular bed filter systems as control
systems in the ANL Component Test and Integration Unit and the Curtiss-Wright
PFBC pilot plant. Installation of such a unit would also be used in testing
and evaluating acoustic conditioning as a pretreatment step in the control
process.
The two most frequently mentioned high-efficiency cyclones for use
in PFBC flue gas cleaning systems are the TAN-JET (Donaldson Co.) and the
Aerodyne SV (Aerodyne Development Corporation)'. These devices are similar
in that each employs injection of auxiliary (secondary) air to create a
constant controlled vortex at the cyclone wall. Collection efficiencies of
90% for 1-pm dust for the Aerodyne cyclone and 90% for 2-ym dust for the
Donaldson cyclone have been reported.
In discussions with representatives of Donaldson Co., it was
determined that the TAN-JET cyclone can be adequately scaled and tested on
the flue gas effluent from the ANL 6-in.-dia combustor. It is planned,
therefore, to proceed with the design, procurement, and installation of a
Donaldson TAN-JET cyclone as a part of the flue gas cleaning studies. This
effort will proceed in parallel with design and installation of acoustic
agglomeration equipment.
ACKNOWLEDGMENTS
Many people have contributed to the progress made in these studies.
We gratefully acknowledge the help given by Mr. L. Burris and Mr. D. Webster
in directing the program. Operation and maintenance of the PDU equipment
was ably done by Messrs. H. Lautermilch, S. Smith, R. Mowry, R. Beaudry, and
J. Stockbar.
We also appreciate the analytical services support given us by Dr. P.
Cunningham and his associates, Messrs. R. Bane, K. Jensen, B. Tani, R. Meyer,
N. Johnson, H. Goodspeed, R. Telford, W. Shinn, Ms. C. Blogg, and Ms. A.
Engelkemeir.
Design and drafting services were provided by Mr. R. Stimac and Mr.
R. Frank and secretarial services by Ms. M. Sobczak.
-------
185
REFERENCES
1. W. F. Bischoff, Jr. and Peter Steiner, Coal Converts S02 to S, Chem.
Eng., 74 (January 1975).
2. D. A. Martin, F. E. Brantley, and D. M. Yergensen, Decomposition of
Gypsum in a Fluidized-Bed Reactor, U.S. Bureau of Mines Report of
Investigation 6286 (1963).
3. T. D. Wheelock and D. R. Boyland, Reductive Decomposition of Gypsum by
Carbon Monoxide, Ind. Eng. Chem. 52, 215 (March 1960).
4. John C. Montagna, John F. Lenc, Gerhard J. Vogel, and Albert A. Jonke,
Regeneration of Sulfated Dolomite from a Coal-Fired FBC Process by
Reductive Decomposition of Calcium Sul fate in a Fluidized Bed, Ind.
Eng. Chem., Process Des. Dev. _16^ (2), 230 (April 1977).
5. R. C. Hoke et al., A Regenerative Limestone Process for Fluidized-Bed
Coal Combustion and Desulfurization, EPA-650/2-74-001 (January 1974).
6. A. Skopp, J. T. Sears, and R. R. Bertrand, Fluid Bed Studies of the
Limestone Based Flue Gas Desulfurization Process, Report No. GR-9-FGS-69,
Esso Research and Engineering Company, Government Research Div., Linden,
N.J. (August 1969).
7. J. S. Gordon et al., Study of the Characterization and Control of Air
Pollution from a Fluidized-Bed Boiler - The 5(?2 Acceptor Process, Pope,
Evans and Robbins Report, EPA-R2-72-021 (1972).
8. W. M. Swift and T. D. Wheelock, Decomposition of Calcium Sulfate in a
Two-Zone Reactor, Ind. Eng. Chem. Process Design Develop. 14_(3), 323
(1975).
9. Pope, Evans and Robbins, Inc., Multicell Fluidized-Bed Boiler Design,
Construction and Test Program, Quarterly Program Status Report, FE-1237-
Q76-2, October-December 1975.
10. Westinghouse Research Laboratories, The Influence of Limestone Calcination
on the Utilization of the Sul fur-Sorbent in Atmospheric Pressure Fluid-
Bed Combustion, Quarterly Progress Report for EPRI, 76-5E3-FBCAL-R3,
August 1-October 31, 1976.
11. M. J. Gluckman, J. Yerushalmi, and A. M. Squires, Defluidization
Characteristics of Sticky or Agglomerating Beds, Proceedings of the
International Fluidization Conference, Pacific Grove, California, June
15-20, 1975.
12. J. C. Montagna, W. M. Swift, G. W. Smith, G. J. Vogel, and A. A. Jonke,
Fluidized Bed Regeneration of Sulfated Dolomite Via Reductive Decomposi-
tion with Coal, Presented at 69th Annual AIChE Meeting in Chicago,
December 1, 1976.
-------
186
13. C. W. Zielke et al., Sulfur Removal During Combustion of Solid Fuels
in a Fluidized Bed of Dolomite, J. Air Pollution Contr. Assoc., 164
(March 1970).
14. M. G. Nagier, The Theory of Recycle Processing in Chemical Engineering,
Vol. 3, International Series of Monographs on Chemical Engineering,
MacMillan, New York (1964).
15. M. Hartman and R. W. Coughlin, Reactions of Sulfur Dioxide with
Limestone and the Influence of Pore Structure, Ind. Eng. Chem. 13(3),
248 (1974).
16. G. A. Hammond and A. Skopp, A Regeneration Limestone Process for
Fluidized Bed Coal Combustion and Desulfurization, Final Report, Esso
Research and Engineering Co., Gov. Res. Div., Linden, N.J. (February
1971).
17. N. P. Phillips, Thermodynamics Screening of Dry Metal Oxides for High
Temperature SO^ Removal, Radian Corporation Technical Note 200-045-10-01
(September 1974).
18. R. B. Snyder, W. I. Wilson, G. J. Vogel, and A. A. Jonke, Sulfation
and Regeneration of Synthetic Additives, Proceedings of the Fourth
International Conference on Fluidized-Bed Combustion, McLean, Virginia,
Dec. 9-11, 1975.
19. D. L. Keairns et al., Fluidized Bed Combustion Process Evaluation,
Annual Report, EPA-650/2-75-027-C (September 1975).
20. R. H. Borgwardt and R. D. Harvey, Environ. Sci. Technol. 6^(4), 350
(1972).
21. Pope, Evans and Robbins, Inc., Multicell Fluidized-Bed Boiler Design,
Construction, and Test Program, Combustion Systems Division, Monthly
Progress Report No. 39, December 1975.
22. E. Bagdoyan, Kennedy Van Saun Corp., private communication to R. Snyder,
ANL (October 1976).
22a. D. L. Graf, Amer. Miner. JTJU , 1 (1952).
23. T. Noda, J. Chem. Ind. Soc. Japan _42, 265B (1939).
24. J. Ryba et al., Zb. Pr. Chemickotechnol. Fak SVST 1969-1970, 115-21
(1971).
25. J. Krajci et al., Zb. Pr. Chemickotechnol. Fak SVST 1969-1970, 131-143
(1971).
26. J. A. Murray, Summary of Fundamental Research on Lime, National Lime
Association, Washington, D.C. (1956).
27. K. P. Kacher et al., Zem. Kalk-Gips 25(1), 37-41 (1972).
-------
187
28. P. S. Mamykin and A. V. Ivanova, Ogneupory 36.UO), 32-36 (1971).
29. D. R. Glasson, J. Appl. Chem. 1]_, 91-96 (1967).
30. W. R. Bandi and G. Krapf, Thermochimica Acta 1^6, 221-243 (1976).
31 A. E. Potter, Am. Ceram. Soc. Bull. 4*3, 855 (1969).
32. M. Hartman and R. W. Coughlin, Ind. Eng. Chem., Process Des. Develop.
13(3), 248-53 (1974).
33. W. T. Reid, External Corrosion and Deposits - Boiler and Gas Turbines,
American Elsevier Publishing Company, N.Y., p. 140, 1971.
34. H. J. Gluskoter and 0. W. Rees, Chlorine in Illinois Coal, Circular
372, Illinois State Geological Survey, Urbana, Illinois (1964).
35. K. E. Phillips, Energy Conversion from Coal Utilizing CPU-400 Technology,
Quarterly Report, July-Sept., 1976, FE-1536-Q7, Combustion Power Co.,
Inc., California.
36. N. H. H. Small, H. Strawson, and A. Lewis, Recent Advances in the
Chemistry of Fuel Oil Ash, Proceedings of the Conference on Mechanism
of Corrosion by Fuel Impurities, Butterworths, Scientific Publications,
London, p. 240, 1963.
37. P. D. Miller, H. H. Krause, J. Zupan, and W. K. Boyd, Corrosive Effects
of Various Salt Mixtures under Combustion Gas Atmospheres, Corrosion
28/6), 222 (1972).
38. A. J. B. Cutler, W. D. Halstead, J. W. Laxton, and C. C. Stevens, The
Role of Chloride in the Corrosion Caused by Flue Gases and Their
Deposits, Trans. ASME, J. Eng. Power 93, 37 (1971).
39. P. A. Alexander, Laboratory Studies of the Effects of Sulfates and
Chlorides on the Oxidation of Superheater Alloys, Proceedings of the
Conference on Mechanism of Corrosion by Fuel Impurities, Butterworths,
Scientific Publications, London, paper 40, p. 571, 1963.
40. L. E. Paulson and W. W. Fowkes, Changes in Ash Composition of North
Dakota Lignite Treated by Ion Exchange, Bur. Mines RI 7176, 1968, 18 pp.
41 P. H. Tufte and W. Beckering, A Proposed Mechanism for Ash Fouling
Burning Northern Great Plains Lignites, J. Power, Trans. ASME, 405
(1975).
42. G. H. Gronhovd, W. Beckering, and P. H. Tufte, Study of Factors
Affecting Ash Deposition from Lignite and Other Coals, ASME Winter
Annual Meeting, Los Angeles, Calif. Nov. 16-20, 1969, 9 pp.
43. C. P. Rao and H. J. Gluskoter, Occurrence and Distribution of Minerals
in Illinois Coals, Illinois State Geological Survey, Circular 476
(1973).
-------
188
44. W. M. Fanner, Measurement of Particle Size, Number Density, and Velocity
Using a Laser Interferometer, Applied Optics 10, 2603 (1972).
45. W. E. Ranz and J. B. Wong, Impaction of Dust and Smoke Particles,
Ind. Eng. Chem. 44_(6), 1371 (1952).
46. D. L. Keairns et al., Fluidized Bed Combustion Process Evaluation,
Environmental Protection Agency Report No. EPA-650/2-75-027-C (1975).
47. R. Razgaitis, An Analysis of the High Temperature Particulate Collection
Problem, Argonne National Laboratory, Argonne, 111., ANL<-77-14 (in
preparation).
48. S. Ergun, Fluid Flow through Packed Columns, Chem. Eng. Prog. 48(2),
89 (1952).
48. S. Saxena, private communication (1977).
50 S. D. Scott, A New Approach to the Acoustic Conditioning of Industrial
Aerosol Emissions, J. Sound Vib. 43(4), 607-618 (1975).
-------
189
APPENDIX A
PHYSICAL AND CHEMICAL PROPERTIES OF COALS AND SORBENTS
USED IN SORBENT REGENERATION STUDIES
-------
190
Table A-l. Particle Size Distribution and Chemical and
Physical Characteristics of Arkwright Coal
Sieve Analysis
U.S. Sieve No. % on Sieve
+14 0.0
-14 +25 8.0
-25 +35 14.2
-35 +45 12.3
-45 +80 24.7
-80 +170 17.9
-170 23.0
Mean Particle Dia: 323 \im
Proximate Analysis, wt %
As Received Dry Basis
Moisture 2.89
Volatile Matter 38.51 39.66
Fixed Carbon 50.92 52.43
Ash 7.68 7.91
100.00 100.00
Sulfur, wt % 2.82 2.90
Heating value,
Btu/lb 13,706 14,114
Ultimate Analysis^ wt %
Carbon 77.14
Hydrogen 5.23
Sulfur 2.90
Nitrogen 1.66
Chlorine 0.19
Ash 7.91
Oxygen (by difference) 4.97
Fusion Temperature of Ash
Reducing Oxidizing
Atm, °C Atm, °C
Initial deformation 1104 1160
Softening (H = W) 1177 1216
Softening (H = 1/2W) 1193 1243
Fluid 1232 1271
-------
19.1
Table A-2. Particle Size Distribution and Chemical and
Physical Characteristics of Triangle Coal
Sieve Analysis
U.S. Sieve No.
% on Sieve
+14
-14 +25
-25 +35
-35 +45
-45 +80
-80 +170
-170
Mean Particle Dia:
0.0
25.0
22.0
26.0
24.0
2.0
0.3
576 ym
Moisture
Volatile Matter
Fixed Carbon
Ash
Sulfur, wt %
Heating value,
Btu/lb
Proximate Analysis, wt %
As Received Dry Basis
3.46
31.47
55.69
9.38
100.00
0.98
32.60
57.68
9.72
100.00
1.02
13,053
13,521
Carbon
Hydrogen
Sulfur
Nitrogen
Chlorine
Ash
Oxygen (by difference)
Ultimate Analysis, wt %
76.11
4.99
.1.02
1.30
0.22
9.72
6.64
Initial deformation
Softening (H = W)
Softening (H = 1/2 W)
Fluid
Fusion Temperature of Ash
Reducing
Atm. °C
1383
1444
1485
1510+
Oxidizing
Atm, °C
1430
1480
1510+
1510+
-------
192
Table A-3. Particle Size Distribution and Chemical
Characteristics of Tymochtee Dolomite
Sieve Analysis
U.S. Sieve No. %
+14
-14 +25
-25 +35
-35 +45
-45 +80
-80 +170
-170
Average Particle Dia:
on Sieve
0.4
48.6
19.9
18.8
11.7
0.4
0.4
750 vim
Constituent Chemical Analysis, wt %
Ca
Mg
C02
Si
Al
Fe
H20
Derived Composition
CaCOs
MgC03
20.0
11.3
38.5
2.3
0.87
0.29
0.2
wt %
50.0
39.1
-------
193
Table A-4. Chemical and Physical Characteristics
of Sewickley Coal
Proximate Analysis, wt %
Moisture
Ash
Volatile
Fixed Carbon
Sulfur, wt %
Heating value,
Btu/lb
As Rec'd
1.08
12.73
39.02
47.17
100.00
4.33
13,018
Dry Basis
__
12.87
39.45
47.68
100.00
4.38
13,610
Ultimate Analysis, wt %
Moisture
Carbon
Hydrogen
Nitrogen
Chlorine
Sulfur
Ash
Oxygen (dif)
As Rec'd
1.08
70.75
4.93
1.10
0.01
4.33
12.73
5.07
100.00
Dry Basis
„
71.52
4.98
1.11
0.01
4.38
12.87
5.13
100.00
Fusion Temperature of Ash
Initial deformation
Softening (H = W)
Softening (H = 1/2W)
Fluid
Reducing
Atm, °C
1115
1170
1240
1295
Oxidizing
Atm, °C
1185
1250
1320
1360
-------
194
Table A-5. Analysis of Greer Limestone, As Fed
CaO
MgO
Fe203
Si02
A1203
S
Others
Loss on calcination 37.52
-------
195
TECHNICAL REPORT DATA
(Please read htttruciions on the reverse before completing/
1. REPORT NO.
EPA-600/7-77-138
3. RECIPIENT'S ACCESSION NO.
4. TITLE AND SUBTITLE
Supportive Studies in Fluidized-Bed Combustion
5. REPORT DATE
December 1977
6. PERFORMING ORGANIZATION CODE
7. AUTHOR jonke G. Vogel I. Johnson,S. Lee,J. Lenc,
A. Lescarret, J. Montagna. F. Nunes, J. Shearer, R. Sny-
der.G. Smith.W. Swift.F. Teats.C.Turner.I. Wilson
8. PERFORMING ORGANIZATION REPORT NO.
9. PERFORMING ORGANIZATION NAME AND ADDRESS
Argonne National Laboratory
9700 South Cass Avenue
Argonne, Illinois 60439
10. PROGRAM ELEMENT NO.
EHE623A
11. CONTRACT/GRANT NO.
EPA Interagency Agreement
IAG-D5-E681
12. SPONSORING AGENCY NAME AND ADDRESS
EPA, Office of Research and Development
Industrial Environmental Research Laboratory
Research Triangle Park, NC 27711
13. TYPE OF REPORT AND PERIOD COVERED
Annual: 7/76-6/77
14. SPONSORING AGENCY CODE
EPA/600/13
is. SUPPLEMENTARY NOTES IERL-RTP project officer is Walter B. Steen, Mail Drop 61, 919/
541-2825. Previous report in this series was EPA-600/7-76-019.
. ABSTRACT
Tne report gjves results of studies supporting the development of atmos-
pheric and pressurized fluidized-bed combustion (FBC) of coal. It includes laboratory
and bench-scale studies to provide needed information on combustion optimization,
regeneration process development, solid waste disposal, synthetic SO2-sorbent stu-
dies, emission control and other tasks. It includes characterization of a variety of
limestone and dolomites from various parts of the U.S. for suitability in FB combus-
tors. Reduction in solid waste volumes to reduce the environmental impact of the
waste sulfated limestone is a major goal of this program. These studies are to supply
data essential for the application of FBC units to public utility and industrial systems.
The report gives information on: 10- cycle combustion-regeneration PDU experiments
using Greer limestone and, Tymochtee dolomite, bed def luidization , flowsheet develop-
ment, preparation of synthetic SQ2-sorbents containing metal oxides, limestone char-
acterization, coal combustion reactions, the enhancement of limestone sulfation by
NaCl, evaluation of on-line particle size analyzers, and status of flue gas cleaning
studies.
17.
KEY WORDS AND DOCUMENT ANALYSIS
DESCRIPTORS
b.lDENTIFIERS/OPEN ENDED TERMS
c. COSATI Field/Group
Air Pollution
Coal
Combustion
Fluidized-bed
Processing
Regeneration
Waste Disposal
Sulfur Oxides
Limestone
Dolomite (Rock)
Sodium Chloride
Air Pollution Control
Stationary Sources
Fluidized-bed Combus-
tion
13B
2 ID
21B
13H,07A
07B
08G
18. DISTRIBUTION STATEMENT
Unlimited
19. SECURITY CLASS (This Report)
Unclassified
21. NO. OF PAGES
210
20. SECURITY CLASS (This page I
Unclassified
22. PRICE
EPA Form 2220-1 (9-73)
------- |