EPA-6Q0/R-95-015d
February 1995

PROCEEDINGS: 1993 S02 CONTROL SYMPOSIUM

Volume 4. Sessions 1, 8A, and 8B

For Sponsors;

U.S. Environmental Protection Agency
Brian K, Gullett

Air and Energy Engineering Research
Laboratory

Research Triangle Park, NC 27711

Electric Power Research Institute

Ruseli B. Owens
3412 Hillview Avenue
Palo Alto, CA 94304

U.S. Department of Energy

Charles J. Dnimmond
Pittsburgh Energy Technology Center
P.O. Box 10940
Pittsburgh, PA 15236


-------
EPA REVIEW NOTICE

*

This report has been reviewed by the U.S. Environmental Protection Agency, and
approved for publication. Approval does not signify that the contents necessarily
reflect the views and policy of the Agency, nor does mention of trade names or
commercial products constitute endorsement or recommendation for use.

This document is available to the public through the National Technical Informa-
tion Service, Springfield, Virginia 22161.

ABSTRACT

These four volumes document more than 100 presentations at the 1993

SO2 Control Symposium in Boston, MA, August 24-27, 1993. The
presentations covered a wide range of topics: industry's strategies
for dealing with the Clean Air Act Amendments of 1990, including
Phase I strategies, the emission allowance trading system, and re-
trofit construction; additives, materials, and operating issues for
wet flue gas desulfurization (FGD); clean coal demonstration pro-
grams; the effect of FGD systems on air toxics; dry FGD technologies
of spray drying and furnace sorbent injection; applied sulfur dioxide
(SO2) control research results, and emerging acid rain control tech-
nologies; and waste disposal issues. The presentations covered re-
sults obtained from full-scale demonstration/operation to pilot and
bench scale work.

Copyright © !993, EPRI TR-103289-VI, ¥2, and ¥3, Proceedings.
1993 S02 Control Symposium, ¥olumes 1, 2, and 3. Since this
wor was, in part, funded by the U, S. Government, the Govern-
ment is vested with a royalty-free, non-exclusive, and irrevoc-
ab e license to publish, translate, reproduce, and deliver that in-
formation and to authorize others to do so.

XI


-------
CONTENTS

Paper	VOLUME 1

SESSION 1: CLEAN AIR ACT REGULATORY STRATEGIES

1	Synergies and Conflicts in Multimedia Pollution Control

Related to Utility Compliance With Title IV of The Clean
Air Act Amendments of 1990

D. South, K. Bailey, Technology and Environmental Policy
Argonne National Laboratory

2	Analysis and Simulation of SO2 Allowance Trading

C.	Goldburg and L. Lave, Graduate School of Industrial
Administration; Carnegie Mellon University

3	A Synthesizing Framework for Toxics Risk Management
L. Levin; Electric Power Research Institute

4	Clean Air Technology (CAT) Workstation Case Study
W. DePriest, A. Maurer, J. Wroble, Sargent & Lundy;

D.	Knue, The Cincinnati Gas & Electric Company f
R. Rhudy, Electric Power Research Institute

SESSION 2: PHASE I DESIGNS

5	Status of the Flue Gas Desulfurization System At
Indianapolis Power & Light Company Petersburg
Station Units 1 and 2

C. Rutledge, W. Watson, S. Wolsiffer, Indianapolis
Power £ Light Company; L. Burr, M. Heimendinger,
C. Wedig, Stone & Webster Engineering Corporation

6	Limestone Forced Oxidation FGD System for TVA's
Cumberland Units 1 & 2

J. Buckner, Tennessee Valley Authority, I. Brodskyf
Raytheon Engineers and Constructors/ L. Burnham, ABB
Environmental Systems

7	The Clean Air Act Amendments and The Conemaugh Station -

Phase I Compliance

C. Altin, Ebasco Services Incorporated; J, Campbell,
Pennsylvania Electric Company

8	Ghent Unit 1 FGD System Retrofit Project

S. Johnson, Sargent & Lundy

iii


-------
Paper

Page

9	Henderson Station Two FGD System Retrofit Project

F. Campbell, J. Mcllroy, Burns & McDonnell Engineering

Company; J. Garner, Henderson Municipal Power and

Lighti R. Phillips, Big Rivers Electric Corporation	9-1

10	The Navajo Scrubber Project FGD Process Evaluation and
Selection

J. Lusko, Salt River Project; N. Sekhar, C. Harrell,

Stone and Webster Engineering Corporation	10-1

11	Design of the Gibson Unit 4 FGD System

S. Katzberger, Sargent & Lundy; R. Richard, PSI Energy,

JncJ. Snapp, PSI Energy, Inc.	11-1

12	Engineering and Design Guidelines for Duct Injection
Retrofits

R. Claussen, C. Martin, Raytheon Engineers &

Constructors, Inc.; P. Van Smith, W. Oberjohn,

Babcock & Wilcox/ G. Weber, Energy & Environmental

Research Center	12-1

SESSION 3A: ADDITIVES FOR HIGH EFFICIENCY FGD

13 Evaluation of High SO2 Removal Efficiency Upgrade
Options: EPRI High Sulfur Test Center
G. Stevens, W. Morton, O. Hargrove, Jr., Radian
Corporation/ D. Owens, Electric Power Research

Institute	13-1

14	High Efficiency SO2 Removal Tests at Tampa Electric
Company's Big Bend Unit 4

J, Smolensk!, Tampa Electric Company; J. Phillips, A.
Espenscheid, T. Shires, Radian Corporation	14-1

15	Options For Increasing SO2 Removal and Improving the
Water Balance at Hoosier Energy's Merom Station

P. Reynolds, J. J ember g, Hoosier Energy}

J. Noblett, Jr., J. Lundeen, A. Jones, Radian
Corporation; D. Owens, Electric Power Research

Institute	15-1

16 Results of High Efficiency SO2 Removal Testing at
the Southwestern Electric Power Company's Henry W.

Pirkey station

G. Blythe, J. Phillips, Radian Corporation; T. Slater,
Southwestern Electric Power Company	16-1

iv


-------
Paper

Page

17	The Reduction of Stack Emissions at LaCygne Unit 1

G.	Burnett, D. Dahl, Kansas City Power and Light/ P. Dyer,

C.	Weilert, S. Bjorklun, D. Miller, Burns & McDonnell
Engineering Company	17-1

18	Development of a Predictive Model for Organic Acid
Consumption in Wet Limestone PGD Systems

M. Stohs, T. Carey, Radian Corporation/ D. Owens,

Electric Power Research Institute	18-1

19	Ammonia Scrubbing of SC>2 Comes of Age With Insitu
Forced Oxidation

A. Saleem, GE Environmental Systems; K. Janssen,.

Dakota Gasification Company/ P. Ireland,

Raytheon Engineers & Constructors	19-1

SESSION 3B: MATERIALS FOR FGD

20	Materials of Construction PSI Energy Gibson #4
FGD System

R. Richard, PSJ Energy, Gibson Generating Station	20-1

21	FGD Materials Guideline and Software

P. Radcliffe, Electric Power Research Institute	21-1

22	Materials for FGD Retrofits FGD System Retrofits Require
High Material Reliability

A. DoVale, Jr., tf. Ford, G. Krause, Wheelabrator Air
Pollution Control	22-1

23	The Evolution of Stainless Steel and Nickel Alloys in
FGD Materials Technology

R. Ross, Jr., Nickel Development Institute	23-1

24	Why Fiberglass-Reinforced Plastic (FRP) is Being
Specified in Flue Gas Desulfurization (FGD)

Applications

D.	Kelley, The Dow Chemical Company	24-1

25	Effect of Environment on Coatings in FGD Ductwork

H.	Rosenberg, Guild Associates, Inc.} B. Syrett,

Electric Power Research Institute	25-1

26	Selecting Tank Linings for Flue Gas Desulfurization
Applications

R. Mosier, R. Petersen, Burns & McDonnell

Engineering Co.	26-1

v


-------
Paper

Page

27	Conclusions From a Decade of FGD Materials Failure
Analysis

P. Ellis II, Radian Corporation; P. Racial if fe,

Electric Power Research Institute	27-1

VOLUME 2

SESSION 4A: CLEAN COAL DEMONSTRATIONS

28	Demonstration of Bechtel's Confined Zone Dispersion
Process at Pennsylvania Electric Company's Seward
Station Project Status

J. Battista, Jr., Pennsylvania Electric Company;

A. Rubin and J. Abrams, Bechtel Corporation/

A. Baldwin, DOE's Pittsburgh Energy Technology	28-1

29	Project Update: Advanced FGD Design For Northern
Indiana Public Service Company's Bailly Generating
Station

D. Vynazal, Pure Air; B, Wrobel, Northern Indiana
Public Service Company (NIPSCO); T. Sarkus,

U.S. Department of Energy	29-1

30	SNOX Demonstration Project Performance Data One Year
Interim Report

D. Steen, S. Durrani, Ohio Edison Company/ D. Borio, D.

Collins, ABB Environmental Systems	30-1

31	Design Improvements and Operation of the Passamaquoddy
Technology L.P. - DOE Clean Coal Demonstration on a Coal
Fired Cement Kiln

G. Morrison, Passamaquoddy Technology, L.P.	31-1

32	SOx Emission Control With the SOx-NOx-ROX BOX

A. Holmes, K. Redinger, G. Amrhein, Babcock & Wilcox -
Research and Development Division	32-1

33	Plant Yates ICCT CT-121 Demonstration Results of
Parametric Testing

D. Burford, Southern Company Services; I. Pearl, Radian
Corporation/ S. Zee, Georgia Power Company	33-1

34	LIFAC Sorbent Injection for Flue Gas Desulfurization
J. Servol, ICF Kaiser Engineers, Inc.; J, Viiala,

T. Pokki, Tampella Power Corp.; I. Huffman, Richmond

Power & Light	34-1

vi


-------
Paper

Page

35	10-MW Demonstration of the AirPol Gas Suspension Absorption
Flue Gas Desulfurization Process

T. Burnett, V. Norwood, E. Puschaver, Tennessee Valley
Authority; F. Hsu, B. Bhagat, AirPol, Inc./ S. Marchant,
G. Pukanic, U.S. Department of Energy	35-1

SESSION 4B - APPLIED RESEARCH

36	Limits and Potentials of the Dry Injection Process

I. Bjerle, Z. Ye, W. Wang, Univ. of Lund (Sweden)	36-1

37	Plasma Technologies Applied to Air Pollution Control

D. Helfritch, Research-Cottrell Companies	37-1

38	On The Kinetic Modeling of the FGD Reaction at Low
Temperatures

I. Ortiz, A. Garea, I. Fernandez, A. Oliv&n, J. Viguri,

A. Irabien, Dpto. Quimica. E.P.S.I. Universidad de

Cantabria	38-1

39	Preparation of ADVACATE Reagent in a Flow Reactor

K. Kind, G. Rochelle, Univ. of Texas—Austin	39-1

40	Catalytic Reduction of Sulfur Dioxide to Elemental
Sulfur

Y. Jin, Q. Yu, S. Chang, Univ. of California—Berkeley	40-1

41	Pilot-Scale Testing of a Circulating Bed Absorption
FGD Process

J. Neathery, J. Schaefer, J. Stencel, Univ. of Kentucky;

T. Burnett and V. Norwood, Tennessee Valley Authority	41-1

42	Fly Ash Hydration With Quicklime for Improving
Sorbent Utilization and SO2 Removal in Spray
Dryer Absorbers

J. Sanders, T. Keener, J. Wang, University of

Cincinnati	42-1

43	Experimental Study of Calcium-Based Sorbent Reactivities
in a Drop Tube Reactor

A. Wang, S. Khang, T. Keener, Univ. of Cincinnati	43-1

Vll


-------
Paper

Page

SESSION 5A: DRY FGD TECHNOLOGIES

44	Performance Results from the Tangentially Fired LIMB
Demonstration Program at Yorktown Unit No. 2

J.P. Clark, M.R. Gogineni, R.W. Koucky, A.F. Kwasnik,

ABB Power Plant Laboratories? C,H. Francis,

Virginia Power Company; D.G. Lachapelle, EPA/AEERL

(AEERL-P-1086)	44-1

45	In Furnace Sorbent Dry Injection for Simultaneous
SC>2/NOx Removal

G. Pedrelli, U. DeRobertis, S. Pasini, A. Bianchi,

M. Cioni, ENEL S.p.A. - DSR/CRTN (Italy)	45-1

46	Gas Reburning-Sorbent Injection Demonstration Results
R.T. Keen, C.C. Hong, J.C. Opatrny, T.M. Sommer,

B.A, Folsom, R. Payne, Energy and Environmental

Research Corporation	46-1

47	Supported Sorbents

S. Nelson, Jr., Sorbent Technologies Corporation	47-1

48	Pilot-Scale Testing of Spray Dryer/Pulse-Jet Fabric
Filter FGD Technology for Medium- and High-Sulfur
Coal Applications

G.M.	Blythe, J.M. Davidson, J.J?. Peterson, Radian

Corporation/ R.G. Rhudy, Electric Power Research

Institute	48-1

49	Installation and Initial Operation of LIFAC at
Shand Power station

M.E. Ball, Saskatchewan Power Corporation (Canada)/

T.A. Enwald, Tampella Power Inc. (Finland)	49-1

50	Operation Results of the First Commercial Dry
Desulfurization Plant in Hokkaido Electric Power Co.

H.	Nagashima, M. Konno, Hokkaido Electric Power Co.;

N. Arashi, Hitachi Research Laboratoryj

0. Kanda, Kure Works, Babcock Hitachi (Japan)	50-1

51	Commercial Operating Experience with Advanced-Design,

Circulating Fluid Bed Scrubbing

R.E. Graf, Graf-EPE GmbH; B. Huckriede, H. Kessler,

S. Zimmer, Wulff GmbH (Germany)	51-1

viii


-------
Paper

Page

VOLUME 3

SESSION 5B: NET FGD PROCESS ISSUES (PART I)

52	Selection Considerations for Controlled Oxidation
Wet FGD	.

W. Nischt, D.W. Johnson, H.G. Milobowski,

Babcock & Wilcox	52-1

53	Prospects for Inhibited Oxidation FGD Systems

R.E. Moser, Electric Power Research Institute	53-1

54	Flue Gas Desulfurization Cycling Guidelines

W. DePriest, W.J, Rymarczyk, Sargent & Lundy/

P. Radcliffe, Electric Power Research Institute	54-1

55	No paper is associated with this number.

56	Results of the Babcock & Wilcox Limestone Wet FGD
Pilot Program at EPRI's High Sulfur Test Center

P.A. Bhat/D.W. Johnson, Babcock & Wilcox, Barberton;

B.J. Jankura, Babcock & Wilcox, Alliance Research

Center	56-1

57	Pilot Plant and Full-Scale Experiences with On-Line

Particle Size Measurement for Controlling Grind for
Wet FGD Systems

G.B. Maybach, Electric Power Research Institute;

T.W. King, Gilbert Commonwealth;

T. Braden, Outokumpu Mintec	57-1

58	Mist Elimination System Design and Specification for
FGD Systems

A.F. Jones, K.E. Mclntush, J.E. Lundeen, Radian

Corporation; R.G. Rhudy, Electric Power Research

Institute; C.F.P. Bowen, NELS Consulting Services, Inc.	58-1

59	Economic Impact of Chloride Removal from Wet Limestone
FGD Systems

J.L. Phillips, W.M. Horton, Radian Corporation;

D.R. Owens, Electric Power Research Institute;

J.G. Patterson, Jr., Tennessee Valley Authority	59-1

ix


-------
Paper

Page

SESSION 6A: AIR TOXICS REMOVAL IN PGD SYSTEMS

60	Control of Air Toxics from Coal-Fired Power Plants Using
FGD Technology

J.G. Noblett, Jr., F.B. Meserole, D.M. Seeger, Radian
Corporation, D.R. Owens, Electric Power Research Institute 60-1

61	ABB's Investigations into the Utilityair Toxics Problem

J.D. Wesnor, ABB Environmental Systems	61-1

62	Air Toxics Control by Spray Dryer Absorption

K. Felsvang, Niro - U.S.A.;

R. Gleiser, Joy Environmental Technologies, Inc.;

G. Juip, Northern States Power Company;

K.K. Nielsen, Niro (Denmark)	62-1

63	The Effect of Sorbent Injection Technologies on
Emissions of Coal-Based, Metallic Air Toxics
B.K. Gullett, Air and Energy Engineering Research
Laboratory (EPA); K. Raghunathan, Acurex Environmental
Corporation (AEERL-P-1090)	63-1

64	Air Toxics Monitoring Project: Electrostatic Precipitator
(ESP) Versus SOx/NOx/ROX BOX (SNRB)

J. Czuczwa, K.E. Redinger, Babcock & Wilcox - Research

and Development Division; P. Chu, Electric Power

Research Institute; F.J. Starheim, Ohio Edison	64-1

65	The Semi-Dry Desulphurisation Within the Three-Stage
Flue Gas Cleaning "System Dusseldorf"

B. Kassebohm, Stadtwerke Dusseldorf; S. Streng,

Lentjes Umwelttechnik AG	65-1

66	Evaluation of Fourier Transform Infrared (FTIR)

Technology for Continuous Emissions Monitoring at
Coal-Fired Electric Utilities

J. Palank, Baltimore Gas and Electric Company;

T. Dunder,	G. Plummer,	Entropy

Environmentalists, Inc.; R. Glover, Electric
Power Research Institute; R. Squires,

Electric Power Research Institute	66-1

67	Air Toxics Research Needs: Workshop Findings
S.A. Benson, E.N. Steadman, Univ. of North Dakota;

A.K. Mehta, Electric Power Research Institute; C.E.

Schmidt, Pittsburgh Energy Technology Center (DOE)	67-1

x


-------
Paper

Page

SESSION 6B: WET FGD PROCESS ISSUES (PART II)

68	An Engineering Workbook for the Estimation of Plume
Opacity

T .H. DeFries, A.F. Jones, Radian;

R.G. Rhudy, Electric Power Research Institute	68-1

69	SOg Removal from Flue Gas by Sorbent Injection EPRI
HSTC Phase II Test Results

<7.1?. Peterson, A.F. Jones, F.B. Meserole, Radian
Corporation•, R.G. Rhudy, Electric Power Research
Institute	69-1

70	Process Optimization Efforts at the Bruce Mansfield
Plant

D.M. Gausman, R. Mark Golightley, Ohio Edison Co.	70-1

71	Operation and Maintenance Update of the Wm. H. Zimmer
Generating Station FGD System

Wm. D. Brockman,	Cincinnati Gas & Electric

Company; M.G. Milobowski, R.W. Telesz, The Babcock &

Wilcox Company; G.K. Van Buskirk, Koch Engineering

Company, Inc.	71-1

72	Results of Magnesium-Lime Clear Liquor Scrubbing Tests

at the Miami Fort Station of Cincinnati Gas & Electric
Company

Wilhelm, Codan Associates; J?. Moser, EPRI;

M. Stohs, Radian CorpB. Lani, Dravo Lime Co.;

D. Horn, The Cincinnati Gas & Electric Co.	72-1

73	Experimental Approach and Techniques for the Evaluation
of Wet FGD Scrubber Fluid Mechanics

T.W. Strock, P. Dykshoorn, M.J. Holmes, Babcock &

Wilcox - Research & Development Division;

W.F. Gohara, Babcock & Wilcox - Environmental

Equipment Division	73-1

74	Update on Electric Power Research Institute's (EPRI)

FGDPRISM Process Simulation Model (Version 2.0}

J.G. Noblett, Jr., T.M. Shires, Radian Corporation/

R.E. Moser, Electric Power Research Institute	74-1

75	Status of Commercial FGD Applications in Europe
I?. We Her, STEAG Aktiengesellschaft;

w. Ellison, Ellison Consultants	75-1

XI


-------
Paper

Page

VOLUME 4

SESSION 7: POSTER PAPERS

76	Advanced Flue Gas Treatment System using Lilac Absorbent
Prepared from Fly Ash

S. Nakamura, T. Ueno, Hokkaido Electric Power Co., Inc.;
A. Tatani, S. Kotake, Mitsubishi Heavy Industries Ltd.	76-1

77	Development of High Efficiency Wet Limestone Gypsum Flue
Gas Desulfurization System

S. Kotake, Y. Tsuchiya, T. Higashi, K. Iwashita,

Mitsubishi Heavy Industries Ltd.	77-1

78	Chemistry of Limestone Slurry Scrubbing

R.S. Agarwal, G.T. Roche 1le, Univ. of Texas—Austin	78-1

79	Prediction of the Performance of Furnace Sorbent

Injection on a Coal-Fired Utility Boiler
G.H. Newton, D.K. Moyeda,

Energy & Environmental Research Corporation	79-1

80	Scrubber Renewal Project at Texas Municipal Power Agency
C.V. Weilert, S.M. Ashton, F.W. Campbell, R.D. Norton,

Burns & McDonnell Engineering Company;

R. Hamilton, Texas Municipal Power Agency	80-1

81	The Effects of Increasing S(^ Removal on Near-Plant

Opacity of Coal Fired Utility Boilers

C.V. Weilert, P.N. Dyer, Burns & McDonnell Engineering
Company	81-1

82	Economic Comparisons of Emerging SO2 Control
Technologies

R. Martinelli, T.R. Goots, P.S. Nolan, Babcock &

Wilcox	82-1

83	Case Studies: Early Reports on Phase One CEM Projects
J. Passmore, S. Brodmerkle, S. Voss,

Burns & McDonnell Engineering Company	83-1

84	Secondary Dewatering of FGD Slurry with Centrifuge at
Elrama Power Station

F.J. Bickerton, Jr. , Duquesne Light Company} W-C. Yu,

Conversion Systems, Inc.1 R.R. Bevan, D.L. Milligan,

Alfa Laval Separation, Inc.	84-1

Xll


-------
Paper	Page

85	Simultaneous Removal of Acidic and Volatile Organic
Gases by Spray Dryer Sorber (=SDS)

J-K. Koo, K.R. Baek, Korea Advanced Institute of
Science and Technology; J-K. Kim, Samsung Heavy

Industries Co., Ltd.	85-1

86	Numerical Models for SC>2 Absorption by Duct Injection
W.J. Oberjohn, S.W. Burge, Babcock & Wilcox - Research

and Development Division	86-1

87	Optimizing Electric Utility Air Toxics Compliance with

Other Titles of the Clean Air Act

A,P. Loeb, D.W. South, Argonne National Laboratory	87-1

88	Modified Hydrate Production at Dravo Lime Company
J. College, M.L. Cupp, M. Babu, D. Stowe, Dravo

Lime Co.	88-1

89	Comparison of Worldwide Emission Control Strategies
and Their Effects on Plant Availability and Costs

S. Stallard, Black & Veatch; C. Reese, RWE Energie AG;
L. Salvaderi, ENEL; H. Schlenker, RWE Energie AG	89-1

90	Demonstration of Promoted Dry Sorbent Injection in
a 75 MWe Tangentia1ly-Fired Utility Furnace

D. Rodriguez, C. Gomez, Intevep SA; JR. Payne, P.M. Maly,

Energy and Environmental Research Corporation	90-1

91	Development of Process to Simultaneously Scrub NO2 and
SO2 from Coal-Fired Flue Gas

V.M. Zamansky, R.K. Lyon, A.B. Evans, J.N. Pont, W.R.

Seeker, Energy and Environmental Research Corporation;

C.E. Schmidt, Department of Energy	91-1

92	The NOXSO Clean Coal Technology Project: Commercial

Plant Design

J.B. Black, M.C. Woods, J.J. Friedrich, C.A. Leonard,

NOXSO Corporation	92-1

93	Investigation of Sorbent Regeneration Kinetics in the
Copper Oxide Process

J.M. Markussen, H.W. Pennline, U.S. Department

of Energy	93-1

94	Removal of Particulate Matters and Air Toxics Together
with SO2 in Chiyoda Thoroughbred 121 FGD Systems

H. Yanagioka, M. Uchida, Chiyoda Corporation	94-1

xiii


-------
Paper

SESSION 8A: EMERGING TECHNOLOGIES

95	A Final Report on the Moving-Bed Limestone Emission
Control (LEG) Process Pilot Plant Program

M.E. Prudich, Ohio University; K.W. Appall,

J.D. McKenna, ETS, Inc.

96	Tung FGD Pilot Plant Performance

P. Strangway, Niagara Mohawk Power Corporation;

S. Tung, Raycon Research & Development;

R.J. Keeth, Raytheon Engineers & Constructors, Inc.

91 Integrated Flue Gas Treatment by Wet FGD Operating in
a Water-Condensing Mode

J.P. Heaphy, J.C. Carhonara, Consolidated Edison Company
W. Ellison, Ellison Consultants

98	10-MW Demonstration of the ADVACATE Flue Gas Desul-
furization Process

L.R. Lepovitz, C.A. Brown, Radian Corporation;
T.E. Pearson, J.F. Boyer, ABB Environmental Systems;
T.A. Burnett, V.M. Norwood, E.J. Puschaver, Tennessee
Valley Authority; C.B. Sedman, U.S. Environmental
Protection Agency/AEERL; B. Toole~0'Neil, Electric
Power Research Institute (AEERL-P-1087)

99	NOXSO S02/N0x Flue Gas Treatment Process; Proof-of-
Concept Test

J.L. Haslbeck, M.C. Woods, W.T. Ma, S.M. Harkins,
J.B. Black, NOXSO Corporation

100	SOXAL™ Demonstration Project at Niagara Mohawk's
Dunkirk Steam Station

D. Hurwitz, C. Denker, P. Birdsall, J. Soltys,

Allied Signal, Inc.; P. Strangway, Niagara Mohawk
Power Corporation

101	Optimization of Advanced Coolside Desulfurization
Process

M.R. Stouffer, W.A. Rosenhoover, J.A. Withum,
J.T. Maskew, CONSOL Inc., Research & Development

xiv


-------
Paper

Page

SESSION 8B: WASTE UTILIZATION AND DISPOSAL

102 Economic Analysis of FGD By-Product Disposal
Alternatives

D.L. Forster, Ohio State Univ.;

J. Rausch, Cornell Extension Service

102-1

103 Conversion from Disposal to Commercial-Grade Gypsum
S.K. Conn, Owensboro Municipal Utilities;

M.G. Vacek, Sargent & Lundy/

J.T. Morris, Jr., Wheelabrator Air Pollution Control

103-1

104	Plant Growth and Soil Properties Responses to

Additions of Dry Flue Gas Desulfurization
By-Products

Pf.A. Dick, R.C. Stehouwer, P. Sutton, J.M. Bigham,

R. Lai, S.J. Traina, E.L. McCoy, R. Fowler, Ohio

State Univ.	104-1

105	The Stabilization of Orimulsion Spray Dryer Waste for
Landfill Disposal

S. Kuchibotla, Law Engineering, Inc.!

E.H. Kalajian, C~S. Shieh, Florida Institute of

Technology; K.R. Olen, Florida Power & Light Co.	105-1

106	When the Regulations Get Tough: Advanced Treatment
of FGD Slowdowns

M.K. Mierzejewski, Infilco Degremont Inc.}

D.C. Ciszewski, K.R. Minnich, Aqua-Chem, Inc.	106-1

107	Development of Waste Water Concentration &

Solidification System for Wet Flue Gas
Desulfurization Plant

S. Tsuhouchi, T. Miwada, Chubu Electric Power
Co., Inc..; S. Kotake, N. Ukawa, Mitsubishi Heavy

Industries Ltd.	107-1

108	FGD Waste Compositions as Landfill Liner

C.L. Smith, Conversion Systems, Inc.	108-1

xv


-------
ADVANCED FLUE GAS TREATMENT SYSTEM USING LILAC ABSORBENT

PREPARED FROM FLY ASH

H. Nakamura
T. Ueno

Hokkaido Electric Power Co., Inc.
Satozuka 461-6, Toyohira-ku
Sapporo Japan 004

A. Tatani
S. Kotake
Mitsubishi Heavy Industries Ltd.
5-1, Marunouchi 2-ehome, Chiyoda-ku
Tokyo Japan 162

Abstract

The FGD processes called LILAC processes have been developed jointly by Hokkaido
Electric Power Co., Inc. and Mitsubishi Heavy Industries Ltd. In this paper, the
outlines of the processes and the results of the demonstration teste are presented. The
processes are characterized by the use of a highly active absorbent which is prepared
from lime, gypsum and flyash by aging in hot water. For development of the spray dry
system of LILAC process, improvement of the atomizer as well as the absorbent was a
key point. Hie results of demonstration test for spray dry system and duct injection
system are described in terms of absorbent preparation and designing the spray dryer
including atomizer.

As the results of the demonstration test, it is revealed that the spray dry system of
LILAC process is capable of achieving a high S02 removal efficiency and the duct
injection system of LILAC process has an ability of simultaneous S02 and NOX
removal.

The features of different types of FGD system are compared in terms of S02 removal
efficiency, utilities, and construction space and costs. The LILAC FGD processes are
found to be flexible and can be applied to a wide variety of the FGD requirements.

76-1


-------
introduction

Hokkaido Electric Power Co., Inc. (HEFCO) found that a solid compound prepared
from coal flyash, lime arid gypsum possesses a high desulfurization ability, and with
the use of this absorbent, a dry type of the commercial desulfurization plant was
established at Tomato-atsuma power station in March 1991.

On the basis of this active absorbent, HEPCO and Mitsubishi Heavy Industries Ltd.
(MHI) have jointly developed a new FGD system in which the preparation of the
absorbent and the construction of the process are modified for versatile applications.
The newly developed absorbent is called LILAC (Lively Intensified Lime-Ash
Compound), and the FGD process using the absorbent is called LILAC process. A
demonstration test for a spray drying LILAC process using the absorbent in the slurry
form for the flue gas from a commercial coal-fired boiler has started since October
1991.

Another LILAC process of a duct injection process in which powder type of the
absorbent has been performed since April 1993. These two kinds of processes have
been tested in the same pilot plant constructed at Tomato-atsuma power station.
In this paper, we show the outline of the LILAC processes as well as the results of the
demonstration test.

LILAC absorbent

The absorbent is prepared from a mixture containing flyash, lime and gypsum or spent
absorbent The spent absorbent is a by-product that is formed in the reaction of LILAC
absorbent with S02.

There are two types of the preparation methods of LILAC absorbent. A general
preparation method is shown in figure 1. First, raw materials are mixed and hydrated
in hot water at 95°C for about 15 min., followed by aging at 95°C for 12h. During the
aging period, porous compounds such as calcium silicate and ettringite are formed. A
slurry of the absorbent is obtained after the hydration process. Dehydration and drying
the slurry absorbent produces the powdery absorbent which is used in the duct
injection system of LILAC process.

In addition an absorbent containing a large amount of Ca (85 ~ 95 % as Ca(OH)2)
which is prepared without aging process of 95°C and 12h, but hydration of only ISmin
during mixing of the raw materials was developed to lower equipment cost.

These LILAC absorbents that are prepared by the methods mentioned above are
adopted in demonstration test.

•	LILAC-50S = 50% of Ca(OH)2 contents, 15min hydration and 12h aging, slurry form.

•	LILAC-50P = 50% of Ca(OH)2 contents, 15min hydration and 12h aging, powdery

76-2


-------
form.

•	LILAC-85S = 85% of Ca(OH)2 contents, only 15min hydration, slurry form

•	LILAC-85P = 85% of Ca(OH)2 contents, only 15min hydration, powdery form.

LILAC system

Pilot plant

A pilot plant capable of treating 10000 m3N/h of flue gas from No.2 coal-fired boiler
unit of Tomato-atsuma power statical (HEPCO) was constructed and put into
operation in October 1991 to test the spray drying process and. duct injection process.
Figure 2 is a simplified process flow diagram for the pilot plant of the LILAC system.
Flue gas is extracted from the outlet duct on the unit 2 boiler unit having 600 MW
generating capacity. The normal S02 concentration of the flue gas from this boiler unit
is about 400 ppm. The S02 concentration of the LILAC pilot plant inlet gas can be
raised by adding a pure $02 gas to allow testing at pilot plant inlet S02 levels of 2000
ppm and greater.

In the test of the spray dry system, flue gas enters the spray dryer, where it contacts a
finely atomized slurry form of LILAC absorbent. The water content of the slurry
evaporates, cooling the flue gas, and at the outlet of the spray dryer, the temperature of
gas is slightly above the dew point of the gas. Simultaneously, the LILAC absorbent
reacts with flue gas S02.

The duct injection system is also tested at the pilot plant. The powdery form of the
absorbent is injected in the outlet duct on the spray dryer. In the both systems, the
spent absorbent that is powdery form is captured by electrostatic precipitator or bag
filter and a part of the spent absorbent is supplied to the absorbent preparation unit as
a gypsum source to produce LILAC absorbent.

Characteristics of the equipments

In order to obtain a high S02 removal performance in the spray dry FGD system, it is
indispensable to develop a sprayer of high performance as well as a high active
absorbent. The sprayer is required to scrub a large volume of a low approach
temperature gas in stable operation without causing accumulation of solids in the
spray dryer. A rotary atomizer and spray dryer were designed to meet the
requirement.

The important points in the development of the rotary atomizer are the size and
materials of the atomizer. A small size atomizer installed at the top of the spray dryer
permits easy maintenance access, simplified top frame structure and reduced electric

76-3


-------
power consumption. Adoption of a high frequency motor made it possible to reduce
the size to one half that of a conventional one. The abrasion resistant materials for the
atomizer were selected through abrasion resistance tests which also had to be highly
precise and cost effective. To simulate the deposition and accumulation of solids on the
top of the disk, a fluid dynamics analysis was conducted by taking the velocity and
angle of gas flow and the turning angle of the gas introduced from the periphery as the
parameters. On the basis of the analysis, the optimum parameters were set for sprayed
particles so as not to let them adhere on the spray dryer wall.

The atomizer was also subjected to visibility tests with Tuft method and simulated gas
using a testing unit at the Takasago Research Center of MHL The results indicated that
even a slight drift in gas flow causes local deposition of solids and acceleration of
deposit with the elapse of time. This problem could be completely resolved by
adopting an air curtain system.

The important points in developing spray dryer are the proper gas particle mixing and
avoidance of particle impingement on the wall. Uniform and adequate mixing of the
sprayed particles and gas is required. Non-uniform mixing causes deposition and
accumulation of solids inside the spray dryer. In addition, existence of diluted spot of
the sprayed particles reduces the efficiency of sulfur removal. The impingement of the
sprayed particles was avoided by improving the gas introduction method and shape of
the tower. The impingement of the sprayed particles on the wall prior to evaporation
of water makes it impossible to continue operation because of rapid accumulation of
solids. To deal with this subject, computational fluid dynamics (CFD) analysis was
conducted for ideal gas introduction method and shape of the tower. The CFD analysis
is useful in designing a spray dryer of small diameter with high speed rotary atomizer,
and could be applied to the spray drying FGD system in the planning of the plants
which have space limitation.

Demonstration test results

The demonstration test for spray dry system and duct injection system has been carried
out with the pilot plant. As a results of the test, the features of both LILAC processes
have been revealed.

Spray dry system

The test results showed that mixing of the sprayed particles and flue gas was sufficient
to obtain high efficiency of SO2 removal, and that a stable continuous operation was
performed without the build up of solids. It is confirmed consequently that the CFD
analysis adopted in designing the spray dryer has promising scope for its practical
application.

76-4


-------
The relation between S02 removal efficiency and Ca/S molar ratio for three kinds of
absorbent is shown in figure 3. Hie hydrated lime absorbent which is commonly use in
a wet system showed the S02 removal efficiency of about 55% under the conditions of
the approach temperature 15 -17 °C, Ca/S molar ratio 1.2 and no recycled solids.
LILAC-50S absorbent achieved much higher efficiency of about 75% under the same
condition. Hie efficiency is comparable to that of the wet system. LILAC-85S absorbent
showed the ability in between those of LILAC-50S and hydrated lime. Although the
ability of LILAC-85S absorbent is not high as compared to the IILAC-50S absorbent, it
has the advantage of a simplified preparation method in which no further aging
period is required after hydration during mixing of raw materials for 15min.

The relation between SO2 removal efficiency and inlet SO2 concentration is shown in
figure 4. The SO2 removal efficiency of ULAC-50S absorbent is high and keeps
constant with an increase in the inlet SO2 concentration. The SO2 removal efficiencies
for both LILAC-65S and hydrated lime decreased as the inlet SO2 concentration
increased. As shown in figure 4, when a higher S02 removal efficiency is desired
LILAC process is more feasible than hydrated lime process. Selection of the absorbent
depends upon the SO2 removal efficiency required.

The approach temperature defined as the outlet temperature subtracted by the
temperature of the dew point of the gas also influences the SO2 removal efficiency.
The relation between S02 removal efficiency and the approach temperature is shown
in figure 5. The SO2 removal efficiency becomes high as the approach temperature
becomes low.

Duct injection system

Duct infection system using IILAC-50P and LTLAC-85P absorbent have been also
tested in the pilot plant. In the system, not only SO2 removal but also NOX removal
occur. Figure 6 shows SO2 and NOX removal efficiencies as a function of Ca/S molar
ratio. Both SO2 and NOx removal efficiencies increased with the high Ca/S molar
ratio. At the Ca/S molar ratio of 2.9, the S02 and NOX removal efficiencies reached 75
and 55 %, respectively. Simultaneous removal of SO2 and NOX is an important
advantage for the powdery LILAC absorbent. A duct injection system employing the
powdery LILAC absorbent is therefore equivalent to the combination of SO2 and NOX
removal facilities. In order to make the duct injection system more competitive and
reliable, further test program is being prepared and such test results will be presented
in future issues.

Comparison with other FGD systems

For selecting the optimum FGD system, the points to be taken into account are (1)
capacity of the equipment, (2) fuel type, (3) absorbent cost, (4) SO2 removal efficiency
required, (5) plant availability, and (6) disposable or salable by-products. Features of

76-5


-------
different systems, the wet limestone/gypsum FGD system, and spray dry systems
using hydrated lime, LILAC-85S and LILAC-50S, are compared in terms of SO2
removal performance, utilities, construction space, and construction cost

Hie wet limestone/gypsum system is suitable for large size plants requiring 90 to 95 %
SO2 removal efficiency, and is also applicable for fuels such as coal, heavy oil,
orimulsion, etc., which have a wide range of sulfur content Because of these merits,
the wet limestone/gypsum system is widely adopted at present in spite of its high cost.

The spray dry system using lime is suitable in a place where supply of quick lime or
hydrated lime is available at a stable price since the capital cost is much lower than of
the wet system.

The spray dry system using LTLAC-50S absorbent has 80 to 90 % SO2 removal
efficiency at almost the same Ca/S molar ratio as that of the wet limestone/gypsum
system. The spray dry system using LILAC-85S absorbent has the advantage of
achieving the SO2 removal efficiency higher than that of the spray dry system using
hydrated lime.

Conclusion

The LILAC FGD system is applicable to various types of flue gas treatment for its
flexibility in the preparation of different forms of the absorbent. This system is capable
of achieving a high SO2 removal efficiency without waste water treatment which is
usually requited in a wet type FGD system like the limestone/gypsum method.
Because of omission of the waste water treatment unit, the LILAC system is composed
if simplified equipment, and therefore, its capital and operational costs are low. In
addition, it permits simultaneous SO2 and NOx removal with the duct injection system
using the powdery LILAC absorbent.

Figure. 1 Preparation of LILAC absorbent

76-6


-------
Mixing Aging
Tank Tank

Spent Absorbent (SA)

Figure 2. Simplified Process How Diagram for The Pilot Plant

J

as
u

53
>
O

Pi

C4

Q

cf>

40

S02 conc. = 2000ppm
App. temp. = 15-17 °C

0.5

T TT A rAC

iiLAC—50b

TILAC-SSS

CaOffl2

1.0	1.5

Ca/S Molar Ratio

2.0

Figure 3. S02 removal rate vs Ca/S

76-7


-------
90

Si

*?d
>
o

g

4)

s

Cfi

80 -

70 -

60 -

50 •

Ca/S= 1.3

App. Temp. = 15-17 °C

40

LILAC-50S

Ca(OH)2

0	1000 2000 3000

Spray Dryer Inlet S02 (ppm dry)

Figure 4. Effect of inlet S02 concentration on S02 removal

Figure 5. Effect of approach temperature on S02 removal

76-8

A


-------
100

Ca/S

Figure 6. Typical S02 and NOx removal performance of duct injection system

Tablel. System comparison

FGD System

Wet System

Spray Dry System

Absorbent

CaC03

Ca(OH)2

1JLAC-85S

IJLAC-50S

S02 Removal rate (%)

90

65

70

80

Ca/S

1.05

1.5

1.4

1.1

Approach temp. (°C)

-

12

12

10

Outlet gas temp. (°C)

90

-

—



Utility Corisump.









Electric Power (KW)

1600

820

880

960

W ater (t/h)

16.8

14

14

14

Absorbent (t/h)

2.5

2.0

1.9

1.5

Other raw materials

none

none

Flyash, SA

Flyash, SA

Required area (m2)

270

190

. 200

210

Initial cost

100

65

70

80

Basis of plant: S02 cone = 2000 ppm, Gas volume = 300,000 m3N/h

76-9


-------
DEVELOPMENT OF HIGH EFFICIENCY WET LIMESTONE GYPSUM FLOS GAS

DISTOFDRIZATION SYSTEM

S. Kotake
Y. Tsuchiya
T. Higashi
K. Iwashita

Mitsubishi Heavy Industries Ltd.
2-5-1 Marunouchi
Chiyoda -ku, Tokyo ,100 Japan

ABSTRACT

Construction of new coal-fired thermal power generation plants
has become increasingly difficult year by year due to their
potential impact on the environment. In order to overcome this
problem, Mitsubishi Heavy Industries Ltd., has developed a
high efficiency wet limestone gypsum flue gas desulfurization
system which is featured for its exceedingly high dedusting
efficiency (particulate concentration of less than 10mg/m3N in
stack gas), lower electricity consumption (90% of the
conventional system), and smaller space requirement. These key
advantages were confirmed by the demonstration test conducted
jointly with Chubu Electric Power Co., Inc. with a 15,000
m3N/H capacity pilot plant employing GGH upstream of the
electrostatic precipitator and by the actual plant operations
of horizontal shaft variable pitch vane absorber recirculation
pumps, arm rotating spargers and etc.

OUTLINE OF THE SYSTEM

Recently for coal-fired thermal power generating plants in
Japan, high-dedusting performance requirement for FGD system
has been increasing to prevent potential impact on
environment, while maintaining lower operating cost and lower
space requirement.

As shown in Fig. 1, conventional FGD system with regenerative
Gas Gas Heater (GGH) was allowed to emit dust below 30mg/m3N
due to high dust leakage at regenerative GGH. In Hekinan Power

77-1


-------
Station of Chubu Electric Power Company recently under
commercial operation, nonleak type GGH and Wet type
Electrostatic Precipitator (WEP) were adopted to cope with,
severe requirement for low dust emission of less than 10mg/m3N.

In order to realize more feasible FGD with high dedusting
performance, Mitsubishi Heavy Industries Ltd. has developed
the High Efficiency Wet Limestone Gypsum Flue Gas
Desulfurization System. High dedusting performance of this
advanced system was confirmed by the demonstration, test
conducted jointly with Chubu Electric Power company with a
15,OOOm^/H capacity pilot plant. Other key technologies except
high dedusting performance, which promote features of the
advanced system, were developed separately from this pilot
plant test, and some of them have been already applied for
commercial plants.

In addition waste water discharge can be eliminated by the
combination with the waste water concentration and
solidification(WCS) system, details of which are described in
the other paper.

System flow for this system is shown in Fig 2, where nonleak
type GGH is placed at the upstream of dry ESP, and the gas
temperature through dry ESP is brought down to 100-90 °C instead
of 140~130°C of conventional practice.

High Dedusting Performan ce

With this low gas temperature, ESP has remarkably high
dedusting performance due to low electrical resistivity of
dust, and dust contents in outlet gas can be kept stably
within 30mg/m3N for variety of coal types. The gas temperature
is lower than SO3 dew point, thus most of the S03 in sulfuric
acid form is removed with dust in ESP. Dedusting performance
of the desulfurization unit is also remarkably high because
particle size at ESP outlet is coarser than the conventional
system. Combined effect of high dedusting performance of ESP
and absorber enables lowering of the dust concentratation at
the stack to less than 10mg/m3N without WEP.

Lower Electricity Power Consumption

Power consumption for boosting up flue gas by IDF and
BUF is totally lower than conventional system due to the
following reasons;

a.	Power consumption for induced draft fan (IDF) is
equivalent to that of conventional system because

reduced flue gas volume due to low temperature
compensates the additional draft loss of nonleak GGH
(heat extractor).

b.	Power consumption for FGD boost up fan (BUF) is lower
than the conventional one because of the reduction of
the draft loss of nonleak GGH (heat extractor).

77-2


-------
In addition to the reduction of power consumption for IDF and
BUF, the following key technologies of desulfurization unit
enable far lower power consumption, which are described later
in detail.

•	Arm rotating sparger (ARS)

*	Horizontal shaft variable pitch vane absorber
recirculation pump

Energy saving control by CaC03 sensor

Simple and Compact Absorber

Single loop co-current type grid-packed tower is suitable for
low dust condition at dry ESP outlet, due to the following
reasons:

a.	Lower power consumption and compactness in comparison
with other type of absorber.

b.	By-product gypsum can be wallboard grade (95% or higher
purity) even when low sulfur coal is used because

fly ash amount removed at absorber is far lower than the
conventional system as shown in Figure 3.

No Waste Disposal

Sludge from the waste water treatment is mixed with a part of
the by-product gypsum and then reused as the gypsum for cement
industries. Thus, the treatment of waste matters becomes
unnecessary due to reduced flyash inflow in absorber.

DEMONSTRATION TEST RESULT FOR HIGH DEDDSTING PERFORMANCE

Demonstration test was conducted during 1988-1990 with a pilot
plant installed at Chubu Electrics Shin-Nagoya Thermal Power
station. The pilot plant have 15,000m3N/H capacity with which
system performance for both the conventional and advanced
system for dedusting performance comparison was conducted.

Dry ESF

In the advanced system, dust electric resistivity could be
kept lower than conventional system due to low gas temperature
which consequently deleted back corona as shown in Figure 4.
As a result, stably high dedusting performance was achieved
and the dust concentration at the ESP outlet was below 30mg/m3N
for various types of coal as shown in Figure 5.

It was confirmed in the test that such problems as increased
corrosion and dust adherence in ESP due to low gas temperature
were equivalent to those in the conventional system.

77-3


-------
Desulfuriza tion Unit.

Dedusting performance of desulfurization unit was also
remarkably high as shown in Figure 6, the reasons of which
could be as follows. In the advanced system, dedusting
performance of ESP is sufficiently high due to low gas
temperature. Thus, the ISP outlet dust mainly comprise the
coarse particles formed by aggregation of dust once adhered to
the collection electrodes and partly rescattered by the
rapping effect etc. Figure 7 shows that the absorber inlet
dust particle size for advanced system is coarser than that
for the conventional system.

Gas-Gas Beater

In the advanced system, dust concentration at GGH (Heat
Extractor) is approximately 100 times that of the conventional
system, hence performance deterioration by dust deposit on the
heat transfer tubes was anticipated. As a preventive measure
for this. Steel Shot Cleaning System (SSCS) was provided in
the pilot test plant. SSCS was developed as an alternative to
conventional pneumatic/steam soot blower and has already been
adopted for actual plants.

The highly effective dust removal performance of SSCS ensures
stable heat transfer and pressure loss over a prolonged period
as shown in Figure 8. Dust removal effect has been proved for
SSCS even under high dust concentration with appropriate
scatter of steel shot as shown in Figure 9.

Principal Equipment of Desulfurization Unit

The following key technologies of desulfurization unit: have
been developed separately from the demonstration test for high
dedusting performance, and have been employed in actual plants
already. These technologies comprise Mitsubishi High
Efficiency System and promote the feature of this advanced
system i.e. lower electricity consumption and lower space
requirement.

Co-current Absorber

Single loop co-current grid-packed absorber has been employed
in a large number of plants due to its lower power consumption
and compactness in size. (ref. Figure 10 and Figure 11)

In the co-current flow, as the absorbent liquid comes in

contact with flue gas flowing in the same direction, the gas
pressure loss becomes small even when the gas flow velocity is
made to be more than 4.5m/s which is 1.5 times that of the
counter-current flow systems. This enables a compact and
energy saving absorber to be designed. Moreover, since the
recirculating slurry is discharged without getting pressurized
by the spray pipe (atomizing back pressure 0.2kg/cnrG), the
pump head compared to the pressurized spray system (atomizing

77-4


-------
back pressure l.Skg/cm^) is lower to minimize the pump power
consumption.

Arm rotating sparger (MRS)

ARS is a Mitsubishi developed efficient oxidation air sparger.
In its principle, air is blown through the rotating blades of
the absorber agitator, and utilizing the water stream
generated by the rotating blades, the air is effectively
dispersed in fine bubbles to produce improved gas-slurry
contact. As a result, oxidation is carried out efficiently
in a small liquid retention tank with a small volume of air.
Further, with the rotation of blades and dispersion of air
bubbles a state of complete mixing between air and slurry
occurs in the absorber liquid retention section which prevents
any accumulation of solids at the absorber bottom while
absorption and neutralization reactions get promoted to
generate high grade gypsum as the by-product, (ref. Figure 12)

Borizontal Shaft Variable Vane Pitch Absorber Recirculation
Pump

Absorber slurry recirculation pumps are core equipment of an
FGD plant in that they efficiently carry out desulfurization
control. In the desulfurization system, stipulated outlet S02
concentration is required to be achieved economically under
various operation conditions, and this need can be satisfied
with the large size horizontal shaft variable vane pitch pump
developped by MHI. (ref. Figure 13 and Figure 14)

In this system by varying the pump vane pitch, the flow rate
can be controlled linearly to enable economical control of the
desired outlet S02 concentration instead of the conventional
stagewise control by changing the number of operating pumps.
Further, as capacity of these pumps can be made as large as 3
times that of the conventional pumps, the number of pumps can
be reduced with consequent saving in installation space while
operation with good flow rate control can be obtained together
with reduction in maintenance works.

Energy Saving Desulfurization Control System

h continuous automatic calcium carbonate slurry analyzer has
been developed and has already been installed at commercial
plants. This analyzer enables energy saving desulfurization
control and improved follow-up ability for boiler load as
shown in Figure 15 and 16.

77-5


-------
SUMMARY OF ADVANTAGES

Based on the pilot; test result, advantages of the advanced
system are evaluated in comparison with conventional system as
shown below:

System Type
WET ESP
GGH Type

Dust at Stack Inlet
Power Consumption
Plant Area
Treatment of waste

Conventional
not installed
Regenerative
< 30mg/m3N
95-97%
90-95%
necessary

Conventional
installed
Nonleak
^ 1 Omg /iHjK
100%

100%
necessary

advanced
not installed

Nonleak
< 10mg/m3N
85-90%
80-85%
unnecessary

It is concluded that;

1.	Stably high dedusting performance can be obtained without
wet ESP.

2.	Power consumption can be reduced due to power reduction of
IDF/BUF and adoption of other technologies.

3.	Waste water discharge can be eliminated by the WCS system.

4.	Plant area can be reduced due to omission of wet ESP,
adoption of single loop co-current type absorber, and the
WCS system.

77-6


-------
(1) Conventional system (Low dedlisting performance)

Regenerative rotary type GGH	Final dust emission

* | \ <30 mg/m3N

Boiler

DeNOx

AH

—*-[ ESP

GGH

Absorber

GGH

(2) Conventional system (High dedusting performance)

(3) Advanced system

Non-leakage type GGH

| Boiler |—>¦

DeNOx

AH

GGH

ESP



Absorber



GGH



	>¦





~^j \ <10mg/m3N

Figure 1 FGD System and Final Dust Emission

Gas temperature (°C)

135

Fly Ash (mg,'m3N)

20,000

SOj; 

1,000

S03 (mg/mSN)

<100

1 Boiler { 0| DeNOx j O j AH J O

l

	i

Treated Water

Wastewater
Concentration &
Solidification System

Land Fill Solid

Gypsum for Cement

Figure 2 System Flow of Advanced System

77-7


-------
Remarkable Deduction of Removed Fly Ash in Absorber ft-

Conventional System
(Dual-Loop)

Advanced System
(Single-Loop)

145mg/m3N

Quencher

Single Loop System

Recover)
as Fly ash

sludge

25mg/m3N

Absorber



(Example for 1000MW)

1.2%Sin Coal

Wall Board Grade
(>95wt%)

Waste

Cement Grade
(>85wt%)

Fly Ash
Sludge

(Conventional 0.4)

Unit:T/H

0.3%S in Coal

Wail Board Cement
Grade	Grade

(>95wt%) (>85wt%)

Waste

Fly Ash
Sludge

(Conventional 0.4)

Figure 3 Byproduct Gypsum for Advanced System

77-8


-------
Electrical

Resistivity
(fl-cm)

1013 -

1Q12

1011

1010

High

Temperature
/ ESP

Critical Resistivity
for Back Corona

100	200	30

Advanced
System

Gas Temperature fC)

400

Figure 4 Electrical Resistivity of Ash vs Gas Temperature

Dust
correction
Efficiency
(%)

f

99.9

99.5

99

90

Advanced Conventional
System	System	

(Variable Range
depending on
yCoal type

J	I	L.

100	150

-» Gas Temperature (°C)

Ficrure 5 Dust Correction Efficienty vs Gas Temperature

77-9


-------
Figure 6 Absorber Dedusting Performance for
Conventional/Advanced System

Acumulative
Weight
R%

Key

Coal

O

P

A

A

0

B

O

C

~

D

Acumulative
Weight
R%

99
99.S

Conventions I System

Key

Coal

A

A

O

B

O

C

~

D

Dust Particle Size (Pm)	Dust Particle Size (Mm)

Figure 7 Absorber Inlet Dust Particle Size

77-10

A


-------
100

AP/APo
(%)

100

U/U0(%)

50
0

»

«

Pressure drops

t t

•

*

1

• '

*

•

r *

«



<1 t

• * • •• •

t

•

Heat transfer rate

i i

5



50

100 ISO
Operating Time (days)

200



Figure 8 Change in Heat Transfer Rate and Pressure Drop

Conventional system

100

U/Uq{%)

90 -

80

High performance system

70

I

300	600	900

Scatter rate of steel shot (kg/m* h)

Figure 9 Dust Removal Characteristics of SSCS

77-11


-------
Gas Inlet

n

Spray Pipe

Absorber Mist Eliminator

Gas Outlet

Dedusting Pump

Absorber	^S!eS	-.ARS

Recirculation Pump Absorber

Oxidation Air Blower

Figure 10 Structural Diagram of Absorber

Figure 11 Photograph of Absorber

(Chubu Electric Power Co.,
Hekinan P/S No.l unit)

77-12


-------
Features

1.	Energy saving type with 50%
oxygen utilization

2.	Highly efficient air-liquid
contact permits small absorber
liquid retention volume

3.	Compared with fixed air sparger
system, the number of air
discharge nozzles is reduced to
1/40 with complete elimination
of complicated air pipings.

4.	Easy washabiiity of air pipes
fully prevents any plugging due
to solids build-up.

Principle |~

Dispersion of oxidation air by the
agitator

•	Generation of fine air bubbles

•	Dispersion over an wide area

Effective oxidation with small
volume of air

Figure 12 Structural Diagram of ARS

77-13


-------
Features

1. Reduction in pump running costs during intermediate load
operations

Absorber
recirculation
slurry flow rate

Fixed vane
pitch pump

(changing'
Operation oygp

ra"9e number of
operating
pumps)

Reduction
in power
consumption

Boiler Load (IV1W)

Variabievane pitch pump

(Control of vane angle)
	^

2. Reduction in pump number with
enlargement of pump size

Reduced installation
area and maintenance

Conventional fixed vane pitch pump 	 Max: 5,000 m3/h

Variable vane pitch pump 			 Max: up to 15,000 m3/h

Figure 13 Features of Horizontal Shaft Vane Pitch Pump

Vane -11
Control Actuator

Operatng Lever

Discharge

Casing

~^p^Runiier Vane

n 'W~

„	jr.	/ \ k,._

Vane	/ \	i ^	*

Control Rod J	% ky. _

I ^ gSv	i _ *.»•.,y	__A

ill Shafggatase/'—¦—i

X^d 1-3L/' *'"'1® J ^ : Sucti01

Beariflg^"^ T /2k " % 1

Slide BeariflK	\P Suction Liner

Suction

Bearing-



Slide Bearing Box

Thrust Bearing-?.

^Runner Boss

Figure 14 Construction of Horizontal Shaft
Variable Vane Pitch Pump

77-14


-------
Function p					—				

Control of Outlet SO2 Concentration

SO2 Concentration at absorber outlet is controlled by controlling the
flow rate of recirculating slurry in the absorber with the variable vane
pitch pumps. At this time, by controlling the concentration of
unreacted CaC03 in absorbent slurry the calcium utilization rate can be
maintained while keeping the slurry recirculation amount to the
minimum.

^To Gypsum Dewatering
Equipment

Limestone

Figure 15 Outlet S02 Control System

77-15


-------
Reduction in Operation Cost

During low boiler load, absorber power consumption can be effectively
reduced by decreasing the absorber recirculation slurry amount.

Absorber Power
Consumption
(pumps + fan)





Boiler Load

Improvement in speedy follow-up ability of boiler loads

Even for partial loads, a speedy and steady follow-up of boiler load is
possible by only changing the absorber recirculation amount while
temporarily stopping the limestone supply as the concentration of
unreacted CaCO^ in absorbent slurry is maintained at a constant

Boiler load

Absorber recirculation
slurry flow rate

Unreacted CaCCb
conc.

Absorber outlet S02
conc.

^ jjji-^ :=:	il^^l j-:: ^

; ;v:j	rH.ill;i l: | ;:i;; ;Vr- 1-l'S i; ? ^

Chain lines indicate conventional control during constant pH control and
fixed number of operating pumps

Figure 16 Features of Energy-saving Desulfurization Control

77-16


-------
Chemistry of limestone Slurry scrubbing

R, S. Agarwal
G. T. Rochelle
Department of Chemical Engineering
University of Texas at Austin
Austin, Texas 78712-1062

Abstract

Limestone slurry scrubbing is the dominant technology for the removal of SO2 from
flue gases in coal based power plants. A fundamental, rate-based model to simulate
the process has been developed at the University of Texas at Austin. This model
incorporates principles of mass transfer with chemical reaction and electrolyte
thermodynamics. The model has been used to simulate the effects of important
operating variables such as inlet SO2 concentration, ratio of liquid to gas flow rate,
hold tank residence time and limestone utilization. This paper uses the model to
explain the behavior of the system due to changes in the above operating variables.

Introduction

Development of the limestone slurry scrubbing model at the University of Texas
has been ongoing for 10 years. The model for gas/liquid absorption of SO2 in CaC03
slurry was written by Mehta (1). Chan (2) incorporated the gas/liquid mass transfer
of both CO2 and O2, along with the mass balance for the solid species. The model
configuration was altered to include a hold tank and the thermodynamic speciation
was modified to use pseudo-equilibrium concentrations. Chan also used the model
to predict the effects of chlorides (2). Gage (3) incorporated the effects of limestone
type and grind as well as the effect of the presence of sulfite in solution on the
dissolution of limestone. Since then the model has been modified for parameter
estimation from field data. Many changes have been implemented to improve
robustness and speed of convergence.

Submitted for presentation at die 1993 SO2 Control Symposium, Boston, Mass., August 26, 1993.

78-1


-------
Other Models

The Electric Power Research Institute (EPRI) has funded development of a similar
model (4). The Flue Gas Desulfurization Process Integration and Simulation Model
(FGDPRISM) is based on many of the concepts developed at the University of Texas.

FGDPRISM is based on the same thermodynamic data base, and similar gas/liquid
mass transfer models. FGDPRISM does not incorporate the effects of limestone type
and grind or the effects of sulfite concentration on limestone dissolution in the
same way as the UT model (3). The UT model also incorporates calculations for
sulfite dissolution and gypsum crystallization (5), which are not present in
FGDPRISM.

FGDPRISM uses a droplet size distribution model for the prediction of the effect of
spray on mass transfer (4). FGDPRISM has been used to predict the effects of (L/G),
chlorides, pH and DBA on operating systems of Spray scrubbers, after fitting the

model to available data (4).

The UT model has been used to predict the effects of limestone type and grind on
the overall system chemistry (3) and to simulate the effects of chlorides (2).

Model Description

The UT model simulates the limestone slurry scrubbing system as a staged absorber
with a hold tank (Figure 1). Upward moving gas is counter currently contacted with
limestone slurry, causing absorption of SO2 and O2 and some desorption of CO2.
The absorbed SO2 reacts in solution to form sulfite species, some of which are
oxidized by the absorbed oxygen to form sulfate. The solution from the scrubber is
then taken into the hold tank where a fresh feed of limestone is added and the
sulfites and sulfates removed by desupersaturation.

The UT model calculates the absorption/desorption of gases as well as the
dissolution of solid species using rate-based mass transfer models. Models for
calculating the crystallization of gypsum, calcium sulfite and mixed crystal are based
on the work of Tseng and Rochelle (5). The equilibrium constants for ionic
reactions have been calculated based on the work of Lowell (1970). A detailed
description of all the assumptions used in the above models as well as the
convergence procedure for calculations is given below.

Solution Thermodynamics

Equilibrium is assumed among ionic species in the bulk liquid as well as
solid/liquid and gas/liquid boundary layers. A rigorous calculation of the
equilibrium compositions is done based on temperature dependent equilibrium
constants compiled by Lowell et al. (6) for 41 FGD species. The effect of ionic
strength on activity coefficients is calculated using the extended Debye-Huckel

78-2


-------
theory. These calculations are contained in the Bechtel-modified Radian
Equilibrium Program (7).

-11/2

logTi = AZj2 ( 1+Ba.°Ii/2 + -biI) + UiI

I = Ionic strength

1 n

-2

j=l

Yi = activity coefficient
B, aj0, bi, Uj = constants

Chan and Rochelle (8) have simplified the solution equilibria and reduced the
solution species to 17 (from 41) by eliminating ion pairs (CaSOj0, CaCC>30, etc.). The
concentrations, diffusivities and equilibria for the pseudo species are defined based
on the values of the these constants for the component species. In case of the
bicarbonate pseudo species concentration, equilibrium constant and diffusivity are
calculated as below.

[HCO3"] pseudo = EHCO3-] + [CaHC03+] + [MgHCQ^]

tr — P^4*] t^03=]pseudo
%seud0" [HCQrlpseudo

2D| G

DPseudo= SCJ

These definitions include the effects of activity coefficients in the equilibrium
constants and are valid as long as the ionic strength of the solution and the Ca++
concentration do not change significantly.

Temperature dependent parameters for CaHCC>3+, CaC03°, and the solubility
product of calcite were updated by Gage (3) to the values given by Plummer and
Busenberg (9).

34765 05

log KCaHC03+ = 1209.12 + 0.31294T 		- 478.782 logT

35512 75

log KcaC03= = -1228.732 - 0.2944T + —j1— + 485.818 T

2839.319

log Kcaicite = -171.9065 - 0.077993T +	^	+ 71.595 logT

78-3


-------
Three bisulfite equilibrium constants have also been added by Gage (3)

KcaHS03+ = 0.073
KMgHS03+= 0.15
KNaHS03+ = 2-53

Calculation for Gas/Liquid Mass Transfer

The two film theory of mass transfer is used for the calculation of gas/liquid mass
transfer. In the liquid film the ionic species are assumed to be at equilibrium. To
account for the transport of species having unequal diffusivities an approximation
of surface renewal theory is applied to the liquid film (10). The flux is calculated as
prescribed by film theory, however the diffusivity (D) is replaced by square root of its

value in film theory (%fD).

The Flux for SO2 is given in terms of the total enhancement factor, Etotal/ by:

Flux SO2 = kl,S02° Etotal ([SC>2l - [SC^y
The total enhancement factor consists of contributions from several reactions;

Etotal® 1 + EH+ + Es03- + EhC03"+ ES04=

The most important reaction is hydrolysis of SO2 to produce hydrogen ion and
bisulfite:

Hydrolysis:	S02 + H20 H+ +HSO3-

Eh+	(EHii-[H+]b)

Dso2 ([S02]i-[S02]b)

Dissolved alkaline species such as sulfite and bicarbonate react directly with SO2 to
produce bisulfite:

Alkalinity:	S03= + S02 + H20 « 2HSO3-

_ ^ /Ds03=([S03=lb-[S03=l)

Dso2 ([sOzli-tSOzlb)

HCO3- + S02 <=> HSO3- + co2

. /Dhcos-mcos-h - [HCOs-D
Ehc°3"" V Dsoa ([SOzli-ISOik)

Buffer couples such as sulfate/bisulfate and oxygen also react with S02 to give the
last contributing term:

Bisulfate Buffer:	S04= + S02 + H20 <=> HSO4- + H503"

Oxidation:	S04= + S02 +1 /2 02 + H20 2 HSOf

78-4

A


-------
tj - -v I PSQ4~ ([SC>4=jb " [S04~|)

Eso4"-"\/ Dg02 ([S02]i-[S02]b}

The addition of Buffer species such as dibasic acids with pKa values between 3 and 6
is also known to enhance SO2 removal to a significant extent (7).

Dibasic Buffer:	A" + S02(aq) + H20 HS03" + HA

•e _ a / Pa- ([A=3b - EA~]j)
HA— yj Dso2([S02]i-[S02lbX

The absorption of CO2 is assumed to be completely liquid film controlled and the
hydrolysis reaction of CO2 is assumed to be at equilibrium (8):

CO2 + H20 <=>H+ + HCO3-

As in the case of C02, the absorption of 02 is assumed to be liquid phase controlled.
Oxygen is assumed to react instantaneously at the gas/liquid interface with sulfite
forming sulfate. The flux of oxygen is enhanced over physical absorption due to its
fast reaction in the boundary layer.

Flux = 2E kfHo.Pa

Ho2= Henry's constant for oxygen = 1177 (atm)/M, at 55 °C.

To calculate the driving force for gas/liquid mass transfer at specified liquid and gas
compositions, the model requires compositions at the gas/liquid interface . These
are calculated by numerical solution of a set of algebraic equations involving mass
balances at the interface, solution equilibria and assuming zero charge flux at the
gas/liquid interface. The zero charge flux assumption is valid for systems showing
small potential gradients in solution (11). The diffusivity values (5,9,12) of the
individual ionic and non ionic species are tabulated by Gage (3).

Since the gas is plug flow, gas phase absorption/desorption is calculated by
integration of a set of three ordinary differential equations. The Livermore Solver
for Ordinary Differential Equations (LSODE) is used to integrate the flux at the gas
liquid interface and calculate net absorption/desorption (P^i - Pj^0) per stage.

dPk _ kf £ VpjkACjk
h VDscfe

78-5


-------
Ng

k]° £ -%/l^lk ^ Cjjc

"5 ^

Pk,i " Pk,o =

J

0

k = SO2, CO2 and O2
A C = Cinterface ~ Qjulk

k^aZ
NS= G

In case of SO2

jk - SO2, S03= S04=, HSO3-, HSO4-
Solid/Liquid Transport

The dissolution of CaSC>3 and limestone is calculated using steady state mass
transfer. Compositions at the solid/liquid interface are calculated by numerical
solution of the algebraic equations involving mass balances, equilibria and
assuming zero charge flux at the interface (S), The calculated compositions are
verified using boundary conditions at the solid surface (i.e. solubility product and
continuity of flux). Since the liquid on each stage is assumed to be well mixed, the
dissolution/ crystallization flux for solids is uniform over the stage. The dissolution
rate of solids is calculated by an expression of the following form.

A(solid) , _

~~= EDjACj

aires

where

Atres - residence time per stage (s)

K' = dissolution rate constant

j = component ionic species (HCO3-, CC>3= in case of CaCC>3)

The dissolution models for limestone account for the effects of limestone type and
grind via the limestone dissolution constant (3). The effect of sulfite concentration
on the limestone dissolution is accounted for by an empirical model that uses a
combined mass transfer and reaction kinetics model. The presence of sulfite alters
the boundary condition at the limestone surface. The calculated flux is equated to
the boundary condition given below.

78-6


-------
_ ([CaCOslea-lCaCQsIs?0-5
[CaCC^o]s[CaS03°]s

The empirical crystallization models are based on the work of Tseng and Rochelle
(5), and apply to the crystallization of solid calcium sulfite and calcium sulfate.

ACCaSOs)	(RScaSOa -1)2

Atres _Kcry2 RSCaSOi

A(CaS04-i /2H2O)	A(CaSOs )

	7-		 0.2 (RScaSOJ	77	

Atres	Atres

A(CaS04-2H20)

At

•res

= K'cryl 0RSCaSO4 • 2H2O " D

Since the individual dissolution/crystallization rates are uniform over each stage,

the dissolution/crystallization constants can be rewritten as K = K' Atres. Hence, in
general solid dissolved = K (flux at the solid surface)

Stage Calculations

The composition of the liquid and the gas streams at the bottom end of each stage
are calculated from the composition of the liquid and gas at the top of the stage. The
stage model assumes that the liquid is completely backmixed and that the gas is in
plug flow. The mass balances for each stage are written in terms of four species.

[Speciesjout - [Species^ = Species Absorbed + Species Dissolved

Species	Absorption	Dissolution/ Crystallization

S(VI)	Z^-Po^o)	WRS 0,50.-1)

SCIV)	E< PS02,i - Ps02,o)	**0, < ®)A ISO*® -

o	CRScaSCte-D2

-2l
-------
Convergence Procedure

The calculations are begun at the top stage based on an initial guess for the solution
and gas compositions.

Stage Convergence.

1.	Based on a guess of the outlet solution composition gas absorption / desorption
and solids dissolution/crystallization are calculated.

2.	Mass balances are solved iteratively to calculate the outlet solution composition
from the stage.

Scrubber Convergence.

3.	Steps 1 and 2 are repeated till the bottom stage is reached. At the bottom stage
the calculated and specified SO2 compositions are compared. If the specified
tolerances are not met then the gas composition is updated and 1 through 3 are

repeated.

Overall Convergence.

4.	Hold tank convergence is calculated in the same manner as individual stage
convergence.

5.	Overall convergence is reached when the calculated hold tank outlet
concentrations are within the specified tolerance. If fee required tolerance is not
met then the inlet solution to the top stage is updated and steps 1 through 5 are

repeated.

Base Case Conditions

This work focuses on using the model to explain effects of utilization, L/G,
residence time and inlet SO2 on system performance. The base case values for
solution and solid compositions and most operating variables were obtained from a
limestone type and grind study conducted in a turbulent contact absorber at
EPA/RTP (3,14).

Number of Stages

3

CaSC>4 dissolution constant

3e5

Scrubber residence time (s)

9

CaSC>3 dissolution constant

3e5

Hold tank residence time (s)

540

Inlet slurry T (°K)

333

G flow rate (acfm)

300





L/G (1/gmol)

0.2

kj°/kg (cm3-atm/gmol)

200

Inlet SO2 (ppm)

2900

Ng

6.9

Chloride (mM)

56

Ni ( hold tank)

0.6

Limestone Utilization (%)

77

L/G (hold tank) (1/gmol)

45

Solids (%)

10

Oxidation (%)

8.4

78-8


-------
Results

The UT model was used to simulate the effects of inlet SO2 partial pressure,
residence time in the hold tank, limestone utilization, L/G, and inlet solution
composition on the performance of the slurry scrubbing system. SO2 penetration,
pH, and other solution compositions were calculated over a range of each of these
variables. The fractional penetration (p) is the ratio of outlet gas SO2 concentration
to the inlet concentration or 1-fractional removal (f). At complete gas film control
Ng = -lnp.

Effect of Inlet SO2

As the SO2 partial pressure at the inlet to a stage increases, penetration also increases
although the net absorption of SO2 (ppm) increases. Figure 2 shows the effect of
inlet SO2 concentration at 1900, 3200 and 5000 ppm on penetration. Total SO2 flux
equals the sum of fluxes due to physical absorption and reaction with SC>3=, HCO3",
HSO4-, buffer acid (HA) and H2O (hydrolysis). At high values of SO2 partial
pressure, the contribution of the enhancement reactions (Etotal) is smaller in
comparison to the contribution of physical absorption (Table 1). As the partial
pressure of SO2 decreases from 5000 ppm (bottom stage, high inlet SO2) to 626ppm
(top stage, low inlet SO2) the total enhancement increases (2.08 to 12.58). This is a
combination of two factors, a smaller driving force for SO2 absorption and lower
levels of H+ (0.76 vs 0.25 mM) at the gas liquid interface. It can be expected that
further decrease of SO2 partial pressure would decrease penetration until the gas
film control limit is reached. Figure 1 shows this limit at Kg/kg = 1.0 on the Y axis.

Figure 2 also shows the effect of partial pressure on the extent of gas film resistance
and penetration for each stage in a 3 stage turbulent contact scrubber. As the partial
pressure decreases ( from the bottom to the top stage), enhancement increases in
each case causing a greater tendency towards gas film control.

For each stage the overall gas film mass transfer coefficient, Kg is given by:

PS02 int" P*S02
kg" PS02 "P*S02

The fraction gas film resistance, Kg/kg, approaches 1.0 as the SO2 concentration
decreases. An estimate of penetration can be made by calculating Kg at the inlet

conditions of each stage:

Ke

-Inp=Ng^

The values of penetration calculated by the model are in fairly good agreement with
the above, proving that equilibrium concentration of SO2 in the bulk liquid is
unimportant.

78-9


-------
Effect of Limestone Utilization

As the limestone utilization decreases the pH levels rise and the penetration
decreases (Fig. 3). Limestone utilization is the moles of SO2 absorbed per mole of
limestone fed to the system. With a constant total slurry solids concentration,
decreasing utilization increases the CaCC>3 solids concentration and surface area
available for dissolution. The model also accounts for the effect of utilization on the
limestone particle size distribution. With increased surface area, the steady-state pH
required to dissolve approximately the same amount of CaCC>3 will be greater
throughout the system. Figure 4 shows that the pH of the scrubber feed increases
from 5.25 to 5.95 when the utilization decreases from 95 to 65% with 540 seconds
residence time. Table 2 shows that the bulk solution pH leaving the scrubber also
increases by 0.5 pH units over the same range of utilization.

The primary effect of pH is as an indicator of other processes that reduce the bisulfite
concentration in the scrubber. Scrubber bisulfite determines hydrogen ion
concentration at the gas/liquid interface through the equilibrium:

S02(1) + H20 «• HSO3- + H+

Bisulfite in the feed solution is reduced from 5.85 mM to 1.18 mM by higher pH at
65% utilization than at 95%. Bisulfite in the bottom stage is reduced from 22.6 mM
to 15 mM as an accumulation of reduced feed bisulfite and reduced dissolution of
calcium sulfite. The interface hydrogen ion concentration determines the
contribution of the above hydrolysis reaction to the SO2 enhancement. The
contribution of H+ to the SO2 enhancement factor in the bottom stage increase by
34% from 95 to 65% utilization. The combined effect of reduced Ps02 and reduced
bisulfite increases the H+ contribution by 65% on the top stage.

In the scrubber greater pH increases dissolved alkalinity as sulfite. The
enhancement due to sulfite in the bottom stage varies from 0.16 (8%) at 95%
utilization to 0.32 (10%) at 65% utilization. This effect is significant but not
overwhelming.

In the hold tank greater pH results in greater dissolved alkalinity as bicarbonate, but
not as sulfite. Sulfite concentration is reasonably stable at about 1 mM, which is
near the saturation value for the calcium sulfite ion pair. Greater bicarbonate
alkalinity carries over through the top stage of the scrubber where bicarbonate
alkalinity contributes 12% of the SO2 absorption at 65% utilization, but only 3% at
95% utilization. Even with greater pH, bicarbonate alkalinity is never important in
the middle or bottom stages of the scrubber.

Effect of Residence Time in the Hold Tank

The effect of reducing residence time from 540 to 100 seconds in the hold tank is to
decrease scrubber feed pH (6.05 to 5.74) because a greater driving force is needed to
dissolve about the same quantity of limestone. The concentration of bicarbonate

78-10


-------
decreases (2.85 to 1.52 mM) and the bisulfite increases (1.18mM to 2.32 mM),
resulting in somewhat greater SO2 penetration (0.17 vs 0.14). The decrease of
residence time has a significant effect on the performance of the first stage alone,
where the penetration is reduced from 0.43 to 0.36. Thereafter sufficient amounts of
limestone are dissolved in the following stages to maintain nearly equal levels of
penetration for both cases (100s vs 540s residence time). At 540 seconds residence
time, the hold tank is near equilibrium with respect to limestone dissolution and
further increase of the residence time (1000s, 2000s) produces very small changes in
penetration and pH.

Figure 4 shows the SO2 penetration as a function of utilization, but calculated at the
inlet solution composition with a constant utilization of 77%. Comparison of the
penetration at these conditions shows that the effect of changing utilization without
changing solution composition is about half of the total effect of allowing the hold
tank pH and solution composition to change when excess limestone is added. In
other words the scrubber performance depends on both the solution (bisulfite,
bicarbonate) and solids composition (limestone utilization) of the scrubber feed.

Comparison of the effect of residence time to the effect of utilization shows an
interesting phenomenon. Decrease of the hold tank residence time leads to a lower
level of dissolved sulfite and bicarbonate at the first stage. As expected this leads to a
smaller enhancement and greater penetration of SO2. After the first stage there are
very small differences in the extent of penetration and the levels of PH. It is
reasonable to say that in this case the extent of improvement in performance is
restricted to the first stage itself. The decrease of utilization results in a greater
penetration at all the stages of the scubber for reasons explained before.

Effect of Liquid-to-Gas Ratio (L/G)

The effect of L/G was simulated by adjusting the liquid rate with constant hold tank
volume and constant liquid phase residence time of 9 seconds in the scrubber.
Increased L/G reduces SO2 penetration because it provides additional solution
capacity and increases the mass transfer capability of the scrubber. Both of these
effects are quantified in Figure 5. With Ng constant at 6.9, only the effect of solution
capacity is quantified. With NS proportional to L/G, the combined effect of both
mechanisms is simulated. At low L/G, the two effects appear to be of equal
importance. At higher L/G, the effect of L/G on Ng is more significant.

Increase of the L/G causes a decrease in the level of bisulfite at each stage (19.6 vs
14.46 mM for bottom stage) due to the dilution caused by the additional solution
present. The decrease of bisulfite levels and greater enhancement due to SC>3= and
H+ lead to better performance at higher L/G. Table 3 shows that both Eh+ and Es03=
increase in both the top and bottom stages as the L/G increases.

Conclusions

78-11


-------
The model is a useful tool for improving the understanding of the limestone slurry
scrubbing process. The model calculations provide a means of determining the
relative contributions of the individual rate based processes at each stage of the
process.

1.	The UT model correctly predicts that SO2 penetration decreases with decreasing
inlet SO2 and limestone utilization and increasing L/C and hold tank residence
time. As the inlet partial pressure decreases the model predicts a decrease of
Kg/kg i.e. a shift from liquid film control to gas film control.

2.	At the conditions of these simulations, the hydrolysis of SO2 to bisulfite was the
primary reaction enhancing SO2 absorption. Bicarbonate and sulfite alkalinity
was usually not a significant factor. The level of bicarbonate in the bulk solution
is also an important indicator of the capacity of the solution for absorption of

S02.

3.	Reduced limestone utilization enhances SO2 penetration by reducing bisulfite in
the scrubber feed and by effecting additional limestone dissolution in the
scrubber.

Notation



a =

interfacial area of contact (cm2/cm3)

C

Concentration (gmole/1)

Gb =

Bulk concentration (gmole/1)

Q =

Concentration at the solid surface (gmole/1)

Dj =

Diffusivity (cm2/s)

Ej =

Enhancement factor for jth species

f

fraction of SO2 absorbed

G

Gas flow rate (gmol/s)

I

Ionic Strength of solution

KcaCOs =

Dissolution constant for CaCOs

Kcrvl =

Crystallization constant for gypsum

Kcry2 =

Crystallization constant for CaSC>3

kg

Gas side mass transfer coefficient (gmol atiir1 cm-2 s*1)

kf =

Liquid side mass transfer coefficient (cm s*1)

Kpseudo =

Pseudo equilibrium constant

Ksp =

Solubility product

L

Liquid flow rate (1/s)

Mj =

Molar concentration of species in solution

Ng =

Number of gas transfer units (Scrubber)

N] =

Number of liquid transfer units (Hold tank)

p

SO2 penetration

78-12


-------
P	-	Partial Pressure (atm)

RS	=	Relative Saturation

Z	=	liquid holdup (cm of H?0)

Zj	=	Ionic charge of species

Greek	Symbols

A	=	Difference between bulk and interface

h	=	Summation of i species

Yi	=	Activity coefficient

References

1.	Mehta, R. R., Modeling of SO2 Removal and Limestone Utilization in Slurry
Scrubbing Systems with Forced Oxidation, M.S. Thesis, University of Texas,
Austin (1982).

2.	Chan, P. K. and C T. Rochelle, "Modeling of SO2 Removal by Limestone Slurry
Scrubbing: Effects of Chlorides/'" presented at EPA/EPRI Symposium of Flue
Gas Desulfurization, New Orleans (1983).

3.	Gage, C. L., Limestone Dissolution in Modeling of Slurry Scrubbing for Flue
Gas Desulfurization, Ph.D. Dissertation, University of Texas, Austin (1989).

4.	Noblett, J. G., D. P. DeKraker and R. E. Moser, "FGDPRISM, EPRTs Process
Model - Recent Applications/' presented at 1991 SO2 Control Symposium,
Washington, D.C., December 3-6,1991."

5.	Tseng, P. C. and G.T. Rochelle, "Dissolution Rate of Calcium Sulfite
Hemihydrate in Flue Gas Desulfurization Processes/' Envir. Prog., 5, 34-50
(1986).

6.	Lowell, P. S., D. M. Ottmers, K Schwitzgebel, T. I. Strange, and D. W. Deberry,
"A Theoretical Description of the Limestone liijection-Wet Scrubbing Process,"
USEPA, AFID 1287, PB 1931-029 (1970).

7.	Epstein, Mv "EPA Alkali Scrubbing Test Facility ; Summary of Testing through
October 1974," EPA-650/2-75-047, NTIS No. PB-244 901 (1975).

8.	Chan, P. K. and G. T. Rochelle, '"Limestone Dissolution - Effects of pH, CO2, and
Buffers Modeled by Mass Transfer," ACS Symp. Ser., 188, 75-97 (1982).

9.	Plummer, L. N. and E. Busenberg, "The Solubilities of Calcite, Aragonlte, and
Vaterite in CO2-H2O Solutions between 0 and 90° C, and the evaluation of The
Acqueous Model for the System CaC03~C02-H20," Geochim. Cosmochim. Acta.
46,1011-1040 (1982).

10- Chang, J. C. S. and J- H. Dempsey, "PUot Plant Evaluation of By-Product Dibasic
Acids as Buffer Additives for Limestone Hue Gas Desulfurization Systems,"
presented at the EPA/EPRI FGD Symposium, Hollywood, Florida (1982).

11.	Rochelle, G. T., "Comments on absorption of SO2 into aqueous systems/'
Chemical Engineering Science, 47, No. 12, pp. 3169-3171,1992.

12.	Mehta, R. R. and G. T. Rochelle, "Modeling of SO2 Removal and Limestone
Utilization in Sluny Scrubbing Systems with Forced Oxidation," presented at
the AIChE National Meeting, Houston (1983)

13.	Lasdon, L. S. and A. D. Warren, "GRG2 User's Guide," available from,
Department of General Business, School of Business Administration,

University of Texas at Austin, Austin, Texas 78712 (1986).

14.	Dempsey, J. H., J. C. S. Chang, and J. A. Mulholland, "Operation of EPA Owned
Pilot SO2 Scrubber and High Temperature Baghouse," Acurex-RTP Progress
Report-39:EPA Contract 68-02-3648 (March, 1979).

78-13


-------
FLUE GAS

LIMESTONE-—•>

AIR

SCRUBBER

SLURRY

Solids
CaC03,
CaS03
0aS04

Solution
Ca++
S(IV)
S(VI)

co3=

HOLD TANK

-~BLEED

Imp Imh Immt Imm Smt

Figure 1. Simplified process flow diagram of the limestone slurry scrubbing system.

1000 2000 3000 4000 5000
PS02( PPm> at stage inlet

Figure 2. Performance of stages in a three-stage TCA column, Ng = 2 .3/stage.

78-14


-------
Limestone Utilization (%)

Figure 3: Effects of limestone utilization on penetration and pH and hold tank
residence times of 100s and 540s.

60 65 70 75 80 85 90 95 100

Utilization (%)

Figure 4. Effect of limestone utilization with constant inlet pH.

78-15


-------
Figure 5. Scrubber performance as a function of liquid rate with a constant and
variable number of gas phase transfer units (Ng).

Table 1.

Effect of Met SO2 Partial Pressure on Calculated Enhancement Factors

Case I.D.
Stage
L/G (1/gmol)

Utilization
(%)
tres

(hold tank, s)
Ng/ stage
Ysoe (ppm)
Penetration

Low Inlet SO2

High Inlet S02

Base Case

Overall Top Bottom. Overall Top Bottom Overall Top Bottom

0.2
77

540

0.08

2.3

626 1880
0.23 0.61

0.2
77

540

0.33

2700
0.60

2.3
4950
0.77

02
77

540

2.3

1260 2870
.19 .43 .70

(Continued)

78-18


-------
Table 1, (Cont.)

Effect of Met SO2 Partial Pressure on Calculated Enhancement Factors

CaseLD.

Low Inlet SO2

High Inlet SO2



Base Case



Stage

Overall

Top

Bottom

Overall

Top

Bottom

Overall Top Bottom

Bulk



















Compositions



















pH

6.05

5.49

4.57

5.76

4.84

4.09

5.93

5.24

4.37

H+interfaee



0.25

0.48



0.76

0.70



0.49

0.56

(mM)



















HSO3- (mM)

1.04

3.35

11.00

2.30

9.10

25.00

1.50

5.48

16.67

S03= (mM)

1.12

1.00

0.42

1.28

0.61

0.30

1.22

0.91

0.37

HCO3- (mM)

2.28

0.47

0.05

2,48

0.18

0.02

2.41

0.33

0.03

Flux (ppm)



304

371



606

612



426

445

Etotal



12.58

3.38



3.81

2.08



6.51

2.52

Eh+



4.35

1.75



2.00

0.88



2.80

1.27

EsOa=



4.25

0.37



0.39

0.09



1.47

0.21

EhCOs"



2.82

0.06



0.14

,0.01



0.62

0.02

Table 2.

Effect of Change of Limestone Utilization and Hold Tank Residence Times

Case LD.	Low U	High U	Low U, low tres

Stage

Overall

Top

Bottom

Overall

Top

Bottom

Overall

Top

Bottom

L/G (1/gmol)

0.2





0.2





0.2





Utilization

65





95





65





(%)



















tres

540





540





100





(hold tank, s)



















Ng/ stage





2.3





2.3





2.3

YS02 (ppm)



1150

2881



1570

2880



1150

2881

Penetration

0.14

0.36

0.67

0.34

0.62

0.76

0.17

0.43

0.68

Bulk



















Compositions



















pH

6.06

5.41

4.60

5.32

4.65

4.14

5.74

5.25

4.59

(Continued)

78-17


-------
Table 2. (Cent. )

Effect of Change of Limestone Utilization and Hold Tank Residence Times

CaseLD.



Low U





High U



Low

U, low tres

Stage

Overall

Top

Bottom

Overall

Top

Bottom

Overall

Top

Bottom

H+interface



0.42

0.59



0.44

0.46



0.40

0.57

(mM)



















HSOs" (mM)

1.18

4.68

15.00

5.85

10.80

22.60

2.32

5.80

15.60

S03= (mM)

1.24

1.15

0.57

1.15

0.46

0.30

1.23

0.98

0.59

HCO3" (mM)

2.85

0.49

0.06

0.68

0.08

0.01

1.52

0.32

0.06

Flux (ppm)



432

502



338

362



384

490

Etotal



7.95

2.81



3.75

2.3



5.88

2.76

"tH J-

Eh*



3.28

1.35



1.98

1.02



2.88

132

Es03=



2.19

0.32



0.51

0.16



1.73

033

EhCOs"



1

0.04



0.09

0.01



0.53

0.00

Table 3.	Effect of Change of (L/G)

CaseLD.



Low L/G





High L/G



Low L/G, low Ng

Stage

Overall

Top

Bottom

Overall Top

Bottom

Overall

Top

Bottom

L/G (1/gmol)

0.16





024





0.16





Utilization

77





77





77





<%)



















*res

540





540





540





(hold tank, s)



















Ng/ stage





2.3





2.3





1.84

YSO2 (ppm)



1400

2870



1150

2870



1540

2870

Penetration

0.25

0.50

0.74

0.15

0.39

0.68

0.31

0.56

0.74

Bulk



















Compositions



















pH

5.91

5.08

4.25

5.95

5.34

4.47

5.94

5.14

4.35

H+taterface



0-46

0.5



0.47

0.60



0.54

0.55

(mM)



















HSO3" (mM)

1.61

6.76

19.60

1.40

4.57

14.46

1.48

6.27

17.80

S03= (mM)

1.24

0.79

0.34

1.20

0.97

0.41

1.24

0.84

0.38

HCO3- (mM)

2.64

0.24

0.02

2.24

0.42

0.04

2.60

0.29

0.03

Flux (ppm)



396

393



432

483



457

441

Etotal



• 5.18

2.35



7.83

2.78



5.36

2.40

Eh*



2.57

1.10



3.61

1.40



2.78

1.25

Es03=



1.06

0.18



1.83

0-23



1.05

0.21

%CQT



0.38

0.02



0.85

0.03



0.41

0.02

78-18


-------
PREDICTION OF THE PERFORMANCE OF FURNACE SORBENT
INJECTION ON A COAL-FIRED UTILITY BOILER

Gerry H. Newton
1450 N. Dixie Downs Rd. #11
St. George, UT 84770

David K. Moyeda

Energy & Environmental Research Corporation
18 Mason
Irvine, CA 92718

Abstract

The control of sulfur dioxide emissions from coal-fired boilers by injection of
calcium-based sorbents into the upper regions of the furnace is a well-know process
which offers utilities flexibility in achieving compliance with the Clean Air Act. In
general, the furnace sorbent injection process offers 50 percent control of sulfur
dioxide emissions; however, recent advancements in sorbent technology indicate
that control levels of 80 percent may be realizable in practical applications.

Due to the sensitivity of the process to the thermal characteristics of the flue gas,
successful application of the technology to utility boilers requires knowledge of the
flow and temperature fields in the upper furnace and prediction of the impacts of
these parameters on sorbent performance. The design approach described in this
paper utilizes subscale physical flow models and heat transfer models to predict the
furnace flow and temperature characteristics. A model of the sulfation process is
then used to assess and optimize the process.

The approach used to model SO2 capture by the furnace sorbent injection process
and its use in the design of the furnace sorbent injection system for a 71 MWe coal-
fired utility boiler are described. Predictions performed during the design of the
furnace sorbent injection system matched data measured when the full scale system
was put on-line. Model predictions and data for both normal and promoted
hydrated limes are compared as a function of calcium-to-sulfur ratio and load.

79-1


-------
Introduction

Furnace Sorbent Injection (FSI) involves injecting calcium-based sorbents such as
hydrated lime, limestone, or dolomite into the upper regions of the boiler furnace
where the sorbents react with SO2 in the flue gas to form calcium sulfate (CaSC>4).
The significant amount of small- and full-scale research into this process has
shown that optimal sulfur removal is influenced by the temperature at the point of
injection, the rate of temperature decay, or quench rate, and the mixing process
associated with injection of the sorbent. Therefore, the design of FSI systems
requires the use of several tools which include heat transfer models, physical flow
models, and a model of the sulfation process. Heat transfer models provide
temperature profiles within the furnace while physical flow models characterize the
flow patterns and allow the impacts of injector configurations to be determined.

In evaluating the FSI process, a sulfation model can serve three primary purposes.
First, it can aid in the selection of injection locations. The optimum injection
temperature is typically considered to be 2300°F. Varying quench rates and residence
times within the furnace may shift the optimum injection temperature.
Temperatures at a particular location will vary as load is varied. In addition,
physical restrictions may prevent the installation of sorbent injectors at the
optimum injection temperature. A sulfation model will allow the influence of
these parameters on SO2 capture to be determined when selecting the elevation and
location of the injectors. Second, a sulfation model can be used to determine the
expected level of SO2 capture in the furnace. This allows the applicability and cost
effectiveness of the FSI process to be determined for a particular boiler. Finally, a
sulfation model can be used for data interpretation and system optimization
following installation of an FSI system.

Energy and Environmental Research Corporation (EER) designed and installed an
FSI system on a 71 MWe tangentially fired utility boiler firing a high-sulfur Illinois
coal. Between 1991 to 1992, a series of sorbent injection tests with a commercial
hydrated lime and a promoted hydrated lime were completed, hi this paper, the
approach used to model SCb capture by normal and promoted sorbents is discussed,
and predicted levels of SO2 reduction are compared to measured levels.

FSI System Design Methodology

The complete design of a sorbent injection system consists of three primary steps:

*	Field measurements of furnace gas temperatures and flow fields
in the injection region.

•	Design of the injection system. This involves selection of the
injection elevation and the design of the sorbent injectors.

79-2


-------
• Detailed engineering design of the injection equipment and
retrofit modifications.

The design of the injection system requires detailed specification of the sorbent
injectors, including number of injectors, size, injection velocity, and locations. To
specify injector characteristics needed to achieve optimal sulfur capture, EER has
developed a generalized design methodology as a result of a number of previous
projects1. This methodology involves the application of various experimental and
analytical tools to adapt the FSI process requirements to the specific boiler geometry
and operating parameters.

The overall design methodology is shown in Figure 1. As indicated in this figure,
the design of an injection system consists of three key components: isothermal flow
modeling to specify the furnace flow field and injector characteristics; heat transfer
modeling to specify the upper furnace thermal characteristics and to assess
performance impacts; and sulfation modeling to develop performance predictions.

isothermal Flow Modeling

A reduced scale acrylic model of the boiler is constructed based upon design
drawings and current operational data. Measurements are made in the model to
fully characterize the flow field for use by the heat transfer model and to determine
whether any unusual features exist which would impact the FSI system design.
Model data are compared to field velocity measurements to ensure that a reasonable
representation of the boiler has been achieved.

The appropriate injection elevations are selected based on the temperature profiles
calculated by the heat transfer model and predictions of the sulfation model based
on these temperature profiles. Preliminary specifications for the number of
injectors, their location, size and velocity are estimated based on empirical
correlations and computer models of jet mixing. The injectors are installed in the
isothermal model and dispersion and penetration measurements are performed
using various flow visualization and tracer measurement techniques. The injector
design is then optimized based on these measurements to ensure rapid and
complete mixing of the sorbent.

Heat Transfer Modeling

Computational heat transfer models are applied to the specific boiler geometry and
operating conditions and the flow field from the isothermal model. Details of the
heat transfer models may be found elsewhere2. Model predictions of furnace
temperatures and steam side performance are compared against field measurement
data to verify operation for baseline conditions. The model is then used to provide
temperature profiles for use in the design of the FSI system and to determine the

79-3


-------
impact of the FSI system oil boiler performance over the desired operating range for
the sorbent Injection system.

Sulfation Modeling

The model used to predict SO2 capture is a distributed pore model based on the
description of a porous calcine particle given by Christman and Edgar3. Sulfation on
both the internal pore surfaces and on the external surfaces is considered. The
model considers diffusion of SO2 to the particle and through the pore structure,
diffusion through the product layer on both the particle surface and in the pore
structure, and the reaction at the CaO/CaSC>4 interface. The model considers a
distribution of pore sizes and particle sizes. The influence of oxygen concentration is
considered, but is typically important only at concentrations much lower than found
in a typical utility boiler furnace.

Hie in—situ porosity of a sorbent has been shown to control the extent of SO2
capture4. The porosity developed when a sorbent is injected into a sulfation
environment is shown in Figure 2. When sorbent is injected into a furnace, the
finite time required for the sorbent injector jet to mix with the furnace gases and the
quench rate within the upper regions of the furnace dictate that the sorbent will
mix, calcine, and sulfate oyer a range of temperatures. The model calculates an
average porosity based on the jet mixing time and the input temperature profile.
This approach for calculating an average porosity is illustrated in Figure 2 for the
case where the jet mixing time and the temperature profile dictate that the sorbent is
mixed for the range of temperatures from 2300°F to 2100°F.

The maximum porosity which a sorbent will develop at the optimum injection
temperature will vary with the particular type of sorbent4. The porosities illustrated
in the curve in Figure 2 were determined by measuring the porosity of the sulfated
sorbent and calculating the in—situ porosity using the measured level sorbent
utilization (based on gas phase measurements). In practice, this is time consuming
and expensive. The maximum porosity for a particular sorbent is therefore
determined by measuring the extent of SO2 capture in a well characterized
experimental reactor and back-calculating the maximum porosity using the
sulfation model.

When temperatures and gas velocities are relatively uniform across the width of a
furnace, a single temperature profile may be adequate. When these parameters are
not uniform across the width of a furnace, a more complex set of sulfation model
predictions are required. This involves performing individual sulfation calculations
which consider the different time-temperature profiles for each of the sorbent jets.
This approach was used to successfully match data taken during testing of the FSI
process at on a 60 MWe utility boiler5. Further refinements in the sulfation model
predictions are possible if dispersion data are available, from the isothermal flow

79-4


-------
model, which indicate the degree of dispersion achieved with a particular sorbent
injection scheme.

In the design process, predictions of SO2 capture are made during various stages.
Initially, before the field measurements have been made, predictions of SO2 capture
are performed based on whatever data are available from the boiler operator
(usually dimensions and temperatures as a function of load). These predictions
provide a rough indication of the level of SO2 capture that may be expected for the
particular furnace. Once the field measurements have been made and temperature
profiles have been calculated by the heat transfer model, the sulfation model is
utilized to provide predictions as a function of elevation. These predictions are then
used in selecting the appropriate injection elevation. A final set of predictions are
performed as a function of various operating parameters when the design of the
sorbent injectors is complete.

Full-Scale Injection System Design

Energy and Environmental Research Corporation (EER) designed and tested a
furnace sorbent injection system on Hennepin Station Unit 1 as part of a
demonstration program under Round 1 of the DOE Clean Coal Technology
Program6. Hennepin Unit 1 is a 71 MWe tangentially fired boiler owned and
operated by Illinois Power. Unit 1 is a balanced draft unit supplied by Combustion
Engineering and is fired with a 3 percent sulfur Illinois bituminous coal. Each of the
four burners contains three levels of coal nozzles. Steam temperature is
accomplished by tilting the coal nozzles up or down and/or by using the spray
attemperators mounted in both the superheater and reheater. A gas reburning
process was installed at the same time the FSI system was installed to reduce NOx
emissions.

Initially, a subscale flow model was constructed to study the overall furnace flow
patterns. An illustration of the isothermal flow model is shown in Figure 3.

Based upon the measured flow field characteristics and boiler design data, mean
furnace time-temperature profiles were calculated using a two-dimensional heat
transfer model. Figure 4 shows the temperature profiles predicted for loads of 45 and
70 MWe. These profiles was used to predict the impact of injection elevation on SO2
capture, as illustrated in Figure 5. These results indicate that the optimum injection
location for both full and low-loads is near the elevation of the upper furnace arch,
or boiler nose.

Based upon the results of the preliminary sulfation model predictions, the
isothermal flow model was used to study various injection systems. In total, six
sorbent injection configurations were installed on the isothermal model. These
injection configurations included injectors on the front and side walls. To evaluate
sorbent mixing, tracers were injected though the simulated sorbent injectors and
dispersion measurements were performed at the vertical plane at the entrance to

79-5


-------
the convective sections. Measurements taken in the isothermal model indicated
that injection schemes which utilized jets on the front wall and jets on each of the
side walls yielded the best dispersion of the sorbent. Dispersion data for an injection
scheme which utilized four jets on the front wall and one jet on each of the side
waUs are shown in Figure 6.

To assess the performance of two final injection schemes developed with title
isothermal flow model, the sulfation model was used to develop predictions of SCb
capture as a function of calcium-to-sulfur ratio at the nose elevation as shown in
Figure 7. The dispersion measurements were then used to calculate localized
sorbent concentrations across the furnace and local levels of SO2 capture were
determined from Figure 7. These local levels of SO2 capture were normalized by the
local velocity to obtain an overall level of SO2 capture as described in a previous
paper7. Predictions of SO2 capture are also shown in Figure 7 for the two injection
configurations which provided the greatest levels of sorbent dispersion. The
injection scheme which utilized four jets on the front wall and two side jets (one jet
on each of the side walls) provided the greatest level of predicted SO2 capture. This
injection scheme was installed on Hennepin Unit 1.

Results

Comparisons of sulfation model predictions and data from the frill scale testing of
the FSI process are presented for two types of sorbents: a Linwood hydrated lime and
a PromiSOx™ sorbent. These two sorbents were tested during two separate series of
tests. The Linwood hydrated lime was tested during August and November-
December, 1991, while the PromiSOx™ sorbent was tested in late 1992s. Testing was
performed with the gas reburning system in operation.

The sorbent feed rate required to achieve a given level of SO2 reduction determines
the cost effectiveness of the FSI process. Sulfation model predictions for 45 and 70
MWe loads are compared to field results for the Linwood hydrated lime in Figure 8
as a function of calcium-to-sulfur (Ca/S) ratio. The predictions illustrated in Figure
8 were performed assuming that mixing of the sorbent in the furnace was perfect
This assumption is reasonable since the levels of SO2 capture achieved with the
Linwood hydrated lime were not great enough so that imperfect mixing would be
expected to have a significant effect on the final level of SO2 capture. Overall, the
close correspondence between the predictions and field results verifies the design
methodology and indicate that the sulfation model can be used as design tool for
full-scale applications of the FSI process.

Predictions for the PromiSOx™ sorbent are compared to full scale data taken at 45
MWe in Figure 9. Predictions which assume perfect mixing over predict the
measured levels of SO2 capture at the highest calcium-to-sulfur ratios (-2.5). The
solid line in Figure 9, which represents predictions which consider the effects of
sorbent dispersion based upon the dispersion data from Figure 6, more closely

79-6


-------
matches the data taken at the higher calcium-to-sulfur ratios. These results suggest
that the reactivity of the PromiSOx™ sorbent is great enough so that localized
regions of the furnace are depleted of SO2 and the overall level of SO2 capture
decreases when the calcium-to-sulfur ratios are increased beyond a value of
approximately two.

Sulfation model predictions are compared to levels of SO2 capture by the Linwood
hydrated lime as a function of load in Figure 10. The model indicates that a slight
dependence on load should occur. Hie scatter of the full scale data do not allow this
to be observed. Predictions of SO2 capture as a function of load are illustrated in
Figure 11 for the PromiSOx™ sorbent. The greater utilization of the PromiSOx™
sorbent mate the dependence on load more apparent.

Summary

Sulfation modeling plays a key role in the design of Furnace Sorbent Injection (FSI)
systems for full-scale coal-fired utility boilers. Hie sulfation model serves several
key purposes: 1) to aid in the selection of injection locations by allowing the
influence of quench rates and temperature profiles on SO2 capture to be calculated;
2) to determine the expected level of SO2 capture; and 3) to provide data
interpretation and system optimization. Sulfation modeling,, when coupled with
the overall design methodology described in this paper, serves as a useful design
tool for the application of the FSI process.

Predictions of SO2 capture by the sulfation model were compared to full scale data
taken for FSI application to a 71 MWe utility boiler for a Linwood hydrated lime and
a PromiSOx™ sorbent. Predictions of SO2 capture matched data taken with the
Linwood hydrated lime as a function of calcium-to-sulfur (Ca/S) ratio. The SO2
capture data taken with the PromiSOx™ sorbent were matched by the sulfation
model except at the higher calcium-to-sulfur ratios where it was necessary to
consider mixing. Predictions by the sulfation model indicate that there should be a
dependence on load. The scatter in the data taken with the Linwood hydrated lime
did not allow this to be observed. Utilizations of the PromiSOx™ sorbent, however,
exhibited this dependence.

Acknowledgments

The authors wish to thank Tony Marquez and Quang Nguyen for their aid in
preparing this paper and to acknowledge Intevep SA, the U.S. Department of Energy
(DOE), the Gas Research Institute (GRI), and the State of Illinois Department of
Energy and Natural Resources (ENR) who were sponsors of the demonstration
programs used to obtain the full-scale data.

PromiSOx™ is a trademark of Intevep SA (an affiliate of Petroleos de Venezuela).

79-7


-------
References

1.	Payne, R., B. A. Folsom, W. Bartok and D. K Moyeda, " Demonstration of Gas
Rebuming/Sorbent Injection N0X/S02 Control Technology on Three Utility
Boilers," Paper presented at the AIChE Summer National Meeting, Denver,
Colorado (August 1988).

2.	Payne, Rv K. T. Wu and D. K. Moyeda, "Three-Dimensional Model
Evaluation of Time/Temperature Histories of Combustion Products in a Coal
Fired Utility Boiler/' Paper presented at the Fall Meeting of the CEA Thermal
and Nuclear Section, Saskatoon, Saskatchewan (October 1988).

3.	Christman, P. G. and T. F. Edgar, "Distributed Pore-Size Model for Sulfation
of Limestone," AIChE J., 29,388 (1983).

4.	Newton, G. H., S. L. Chen and J. C. KramHch, "The Role of Porosity Loss in
Limiting SO2 Capture by Calcium Based Sorbents," AIChE J., 35, 6,988 (1989).

5.	England, G. C, D. K. Moyeda, R. Payne, B. A. Folsom, B. Toole-OTMeil, D. G.
Lachapelle and I. A. Huffman, "Prototype Evaluation of Sorbent Injection on
a Tangentially Fired Utility Boiler/' Presented at the 1990 SO2 Control
Symposium, New Orleans, Louisiana (May 1990).

6.	Moyeda, D. K., B. A Folsom, T. M. Sommer and Q. Nguyen, "Demonstration
of Combined NOx and SO2 Emission Control Technologies Involving Gas
Rebuming/' Paper presented at the 1991 AIChE Annual Meeting, Pittsburgh,
Pennsylvania (August 1991).

7.	Cetegen, B. Mv T. R Johnson, D. K. Moyeda and R. Payne, "Influent of
Sorbent Injection Aerodynamics on SO2 and Simultaneous S02/N0X
Control," Presented at the EPA/EPRI Joint Symposium on Dry SO2 and
Simultaneous SC^/NOx Control Technologies, Raleigh, North Carolina 0une
1986).

8.	Rodriguez, D., C. Gomez, R Payne and P. Maly, "Demonstration of Promoted
Dry Injection in a 75 MWe Tangentially Fired Utility Furnace," Presented at
the EPA/EPRI SO2 Control Symposium, Boston, Massachusetts (August 1993).

79-8


-------
Figure 1. Design methodology for furnace sorbent injection.

Figure 2. In-situ porosity of hydrated lime in a sulfation environment4.
Shaded area represents range of porosities which develop as
the finite mixing time results in the sorbent being exposed to
a range of calcination temperatures.

79-9


-------
Simulated
Burner

Conveetive

Pass

Sections

Hopper

Figure 3. Sketch of 1:12 scale model of Hennepin Unit 1.

79-10

A


-------
3000

2500

2000

1500



I
1

Hennepin Unit #1

1 7QMW

Nose

•

A -

45 MW

V 1

\!



-

i
I
1
1

V\

.... i .

1
1

1

..I....

		 • A, i . . .:

-1.5 -i.o -o.5 ao 0.5
Time (s)

1.0 1.5 2.0

Figure 4. Thermal profile predicted by the 2-D heat transfer model.

i •111 • • • i1 • • i
LInwood Hydrated Lime |

60

50

S 40

fr

u

g 30

20

10

45 MW

¦ • i ... i	it¦•

15 17 19 21 23 25 27 29 31 33 35
Elevation (m)

Figure 5. Predicted levels of S02 capture as a function of injection location.

79-11


-------
Figure 6a. Optimal sorbent injection configuration.

Std. Dev. = 0.52

Figure 6b. Sorbent dispersion profile.


-------
Ca/S Molar Ratio

Figure 7. Effect of sorbent injector scheme on SOz capture.

79-13


-------
<0

70
60
SO
40
30
20
10

I ' 1 ' ' I 1 1 !¦''¦!
45 MW

70 MW

Boiler Load:
A 45 MW A 60 MW

~	50MW ~ G5MW

•	SIIW O 70 MW
O 58MW V75MW

(M) OS 1.0 1.5 ZJ0
Ca/S Molar Ratio

2.5

3J0

100

80

S 60

3
a
a
0
«

O

5 40

20

1 1 » 1

t	T	*	¦"'"i""'	*		¦	*		r	'	'	

Perfect Affixingyr ,



/ yr Aspersion

"

/ / Considered

r. , . . i . . - .

i .... i .... i i ... .

OS

as

1.0 1.5 2.0 2J5 3.0
CVS

Figure 8. Sulfur removal achieved
with Linwood hydrated
lime.

Figure 9. Sulfur removal achieved with
PromiSOx™ sorbent.

40	50	eo

Load(MW)

70	80

SO

Load (MW)

Figure 10. Effect of load on calcium
utilization for Linwood
hydrated lime.

Figure 11. Effect of load on calcium
utilization for PromiSOx™
sorbent.

79-14


-------
SCRUBBER RENEWAL PROJECT AT TEXAS MUNICIPAL POWER AGENCY

Carl V. Weilert
Steve M. Ashton
Fred W. Campbell
Richard D. Norton
Burns & McDonnell Engineering Company
4800 E. 63rd Street
Kansas City, Missouri 64130

Ray Hamilton
Texas Municipal Power Agency
P.O. Box 7000
Bryan, Texas 77805

Abstract

Texas Municipal Power Agency owns and operates title wet limestone FGD (scrubber)
system at the 443-MW Gibbons Creek Station. The scrubber system started up in
1982. In late 1990, TMPA developed a 5-year plan for renewal of the scrubber to
improve reliability and reduce operation and maintenance costs. Upgrades from the
original materials of construction are an important component of the plan. To date,
implementation of the scrubber renewal plan has included the following:

•	Retrofit of reinforced concrete covers to the reaction tanks for three absorbers.

•	Replacement of the carbon steel tower hood and outlet dampers on one absorber
with new segmented tower hood and dampers fabricated of 317LM stainless steel.

•	Replacement of the carbon steel finned-tube flue gas reheater in one absorber with
new modular-design reheaters of 317LM stainless steel construction.

This paper describes aspects of the fast-track design and construction of these

upgrades which may be of interest to other FGD system owners contemplating
similar improvements.

Introduction

Background

The Gibbons Creek Steam Electric Station, located near Carlos, Texas, consists of a
single lignite-fired unit rated at 443 MW. The fuel for the unit is supplied from a
captive mine located adjacent to the plant site. Air pollution control equipment

80-1


-------
serving the plant includes an electrostatic precipitator and a wet limestone FGD
(scrubber) system.

As shown in Figure 1, the scrubber has three spray tower SOz absorbers. Each
absorber tower has five levels of sprays which provide counter-current contact
between the flue gas and the recirculating slurry. The recirculating slurry drains
from each absorber tower to a pair of reaction tanks which provide for limestone
dissolution and solids precipitation.

Each scrubber tower is provided with an in-line steam coil reheater to raise the
temperature of the outlet flue gas to a point safely above the dew point Partial
reheat of the flue gas is also provided by flow through the system bypass damper,
which is automatically modulated to control the stack S02 emission level at or below
the allowable limit of 12 pounds per million Btu.

Need for Scrubber Renewal Project

The scrubber system at Gibbons Creek was started up in 1982. As TMPA gained
experience over the years in operating and maintaining the system, the primary
emphasis was placed on maintaining those system components directly related to S02
removal efficiency performance, such as recycle pumps, spray headers and nozzles
and process instrumentation.

An assessment of the condition of the scrubber system conducted by TMPA in late
1990 indicated that, to protect the investment in the scrubber system and to maximize
its useful life, it would be necessary to address, in a comprehensive fashion, the
maintenance needs of those system components not directly related to performance.
Key needs identified during this assessment included:

» The repair of cumulative effects of corrosion, both inside and outside the flue gas
flow path. Components affected included ductwork, dampers, support steel,
walkways and handrail, and reaction tanks.

•	The upgrade of original materials of construction as necessary to prevent
recurrence of the corrosion problems.

•	The redesign of the reheaters to improve maintainability.

Staged Implementation Plan

The Gibbons Creek Station is the only base-load generating capacity owned by
TMPA. Consequently, it was not economically feasible for TMPA to schedule an
extended outage to simultaneously address all the recommended components of the
scrubber renovation. Instead, a plan to implement the changes gradually over a 5-
year period was developed and approved. The staging of the scrubber renovation
project allowed that work which required a unit outage to be scheduled in

80-2


-------
accordance with TMPA's long range outage planning. In addition, it allowed TMPA
to spread the cost of the renovation over a longer period to optimize cash flow.

The staged implementation of the scrubber renovation project required that work be
planned for compatibility with two-week unit outages. TMPA typically extends the
time available to perform work on a given scrubber tower by isolating one tower for
internal maintenance for one week prior to and one week following a unit outage.
However, since the work recommended included replacement of the outlet isolation
dampers for each tower, it was necessary to insure that removal and replacement of
the dampers could be accomplished within a two week period.

Implementation of the five-year scrubber renovation program at TMPA's Gibbons
Creek Station began in the fall of 1991. The next three sections of this paper describe
the key work components which have been accomplished to date.

Reaction Tank Cover Retrofit

As shown in Figure 1, each absorber tower at Gibbons Creek is served by a primary
reaction tank and a secondary reaction tank. The primary reaction tanks are each 58

feet in diameter and 30'-6" high. The secondary reaction tanks are each 29 feet in
diameter and 3G'-6" high.

Purpose

In August 1991, Burns & McDonnell inspected and evaluated the condition of the
absorber support steel at TMPA's Gibbons Creek Station. The original protective
coating system had failed in several areas and the steel was beginning to corrode.
The worst areas were above the sulfur dioxide absorber reaction tanks. Fumes from
the open-top reaction tanks had been attacking the steel near the surface of the tanks,
including walkways and handrails necessary for safe access to the scrubber system.
In order to prevent further corrosion, Burns & McDonnell recommended that covers
be installed on the reaction tanks, and that the steel which had been attacked by
corrosion be cleaned and coated with an epoxy coating system.

In November 1991, TMPA hired Burns & McDonnell to design retrofit covers for all
six reaction tanks. Several options had been considered for the material of
construction for the reaction tank covers, including fiberglass and concrete. After
studying the different options and visiting other power plants that have retrofit
covers on absorber reaction tanks, TMPA decided that their covers would be
constructed using reinforced concrete. One disadvantage is that concrete is heavier,
and therefore requires more support. However, concrete covers were chosen because
they are more economical and easier to maintain.

80-3


-------
Unique Design Considerations

Open-top reaction tank designs in utility FGD service are rare today. However, the
over-hung tower design with open-top reaction tanks was used by Combustion
Engineering on a number of FGD systems designed in the late 1970s and installed in
the early 1980s. Designing covers for retrofit to these tanks is a challenge from many
standpoints. At Gibbons Creek, unique design considerations were involved in the
reinforced concrete design, the support steel design, the foundation design, the vent
system, and the drainage system. In addition, access for maintenance and equipment
repair had to be taken into consideration.

Reinforced Concrete Design. Hie reinforced concrete design for the cover slab was
the first issue to address. Due to the characteristics of the sulfur dioxide scrubbing
slurry in the reaction tanks, the reinforced concrete needed a high resistance to
sulfate. This was provided in two ways. First, a type V cement was utilized in the
concrete mix design. Second, a coal tar epoxy coating was specified to be applied to
the underside of the slab after it cured. Due to schedule constraints, the coating
application had to be delayed by approximately six months. The uncoated concrete
held up well during operation in the interim.

As with all concrete, the tank cover slab does have the possibility of developing
cracks. However, it was designed using working stress design so that crack width
could be minimized. For this reason, the number of options for the size and spacing
of reinforcing bars was limited. The final design used an 8-inch thick slab with 3/4"-
diameter reinforcing bars running perpendicular to each other at an 8-inch spacing.
The reinforcing bars had a concrete cover of two inches. As an extra precaution, the
reinforcing bars were epoxy-coated. Slab details are shown on Figure 2.

Finally, due to the constraints of the fast track schedule and the tank orientation, a
concrete mixture needed to be specified to reach 85% of its design strength in seven
days and have a large enough slump so that it could be pumped over 30 feet
vertically. In order to achieve this, Burns & McDonnell specified a high range water
reducing admixture (sup erplas ticizer).

Support Steel. The reaction tank covers were designed to carry their self weight in
addition to a live load of 50 pounds per square foot. This caused an additional load
of 35,000 pounds of live load and 70,000 pounds of concrete on each secondary
reaction tank and 140,000 pounds of live load and 280,000 pounds of concrete on each
primary reaction tank. The only existing steel available to support these loads were
the beams which supported the mixers, platform steel, and the tank walls. This
existing steel was not sufficient to support a load of this magnitude. Therefore, new
support steel had to be added. The new steel is shown on Figure 3. On the
secondary reaction tanks, the only new support steel needed was four columns
welded to the sides of the reaction tanks. These were W12 x 45 columns located at

80-4


-------
the ends of the existing tank agitator support beams. On the primary reaction tanks,
a total of six beams were added on each reaction tank to support the covers: two
W30 x 235 beams running east to west and four W12 x 87 beams running north to
south. In addition, eight columns were added to each primary reaction tank - one at
the interface of each beam at the tank wall. No internal columns were used.

Foundations. The existing foundation for each of the reaction tanks was a 1-foot
wide by 5-foot deep ring wall foundation. This type of foundation did not have the
necessary bearing area to support all of the new load due to the new covers. To
solve this problem, the width of the ring wall was increased as shown on Figure 3 by
doweling into the existing foundation.

Vent System. Two options were considered for venting the reaction tanks. The first
was to release the vapors directly to the atmosphere. The second was to vent directly
into the absorber. From a structural standpoint, venting to the atmosphere was the
preferred choice. Direct venting to the absorber was not originally implemented due
to the concern that a high positive or negative pressure could build up in the reaction
tank beneath the cover slab. If this happened, it might cause serious deformations in
the tank walls or the concrete covers.

The tank covers were initially configured so that the reaction tanks vented directly to
the atmosphere. The main problem with venting the reaction tanks to the
atmosphere is the condensation which can form on nearby surfaces. This problem
becomes more severe as the ambient temperature drops. During the first winter of
operation, the discharge of condensate from the vents became a nuisance, and TMPA
requested that a solution be found. At Bums & McDonnell's direction, a test was
recently performed on the reaction tanks to verify that they could be vented directly
into the absorber towers. The vents were blocked off and the pressure of the vapor
above the liquid level of the reaction tanks was monitored. This test proved that the
reaction tanks would normally generate a positive pressure greater than the operating
pressure of the absorber. This will ensure that there is no badcflow of flue gas
through the vent system. Therefore, one of TMPA's next projects will be to reroute
the vent system into the absorber. Different options for limiting the maximum
pressure to which the tank will be exposed have been proposed by Burns &
McDonnell and are currently being evaluated by TMPA.

Drainage System. Because there is a possibility of sludge accumulating on top of
the reaction tank covers, a drainage system was incorporated into the design. As
shown on Figures 2(a) and 3, drains were provided through the perimeter curb at the
top of the concrete slab at eight locations on each primary reaction tank and four
locations on each secondary reaction tank. The concrete covers slope toward the
drains. Part of TMPA's routine maintenance program is to hose the sludge off the
top of the reaction tanks. Any sludge is flushed through the drainage system to
trenches located near the base of the reaction tanks.

80-5


-------
Penetrations Through the Concrete Cover. Hie design for retrofit of the covers to
the reaction tanks required provisions for multiple penetrations through the concrete
slab. These included a 7'-0" diameter penetration for the main recirculating slurry
downcomer from the absorber tower, as well as many small pipe penetrations. To
accommodate these penetrations, a split pipe sleeve design was used. Each pipe
sleeve was fabricated in two pieces split along the axis of the sleeve. The two halves
were fit around the existing pipe and field welded to form a continuous sleeve.

Then, during the pouring of the concrete tank cover slab, the pipe sleeves ware
incorporated into the slab. After the covers were constructed a zippered casing seal
was placed around each pipe to prevent leakage of vapors from the tank. A typical
pipe penetration is shown on Figure 2(c).

Access for Maintenance and Equipment Repair. Once the covers were in place,
there would be no way in which to remove the tank agitator blades for maintenance.
Therefore, larger access doors had to be put into the bottom of the reaction tanks.
These doors were designed and fabricated for installation in the tanks during the first
unit outage following the construction of the tank covers. Installation of the new
doors was completed on all six tanks during the November 1992 outage.

Erection Procedures for Fast-Track Construction Schedule

The reaction tank covers were constructed without a power plant outage. In order to
construct each pair of tank covers, the respective absorber tower was taken off line.
However, the amount of time that one absorber could be off line was limited to about
two weeks. From the moment the absorber was taken off line, TMPA crews were
working around the dock in order to meet the aggressive schedule.

Erection Sequence. The work sequence used in the construction of the tank covers
for each absorber, is listed below, with additional explanation of procedures. Note
that the foundation work did not require a tower outage and was not part of the fast-
track schedule. Steps 2 through 11 below were typically accomplished in about 10 to
12 days.

1.	Construct the foundations.

2.	Take the absorber off line.

3.	Drain and clean out the reaction tanks.

4.	Erect the support steel. The large main support beams were brought into the
tanks through the existing access door in the tank wall just above grade.

5.	Erect the shoring.

6.	Construct the forms.

7.	Place the reinforcing steel and pipe sleeves.

8.	Place the concrete. Concrete trucks were parked as close to the base of the tanks
as possible. Concrete was pumped though a hose over 30 feet vertically to reach
the top of the tank for placement.

9.	Cure the concrete. A special type concrete was used to reach 85% of its design

80-6


-------
strength within seven days.

10.	Remove the forms and shoring.

11.	Refill the tanks and bring the absorber back on line.

12.	Coat the bottom surface of the concrete. Note this was done during a unit
outage approximately six months after completion of the tank covers.

13.	Install the new larger access doors in the tank walls for future agitator blade
maintenance. This was done at the same time as the coating application.

Tower Hood and Damper Redesign and Replacement

Figure 1 shows the orientation of the absorber towers at Gibbons Creek. The tower
hood and dampers are located directly above the steam coil reheaters. The outlet
dampers are used to isolate the absorber tower for on-line maintenance. The
environment downstream of the isolation dampers on the center "B" tower is a
mixture of unscrubbed (bypass) flue gas and reheated scrubbed gas from the other
two towers. The original material of construction for the absorber outlet tower hood
and double-louver outlet isolation dampers was carbon steel. Years of exposure to
this environment had caused severe deterioration of the dampers due to the effects of
corrosion. Tower hoods also were subject to localized corrosion, especially at cold
spots on the insulation and lagging system. As part of the scrubber renewal project it
was determined that the tower hood and dampers should be replaced.

Purpose

The replacement of the tower hood and dampers was intended to meet the following

requirements:

•	Restore the capability for reliable tower isolation .

•	Upgrade the materials of construction to improve corrosion resistance and extend
useful life.

•	Incorporate improved blade design for more positive sealing, with and without the
use of seal strips.

•	Utilize the existing seal air supply system.

•	Renew the insulation and lagging system.

•	Provide for maintenance access to the reheater steam coils.

Unique Design Considerations

The principal design challenges for the tower hood and damper replacement resulted
from the location of the "B" tower with relation to the overhead ductwork, the design
of the structural support system for the absorber tower below, and the need to
provide for future maintenance access to the reheaters. The B absorber tower is
located in the center of the FGD system, and the bypass duct for the FGD system is

80-7


-------
located directly above the damper. This eliminated any possibility of crane access to
the top of the tower for removal and replacement of the tower hood and dampers,
and restricted the weight of the heaviest component which would be lifted during the
demolition and reconstruction to five tons or less.

Additionally, the location of the main structural supports for the absorber tower
limited the size of components which could be moved to or from the work area. The
rectangular absorber tower is supported from above by eight large hanger rods, two
along each side. Due to the existing equipment layout, access to the work area for
the B tower required an approach from the south. The clearance between the hanger
rods on the south side of the absorber tower is 13'-6".

Finally, the requirement for providing future maintenance access to the reheaters
below meant that it must be possible to easily remove a section of the tower hood
large "enough to allow removal and replacement of a reheater section. The original
tower hood design did not incorporate this feature, and the replacement of reheater
tubes had been done on a single tube basis, requiring cutting an access hole through
the south face of the tower hood for tube removal and replacement, then patching the
hole again when the maintenance work was completed.

The solutions to these design challenges, as utilized in the tower hood and damper
replacement, are described in the following sections.

Segmented Design. Design specifications developed for both the tower hood and
the dampers required these components to be shop fabricated in sections which
would meet the weight and size constraints imposed by the unique requirements of
the B tower location and arrangement as described above.

Each of the two 12'-6" x 16'-6" dampers in the double-louver damper pair were
specified to be fabricated in two sections which could then be bolted together in the
field. As shown on Figure 4, the splice between the two sections of each damper was
designed for field assembly using a bolted, gasketed connection. The overlapping
blade edge design which was specified insured that the two sections could be joined
in the field to form a single functional damper. The damper blade linkage was also
designed to be joined along the splice line. Demonstration of the ability of the
damper assembly to meet these functional requirements was required at the time of
the damper factory test.

The tower hood was redesigned to be shop fabricated in eight trapezoidal flanged
plate sections complete with external stiffeners. As shown on Figure 4, the design
allowed the individual sections to be lifted into position then assembled with no field
welding required. Mating flanges on the adjacent sections permitted field assembly
of the gas tight tower hood using bolting and gasketing. Specially designed brackets
allowed bolted connection of external stiffeners at the corners of the hood to insure
maximum structural integrity. As with the dampers, the specification required a trial
fit up in the fabrication shop to insure that the field assembly would work.

80-8


-------
Two special spool pieces were incorporated into the redesign of the tower hood as
shown on Figure 4. Although they have the same inside duct dimension as the
sectionalized damper (12'-6" x 16'-6") these spool pieces were designed to be
fabricated as a single piece. Design calculations indicated that their weight would
not exceed the maximum lift capability and that their size would allow passage
through the tower hanger rod clearance space. As shown on Figure 4, one of these
spool pieces was for connection to the existing seal air duct to act as the seal air
chamber between the two louver damper assemblies. The other spool piece was
designed to support the temporary internal monorail system to be used during
reheater installation and replacement as shown on Figure 5. These two spool pieces
also were designed for incorporation into the overall tower hood and damper
assembly by bolted gasketed connections requiring no field welding.

Materials of Construction. Based on its prior success using 317L stainless steel for
patching corroded portions of the original carbon steel tower hood, TMPA selected
317LM stainless steel (UNS S31725) as the primary construction material for the tower
hoods and dampers, including the damper frames and blades. Higher grade
materials were used for some damper components, such as blade shafts (Alloy 625)
and blade fasteners and seals (Alloy C276). External stiffeners on the tower hood
plate sections were A36 steel. All gaskets used were EPDM.

Fast-Track Schedule

When originally planned by TMPA, the tower hood and damper replacement was
envisioned for construction during the April 1993 outage. Engineering activities
began in late July 1992 with this schedule in mind. However, an internal inspection
of the B tower during the first week of August revealed that the condition of the
damper was deteriorating, and that it could not be expected to survive until the
spring. TMPA decided that the project schedule must be accelerated to allow
construction to commence on November 1, 1992. The actual schedule was:

Project Schedule - Tower Hood and Dampers

Engineer notified of schedule change
Damper specification out to bid
Damper bids received
Damper contract award

Tower hood drawings and specification out to bid

Tower hood bids received

Tower hood fabrication contract award

Damper factory test

Dampers delivered to site

Outage begins

Tower hood spool sections delivered to site
Damper installation complete

November 1
November 11

September 28

October 5
October 7
October 24
October 28
October 30

August 11

August 20
August 26
August 28

80-9


-------
Tower hood plate sections shop trial fit up
Tower hood plate sections delivered to site
Tower hood installation complete

November 12
November 15
November 18

Erection Procedures

Due to lack of crane access to the work area on the B tower, TMPA devised a
monorail and hoist system to move material and equipment to and from the work
area. The monorail beam was hung from the support steel for the duct to stack
which extends from the B tower outlet to the chimney. On the south end, the
monorail allowed equipment and material to be lifted from or lowered to grade. A
trolley equipped with a 5-ton capacity air-operated hoist was used to move
equipment and material back and forth on the monorail. Some bracing on the steel
support structure for the duct to stack was temporarily removed to allow the
movement of equipment. On the north end the monorail extended between the two
main support hanger rods on the south side of the B tower. Once the equipment had
cleared the passage between the hanger rods it was positioned and moved into place
by the use of chain hoists and cables secured to the overhead support steel.

The FGD system tower hood and outlet damper replacement for the "B" tower at
Gibbons Creek was successfully completed within a period of three months following
the start of design. Despite the accelerated design, bid, award, fabrication, and
delivery schedules, this portion of the project was completed on time and within
budget. Equipment contractors were ACDC, Inc. for the dampers and Sterling Boiler
and Mechanical for the tower hood fabrication. All demolition and erection field
labor and supervision were performed by TMPA's Construction Department.

Reheater Redesign and Replacement

Each absorber tower in the FGD system at Gibbons Creek was originally equipped
with a finned-tube steam coil flue gas reheater designed to raise the temperature of
flue gas leaving the absorber approximately 50°F. The material of construction of the
reheaters was carbon steel. In recent years, TMPA has experienced increasing
maintenance problems with the reheaters, including tube leaks caused by corrosion.
Due to the all-welded construction of the steam supply and condensate collection
headers located just inside the tower wall, maintenance of the reheaters was difficult
Replacement of damaged or corroded tubes required cutting the damaged tube away
from the header and welding in a new tube, with all work taking place inside the
absorber tower. Several major tube replacement programs had been undertaken, but
as the condition of the reheater deteriorated it was frequently necessary to blank off
leaking tubes at the header. As described earlier, these tube replacements also

Results

80-10


-------
required cutting access openings in the tower hood. TMPA determined that the
reheaters should be redesigned and replaced as part of the scrubber renovation
project for the "B" tower.

Purpose

The redesign and replacement of the steam coil reheaters was intended to meet the
following requirements:

•	Provide a modular design which could be more easily removed and replaced.
Modules to be interchangeable with one another.

•	Upgrade the materials of construction to improve corrosion resistance and extend
expected life.

•	Reuse the existing structural supports inside the absorber tower.

•	Reuse the existing steam sootblowers.

Design Considerations

Modularizaton. As was the case with the tower hood and dampers, the maximum
allowable lifting load played a role in the extent of modularization. A specification
limit of 9,000 pounds was placed on the weight of each module. This resulted in a
six-module arrangement being selected. The layout of the original and modified
reheater designs is compared on Figure 6.

Space Limitations. Because the redesigned reheaters were required to use the same
support structure and the same sootblowers as the existing coils, the extent of
modification possible to the original design was limited. For example, it was
intended that the reheater tubes be sloped to improve condensate drainage. The
original design had no slope on any of the four tube passes, and this was suspected
as one possible cause of the corrosion of the tubes. However, due to the space
limitation, the design intent was compromised to provide a sloped tube on the
bottom pass only. This tube is farthest from the steam supply inlet and most likely
to contain significant condensate flow.

Materials of Construction. With die upgrade of the tower hood and outlet damper
from carbon steel to 317LM stainless steel construction, the reheaters were the last
remaining component in the absorber tower assembly to have been fabricated out of
carbon steel. Based on the corrosion problems TMPA had experienced with carbon
steel reheaters in the past, it was decided that the new modularized reheaters would
be specified to be fabricated from 317LM stainless steel (UNS S31725) tubes and fins.
To provide a comparison with regard to schedule, cost, and predicted performance,
the bid specifications also required a complete bid for a reheater fabricated of carbon
steel tubes and fins.

80-11


-------
The effect on reheater performance of the change to 317LM stainless steel from
carbon steel is significant. The lower thermal conductivity of 317LM stainless steel
compared to carbon steel means that the two reheaters of nearly identical design
except for the material of construction will have notably different performance. In
this case, the reheated flue gas temperature at design conditions was predicted to be
186°F for the 317LM stainless steel option compared to 200°F for the same reheater
design fabricated of carbon steel. However, since the 186°F outlet temperature value
represented a 40°F increase across the reheater, the performance for the stainless steel
design was deemed acceptable.

Maintenance and Access. With the reheater modules weighing in excess of 8,000
pounds, the ability to place and remove modules within the reheater section of the
tower was critical. Burns & McDonnell designed an internal monorail system to be
placed in the duct spool section just below the lower outlet damper. When
placement or removal of a reheater module is required, the south face of the tower
hood is unbolted and moved away. Then, as shown on Figure 5, two lateral
monorail beams are bolted into the spool piece at the top of the tower hood. An
axial monorail beam is then attached, at each aid, to trolleys running on the two
lateral monorail beams. Two more trolleys on the axial beam are used to move
modules in or out of the tower hood. This design allows the monorail system to act
as a bridge crane which can be positioned as required to place or remove any of the
six reheater modules.

Piping Modifications. As shown on Figure 6, the original reheater design featured
single penetrations through the absorber tower wall for the steam supply and
condensate return piping. To make the steam supply and condensate piping systems
compatible with the change to a modular design, and to insure that the modules
would be interchangeable, modifications to the piping were required. The steam
distribution and condensate collection headers were redesigned and relocated to the
outside of the absorber tower, and individual steam supply and condensate return
piping penetrations through the tower wall were provided for each of the six reheater
modules. The piping connections to each reheater module were flanged so that
modules could be easily disconnected for removal as required.

Schedule

As with the tower hood and dampers, the reheaters were originally planned for
installation during an April 1993 outage, based on engineering starting in late July
1992. When the decision was made to accelerate the project schedule for the tower
hood and damper installation, TMPA recognized that a parallel delivery schedule for
the reheaters would not be possible. However, it was hoped that the reheaters would
be available soon after the end of the planned outage. Installation of the new
dampers would allow the reheater installation to proceed with the unit on line by
closing the absorber outlet isolation damper.

80-12


-------
The engineering for the reheater redesign proceeded coincident with that for the
tower hoods and dampers, attempting to achieve a similar fast-track schedule.
However, due to fabrication problems and shop scheduling conflicts, the installation
was delayed by five months.

Project Schedule - Reheater Replacement

Engineer notified of schedule change

August 11,

1992

Specification out to bid

August 28,

1992

Bids received

September 9,

1992

Contract award

September 14,

1992

Reheaters delivered to site





Contracted

January 18,

1993

Actual

June 1,

1993

Reheater installation





Planned

January

1993

Actual

June

1993

Fabrication Problems

The principal problem encountered during fabrication of the reheaters was in the
tube finning operation. The specification required that fins be continuously resistance
welded to the tubes. The reheater supplier's design was based on a combination of
1.75-inch O.D. tubes and 0.75-inch tall fins, with a fin spacing of three per inch. This
fin height and tube diameter, combined with the mechanical characteristics of the
317LM material, proved to be a difficult application for the tube finning machinery to
handle. Production rates were initially very slow as the proper machine settings
were determined by trial and error. Fin breakage during resistance welding
operations led to arc-gouging of the tubes, causing approximately one-fourth of the
tubes to require weld repairs. The excessive fin breakage caused wastage of 317LM
fin material, requiring substitution of more readily available 316L fin material on
seven percent of the tubes when the supply of 317LM material was exhausted.

The tube finning problems resulted in a six-week delay in delivery of the finned
tubes to the reheater fabrication shop. This caused the available "window" in
production scheduling at the fabricator's shop to be missed. Fabrication and
shipment of the finished reheaters was consequently delayed 19 weeks past the
original contract schedule. This represented almost exactly twice the originally
anticipated period from contract award to reheater delivery.

Installation Procedures

The demolition of the existing reheaters and the installation of the new reheater
modules was accomplished by moving equipment to and from the work area using

80-13


-------
the combination of the previously described monorail and hoist systems exterior and
interior to the tower hood. This work was performed with the unit on line, and with
the B tower isolated using the new damper. TMPA crews removed the south face of
the tower hood, demolished and removed the old reheaters, installed fee new
reheaters, piping and supports and replaced the tower hood panels. By working
around the clock, they were able to complete the reheater replacement in less than
five days.

Results

The engineering for redesign to a modular reheater system was completed on a fast-
track schedule coincident with that for the tower hood and dampers. However, tube
fabrication problems and production schedule delays caused the reheater portion of
fee scrubber renewal project to require approximately twice the original schedule.
Equipment supplier for the reheaters was Aerofin Corporation. The finned tubes
were supplied to Aerofin by ESCOA.

Work Remaining

Current plans are for the installation of the tower hood, damper, and reheater
replacements for both the other two towers to take place in the spring of 1994.
TMPA believes that the experience gained in the work on "B" tower, along with fee
relatively easier access to fee two outboard towers, will make it possible to complete
all this work within the time allowed by fee outage planned for April 1994.

Conclusions

With regard to fee reaction tank covers, this project has demonstrated the suitability
of reinforced concrete as an economical design option for the retrofit of covers to
FGD absorber reaction tanks. The use of specialized concrete mix design and fast-
track scheduling during erection has proven the feasibility of tank cover retrofit in a
compressed schedule without requiring the FGD system to be taken off line.

Wife regard to fee tower hood, dampers and reheaters, this project has proven the
value of modularized prefabricated construction as a means of enabling a fast-track
scheduling of scrubber tower renovations. Proper planning during the design stage,
as well as shop fit-up demonstrations for verification of critical connections of
fabricated components requiring field assembly, were especially beneficial in this
case.

80-14


-------
r

ABSORBER

Figure 1. Scrubber System Arrangement at Gibbons Creek Station

80-15


-------
a. Connection at Tank Wall

b. Section Through Support Beam

c. Pipe Sleeve Penetration
Figure 2. Details of Reaction Tank Cover Concrete Slab

80-16


-------
Figure 3. Support Steel and Foundations for Reaction Tank Cover

80-17


-------
16-6"

£



DAMPER



SPLICE LINE (TYP>

r

CONNECTION FOR
SEAL AIR DUCT*

? ~



ACCESS DOOR



40-0"

UPPER LOUVER
DAMPER

SEAL AIR CHAMBER
SPOOL PIECE

LOWER LOUVER DAMPER
MONORAIL ATTACHMENT
SPOOL PIECE

UPPER HOOD
SECTION

LOWER HOOD
SECTION

SIDE VIEW LOOKING WEST	END VIEW LOOKING SOUTH

~ INDICATES LOCATION OF
BOLTED, GASKETED FIELD
CONNECTION

Figure 4, Sectionalized Design of Tower Hood and Dampers


-------
TOWER HOOD

EXPANSION
JOINT

TOWER

1			

TROLLEY



REMOVABLE
LIFTING BAR

BEAMS

I

vr

LIFTING
BEAM



rererei reipg.

Reference: Figure 4, Section "A - A"

REHEATER
MODULE

Figure 5. Use of Temporary Monorail System for Reheater Module Placement

80-19


-------
a. Original Design Plan View

b. New Modular Design Plan View

Figure 6. Comparison of Flue Gas Reheater and Piping Arrangements

80-20


-------
THE EFFECTS OF INCREASING SOz REMOVAL ON NEAR-PLANT OPACITY

OF COAL FIRED UTILITY BOILERS

C.V. Weilert
P.N. Dyer

Bums & McDonnell Engineering Company
4800 East 63rd Street
Kansas City, MO 64130
(816) 333-4375

Abstract

One problem which is common to many high sulfur coal-fired units with wet FGD
systems is near-plant opacity. Near-plant opacity typically occurs in the region
downstream of the point at which the vapor plume dissipates but upstream of the point
at which any NOx plume could have formed. One characteristic of near-plant opacity is
a low opacity monitored at the precipitator outlet combined with a substantially higher
Method 9 opacity reading near the stack.

This paper deals with some of the possible mechanisms which can cause near-plant
opacity by the formation of sulfuric acid aerosol Potential remedies for this type of
near-plant opacity problem are discussed. The data presented is based on a combination
of field tests, Method 9 observations and statistical thermodynamic analysis. Of
particular importance to the reader will be the correlation of S02 emissions and
observed opacity data.

Introduction

Over the years, Burns & McDonnell has had the opportunity to visit many of the power
plants in the United States with wet FGD systems. During several of these visits, we
have noticed stacks which exhibit a plume which persists after the moisture plume has
dissipated. The majority of these plumes are generally pay to blue in color. In
discussing our observations with plant operations personnel, we often discovered that the
stack opacity monitors were indicating an opacity much lower than the apparent plume
opacity, which would be reported by EPA Method 9.

81-1


-------
This paper Is a summary of data and opinion generated by an internal project of Burns
& McDonnell. The purpose of this project is to better understand the impact of
scrubber design and operation on near-plant opacity.

Basic Principle®

Formation of Sulfur Dioxide

Fossil fueled power plants produce sulfur dioxide during the combustion of sulfur
containing fuels. The concentration of S02 produced varies with the sulfur content of
the fuel, the form of the sulfur in the fuel and the combustion environment

Formation of Sulfur Trioxide

Once formed, S02 can be converted to S03 and ultimately sulfuric acid though several
pathways. The first of these is the conversion of S02 by means of a catalyst. Several
components of fly ash are suspected to have such a catalytic effect Some of the more
common elements capable of this are carbon, iron, manganese and vanadium. Other
potential pathways are direct S02 to S03 conversion at high temperature and low
temperature hydrolysis.

The chemistry of converting S02 to S03 as mentioned above is well known. In fact, the
ability to easily convert S02 to SOa is the bask of such processes as fhie gas conditioning
to increase electrostatic precipitator performance, production of sulfuric acid and the
measurement of S02 using EPA Method 6.

Under the proper circumstances, the conversion of S02 - S03 - HjSO,* can result in the
creation of an acid aerosol. Aerosols are relatively small (sub-micron) droplets which
exhibit behavior more like particles than drops. Because of their small size, a relatively
small mass of acid can produce an extremely large number of aerosol droplets.

Opacify

Opacity is the result of particles disrupting the transmission of light by means of
scattering or absorbing a portion of the light. The particles which cause opacity do so
based on their size, concentration and composition. Other factors, such as the distance
light travels through the particle field, also influence opacity.

Vapor Plumes

Vapor plumes can exist as the result of combining two or more gases with different
chemical or physical compositions. In the context of this paper, a vapor plume is formed
when a saturated flue gas stream is released to the atmosphere.

81-2


-------
The characteristics of a vapor plume near the point of release are dependent to a large
extent on the conditions of the atmosphere which surround it. These conditions dictate
the rate of dispersion of the plume, the distance downwind for which condensed water
vapor affects the plume's color, and the overall plume appearance.

As the plume moves farther from the point of origin, other factors such as
photochemistry, and plume reactions with airborne chemicals, can alter both the
appearance and composition of the plume.

The Basis for Near-Plant Opacity

Thermodynamically, it can be shown that sulfur trioxide can be produced as a by-product
of combustion at temperatures less than approximately 2000 °F. Figure 1 shows the
results of a statistical thermodynamic modeling run performed using the CREST1
computer code. This figure shows the temperature regimes where the precursors for
sulfuric acid predominate.

Under the proper conditions, suite trioxide will form sulfuric add. Ibis step is largely
temperature dependent. Once formed, sulfuric acid vapor will condense at a
temperature determined by the acid content and moisture content of the flue gas. This
point is commonly referred to as the acid dewpoint Adds generated in this manner
tend to condense on surfaces and subsequently form pools. Solids in the gas stream
impinge on the condensed add and form deposits. This type of add formation is
probably not a significant contributor to near-plant opatity.

Under certain conditions, sulfuric add will form an aerosol. These conditions are
generally those which rapidly change the gas or its immediate environment An example
of this could be the quenching of flue gas in the absorber or cold end add formation in a
rotating air heater. Because of the aerosol's small size it stands very little chance of
being collected in even the most effident mist eliminators used on utility FGD systems.
The aerosol will likely pass through the system and exit the stack as a near-plant opadty
contributor.

It appears that at least a third path for sulfuric add formation may also exist This
pathway involves the reaction mechanisms which converts S02 directly to sulfuric add.
These reactions could take place in the absorber vessel or as a result of the plume
mixing with photochemical oxidants after release from the plant In either case,
controlling S02 concentration seems to be able to reduce the impact of this contributor
to near-plant opadty.

The important distinction between these three formation scenarios is that the third, S02
conversion in the absorber and/or plume, appears to be somewhat controllable through
proper FGD system design and operation.

81-3


-------
Case Study

Burns & McDonnell was retained by a utility to assist them in reducing the opacity from
one of their coal-fired unite equipped with a wet FGD system. During the early phases
of the project, comments were made by the operating staff which indicated that the
opacity was generally lower during start-up after an outage. Then, within days of going
on-line, the opacity would rise to pre-outage levels. During the project, several
parametric tests were performed on unit operating variables such as boiler oxygen levels
and absorber L/G ratio to try to simulate start-up conditions. There appeared to be
little or no impact on system-wide opacity as a result of these tests.

A second set of tests were proposed which would attempt to tie S02 outlet emissions to
observed opacity. The test methodology required making operational changes which
would produce significant variation in the outlet S02 concentration. These changes
included varying absorber tank pH and adding adipic acid. The results of the absorber
pH testing began to give some confidence that by increasing S02 removal we could
expect some decrease in opacity. This opinion was reinforced by completion of the
adipic acid test.

The adipic acid test was performed during a period when the boiler was firing a fuel
which historically had given a stack opacity above 40 percent In preparing for the test,
absorber tank pH had been increased to 6.0. The test began by taking two sets of
baseline opacity readings with the unit under stable load and the FGD system under
stable operating conditions. These readinp averaged 36 percent The next step was to
rapidly add approximately adipic add directly to each absorber reaction tank to produce
a concentration of 1,000 ppm in the recirculating slurry liquid. During the time of acid
addition, S02 concentration was being monitored at the stack and opacity readings were
being made using EPA Method 9. Figure 2 shows the opacity measured during the test
The data collected show that the stack plume opacity dropped from 36.1 to 31.6 percent
during the test. This drop corresponded to a reduction in outlet S02 emissions from
approximately 90 ppm to 9 ppm at the stack.

The results of the test point to a strong possibility that there may be a low-temperature
mechanism which acts upon S02 in either the absorber or stack environment to
contribute to near-plant opacity. Further, the results of the test support the observations
of the plant staff. As mentioned above, the plant reported low opacity for a brief period
of time after coining up from an outage. The explanation for this seems to be that at
the time of the startup, the reaction tanks were at a pH much higher than normal. This
condition is caused by the dissolution of the limestone with no incoming S02. When the
unit began to operate, the absorber "overscrubbed" S02 until the absorber pH dropped
to the control setpoint. Figure 3 shows a curve relating S02 emissions to opacity which
lends support to the "overscrabbing" theoiy.

81-4


-------
Impact of New FGD Design Parameters on Plume Formation

Based on the work done to this point, we believe that engineers and owners should be
aware of the possibility for near-plant plume formation. The risk seems highest for wet
lime/limestone FGD systems on boiler units burning medium to high sulfur fuel. There
are some design requirements which we believe can help to reduce the potential for this
problem.

Fuel Selection

We see a growing trend in the industry for wide variation in the fuel composition
specified to be used in design of FGD systems. This could potentially lead to an
absorber design which operates within S02 removal requirements (on a percentage basis)
at the extremes of the fuel range, but which creates a noticeable plume near the plant

Sulfur Dioxide Removal

S02 removal would appear to be the operator's sole controllable factor in the reduction
of near-plant opacity. Reliable control of SO^SOj concentrations leaving the scrubber is
important since it has been reported in the literature that S03 concentrations as low as
10 ppm can result in opacities as high as 20 percent For this reason extra flexibility
should be designed into any new wet limestone FGD system to allow for increased S02
removal. Examples of this would include designing for the addition of extra spray
headers, adding more nozzles to improve absorber spray coverage, modifications to
improve limestone grinding and the incorporation of feed and storage systems for
promoting agents.

Reagent Selection

At this time, no attempt has been made to extrapolate the information gathered on the
wet limestone system to any other FGD process or reagent We anticipate expanding
our investigation to these systems in future releases of the report

Control Options for Existing FGD System
Process Modifications

Controlling near-plant opacity which is related to sulfuric acid formation on existing
plants can be a significant challenge. The best advice is to set operations for the
maximum S02 removal. This would normally involve increasing the pH setpoint,
reducing gas flow (air heater leakage) to increase L/G and possibly the use of removal
enhancement additives such as DBA.

81-5


-------
Equipment Additions

There are at least two potential systems which could be added to a wet limestone FGD
system to help control sulfuric add aerosol. The first of these is the injection of alkali
material in the ductwork up-stream of the absorber. Commonly used reagents are lime
(hydrated and high surface area) and sodium carbonate. These reagents should be able
to remove a significant fraction of the add already present in the flue gas.

A second type of system which has been of some benefit is based on direct ammonia
injection. litis type of system has reportedly been successfully installed on a unit to
control near-plant opadty.

Conclusion

Our investigations have lead us to conclude that near-plant opadty resulting from the
discharge of sulfur compounds is a problem that affects a number of older coal-fired
units with wet FGD systems. We also believe that this problem will surface on many of
the Phase I FGD systems currently under construction. During the next year Burns &
McDonnell will continue explore potential corrective actions which can be taken to
mitigate this problem.

References

1. Benton, Dudley J. CREST: Chemical Reactions and Equilibrium Statistical
Thermodynamics Version 4.62 TVA Engineering Laboratory

81-6


-------
"G- SmonDX

Sdlox
-A-- Sir!ox
-*$•- Sul furIcacid

10® =

tfl
I—

CJ
3

f* ' *1	4

O10-'

Q_

U_

o

10-2 -

u
<
Of

LU

a

10~3 =

10

-A.

-5

10

10-B,

500

——"A— A-—A—

/

V

/A
* \

\

V p'

/

/

0

/

X



0

y

"A.,
,0*



0.	-Q~ —G— —O— -©— )

V

4
©

7\

1000

""A.

"A..



1500

TEMPERATURE OFPRODUCTS CR1

2000

2500

Figure 1. Thermodynamic Analysis of Sulfur Based Combustion Products


-------
TIME

Figure 2. Opacity vs. Time During Adipic Add Addition Test

Figure 3. Opacity vs. S02 Emission Concentration

81-8

A


-------
ECONOMIC COMPARISONS OF EMERGING
S02 CONTROL TECHNOLOGIES

Robert Martinelli
Thomas R. Goots

Paul S. Nolan
Babcock & Wilcox
20 S. Van Buren Avenue
Barberton, Ohio 44203

Abstract

Over the last several years, various new S02 removal technologies have been
demonstrated at both pilot scale and Ml scale plants. These new technologies may
compete with more conventional wet and dry scrubbing systems in the near fixture.

This paper will review the economic comparison of Limestone Injection Multistage
Burner (LIMB), Coolside, Limestone Injection with Dry Scrubbing (LIDS™), the
SOx-NOx-Rox Box™ (SNRB), conventional (lime) dry scrubbing, and conventional
limestone wet scrubbing with forced oxidation. Various plant sizes, sulfur contents, and
S02 removal rates win be considered. This comparison is intended to assist the power
generation industry in choosing.the most cost effective technology for a range of
conditions.

Introduction

The recent, successful demonstration of emerging S02 control technologies has been
reported previously.1,2,3 Economic comparisons of the LIMB and Coolside processes
have been made with conventional wet flue gas desulfarization (FGD) systems.4 This
paper includes material from these previous comparisons, in addition to an equivalent
economic comparison of LIDS, and conventional diy scrubbing. Due to the integral NOx
and particulate removal capabilities of the SNRB process, a separate economic
comparison of combined SO2, NOx and particulate removal is presented for dry FGD in
combination with a fabric filter and the Selective Catalytic Reduction (SCR) process.
All of these comparisons are made using an existing boiler with an electrostatic
precipitator as the base.

82-1


-------
LIMB SOz Control System

The LIMB technology involves the injection of a calcium-based sorbent into the fumace
at the upper end of a 1600 to 2300F sulfation temperature window, where it calcines to
active calcium oxide, and then reacts with S02 and oxygen in the flue gas to produce
calcium sulfate. The solid reaetant products and flyash exit the boiler with the flue gas,
and are collected as particulate by a baghouse or electrostatic precipitator (ESP). The
LIMB system included in this comparison uses calcitic hydrated lime as the sorbent to
achieve 60% S02 removal efficiency.

Cooiside S02 Control System

The Cooiside SOa control technology involves pneumatic injection of dry hydrated lime
into the flue gas downstream of the air heater, followed closely in distance by flue gas
humidification. SOa reacts first with the entrained lime particles in the humidifier, then
with the unreacted lime collected in the particulate removal device. Humidification
serves two purposes. First, it activates the sorbent to enhance S02 removal and, second,
it conditions the particulate matter to maintain efficient ESP performance. Spent
sorbent is removed from the gas, along with flyash, in an existing particulate collector
(ESP or fabric filter). Hie sorbent activity is enhanced by dissolving sodium-containing
additives such as sodium hydroxide (NaOH) or sodium carbonate (Na2C03) in the
humidification water. Solids recycling can be used to improve lime utilization, if the
particulate collector can accommodate the increased solids loading. The Cooiside system
included in this comparison uses calcitic hydrated lime to achieve 70% S02 removal.

LIDS S02 Control System

The LCDS system combines the limestone injection technology of LIMB with a dry
scrubber. As with the LIMB system, limestone is injected into the furnace, where it
calcines to active calcium oxide. Some of the calcium oxide reacts with S02 and oxygen
in the furnace. The remaining calcium oxide is conveyed, along with flyash and calcium
sulfate particles, to the downstream dry scrubber module(s) and particulate collectors).
A portion of the solids from the particulate collector hoppers is conveyed to a silo, then
mixed with water in a slurry tank, and pumped to the dry scrubber atomizers. The LIDS
system included in this comparison uses limestone to achieve 90% S02 removal.

SOx-NOx-Rox Box (SNRB) System

SNRB technology allows for the control of SOx, NO*, and particulate (Rox) using sorbent
injection in conjunction with a hot catalytic scrubbing baghouse (Box). S02 removal is
achieved using a calcium based sorbent (sodium based sorbents can also be used)
injected into the flue gas downstream of the economizer. NOx control is obtained by
injecting ammonia into the flue gas stream and then allowing the gas stream to pass

82-2


-------
through a catalyst located in a high temperature baghouse. Particulate is removed using
high temperature filter bags. The SNRB system included in this comparison uses
ammonia and hydrated lime to achieve 90% NO, removal and 90% S02 removal.

Dry Scrubber SOz Control System

The dry scrubber system for the base study is a conventional utility-design system with a
nine second gas residence time and dual-fluid atomizers with provisions for separate lime
slurry and recycle slurry systems. For the base comparison, it was assumed that the
existing precipitator would be adequate; 80% S02 removal is assumed for this case. For
the comparison with SNRB, a reverse air fabric filter (FF) and an SCR system are
included; 90% NOx removal and 90% S02 removal are assumed for this case.

Wet Scrubber S02 Control System

The wet scrubber system for the base study is a conventional limestone forced oxidation
system, utilizing a single absorber tower, with 10 ft/s gas velocity, integral tank,
distribution tray and horizontal mist eliminators. It is assumed that the tower
construction is carbon steel with Hastelloy cladding. The wet scrubber system included
in this comparison uses limestone to achieve 95% S02 removal.

Basis of Comparison

The purpose of this comparison is to present the economics of the various processes over
the range of the economic and technical premises chosen. The results are intended to
assist in evaluation of compliance options for existing plants.

The economic evaluation is based on capital and annual levelized costs for each of the
six technologies. Technical and economic premises were developed using the U.S.
Department of Energy Program Opportunity Notice (PON) DE-PSO1-88FE61530, the
Electric Power Research Institute's (EPRI) TAG™ Technical Assessment Guide (1989),
the design and operating experience from the LIMB project, CONSOL's topical report
on the Coolside process,2 and a review of state-of-the-art technology being used in the
design of dry FGD and wet limestone FGD systems.

Operating experience gained from the 5 MW SNRB demonstration was used to develop
costs for this technology. Operating experience from the LIDS pilot was used to develop
operating costs; capital costs for LIDS are based on components of dry scrubbing and
LIMB. Dry scrubbing, wet scrubbing, LIMB and Coolside have all been demonstrated
on a commercial scale, and therefore operating costs for these technologies are more
easily determined than the pilot scale SNRB and LIDS demonstrations.

The LIMB and Coolside processes were conceived as low-capital-cost technologies for
moderate levels of S02 removal. They are generally targeted for use on relatively small,

82-3


-------
older plants for which conventional wet or diy scrubbers would be difficult to justify.
The other technologies are normally designed for high levels of S02 removal typically
required for newer, larger plants. As a result, comparisons of LIMB and Coolside with
the other technologies are not straightforward, and the inherent limitations must be
considered in the interpretation. Capital costs are presented on a $/kW basis, and
annual levelized costs on a $/ton S02 removed basis, for the five S02 removal processes.

The SNRB process economics are compared to the diy FGD/fabric filter/SCR system
economics. For this separate comparison, a 90% S02 removal efficiency was used for
both systems. Capital costs are presented on a $/kW basis, and annual levelized costs on
a $/ton (S02 + NO*) removed basis where the NOx is calculated as N02.

Similar technical and economic assumptions were used to provide a common basis for
the five S02 removal process designs in order to make the comparisons as valid as
possible. Three reference plant capacities - 100, 250, and 500 MWC - were chosen. The
Eastern bituminous coals selected essentially differed only in sulfur content - 13, 2.5,
and 3.5% by weight An economic comparison, effectively consisting of a budgetary
estimate targeted to be accurate to within 10 to 30%, was then made for each FGD
process with each reference plant/coal sulfur combination. This resulted in nine
separate cases for each FGD process, or a total of 45 S02 cases. In addition, there are
nine cases for each of the two combined S02/NOx/particulate systems, or 18 cases.

Tables 1 through 4 outline the technical and economic assumptions used in determining
costs.

Economic Comparison Procedures

This analysis is intended to reflect total turnkey capital costs, which may be two to three
times the cost of a typical environmental equipment supplier contract. It is presented in
a format which a utility might use in determining the applicability of the processes as
part of developing an overall compliance strategy, rather than as a detailed listing of all
the specific assumptions and costs made for each and every case. Such an approach
recognizes the uncertainty that arises from any of a number of site-specific considerations
that require individual analyses in the final decision making process.

Costs are divided into three major categories: capital, variable and fixed operation and
maintenance (O&M) costs. The capital costs, or total capital requirement (TCR),
consist of the total plant investment (TPI), preproduction costs, inventory, land and
interest during construction. Variable costs include major consumables and disposal
costs. Maintenance costs for both labor and materials, operating manpower costs, and
administration and overhead costs constitute the fixed O&M costs. A constant dollar
levelization technique, as outlined in TAG, negates the effect of inflation on the capital
carrying charges and operating costs. The costs for consumables, utilities, labor and
disposal were derived from TAG and converted to 1993 U.S. dollars.

82-4


-------
Costs

Capital and annual levelized costs are summarized in Tables 5 through 8, for each of the
cases evaluated. Increasing coal sulfur content over the range of 1.5 to 3.5% leads to an
increase in capital costs, although the costs decrease on a $/kW basis as unit size gets
larger. In terms of $/ton S02 removed, the annual levelized figures represent the overall
costs, including O&M, over a basic assumed book life of 15 years. Hie results again
indicate decreasing costs with increasing unit capacity, with the greatest decrease
generally occurring between the 100 and 250 MWe sizes.

Hie lower sorbent utilizations of the LIMB and Coolside technologies are reflected in
the fact that the levelized costs do not decrease as rapidly as those for LIDS, wet
scrubbing or dry scrubbing when coal sulfur content increases. However, lower plant
capacity factors were found to favor the LIMB and Coolside processes, as does shorter
book life.

Although annual levelized costs were found to decrease with coal sulfur content,
relatively low sorbent utilization causes operating costs to increase, la comparison to
wet FGD, where fixed operating costs are significant, the variable operating costs of
LIMB and Coolside reflect the relatively high costs associated with sorbent supply and
ash disposal. As might be expected, operating costs increase with coal sulfur content
more rapidly with either LIMB and Coolside than with wet FGD; the impact would be
less for lower levels of removal at reduced calcium-to-sulfur ratios. As boiler size and
coal sulfur content increase, wet FGD operating costs gradually swing the economics in
favor of this technology.

Conclusion

Hie economics of each technology were determined for nine cases each. Process designs
were based on optimized, commercial, retrofit installations with assumed S02 removal
efficiencies ranging from 60 to 95%. The basic set of reference plants were assumed to
burn 1.5, 2.5, and 3.5% sulfur coal in units of nominal 100, 250, and 500 MWe capacities.
Comparisons made included those of capital costs on a $/kW basis and annual levelized
costs on a $/ton S02 removed basis. For the combined S02/NOx and particulate
removal systems^ annual levelized costs are presented on a $/ton (S02 + NO*) removed
basis.

Comparison of the annual levelized costs for S02 removal show LIMB to be
economically favored for all fuels studied at the 100 MW, size, and for the 250 MWe
case while burning 15% sulfur coal. LIDS is evaluated to have the lowest annual
levelized cost for the 250 MWe case while burning 2.5% sulfur coal, and for the 500 MWe
case while burning 1.5% sulfur coal. The wet FGD system is lowest evaluated for the
250 MWe case with 3.5% sulfur coal, and the 500 MWe cases burning 2.5% and 3.5%
sulfur coaL Annual levelized costs for S02 removal are summarized in Table 6.

82-5


-------
Previous sensitivity analyses showed that lower plant capacity factors favor the LIMB and
Coolside processes, as does shorter book life. Varying the reagent costs has a greater
effect on LIMB and Coolside economics than it does on the other process economics,
primarily because of lower sorbent utilization. These results reflect the relatively higher
operating costs for the LIMB and Coolside processes when compared to the other
processes included in this study.

Comparison of the annual levelized costs for combined S02/N0X and particulate removal
show SNRB to be economically favored for all the 100 MWe cases, for the 250 MWe
cases burning 2.5% and 3.5% sulfur coal, and the 500 MWe case burning 35% sulfur
coal. The conventional dry scrubber, SCR and fabric filter combination is evaluated to
have the lowest annual levelized costs for the remaining cases. Annual levelized costs
for combined S02, NOx and particulate removal are summarized in Table 8.

Analysis of projected SNRB costs indicates that this process is competitive over a wide
range of plant sizes and fuels, compared to a conventional dry scrubber, SCR and fabric
filter system. Based on this analysis, future market studies are planned to examine
SNRB economies for various plant capacities and plant life.

The results of this study suggest that several of the emerging S02 control technologies
ram be an economical solution for a wide range of retrofit conditions. Future analysis is
planned to examine the applicability of these technologies to new power plants.

References

L Goots, T.R., DePero, MX, and Nolan P.S., LIMB Demonstration Project Extension and
Coolside Demonstration, Final report to the U.S. Department of Energy, Office of Fossil
Energy under Agreement DE-FC22-87PC79798, DOE/PC/79798-T27 (NHS
DE93005979), November 1992.

2.	McCoy, D.C, et al} The Edgewater Coolside Process Demonstration: A Topical Report,
report to The Babcock & Wilcox Company under U.S. Department of Energy
Cooperative Agreement DE-FC22-87PC79798, DOE/PC/79798-T26 (NTIS
DE93001722), February 1992.

3.	Nolan, P.S., et aL, Demonstration of Sorbent Injection Technology on a Wall-fired Utility
Boiler (Edgewater LIMB Demonstration), final report to the U.S. Environmental
Protection Agency under Contract 68-02-4000, EPA-600/R-92-115 (NTIS PB92-201136),
June 1992.

4.	Nolan, P.S., DePero, M.J., and Goots, TJR., "Technical and Economic Evaluations of
the LIMB and Coolside Processes," presented to the International Joint Power
Generation Conference, October, 1992.

5.	Electric Power Research Institute, TAG™ Technical Assessment Guide, EPRI Report
P-6587-L, Electric Power Research Institute, Palo Alto, California, September 1989.

82-6


-------


Table 1

Reference Plant Design Information

Plant location

Ohio, near the Ohio River

Plant elevation

500 ft above sea level

Seismic zone

1

Boiler type

Pulverized coal-fired, radiant



boiler

Capacity factor

65%

ESP:



Emission rate

0.1 lb/106 Btu

Specific collection area

400 ft2/10s ACFM

ID fans:



UMB

Adequate

Coolside

Adequate

Dry FGD

Supplemental fans required

LIDS

Supplemental fans required

Wet FGD

Supplemental fans required

SNRB

Supplemental fens required

Plant retrofit factors:



UMB

1.0

Coolside

1.3 for the humidifier,



1.0 for other equipment

Dry FGD

1,3

UDS

1.3

Wet FGD

1.3

SNRB

1.3

Plant size, MWe

100 250 500

Coal flow rate, Ib/h

82,000 205,000 410,000

Main steam flow, Ib/h

634,000 1,585,000 3,170,000

Table 2

Equipment Design Assumptions - S02 Removal Processes



LIMB

Coolside

Dry FGD

LIDS

Wet FGD

S02 removal, %

60

70

80

90

95

Sorbent

Calcific hydrated

Calcitfc hydrated lime

Pebble lime

Limestone

Limestone



lime

and soda ash







Total system AP, in. w.g.

Negligible

1.5

5

5

10

ID fans

Adequate

Adequate

Supplemental

Supplemental

Supplemental







fans required

fans required

fans required

Flue gas reheat

No

No

No

No

No

New wet stack

No

No

No

No

Yes

Total sorbent storage,

7

7

7

31

31

days











Waste product

Flyash, lime.

Flyash, lime, calcium

Flyash, lime, calcium

Flyash, lime, calcium

Disposable

components

gypsum

and sodium sulfites

sulfate, sulfites

sulfate, sulfites

gypsum*





and sulfates







System outlet temperature, F

275

145

145

145

125

Total additional operating

0

4

8

12

16

manpower required











4 As opposed to wallboard quality gypsum

82-7


-------


Table 3



Equipment Design Assumptions - Combined S08/N0X

and Particulate Removal





Dry FGD, SCR & FF SNRB

' S02 removal, %

90

90

Sorbent

Pebble lime

Hydrated iime

NO, removal, %

90

90

Total system AP, in. w.g.

12

12

ID fans

Supplemental

Supplemental



fans required

fans required

Flue gas reheat

No

No

New wet stack

No

No

Total sorbent storage,

7

7

days





Waste product

Flyash, lime,

Flyash, lime,

components

calcium sulfate,

calcium sulfate,



sulfites

sulfites

System outlet

145

275

temperature, F





Total additional operating

12

12

manpower required









Table 4











Economic Premises









LIMB

Coolside

Dry FGD

LIDS

SNRB

Wet FGD

Reference date of cost estimate

June 1993

June 1993

June 1993

June 1993

June 1993

June 1993

Unit book life, yr

15

15

15

15

15

15

Tax life, yr

15

15

15

15

15

15

Levelizing factor for 15 yr

0.139

0.1®

0.139

0.139

ai39

0.139

carrying charges













Construction period, yr

1

1

2 to 3

2 to 3

2 to 3

2 to 3

Indirect costs as percent of total direct capital











General facilities

5

5

5

5

5

5

Engineering

10

10

10

10

10

10

Project contingency

18

18

15

20

20

15

Process contingency

5

5

2.5

10

10

2.5

Table S

Capital Costs, $/kW (S02 Removal)

Plant Size, MWC

Coal Sulfur, wi %

LIMB

Coolside

Dry FGD

LIDS

Wei FGD

100

1.5

95

154

228

303

351

100

2.5

97

158

233

310

358

100

3.5

105

164

235

320

361

250

1.5

47

98

126

163

194

250

2.5

51

104

130

171

200

250

3.5

55

108

133

178

204

500

1.5

32

71

90

112

139

500

2.5

37

78

94

121

144

500

as

41

S3

96

127

148

82-8


-------
Table 6

Anual Levelized Costs, S/ton* SOz Removed

Plant Size, MW„

Coal Sulfur, wt %

LIMB

Coolside

Dry FGD

LIDS

Wet FGD

100

¦ 1.5

800

953

899

953

1289

100

2.5

600

713

657

663

816

100

3.5

528

634

551

540

608

250

1.5

552

711

619

607

760

250

2.5

458

571

490

457

495

250

3.5

421

530

431

390

381

500

1.5

483

594

523

481

572

500

25

418

505

429

381

379

500

3.5

393

484

388

336

297

* 15 yr book life assumed

Table 7

Capital Costs, $/kW Combined SO^/NO*
and Particulate Removal

Coal Sulfur, Dry FGD, SCR

Plant Size, MW,

wt. %

&FF

SNRB

100

1.5

368

293

100

2.5

373

310

100

3.5

375

325

250

15

236

242

250

25

240

255

250

3.5

243

267

500

1.5

180

229

500

2.5

184

240

500

3.5

186

250

Table 8

Annual Levelized Costs, $/ton* S02 + NOx Removed

Coal Sulfur, Dry FGD, SCR

Plant Size, MWe

wt. %

&FF

SNRB

100

1.5

819

782

100

2.5

696

638

100

3.5

647

561

250

1.5

621

646

250

2.5

558

539

250

3.5

489

482

500

1.5

536

607

500

2.5

499

509

500

3.5

498

457

* 15 yr book life assumed

82-9


-------
A


-------
CASE STUDIES: EARLY REPORTS ON PHASE ONE CEM PROJECTS

Jack Passmore
Steve Brodmerkle
Steve Voss

Burns & McDonnell Engineering Company
4800 East 63rd Street
Kansas City, Missouri 64130

Abstract

This paper discusses the efforts of several utilities to meet the monitoring
requirements of 40 CFR Part 75. Four utilities which have Phase I affected units have
been selected as case studies in this investigation.

This paper begins with a brief description of each of the case study utilities. This is
followed by a discussion of the approaches taken by each of these utilities in the
following areas:

•	Contracts

•	CEM Equipment

•	Measurement Locations and Access

•	System Architecture

•	System Reliability

The results the case study CEM projects are then discussed in terms of CEM system
operation, certification, and schedule.

83-1


-------
Introduction

Efforts to meet the monitoring requirements of the Clean Air Act Amendments of
1990 have presented many challenges for utilities over the past three years. The
purpose of this paper is to investigate how several utilities have met these challenges.

Four utilities which have Phase I affected units have been selected as case studies in
this investigation. These utilities are identified as Utility A, Utility B, Utility C and
Utility D for the purposes of this paper. Burns & McDonnell is serving as the
engineer assisting these utilities in the selection, procurement, installation, and
certification of the required CEM systems.

This paper begins with a brief description of each of the case study utilities. This is
followed by a discussion of the approaches taken by each of these utilities in the
following areas:

•	Contracts

•	CEM Equipment

•	Measurement Locations and Access

•	System Architecture

•	System Reliability

The results of the case study CEM projects are then discussed in terms of system
operation, certification, and schedule.

Case Study Utility Information

Utility A has a total of 7 Phase I units located at 3 different stations. Each of the
units are coal fired. Units range in generating capacity from 136 MW to 648 MW.
Total generating capacity for all Phase I units is 2176 MW.

Utility B has a total of 4 Phase I units located at 3 different stations. Two of these
units share a common stack. Each of the units are coal fired. Units range in
generating capacity from 104 MW to 550 MW. Total generating capacity for all Phase
I units is 1199 MW.

Utility C has a total of 7 Phase I units located at 3 different stations. Each of the

units are coal fired. Units range in generating capacity from 75 MW to 600 MW.

Total generating capacity for all Phase I units is 2290 MW.

Utility D has a total of 5 Phase I units located at 2 different stations. Each of the
units are coal fired. The Units range in size from 175 MW to 670 MW. Total
generating capacity for the Phase I Units is 2320 MW.

83-2


-------
Contract Approach

Procurement and installation of the CEM equipment and associated access structures
were accomplished by the following contracts:

Utility A

1.	CEM equipment contract - Included the CEM equipment, equipment shelters, and
startup services.

2.	Structural construction contract - Included fabrication and construction of the
access platforms, ladders, elevators, and shelter pads.

3.	Electrical construction contract - Included installation of the CEM equipment,
power wiring, signal wiring, and lighting.

4.	Certification testing contract - Included all certification testing required by the
regulations.

Utility B

1.	CEM equipment contract - Included the CEM equipment, equipment shelters,
startup services and certification testing.

2.	Elevator contract - Included procurement and installation of the stack elevator.

3.	General Construction contracts - One Construction contract was developed for one
station. A second construction contract was developed for the other three stations.
The construction contracts included electrical work, structural work, and installation
of the CEM equipment.

Utility C

1.	CEM equipment contract - Included the CEM equipment, equipment shelters,
startup services and certification testing.

2.	Elevator contract - Included procurement and installation of the stack elevators.

3.	General Construction Contract - Included electrical work, structural work, and
installation of the CEM equipment.

Utility D

1. CEM equipment contract - Included the CEM equipment, equipment shelters,
startup services and certification testing.

83-3


-------
2. General Construction Contract - Included electrical work, structural work, supply
and installation of the elevator, and installation of the CEM equipment.

In all of the above cases, a separate contract was established for the CEM equipment
One reason for this is that most CEM vendors are reluctant to serve as the general
contractor on a project involving extensive structural work. In addition, most
structural and electrical construction contractors are unfamiliar with CEM systems,
and would probably find the administration of such a specialized subcontractor
difficult. The separate CEM contract approach worked out well for all of the utilities
in this case study.

All of the case study utilities required elevators on some of their stacks. For Utility B
and Utility C, the elevators were purchased and installed by separate stand-alone
contracts. For Utility A and Utility D, the elevators were purchased and installed by
subcontractors to the general construction contracts. In either case, the contracts
required that the elevators be installed by the elevator vendors to insure proper
operation. Separate elevator contracts allowed the elevators to be purchased early.
This prevented long delivery times for elevators from adversely affecting the
schedules.

General construction contracts combining both electrical and structural work were
developed for all but one of the case study utilities. This type of contracting
approach eliminated the engineer's and owner's involvement in structural and
electrical interface problems. Fortunately, these types of problems were rare for the
one utility that decided to use separate electrical and structural contracts.

For most of the case study units, CEM equipment shelters were supplied by the CEM
vendors. For other units, shelters were constructed by the construction contractor.
The advantage of having the CEM vendor supply the shelter is that the CEM
equipment can be fully installed, interconnected, and tested in the CEM vendor's

factory.

Certification testing was included in the CEM equipment contract for all of the case
study utilities except one. This placed total responsibility for providing a certified
system on the CEM vendor.

CEM Equipment
Gas Monitors

All of the case study utilities chose to use dilution extractive systems on their Phase I
affected units. This selection was based on the following:

1. S02 and C02 emissions must be reported as a mass flow rate. Measurement of
flow and pollutant concentration is required for this calculation. Flow is inherently

83-4


-------
measured on a wet basis. If the pollutant is also measured on a wet basis, the mass
flow rate can be easily calculated without knowing the moisture content of the flue
gas. However, if the pollutant is measured on a dry basis, the moisture in the flue
gas must be used as a factor in the calculations. Dilution extractive systems measure
pollutants on a wet basis. This eliminates the requirements to measure or calculate
moisture in the flue gas.

2.	The sample tubing on dilution extractive systems is under positive pressure. This
eliminates the possibility of sample contamination by leaks in the sample tube.

3.	Heat tracing of sample tubes on dilution extractive systems is often not required.
Dilution by dry air lowers the dew point temperature of the extracted sample to a
value below normal ambient conditions in most cases. In some cases, heat tracing is
required on cold days. Heat tracing was installed on ail of the case study units as
insurance against the possibility of condensation problems.

Flow Monitors

Ultrasonic flow monitors were used by all of the case study utilities except for
Utility C. Ultrasonic flowmeters were chosen due to their inherent cross stack
averaging capability and their ability to meet the EPA daily calibration requirements.
However, ultrasonic flowmeters must be installed so that the sound waves cross the
stack or duct diagonally. This usually requires additional access platforms.

Flow is measured on the ducts on two units for Utility C. These ducts are very short
in length. Flow meter vendors were asked to provide assurances that flow could be
measured on these ducts. A pitot tube vendor offered to install their device on one
of the ducts on a trial basis. - The other vendors did not think that their devices
would work in such a difficult application. Flow tests using a pitot tube revealed
that a single measurement point would provide sufficient accuracy in this application.

Data Acquisition Systems

The OEM vendors performed their own application software development for data
collection, recording and reporting of CEM data. Multi-tasking operating systems
were used in all cases.

Measurement Locations and Access

In most cases, probes and flow monitors were located on the stack. This is because
stacks provide the long straight runs for proper flow monitoring. In addition, access
platforms were already available on many of the stacks. In all cases, where probes
were located in the stack, the 40 CFR Part 75 guidelines for 8 and 2 diameters were

met.

83-5


-------
Probes and flow monitors were mounted on the duct on some of the case study units.
This was done on two of the units because long straight runs of duct were available.
If adequate measurement accuracy can be obtained on the duct, then this location is
often better because it is more accessible.

On two stations belonging to Utility C, duct locations were chosen because the stacks
were shared by Phase I and Phase II units. Monitors were located on the Phase I
Unit ducts leading to the stacks in order to keep Phase I emissions separate from
Phase II emissions.

Elevators were either installed, or were already available, on all units that had stack
mounted equipment. In addition, two utilities included weather enclosures around
portions of the stack access platforms. All of the utilities considered easy access to
the stack mounted equipment very important.

System Architecture

Basic system architecture drawings for each utility are presented on Exhibits 1
through 4. Labels used in the drawings are defined as follows:

Analyzers - A set of analyzers used to measure gaseous emissions (S02, NOx, C02)
and flow at a point. The gas analyzers in this case study are all dilution extractive
type systems which perform analyses of gas samples drawn from the sample point.
The term "analyzers" is also used to include the flow monitor electronics.

PLC (Programmable Logic Controller) - Used to control the functions of gas and flow
monitors. For all of the case study projects, each set of analyzers was controlled by
its own PLC.

DAS (Data Acquisition System) - Used to perform the data collection, storage, and
reporting functions required by the EPA.

CRT - This term is used to identify operator interface stations.

In all cases, the CEM shelters contained the analyzers, PLC systems, data acquisition
systems, and operator interface stations. All utilities except Utility D chose to have
individual data acquisition systems dedicated to each set of analyzers. For Utility D,
one data acquisition system is shared by two sets of analyzers in two of the shelters.

Additional operator interface stations were located in the unit control rooms and
engineer's offices. In most cases, the engineer's office stations will be used to
generate the official reports which will be submitted to the EPA.

All four utilities wished to have the capability of receiving data at their corporate
offices. This was established by the use of modems. In most cases, modems were

83-6


-------
attached to the engineers stations. The modems for one utility were attached to the
data acquisition systems in the equipment shelters. In all cases, the modem baud
rates limit the speed at which data can be updated at the corporate offices. Utility D
is planning to generate the EPA reports at the corporate office station.

System Reliability

System reliability was a major concern of each of the case study utilities. The
following is a description of how reliability was enhanced on these projects.

Vendor guarantees: CEM vendors were required to guarantee 95% availability on
their systems.

Analyzer redundancy: None of the case study utilities chose to purchase redundant
analyzers. It was determined that the benefits of total redundancy would not be
worth the cost. Instead, some of the utilities required that cabinet space for spare
analyzers be provided for future use if necessary. Power sources were also sized to
accommodate future redundant analyzers.

Data storage: Backup data storage capabilities were required for all data acquisition
systems. Two hard disk drives were required on each system. Tape or optical disk
backup was also required.

In addition, thirty day data storage capability was required for each PLC. This
would help insure that the data would not be lost in the event of a DAS failure.

Power sources: Two independent sources of 480 V power were fed to an automatic
transfer switch in each analyzer shelter. If the primary source failed, the backup
source would be switched-in to take over.

Uninterruptible Power Supplies: UPS systems were used to power the data
acquisition systems and the PLC systems in all cases. For most of the utilities, UPS
systems were not used to power the analyzers. The reason for this approach is that
the analyzers could tolerate a brief power dip, but the data acquisition systems and
the PLC systems could not. Since the analyzers require more power than the data
acquisition systems or the PLC systems, larger UPS systems would be required. One
utility felt that the addition of UPS systems for the analyzers was justified to prevent
loss of monitoring capabilities during power failures.

HVAC: Most case study utilities chose to use dual HVAC systems in the analyzer
shelters. HVAC system failure could result in temperature conditions that would
adversely affect analyzer or DAS operation.

Air Supply: Backup supplies for dilution air were used in all cases. Bottled
compressed air was connected to the air feed through a pressure regulator and check

83-7


-------
valve arrangement. The gas bottles would supply air in the event of primary air
supply failure.

Dilution Air Conditioning: Backup dilution air dryers and filters were used in all
cases.

In general, the steps taken by each utility to increase reliability were considered
reasonable and cost effective.

Results

The results of many of the decisions made in the planning and design stages will not
be known until the CEM systems have been in operation for many months.

However, some early results can be reported in terms of system operation,
certification, and schedule.

System Operation

At the time of this writing, all of the twenty two case study CEM systems have been
installed. Of these, seventeen systems can be considered mechanically operational.
Additional software development work is required on all of these systems. None of
these systems have been certified. Certification tests must be completed by
November 15,1993.

As mentioned previously, ultrasonic flowmeters were installed at Utilities A,B and D.
Pitot tubes were installed at Utility C. Results of these selections have been mostly
successful to date. A relative accuracy test was performed on the ultrasonic flow
meter on one of the units at Utility A. The ultrasonic flowmeter easily passed the
test by a safe margin.

Problems were encountered in aligning the pipe section connections for the ultrasonic
flow monitor at one of the units at Utility D. The brick stack construction at this unit
made this alignment difficult. No significant problems were encountered in aligning
the connections on steel or concrete stacks. The Utility D flow monitor supplier has
since added an optional mating flange with features that should make alignment
easier.

During*tile" flow tests of pitot tubes for utility C, single point pitot tube
measurements yielded better results than averaged multiple point measurements.
The method of averaging multiple points contributed to these results. During the test,
three pitot tubes were connected so that all of the high pressure ports were tubed to
a common header connected to the high pressure side of a differential pressure
transmitter. Likewise, the low pressure ports were connected to the low pressure
side of the differential pressure transmitter. The differential pressure created
between different points along the flow profile set up a small flow within the pitot

83-8


-------
tube assembly between sensing ports. The pressure drop created by this flow
affected the accuracy of the readings.

Software problems have represented the largest obstacles to becoming fully
operational. Problems included the following:

•	Schedule slippage due to problems and delays in software development

•	DAS lockup during operation.

•	Improper averaging procedures and incorrect use of formulas.

•	Communications problems.

Some of the software development problems were due to delays in the publication of
the final regulations and difficulties in interpreting the requirements of the final
regulations. Other problems such as communications problems were due to vendors
capabilities and resources. In order to avoid these types of problems on future
projects, some of the vendors are now proposing to use third party software
suppliers.

Certification Tests

At the time of this writing, certification tests have been performed on only one of the
case study units. This unit easily passed all of the relative accuracy tests, but failed
to pass the calibration error test on the last day of testing. The cause of the failure is
currently being investigated.

Full certification of the CEM systems cannot be accomplished until the EPA issues the
DAS certification software. This software must be run on the data acquisition
systems to insure that the measured data is being processed in accordance with the
regulations.

Schedule

The following is a summary of the major milestone dates for each of the case study
projects. As mentioned previously, electrical and construction work was combined
into one contract for most of the utilities. Completion dates for installation and
testing are for completion of these activities on all units for each utility.

83-9


-------
UTILITY
A B C D

Start Study
Award CEM Contract

Aug 91 Jan 92 Nov 91	Oct 91

Oct 92	May 92 Feb 92	Mar 92

Feb 93	May 92 May 92	Jun 92

Dec 92	May 92 May 92	Jun 92

Jun 93	May 93 Apr 93	Jun 93

Oct 93	Oct 93 Oct 93	Sep 93

Award Electrical Contract
Award Structural Contract
Complete Installation

Complete Testing (planned)

Total Duration (months)

26

21

23

23

The schedules were affected by the following factors:

•	Delays in publication of final regulations.

•	Large number of projects performed by limited number of CEM vendors.

•	Uncertainties in interpretations of the final regulations.

•	Design changes.

•	Late submittal of shop drawings.

Greater emphasis on timely submittal of CEM vendor shop drawings would improve
schedule performance. Detailed electrical design was delayed on several of the
projects as a result of late submittal of shop drawings by the CEM vendors.

As expected, early results indicate that the total project durations will be shorter for
Phase II units. Phase II projects are benefitting from the experiences gained on the
Phase I projects. In addition, Phase II projects will benefit from having final
regulation available at the onset. Uncertainties in the regulatory requirements should
also be resolved before they adversely affect schedules for Phase II unite.

Conclusion

The monitoring requirements of the Clean Air Act Amendments of 1990 have
presented many challenges for utilities that have Phase I units. Many major design
decisions had to be made while the regulations were still in draft form. In addition,
CEM technology has been in a state of change in order to meet the more stringent
requirements of the new regulations. Little time has been available to investigate
new technology, select qualified bidders out of the growing pool of CEM suppliers,
prepare specifications, and place equipment orders. In spite of these difficulties, the
utilities presented in this case study are well on the road to successful completion of
their CEM projects.

83-10


-------
r

00

to

SAMPLE LINE

STATION
X

UNIT

SAMPLE LINE

STATION HjNIT
Y L

UNIT 1
UNIT 2

DUCT

STACK

STATION
Z

SHELL

UNIT 3
UNIT 4

UNIT 5

DUCT

ANALYZERS

PLC

OAS

SHELTER

CRT

-Tcrt

CONTROL ROOM

ANALYZERS

PLC

DAS

SHELTER

CRT

l_.

'I-'

i_.

CRT - MODEM

ENGINEERS OFFICE

CRT

CONTROL ROOM

CRT - MODEM

ENGINEERS OFFICE

ANALYZERS

PLC

DAS

CRT

ANALYZERS

PLC

DAS

CRT

ANALYZERS

PLC

DAS -

CRT

ANALYZERS

PlC

SHELTER

OAS

m

CRT

I—

I

L_

'1

I I

CRT

- CRT

CONTROL ROOM
UNITS 1 & 2

• -I

L.

I
I

l_.

CRT

	CRT

CONTROL ROOM
UNITS 3 & 4

STACK

SAMPLE LINE

ANALYZERS

PLC

DAS

SHELTER

CRT



CRT

CONTROL ROOM
UNIT 5

CRT MODEM

ENGINEERS OFFICE

n

_L

LEASED
LINES

MODEMS

CRT

CORPORATE OFF'CE



Exhibit 1
CEM SYSTEM ARCHITECTURE
UTILITY A


-------
UNIT

STATION
X

UNIT 2

DUCT

7







ANALYZERS

-- OAS -

J



PLC

	1	

CRT

SHELTER

CRT

CONTROL ROOM

I	

CRT

MODEM

ENGINEERS OFFICE

' ~1

STATION
Y

UNIT 1

DUCT

STACK









}	



ANALYZERS
PLC

- - DAS -

1

EXISTING ROOM UKl

- CRT

CONTROL ROOM

CRT - MODEM

ENGINEERS OFFICE

station
z

UNIT

DUCT

STACK

W]	,	







ANALYZERS

- - DAS -

[-11



PLC

"1'

EXISTING ROOM 1

CRT

CONTROL ROOM

L_.

- CRT



MODEM -

ENGINEERS OFFICE

J



Exhibit 2
CEM SYSTEM ARCHITECTURE
UTILITY B


-------
00

01

W

STATION
X

UNIT 1

STATION
Y

DUCT

STACK

UNIT 2

DUCT

STACK

STATION

UNIT 3

UNIT 1
(PHASE II)

UNIT 2
[(PHASE I)

~ UNIT 1
(PHASE II}

UNIT 2
(PHASE I)

DUCT

STACK

Wi	







ANALYZERS

- - DAS —

r"



PLC

I

SHELTER

¦ —i

r









J	



ANALYZERS

- - DAS -

PLC

1

SHELTER UNI

— — -

CRT



CONTROL ROOM



CRT - MODEM

ENGINEERS OFFICE

]







ANALYZERS

- - DAS -

J



PLC

1

SHELTER

CRT



r'

SHELTER

CRT

CRT

CONTROL ROOb

ANALYZERS



DAS -

	J—

¦ CRT ~

MODEM -

PLC



1



ENGINEERS OFFICE

r'









ANALYZERS

- - DAS -





PLC

CRT

SHELTER

CRT

CONTROL ROOM

CRT

MODEM

ENGINEERS OFFICE

n

L.

MODEMS

CRT

CORPORATE OFFICE

Exhibit 3
CEM SYSTEM ARCHITECTURE
UTILITY C


-------
STATION
Y

UNIT 1

DUCT

UNIT 2.

DUCT

UNIT 3

Oo
w

station

I

UNIT 1

UNIT 2

ANALYZERS

PLC

-1 r
i_i_

MODEM

ANALYZERS

PLC

DAS
TT

J I		

CRT

SHELTER

r

I

TO !
-CORPORATE !

OFFICE |
1	I

	_CRT_

UNIT 1
CONTROL ROOM

CRT

UNIT 2
CONTROL ROOM

i r

I	I

I	I

I TO L.
-I CORPORATE!
OFFICE I

	CRT

UNIT 3
CONTROL ROOM

.J

CRT

ENGINEERS OFFICE

L

CRT

NSTRUMENT SHOP

FROM
DAS:
MODEMS'

SHELTER

TO I
-CORPORATE I
OFFICE 1

l__

CRT

UNIT 1
CONTROL ROOM

	

CRT



UNIT 2

CONTROL ROOM

	CRT

ENGINEERS OFFICE





	CR1

INSTRUMEN

r

T SHOP

Exhibit 4
Cm SYSTEM ARCHITECTURE
UTILITY D


-------
SECONDARY DEWATERING OF FGD SLURRY WITH
CENTRIFUGE AT ELRAMA POWER STATION

Frederick J. Bickerton, Jr.
Duquesne Light Company

Wei-Chung Yu
Conversion Systems, Inc.

Ronald R. Bevan
David L, Milligan
Alfa Laval Separation, Inc.
Sharpies Division

Abstract

Secondary dewatering of FGD slurry from a magnesian lime scrubber is challenging
due to the fineness of the slurry particle size. Field tests of dewatering by a centrifuge
have shown a potential for improving the dryness of the cake.

After extensive testing, a commercial size centrifuge was installed as a secondary
dewatering device and has been in operation since March 1992. The unit is operating
at the Conversion Systems, Inc. facility handling FGD waste generated at Duquesne
Light Company's Elrama Power Station. This paper provides an overview of the
operating experience, process modification, optimization program and the impact on
the final disposal/disposition of by-product from this operation.

Introduction

Beginning in the 1960's, due to clean air legislation, and continuing now beyond the
year 2000 in response to the CLEAN AIR ACT AMENDMENTS of 1990, a large
number of coal burning power stations are expected to scrub the exhaust gases
resulting from coal burning. The scrubbing is intended to reduce the primary
contaminants, sulfur oxides mainly in the form of sulfur dioxide, S02.

84-1


-------
There are many alternatives available for controlling or reducing S02 emissions,
among them - wet scrubbing, switching to lower sulfur coal, coal cleaning, scrubbing
via regenerative processes, dry scrubbing, etc. The present stage of these
technologies and the S02 reduction requirement should result in wet scrubbing, by
limestone or lime, being the dominant S02 control technology through the remainder
of the century. In either limestone or lime scrubbing, a blowdown stream containing 5
to 15% solids will be generated as the scrubber blowdown. Dewatering of this stream
is usually handled in a two stage system, such as thickeners in combination with
rotary drum vacuum filters. This allows the operation to recycle water for reuse and to
generate a solid waste for disposal or sale. Disposal is usually by stabilization and
placement in a landfill.

Magnesian lime is a reagent widely used in the lime scrubbing process. The moderate
content of magnesium oxide (3 to 6% by weight) in this lime enhances the removal
efficiency of S02 due to the increase in concentration of dissolved sulfite species such
as magnesium sulfite1. However, the scrubber blowdown from this system which
contains magnesium is more difficult to dewater than blowdown from a scrubber
without magnesium. One hypothesis to explain this difference is that magnesium may
serve as a crystal modifier by being adsorbed on the surface of CaS03 and causes
crystal defects which inhibit crystal growth2. A 15-19 micron weighted average particle
size, (by Leeds and Northrup Microtrac particle size analyzer), was reported from
samples of scrubber blowdown of one 300-MW power station with Magnesian lime3
with insignificant amounts of fly ash. The weighted average size of Elrama scrubber
blowdown, which includes a significant amount of fly ash, is about 13 micron.

Elrama Power Station

Elrama Power Station is a coal-fired electric generating station, wholly-owned by
Duquesne Light Company (DLC) consisting of four coal-fired boilers and four turbine-
generators with a combined net capacity of 487 MW. It is located along the
Monongahela River in Elrama, PA, about 25 miles southwest of Pittsburgh. In order to
comply with various environmental regulations associated with the operation of coal-
fired power plants, Elrama is equipped with mechanical dust collectors and
electrostatic precipitators to remove fly ash from gases leaving the boilers as well as a
magnesian lime FGD scrubbing system to remove S02 from the flue gas. The removal
efficiency of S02 in the FGD system is about 83 to 85 % and the combined particulate
removal efficiency of the mechanical dust collectors, the precipitators, and scrubbers is
about 99%.

Elrama station generates on average 250,(XX) tons, dry weight basis (DWB), of
scrubber sludge and fly ash per year. DLC has an agreement with Conversion
Systems, Inc. (CSI) to dewater the scrubber sludge and stabilize it utilizing the POZ-O-
TEC® treatment process in CSI's Elrama processing facility. The stabilized material is
disposal of in a nearby landfill. A simplified process flow diagram of the POZ-O-
TEC® process is shown in Figure 1.

84-2


-------
Vacuum Filter Dewatering

The dewatering of the scrubber blowdown consists of two stages. The first stage,
usually called primary dewatering, consists of two 100-foot diameter thickeners. The
sludge is settled under gravity with polymer flocculant to accelerate the settling. The
thickeners are usually capable of thickening the sludge from less than 10% solid
content in the blowdown to about 20 to 30% solid content in the thickener underflow.
The thickener underflow is then pumped into the surge tank for storage prior to
secondary dewatering. Figure 2 shows a schematic diagram for system with vacuum
filters, as well as the centrifuge system.

The secondary dewatering at Elrama until March 1992 exclusively utilized rotary drum
vacuum filters. The solids content in the cake from a vacuum filter is a direct function
of the solids content of the thickener underflow. Figure 3 shows this relationship of
the Elrama material. Typical filtercake solids in the recent years is about 40% for a
typical thickener underflow of 25% solids.

After dewatering, the filtercake is conveyed and mixed with fly ash and other additives
in a controlled manner to form POZ-O-TEC®.

Centrifuge Testing and Selection

From December 1990 to March 1991, DLC and CSI arranged a series of on-site tests
with three (3) centrifuge vendors. The objectives of the on-site testing were:

1.	To determine the dewatering capability of a centrifuge versus a vacuum
filter with actual Elrama material.

2.	To obtain design and scale-up information so that the centrifuge vendors
could recommend to CSI an appropriate commercial size centrifuge for
secondary dewatering at Elrama.

3.	To predict the impact of centrifuge dewatering on the performance of the
FGD system.

A general conclusion, after this series of testing, is summarized below:

1.	The centrifuge produced higher cake solids than vacuum filters, usually
in the range of 7 to 12%. Figure 4 shows the difference of the cake %
total solids between the centrifuge and the vacuum filter at the same feed
conditions during a test.

2.	Higher G force; i.e., higher RPM, results in higher cake solids, within
limits.

84-3


-------
3.	Higher G force; i.e., higher RPM, results in higher recovery, within limits.
Recovery is defined as the total suspended solids (T.S.S.) in the cake
(lbs) divided by the T.S.S. in the feed (lbs). High recovery can be
directly translated into low suspended solids and vice versa.

4.	Centrate is the liquid overflow generated from the centrifuge. With no
polymer addition to the feed, the suspended solids in the centrate, which
is dependent on the feed rate to the centrifuge, was generally higher than
the suspended solids of the filtrate. Typically, suspended solids in the
centrate was up to 3%, while suspended solids in the filtrate was usually
less than 0.5%. This means that the recovery of the centrifuge without
chemical additives may be less than that of the vacuum filter.

5.	The addition of centrifuge polymer, usually at the rate of 0.10 to 0.16 lbs
per ton of sludge (DWB), improved the recovery to about 99%, with
centrate T.S.S. as low as 0.21%.

6.	In addition, preliminary testing on direct dewatering of scrubber
blowdown bypassing the thickeners were performed. Results indicate
better dewatering and recovery than when used for secondary
dewatering. The explanation may be higher feed temperature and larger
particle size.

A model P-7600FGD centrifuge manufactured by Alfa Laval Separation, Inc., Sharpies
Division was chosen for installation at CSI's processing facility. This centrifuge is
designed to process a maximum of 350 GPM of thickener underflow. The vacuum
filters now serve as backup equipment for the centrifuge. This centrifuge was started
up in March, 1992, and has been in operation since then except for periods required
for maintenance and modification.

Centrifuge Operation and Optimization

Since installation of the unit in March, 1992, the centrifuge has dewatered thickener
underflow with typical continuous operation ranges from 250 to 325 GPM, while
operating at a G force of 2572 X G. Polymer has been added to the feed to the
centrifuge as well as to the centrate from the centrifuge, in both cases with the
centrate being returned to the thickener. When polymer has not been used, the
centrate has been returned to either the thickener or the absorber. To date, the best
dewatering performance was seen when the centrate was returned to the absorber. In
all cases, polymer has been added to the thickener, which is the same during vacuum
filter operation.

84-4


-------
The cake solids produced by the centrifuge is a function of the solids concentration of
the thickener underflow. As the feed concentration of the thickener underflow
increases, the solids content of the centrifuge cake also increases. The centrifuge
typically produces a cake between 45 to 50% total solids and has been as high as
56.3% total solids. Figure 5 shows the correlation of centrifuge versus vacuum filter
under the same feed conditions with the centrifuge producing a cake that is
approximately seven (7) percentage points drier than the cake from the vacuum filters.

With no polymer in the feed and the centrate returned to the thickener, a feed rate of
250 GPM has provided an acceptable recovery and dewatering performance. Figure 6
shows the recovery from 6/10/92 to 6/21/92 at 250 GPM. At this feed rate, the
thickeners did not have an overflow loaded with fines. At a feed rate of 300 GPM,
under similar conditions, fines started to build up in the thickeners. With no polymer in
the feed and the centrate returned to the absorber, acceptable recovery and
dewatering performance has been achieved at a feed rate of 300 GPM. Feed rates
higher than 300 GPM are feasible, but have not been tested on a sustained basis.

Optimization of the centrifuge and the associated system will continue and will
include feeding scrubber blowdown directly to the centrifuge.

Benefits of Centrifuge Dewatering

The major benefits for the end user, since the operation of this centrifuge began, are:

1.	Reduction In Disposal Tonnage- Improved dewatering results in a

saving to the Elrama Station of 5.6% of its total disposal tonnage.

2.	Reduction Of Operating Labor- While data is not yet available on
overall operating costs, CSI has seen some savings in plant operating
labor due to the simplification of the secondary dewatering system.

3.	Less Stringent Requirements On Thickener Operations- variability in
the feed solid concentrations to CSI seem to have less influence on the
centrifuge cake solids as opposed to vacuum filtercake solids.

4.	More Homogeneous Cake- The centrifuge appears to produce a much
more homogeneous cake than does the vacuum filter,

5.	Salability Of By-products- The drier and more homogeneous cake
results in a more easily handled by-product. This increases the potential
for beneficial use of the by-product, thus avoiding disposal costs and
savings disposal site capacity for future use.

84-5


-------
Summary

The installation of a centrifuge at Elrama for secondary dewatering is beneficial in
terms of better cake solids and lower disposal costs. The impact on the FGD system,
such as suspended solids in centrate. appears to be manageable. From the cake
solids standpoint, a centrifuge in magnesian lime scrubbing processes appears to be a
better dewatering devise than a vacuum filter.

References

1.	EPA Alkali Scrubbing Test Facility: Advanced Program - Final Report (October
1974-June 1978), NTIS, May, 1980.

2.	J.C.S. Chang and T.G. Brna, "Evaluation of Solids Dewatering for a Pilot - Scale
Thiosorbic Lime S02 Scrubber, "presented at AICHE 1987 National Meeting,
Houston, Texas (April 1987).

3.	L.B.Benson, A.Randolph, and J.H. Wilhelm, "Improving Sludge Dewatering in
Magnesium-enhanced Lime FGD Systems", presented at EPA/EPRI 1st
Combined FGD and Dry S02 Control Symposium (October 1988).

84-6


-------
Scrubber

Scrubber

Slowdown
Tank

Wash Water

a

Thickener or
Hydroiyclone

(Primary
Dewaterirg)



Optional
Wastewater
Treatment
Plant

Rotary Vacuum Filter, Centrifuge,
or Horizontal Belt Vacuum Filter
{Secondary Dewatsring)

Vacuum
Receiver

Tf

Possible
Wastewater
Purge

Fly Ash
Silo

O-Tec

Figure 1: Simplified Process Diagram of POZ-O-TEC® Process

84-7


-------
SEAL WATER

CAKE

Centrifuge System

Figure 2: Vacuum Filtration System Versus Centrifuge System at Elrama

84-8


-------
ELRAMA SLUDGE DEWATERING

VACUUM FILTER

FEED SOLIDS (%TSS)

to 1992

Figure 3


-------
ELRAMA SLUDGE DEWATERINQ

TEST CENTRIFUGE VS VACUUM FILTER

15	20	25	30	35

FEED SOLIDS (%TSS)

Figure 4


-------
55

ELRAMA SLUDGE DEWATERING

CENTRIFUGE VS VACUUM FILTER

50

CO

o

U

45

40

CENTRIFUGE

VACUUM FILTER

35

15

20

Vacuum Filter - 1989 to 1992
Centrifuge — 1992 to 1993

25

FEED SOLIDS (%TSS)

Figure 5

30

35


-------
RECOVERY vs DATA SETS

250 GPM FEED, CENTRATE RETURN TO THICKENERS

6/10/92 to 6/21/92

Data Sets at four hour intervals

30

DATA SETS

Figure 6


-------
Simultaneous Removal of Acidic and
Volatile Organic Gases by Spray Dryer Sorber(=SDS)

Ja-Kong Koo, *Jin-Kyu Kim, Kyung-Ryul Baek
Env. Systems Res. Lab., Dept. of Civil Engr.
Korea Advanced Institute of Science and Technology,
373-1 Kusong-dong Yusong-gu
Taejon 305-701 Korea

^Samsung Heavy Industries Co., Ltd.

Daeduk R&D Center, P.O. Box 43, Daeduk Science Town
Taejon 305-600 Korea

Abstract

To find optimum operating condition for simultaneous removal of acidic
and volatile organic gases, a laboratory scale spray dryer sorber(SDS) is
designed and operated with various experimental conditions to evaluate
the important factors affecting the removal efficiency.

The inside of SDS is divided into three regions of absorption, transition
and adsorption by temperature and humidity control, and removal
efficiency at each region is experimentally estimate! Injected sorbents
were lime slurry, activated carbon and mixture of both.

Target pollutants were SO2 as an acidic gas, and vinyl chloride and
benzene as volatile toxic gases which was relevant to solid waste
incineration

The removal of SO2 was mainly occurred in absorption and transition
region. For organic gases, it was very difficult to remove by lime slurry
only injection, but removal efficiency could be increased by injection of
dry activated carbon into adsoiption region and control of residence time.

Introduction

Dry scrubbing of the flue gas from municipal waste incineration is
becoming an increasingly viable alternative to wet scrubbing[5l It was
reported that benzene, toluene, xylenes, naphthalene, and formaldehyde of

85-1


-------
all the volatile organic compounds (VOCs) commonly emitted from solid
waste incinerators in more or less high concentrations [19,20]. Many
studies on the relation between the removal efficiency of acidic gases
and operating factors in spray dryer sorber were reported, but those on
trace organic gases (PICs ; dioxins, furans, etc.), heavy metals and
simultaneous removal of acidic gases and VOCs were insufficient

A public awareness exists in the potential adverse characteristics of
heavy metals, chlorinated dioxins, other organics and acid gases from
incineration plants[5]. The application potential of aeticvated carbon is at
present emerging from all process, in which highly toxic organic or
inorganic harmful substances are produced. At the present state of
technology, the activated carbon processes are nearly exclusively
employed in the final cleaning stage. A combination of activated carton
process with all complete cleaning stages such as wet washing process,
spray drying processes and dry lime process is possible[22].

Because semivolatile organics(dioxins/furans) move together with
particulate matters, those are removed by particulate matter removal
equipment. However, it is difficult to remove volatile organic gases by
bag filter because it is not adsorbed well on particulate matters.
Therefore, in this research, activated carbon injection into SDS was
considered on pollutants removal charateristie in SDS with the changes
of operating factors.

Spary Dry Scrubbing Process

The incinerator flue gas is adiabatically cooled at air pre-heater and
approached to the adiabatie saturation temperature by water evaporation
from the droplets. The acid gases absorb into water and react with lime
in the droplets. Typically the sluixry droplets are dried to less than a
few percent of free moisture before leaving the spray dryer, and are
transported with the flue gas to the particulate collectors, where the
additional acid gas removal takes place[3, 11, 15].

At high temperature, as the water of droplets evaporates, the lime
particles wholely surrounded by water become the partially surrounded
morphology, and then, completely dried state. Only absojption and
adsorption occur in wholely surrounded sate and completely dried sate,
respectively. At the transition region, both absorption and adsorption
occur in the interval of these two states.

85-2


-------
The type and concentration of sorbent and additives affect on removal
efficiency of gaseous poliutants(S02f HC1) in SDS[17]. Ca(OH)2, CaO,
C&CCk, NZ2CO3, NaHCOa, KOH, etc. as sorbent and CaCk6H20 as
additives are comonly used. Of these sorbents, Ca(OH)2 and CaO are
known as the most economic. CaO is transported to silo tank by
compressed air, then injected paste slaker to make Ca(OH)2
slurry(20-25%) [3,5,12,13,15].

Operating factors such as stoichiometric ratio, residence time, inlet gas
temperature, outlet temperature, and approach to saturation temperature
are very important in the selection of optimal design and operational
criteria[ll,14,16].

Removal Mechanism of Pollutants
Absorption

In the spray dryer, the finely sprayed droplets are contacted with the hot
flue gas in the range of 121-177 °C{250-350 °F). Two events occur due
to the contact of the droplets with the hot flue gas; (1) water of the
droplets evaporates and the flue gas temperature lowers; and (2) sulfur
dioxide is dissolved into the water and reacts with the calcium
hydroxide. The main reaction is shown in equation (1) [3,8,9,10,12].

Ca(OH)z(s) + SOz(g) + H2OQ) —> CaSOsJHzOCs) (1)

Some of the calcium sulfite formed reacts with oxygen( while the particle
is drying) to form calcium sulfate in the reactor.

CaSOs(s) + 1/202 —> CaS04(s)	(2)

Due to the high concentration of carbon dioxide in the flue gas, some
carbonation of the lime also occurs;

Ca(OH)2(s) + C02(g) —> CaCGsCs) + HaO	(3)

Adsorption

CaS03.1/2H20(s), CaS04.1/2H20(s), and unreacted lime arc advantage to
adsoiption of hydrophobic VOCs because they are porous [2]. SO2 in flue
gas becomes SO3 by reaction with oxygen, and sulfuric acid is formed

85-3


-------
by reaction with water vapor. Then it attaches 011 activated carbon
surface. Overall reaction is shown in equation (4) [3],

SO2 + l/202 + H2O	> (H2SQ^ads	(4)

Flue gas treatment system with lime and activated carbon injection can
be classified into three types : injection before SDS, injection into dry
ventuii system, and direrct injection into SDS. Sorbents mixture
composed of 95-97% of lime and 3-5% of activated carbon has been
used in a number of pEot scale plant for retrofit Also, the removal
efficiency of dioxins and mercury is more than 99% and 80%, which is
nearly equal in comparison with the fixed bed of granular activated
carbon[18J.

Experimental Apparatus and Methods
Experimental Equipment

Fig. 1 illustrates a schematic diagram of the spray drying system. Main
devices were composed of spray dryer sorber(carbon steel, 0100mm x
600mm height) for simultaneous removal of acidic gases (SO2, CO2) and
VOCsCCeHe, CHsCHCl), and bag filter (carbon steel+polyamide, 0100mm x
600mm height) for removal of the reaction products.

Dual-fluid nozzle was located on reactor top for well mixing between,
pollutants and sorbent slurry. Dry-bulb temperature and relative humidity
in SDS were measured by thermocouple(K-type) and humidity meter
(Hygrotest 6200) at 6 detection ports, respectively.

To find the removal efficiency at each region of SDS, six sampling ports
were installed at interval of 10 cm. Stainless steel tube and fittings of
1/8 inches were used, and designed to make retention time of gases in
SDS 12 sec. Gas pre-heater(ceramic + Sus 304, 0100mm x 1,000mm
height) with temperature controller(0-400 °C) was designed to keep the
retention time of the gas 30 sec. Dry powered activated carbon was
sprayed into the adsorption region of SDS by compressed air. The
concentrations of SO2, benzene and vinyl chloride are 2000, 275 and 103
ppmv, respectively.

85-4


-------
Experimental Methods and Analysis

To find the effect of slurry types on the simultaneous removal of acidic
and volatile organic gases, the pollutant removal charateristics in case of
lime(10wt% Ca(OH)a) slurry only and mixtures with activated
carbon(5%, 10%, 15% of hydrated lime) were analyzed at each region
absorption, transition and adsorption. The effects of dry powdered
activated carton were investigated, also.

Simulated gas was obtained by mixing nitrogen, air, and pollutant gas
from gas cylinders. Total simulated flue gas flowrate was adjusted to
40 L/min with 7 sec of retention time. Temperatures of reactor and gas
pre-heater were fixed to 200°C by using heating bend. The flowrate of
simulated gas and slurry was controlled by rotameter and flow meter,
respectively.

Thermometer and humidity meter are used at 6 sampling points for the
measurement of size variation of three regions in SDS. SO2 is analyzed
by FTIRCFourier transform infrared spectrometer ; Bomem Michelson,
MB-100). Benzene and vinyl chloride were analysed by GC/F1D
(Hewlett-Packard II). Analysis conditions are followings ; packed column
SP 1100, carrier gas flowrate is 14mL/min, and sampling volume is 20 ft
L.

Results and Discussions
Region Divisions

Inside of the SDS was divided into the three regions(adsoiption,
transition and adsorption) by variation of operating factorsCtemperature,
adiabatic saturation temperature and relative humidity). As shown in Fig.
2, absorption region was increased while adsorption region was decreased
with the increase of slurry flowrate.

If overall water in droplets exists to vapor state over 100°C, surface of
droplet is dried thus adsorption region begins at 9.5, 19 and 30.5 cm
from reactor top, respectively, when lime slurry flowrate is increased to
5, 10 and 20 mL/min. If it is assumed that adsorption region starts
when gas temperature gets to equilibrium, the beginning of the region is
in 3, 4 and 5 sampling ports with the increase of lime slurry flowrate to
5, 10 and 20 mL/min, respectively.

85-5


-------
Because the variation of wet bulb temperature in reactor was little, the
profile of adiabatic saturation temperature was similar to that of dry bulb
temperature, and the size of region was same to it

Relative humiduty was increased to saturation with increase of slurry
flowrate. Supposed that adsorption region begins when relative humidity
gets to equilibrium, its beginning is at 2, 3 and 5 sampling ports. Also,
if absorption and transition region are divided at the point which
relative humiduty is 100%, the boundary begins at 0, 8 and 26 cm with
5, 10 and 20 mL/min of slurry flowrate. That is, the size of absorption
region was increased and size of adsorption region is decreased with the
increase of slurry flowrate. Increase of the transition region size was
little and nearly constant.

SO2 Removal Test
Lime Slurry Injection

Using 10% Ca(OH)2 slurry, SO2 removal efficiency was investigated in
case of 5, 10 and 20 mL/min of slurry flowrate. As shown in Fig. 3, SO2
removal efficiencies at sampling point 1 (reactor top) and 2 were 70-75%
and 80-90%, respectively.

If removal efficiency between each sampling points was defined as the
regional removal efficiency, the maximum regional removal efficiency was
shown in the interval of sampling points 1 and 2, and the regional
removal efficiency was decreased as going to downward. It means that
SO2 removal mainly occurred in absorption region. The removal
efficiencies of absorption and transition region were increased as the
slurry flowrate increased.

SJme Added with Activated Carbon Slurry Injection

As shown in Fig. 3, there was no enhancement of SO2 removal
efficiency by the injection of activated carbon. SO2 removal efficiencies
in three regions were 78-98%, 1-20.7%, 0.6-7.2%, respectively. Therefore,
SO2 was mainly removed in absorption and transition region, and hardly
in adsorption region.

85-6


-------
Operating Factors

The effect of operating factors such as approach to saturation
temperature, relative humidity, slurry flowrate, outlet temperature were
illustrated in Fig. 4 Outlet temperature and approach to saturation
temperature were in reverse proportion to the removal efficiency, while
relative humidity and slurry flowrate were in proportion to that. It means
that relative humidity and slurry flowrate must be maintained high,
and outlet temperature and approach to saturation temperature must be
low for high removal efficiency.

The maximum removal efficiency(97%) could be obtained in this
experiment at the following condition of 20 mL/min slurry flowrate, 100
°C outlet temperature, 50 °C approach to saturation temperature and 42%
relative humidity. If the operating factors are not properly controlled,
operating troubles such as wet bottom, wastewater generation and
corrosion will be occurred.

VOC Removal
Lime Slurry Injection

Fig. 5 shows the results of VOCs removal by lime slurry injection. It
was found that lime slurry could hardly remove VOCs. Benzene was
more easily removal than vinyl chloride because of high volatility of
vinyl chloride.

Lime Added Activated Carbon Slurry injection

Fig. 6 shows the results of VOCs removal by lime added activated caron
slurry injection. Changing the slurry flowrate with 5% activated carbon,
the concentration profile had no vivid trend. In 10 mL/min of slurry
flowrate, adsorption and desoiption were observed. Removal efficiency of
bezene was increased from 3.3-7.3% in lime slurry only injection to
13.3-24% in this case, but it was not satisfactory.

In case of vinyl chloride, the similar feature was observed, but the
removal efficiency was less than benzene in both absorption and
adsorption region.

85-7


-------
Dry Powdered Activated Carbon Injection

To improve the removal efficiency of VOCs in adsorption region, dry
powdered activated carbon was injected. Fig. 7 shows that total removal
efficiencies of benzene and vinyl chloride were 55% and 36%,
respectively, which was much higher than other cases. It was due to
the enlargement of adsorptable surface and the temperature drop by
accompanied air supply.

Effect of Residence Time on VOCs Removal

Residence time was adjusted by controlling the simulated gas flowrate.
As shown in Fig. 8, the removal efficiency of bezene, in case of 10
mL/min of slurry flowrate and powdered activated carbon injection to
sampling point 3, was increased to about 70% by increase of residence
time from 7 sec to 12 sec.

When the amount of injected activated carbon is constant, the increase of
residence time makes the contact time between activated carbon and
VOCs increase, and the removal efficiency is enhanced.

Effect of Sorbent Type on VOCs Removal

As shown in Fig. 9, the removal efficiency of VOCs was highest in the
injection of dry powdered activated carbon, which was better than that
of lime added with activated carbon. Therefore, activated carbon must
be injected in dry state at lower temperature region, or the activated
carbon packed tower can be attached to the SDS effluent of lower
temperature.

Conclusions

From this study, the followings were concluded.

1. The removal of SO2 was mainly performed in absorption and
transition region, but that in adsorption region was much lower
regardless of slurry types.

85-8


-------
2.	Because the removal of SO2 and VOCs was very sensitive to
temperature, careful temperature control was required to enhance the
removal efficiency of acidic and VOCs gaseous pollutants.

3.	Because it was very difficult to remove VOCs by only lime slurry
injection, the removal efficiency could be increased if dry powdered
activated carbon was injected and residence time was properly controlled.

Reference

1.	Adu, M. A., " Control of Residuals and Emissions from Hazardous
Waste Incineration," Proceedings : The Application of US Pollution
Control Technology in Korea. 1989.

2.	Masters, K, Spray Drying Handbook John Wiley &; Sons, 1019.

3.	Kroll. P. J. and Williamson, P., "Application of Dry Scrubbing to
Hazardous Waste Incineration," TAPCA. Vol. 36, No.ll, pp. 1258-1263,

1986.

4.	Chu, P. and Rochelle, G. T,, "Removal of SOz and NOx from Stack
Gas by Reaction with Calcium Hydroxide Solids", TAPCA. VoL39, pp.
175-179, 1989.

5.	Frame, G. B., "A Comparison of Air Pollution Control Systems for
Municipal Solid Waste Incinerators", TAPCA Vol. 38, N0.8, pp.
1081-1087, 1988.

6.	Kiingspor, J. S,, "Improved Spray Dry Scrubbing through Grinding of
FGD Recycle Material," TAPCA. Vol. 37, No.7, pp. 801-806, 1987.

7.	Karlsson, H T., et al., "Activated Wet-Dry Scrubbing of SO2,"
TAPCA. Vol. 33, No.l, pp. 23-28, 1983.

8.	Muzio, L. J. and Offen, G. R., "Dry Sorbent Emission Control
Technology : Part 1. Fundamental Processes," .TAPCA. Vol. 37, pp.642,

1987.

9.	Gelter, J. L., et al., "Modeling the Spray Absoiption Process for SO2
Removal," TAPCA. Vol. 29, pp. 1270-1274, 1979.

85-9


-------
10.	Yuan C, S. and Rood, M. J., "Simultaneous Collection of SO2 and
NOx via Spray Drying : Using Sodium Basal and Calcium Based
Sorbents with Selected Additives," Ph. D. Dissertation, Universitvof
niinr»s TTrhana January 1990.

11.	Bade, K. R., "Application of High-Temperature 3-Phase Equilibrium
Distribution of Pollutant for the Simultaneous Removal of Sulfur
Dioxide and Vinyl Chloride in Dry Scrubber", Master Thesis. Korea
Advanced Institute of Science and Technology. 1990.

12.	Faiter, P. S., "Emission Control Through Dry Scrubbing,"
Environmental Progress. ,K1.5, pp.178-183, 1986.

13.	Rdsinger, A. A. and Gehri, D. C., "Two-Stage, Dry FGD and
Particulate Removal System", JAPCA. Vol.29, No.4, pp.419-421, 1979.

14 Klingspor, J, S., et al, "Acid Gas Emission:Resuts of Spray Diy
Scrubbing Pilot Plant Study", International Conference on Municipal
Waste Combustion. 1989

15.	Donnelly, J. R. and Felsvang, K. S., "JOY/NDRO SDA-FGC systems
North American and European Operating Experence," International
Conference on Municipal Waste Combustion. 9C-39 - 9C-55, 1989.

16.	Steams Roger eng., "Status of Spray-Dryer Flue-Gas
Desulfurization," EPRI-CS-2209. 1982.

17.	Chan, R., "Simutaneous Removal of SO2 and NOx in Sprayer
Scrubbing," Master Thesis. University of minds at Urbana-Champain.
1987.

18.	Clarke, M. J., "A Review of Activated Carbon Technologies for
Reducing MSW Incinerator emissions, "Municipal Waste Combustion
Conference, pp. 975-994, 1991.

19.	Hinshaw, G. D., "Evaluation of Organic Stack Emissions from A
liquid-Injection Hazardous Wate Incinerator, " Proceedings of
A&WMA Specialty Conference, pp.15-30, 1989.

20.	Walker, B. L. and Cooper, C. D., "Air Pollution Emission Factors for
Medical Waste Incinerators, " T.A&WMA. Vol.42, No.6, pp.784-791,
1992.

85-10


-------
21.	Oppelt, E. T., "Air Emission from the Incineration of Hazadous
Waste," Toxicology and Industrial Health. Vol.6, No.5, pp.23-51,

1990.

22.	Shamekhi, R., Research into Activated Carbon Technology on
Harmful Organic Substances, Heavy Metals and NOx Control,"
Proceedings of 83rd Annual Meeting & Exhibition. AWMA.
Pittsburgh, Pennsylvania, 1990.

23.	Tirey, D. A., et aL, "Comparison of Organic Emission from
Laboratory and Full Scale Thermal Degradation of Sewage Sludge,"
Hay-ardmis Waste & Hazardous Material. Vol.8, No.3, pp. 201-218,

1991.

24 Sims, R. C., "Soil Remediation Techniques at Uncontrolled

Hazardous Waste Sites : A Critical Review," T.AWMA. Vol.40,
No.5, pp. 703-732, 1990.

j

	v

1-4-L—i

Preheofer

A

A

A

Ah s(h Cells ai,n«i

©-

| DUST

—< AC 1

—Che

	u:

fcH	

LIME SUPPLY PUMP

	<^WATFR

LIME SLURRY TANK

	SYMBOLS



GAS STREAM



	LIQUID STREAM



-	SOLID STREAM



(n) flow indicates




-------
10 20 30 40 50
Distance from Top. (cm)

Fig. 2. Region Determination by Operating Factors in SDS
a) Temperature b) A.T c) RJK.

85-12


-------
"¦"/~3	i ' 20 r_ 30	<0	50	SD

Distance from SDS Top.fcm)

Fig. 3, Comparisicn of SO2 Bemoval by Lime Slurry and Lime Added
Ativatsd Carbon Slurry Injection in SDS

100-1

SCH

_ 80-

s

•S 70'

o
O

&
X

c

o

J -- -
K
<

60-
50-
40-
30-
20-
10-

•160
-140
-120

y

¦100 |

*a

¦80 a.
E

-60 -

•40

¦20

3S

90	95

Removal efficiency(%>

—rO
100

Fig. 4 Effort of Operating Factors on Removal Efficiency

85-13


-------


i

2

CH2C

j S?1	SPZ ^ SP3	SPA ^ S~5 SPS

' 0 ' Vo ' 20	20	40 ' SO

Distance vrom Tas.t'cnc:}

£0

Fig. 5. VOCs Removal by Lime Slurry Injection

Hydrated lime ©ddted A.C « 5%

10 20 30 40 50
Distance from Top. (cm)

Fig. 6. VOCs Removal by Hydrated Lime Added Activated Carbon
Slurry Injection

85-14


-------
Fig. 7. VGCs Removal by Dry Activated Carbon Powder Injection

Distance -from top. (cm)

Fig. 8. Effect of Residence Time on VOCs Removal

85-15


-------
Gs»*13mfc

Fig. 9. Effect of Sorbent Type on VOCs Removal

85-16

A


-------
NUMERICAL MODELS FOR S02 ABSORPTION BY DUCT INJECTION

W. J. Oberjohn
S. W. Burge

Babcock & Wilcox — Research and Development Division
1562 Beeson Street
Alliance, Ohio 44601

Abstract

This paper describes an empirical correlation and two numerical models that predict
SOz absorption by a lime slurry sprayed into the flue gas between the air preheater and
the solids collection device. A one-dimensional (ID) and a three-dimensional (3D)
model were developed during a U.S. Department of Energy (DOE) program to bring
duct injection technology to die point of commercial readiness. The models were eval-
uated by comparing predictions to data obtained from the DOE 12-MW Duct Injection
Test Facility and from a 6-foot square humidification duct A comparison of ID, 2D,
and 3D model predictions indicates that a ID model is adequate for predicting SO2 ab-
sorption and droplet evaporation. Wall deposition which occurs at low approach to
adiabatic saturation temperatures can be estimated reasonably well with a 2D model,
while a 3D model is required to evaluate non-symmetrical atomizer arrangements and
gravitational effects. S02 absorption and slurry droplet evaporation predictions are
presented for the DOE Duct Injection Reference Plant.

Introduction

The duct injection process is a low capital cost S02 control technology for power plants.
A calcium-based sorbent is injected into the flue gas between the air preheater and the
solids collection equipment The sorbent can be injected dry or as a water-based slurry.
This paper describes mathematical models for predicting S02 removal for the Duct
Spray Drying (DSD) process (Figure 1) in which slaked pebble lime is (optionally)
mixed with recycled solids collected in the ESP to form a slurry. The slurry is sprayed
into the duct through dual-fluid atomizers using compressed air to atomize the slurry.
The water in the slurry cools and humidifies the gas to a specified approach to adiabat-
ic saturation temperature. The adiabatic saturation temperature is the temperature
which results from adding water adiabatically to the flue gas until it becomes saturat-
ed, with the water added at the final mixture temperature. Calcium utilization, which
dominates operating costs, is enhanced by recycling the unreacted calcium captured in
the collection equipment and by operating at duct outlet temperatures that are close to
the adiabatic saturation temperature of the flue gas. Deposition of sorbent and reaction
products on the duct walls at low approach temperatures is the primary factor limiting
calcium utilization.

86-1


-------
In 1988, the DOE Pittsburgh Energy Technology Center (DOE-PETC) initiated a 5-year
Duct Sorbent Injection Technology Development Program1. The goals were to:

•	Use lime-based sorbents to remove 50 to 60% of the S02 produced from coal-fired
power plants

•	Produce an environmentally safe, dry waste product

•	Achieve these results at a cost less than $500 per ton of S02 removed

As the final phase of a five-element program, Raytheon Engineers and Constructors
(RE&Q has developed engineering and design criteria for the full-scale commercial
application of duct injection based on research conducted in the previous phases of the
program and technology developed for dry scrubbers2. As a subcontractor to RE&C,
one of Babcock & Wilcox's tasks was to evaluate and enhance mathematical models of
the duct injection process developed by previous contractors.

This paper summarizes the most significant results of the model development and eval
uation conducted by Babcock & Wilcox (B&W). The models include:

•	A physically based correlation of the S02 absorption data obtained with slurry in-
jection at the DOE Duct Injection Test Facility (DITF)

•	A slurry droplet model for predicting evaporation and S02 absorption

•	A one-dimensional energy and mass balance model of the duct injection process
(DIAN1D) incorporating the slurry droplet model

•	A three-dimensional mass, energy, and momentum balance model of the duct in-
jection process incorporating the slurry droplet model (DIAN3D)*

•	A Lagrangian stochastic droplet dispersion submodel for DIAN3D

In addition to the models, B&W also developed DIANUI, a preprocessor for DIAN1D,
and a postprocessor for DIAN3D.

Model Description

S02 Correlations — Correlations were developed for S02 removal measured be-
tween the duct inlet and the ESP inlet and between the duct inlet and the ESP outlet
during slurry testing at DOE's Duct Injection Test Facility (DITF) at Ohio Power's
Muskingum Power plant located at Beverly, Ohio. The DITF test data, run numbers,
and operating conditions referenced in this paper are taken from the draft report of the
DITF test program3. Hie slurry data at inlet S02 concentrations between 1800 and 2200
ppm during Test Weeks 51-84 were correlated based on a model developed by
Harriott4.

* The three-dimensional model is an enhancement of DIAN3D developed by Adaptive
Research Corporation during an earlier pahse of the DOE development program.

86-2


-------
Figure 1. In-Duct Sorbent Injection Process Schematic

The correlation lias the following form:

Esc>2 = 1 - exp(-X)

where:

X = B1 (Ca/S)/(B2 + Ca/S)ln((Tin - T^/OW - Tas))
and

ES02 =

fraction of inlet S02 removed, moles/mole

Ca/S =

calcium-to-sulfur ratio of the slurry at the atomizer, moles/mole

TW =

inlet temperature

Tout =

outlet temperature

T =

xas

adiabatic saturation temperature

X

overall number of mass transfer units for S02

By employing a number of assumptions, the parameters, Harriott derived expressions
for B1 and B2 from heat and mass transfer considerations. However, Harriott recom-
mended that B1 and B2 be considered as adjustable parameters to be determined from
experimental data. Harriott has also suggested that B2 is proportional to the square
root of inlet S02 concentration5. However, a better fit of the DITF data was obtained
when B2 was assumed to be a linear function of inlet S02 concentration (Yso2, ppm).

The constants determined from a regression analysis of the data set for S02 removal
across the duct as measured at the ESP inlet are:

86-3


-------
B1 = 0.65534 ±0.24465	(95% confidence interval)

B2 = 0.00045486 Ysc^ ±0.00049	(95% confidence interval)

Standard error in Eso2, o = 4.26%; Correlation coefficient, R2 = 0.9955

The correlation of the ESP inlet data matches the test data reasonably well as indicated
by the standard error and the correlation coefficient

The constants determined from a regression analysis of the data set for S02 removal at
the ESP outlet are:

B1 = 1.10990 ±0.62343	(95% confidence interval)

B2 = 0.00074452 Ysoj ±0.00089858 (95% confidence interval)

Standard error in Eso2, o = 9.63%; Correlation coefficient, R2 = 0.9882

Over the range of operating conditions of the D1TF, an examination of the differences
between the correlation and individual data points indicates that S02 removal by slurry
injection can be summarized as follows.

•	Unreacted Ca(OH)2 in the recycle stream is as reactive as fresh Ca(OH)2.

•	The correlation applies equally well over the entire range of approach tempera-
tures, solids concentrations, and recycle ratios included in the final data set

•	The addition of CaCl2 at amounts up to 3.6% as measured in the dry ash at the ESP
outlet has no apparent effect on S02 removal*.

•	S02 removal measured at the ESP outlet averages about 25% greater than at the
ESP inlet for the entire range of test conditions.

•	S02 removal at the ESP inlet and outlet is larger than predicted by the correlations
at high removal rates, and smaller at low removal rates.

•	Data obtained with continuous recycle (rather than batch mode operation) indicate
that S02 removal rates as high as 88% can be obtained at a calcium utilization of
50% and an approach temperature of 25 °F. This is about 14% higher than predict-
ed by the ESP outlet data correlation. However, data scatter at high removal effi-
ciencies adds uncertainty to this conclusion.

Slurry Droplet Model —The slurry droplet model is an extension of the model devel-
oped by Professor Harriott and his coworkers at Cornell University6'7. Sulfur dioxide
and calcium hydroxide react within the slurry drop according to:

S02 + Ca(OH)2 => CaS03 • 0.5 H20 + 0.5 H20

* CaCl2 should be more effective at lower approach temperatures.

86-4


-------
S02 absorption and water evaporation occur simultaneously. The model includes the
resistance to diffusion of a product layer that is assumed to form on the outside of the
droplet, the dissolution of calcium hydroxide in the droplet core and the diffusion of
calcium hydroxide from the droplet core to a reaction front located either at the droplet
surface or within the product layer.

The model describes five stages in the mass transfer processes that occur as a water
drop containing Ca(OH)2 evaporates and absorbs S02 from a flue gas stream (Fig-
ure 2). As water evaporates from the droplet, S02 diffuses through the gas film to the
droplet surface, is absorbed, and diffuses through the water to a reaction front inside
the droplet. Dissolved lime from the solid lime particles diffuses through the water to
the reaction front. The S02 and lime react to form a solid product and water. After
enough water has evaporated from the droplet to make the lime particles in the interior
touch, a layer of dry solid forms which creates a pore diffusion resistance for both wa-
ter and S02. Eventually, all the water evaporates or the core of the droplet may retain
residual moisture if the water contains dissolved salts which reduce the vapor pressure
of the solution. In the absence of water, S02 absorption stops. If water is retained in
the core, absorption continues if unreacted lime is present. During these mass transfer
processes, heat is transferred to the droplet from the flue gas. However, the water
evaporation rate is high enough during most of the process that droplet temperature is
very dose to wet bulb temperature. Some important assumptions about the processes
include:

•	Due to the high slurry viscosity, there is no circulation within the droplet. Conse-
quently, there are concentration gradients in dissolved lime and S02. In addition,
lime particles only move within the droplet due to growth of a product layer.

•	Only the dissolved lime in solution reacts to absorb the S02. Unreacted lime that
may be trapped in a "dry" product layer does not react with the diffusing S02.

•	Recycled Ca(OH)2 is as effective as fresh sorbent and recycled reaction products
are inert. Flyash reactivity is ignored.

The model has been formulated with five physically based, adjustable parameters that
were statistically optimized to achieve agreement with the DITF slurry injection data.

D1AN1D —DIAN1D is a one-dimensional model that describes energy and mass trans-
fer for slurry droplets injected into a flue gas stream by two-fluid atomizers. DIAN1D
incorporates the slurry droplet model described previously. Although DIAN1D is one-
dimensional, it includes an empirical relationship for jet entrainment that accounts for
the rate that flue gas mixes with the two-phase mixture of slurry drops and atomizing air.

DIAN1D determines humidification and S02 absorption assuming no relative velocity
between the droplets and the flue gas. The model is derived by assuming plug flow
downstream of an atomizer as depicted in Figure 3. The atomizing jet is assumed to
expand and entrain flue gas along the duct length. Eventually the jet expands to fill the
duct. The droplets enter the duct at the atomizer, remain within the expansion cone of

86-5


-------
the jet, and move down the duct with the expanding gas. Entrainment and flow within
the expansion cone are special features of DIAN1D usually not included in ID models.

Stage II

diffusion in gas and
liquid important

gas/liquid interface
reaction front

product layer

Stage III

dry zone forms,
particle size fixed

gas/liquid interface
reaction front

product layer

Stage IV

gas film and dry
zone control

.gas/liquid interface and
reaction front

product layer

Stage V

evaporation stops,
reaction in core

gas/liquid interface
reaction front

product layer

Figure 2. Slurry Droplet Model

Figure 3. DIAN1D — One-Dimensional Model
86-6


-------
DIAN1D is based on energy and mass conservation applied to the flue gas control vol-
umes within the jet cone shown in Figure 3. Each control volume contains flue gas and
the slurry droplets that evaporate and absorb S02. Flue gas and droplet conditions are
predicted along the duct using a standard explicit marching procedure. The S02 mass
absorbed and the water mass evaporated to the flue gas are determined by solving the
flue gas and droplet mass and energy equations for each droplet size. The timestep is
adjusted to achieve specified solution tolerances of mass and temperature change for
the droplet. The predicted droplet mass and energy become initial conditions for the
subsequent droplet timestep. The entrainment flow rate into the control volume is cal-
culated as a linear function of distance from the atomizer. The length required to en-
train all the flue gas is a specified input to the model

DIAN3D—DIAN3D is a three-dimensional model that describes the momentum, ener-
gy, and mass transfer occurring in the duct injection process. The 3D model allows
DIAN3D to calculate the effect of atomizer spacing and wall clearance on deposition.
DIAN3D has the ability to model:

•	Multi-dimensional (2D or 3D) gas flow

•	Cartesian or cylindrical polar coordinate systems

•	Laminar or turbulent flow

•	Droplet dynamics using a Lagrangian approach

•	Heat and mass transfer between the gas phase and slurry droplets or inert particles

•	S02 removal

The governing conservation equations for the gaseous phase are formulated in an
Eulerian reference frame and the equations for droplet motion are formulated in a La-
grangian frame of reference. The time-averaged Navier-Stokes equations governing the
transfer of fluid momentum, energy and mass are written in a general form as de-
scribed by Patankar8. The equations are fully elliptical to allow treatment of flows with
thick shear layers or separated flow regions. The effective turbulent viscosity for the
momentum equations is evaluated using the k-e turbulence model9. The diffusion coef-
ficients for turbulent mass and energy transport are related to the effective viscosity by
turbulent Schmidt and Prandtl numbers which are assumed to be constant throughout
the flow field.

The equations of motion for slurry droplets and inert particles are written in a Lagrang-
ian frame of reference. The droplets exchange energy, mass, and momentum with the
gas phase. The transport processes are described by the particle-source-m-cell model10.
Droplet agglomeration is assumed to be insignificant

The partial differential equations representing the gaseous phase are solved numerical-
ly. The solution domain is divided into a finite number of control volumes. Algebraic
equations representing the unknown variables at a point within each control volume
are obtained by integrating the conservation equations over the control volume. The
control volumes for the momentum equations are displaced from the control volumes

86-7


-------
for the energy and mass conservation. This "staggered" grid system is used in conjunc-
tion with a pressure correction equation to solve for a velocity field which conserves
both momentum and mass11. Either the SIMPLE or SIMPLEC12 algorithms can be used
to solve for pressure and velocity.

DIAN3D allows individual atomizers to be represented as point sources of mass, mo-
mentum, and energy thereby eliminating the need for detailed atomizer models requir-
ing a large number of control volumes. Hie momentum source term must be
determined by a separate calculation or from thrust measurements on an actual atomiz-
er. A detailed description of DIAN3D has been provided by Vlachos, et al.13.

Droplet Dispersion Model—Droplet dispersion has an important effect on the duct
injection process. As the slurry droplets move down the duct, they disperse due to the
expanding jets from the atomizers and from turbulence in the flue gas. Realistic predic-
tions of this phenomena are essential for estimating mixing of the slurry drops with the
flue gas, S02 conversion efficiency, and the potential for wall deposition. Initially,
DIAN3D did not include the effect of gas turbulence on dispersion. As a result, the
drops did not spread out to the wall as observed during testing. A droplet dispersion
model was added to account for fluid turbulence.

Hie model is based on a stochastic approach14'15. The gas velocity fluctuations experi-
enced by the droplets arise from turbulent eddies in the flow field. The eddies are a
transient phenomena that develop and dissipate in a somewhat random manner. How-
ever, the velocity fluctuations do not affect the mean flow. The k-e turbulence model is
used to estimate the intensity of the fluctuations along with the length and time scales
of the an eddy. During the interaction with an eddy, droplet motion is predicted from
the solution to the Lagrangian momentum equations. At the end of the eddy lifetime or
once the droplet has moved through the eddy, new velocity fluctuations are generated
and the momentum solution continued. A Gaussian probability distribution with a
mean of zero is assumed for the fluctuating velocities. The time scale of the turbulence
is estimated using procedures described by Milojevic16. The length scale of the turbu-
lence is determined from the time scale and a velocity based on turbulent kinetic ener-
gy. The velocity fluctuations are applied until the eddy lifetime is exceeded or until the
distance traveled by the particle relative to the flue gas is greater than the eddy size.
When the particle moves into an adjacent fluid control volume, either a new eddy is
generated or the remaining time and length scales are adjusted according to the new
local conditions and the velocity perturbations are scaled accordingly. The results pre-
sented in this paper were calculated by generating a new eddy each time a droplet en-
tered a new control volume.

Model Optimization and Evaluation

Slurry Droplet Model—The model of a lime-water slurry droplet is based on funda'
mental heat and mass transfer relationships. However, several assumptions are re-

86-8


-------
quired to describe the physical characteristics of a droplet as it evaporates and absorbs
S02 from the gas stream. Since these characteristics are not directly measured, five
model parameters are defined and have been optimized to fit the DITF data.

The five model parameters are:

•	Fraction of initial inerts trapped in product layer

•	Hindrance factor for diffusion in the product layer

•	Void fraction in core

•	Void fraction in product layer

•	Lime particle diameter

A response surface method was used to optimize the S02 model parameters. A data set
covering the range of DITF operating conditions was generated from a correlation of
the DITF slurry data obtained from Week 28 through the end of the test program. The
operating conditions included all combinations of the following variables subject to the
constraint of a minimum overall calcium-to-sulfur (Ca/S) ratio of 0.9:

-	Approach Temperature - 20°, 35°, SOT - Solids Concentration- 18,28,36%

-	Mass Fraction Recycled - 0,0.3,0.6	- Inlet S02-	2000 ppm

Recycle is defined as the mass fraction of the total solids exiting the duct that is recy-
cled. The overall Ca/S ratio was determined from the approach temperature, the solids
concentration, and the recycle fraction using an iterative heat and mass balance calcula-
tion. The unreacted Ca(OH)2 in the recycle stream was calculated based on S02 remov-
al predicted by the ESP inlet S02 correlation. The drop size distribution was assumed
to be the same for each case and was calculated from a correlation of atomizer test
data17. The mass mean diameter was 43 microns and the droplet streams were divided
into eight size bins.

S02 removal was predicted with DIAN1D at all 27 combinations of each operating con-
dition and at three levels of each model parameter for a total of 6561 predictions. The
middle value of the three levels selected for the model parameters was based on an ini-
tial best estimate. The upper and lower limits were selected so that the optimum pa-
rameters would be expected to fall within these limits. Entrainment was calculated by
DIAN1D based on the assumption that the jet diameter and the entrained mass increase
linearly with distance from the atomizer. The flue gas was assumed to be completely
entrained when the jet diameter equalled the duct diameter. The initial diameter of the
jet was based on a jet velocity calculated from a correlation of atomizer thrust data18
and the mass flow and density of the atomizing air. The DITF duct was represented as
a cylindrical duct with an area equivalent to one-sixth the total duct area (the area asso-
ciated with one of the six atomizers). The response surface method was used to deter-
mine sets of the five model parameters that yielded saddle points in the response
surface of the difference between the predicted and measured S02 removal.

86-9


-------
A set of model parameters that provided title best results is:

•	Fraction of Initial inerts trapped in product layer

•	Hindrance factor for diffusion in the product layer

•	Void fraction in core

•	Void fraction in product layer

•	lime particle diameter

2.83 microns

0.525
0.417
0389
0.443

The hindrance factor and the product layer void fraction have the most significant ef-
fect on predicted SOz removal. The three other parameters have a relatively small ef-
fect. It is likely that physically reasonable values could be selected for these three
parameters, and the model could be optimized with only the hindrance factor and the
product layer void fraction as adjustable parameters.

DIAN1D predictions of the effect of solids concentration on drying time were com-
pared with predictions presented by Harriott19. The drying times are based on the S02
model with the optimized model parameters, except for the void fractions in the core
and product layer which were set at 0.50 to agree with the values assumed by Harriott.
DIAN1D predictions for water and a slurry droplets at an inlet temperature of 300T,
an outlet temperature of 150 T, and an adiabatic saturation temperature of 126 T agreed
well with results presented by Harriott

S02 removal predictions made with DIAN1D and the slurry droplet model were com-
pared to the DITF slurry injection data. Removal across the duct was predicted with a
standard error of 3.3%. The predictions were equally good for data obtained with and
without recycle, justifying the assumption that recycled Ca(OH)2 is as reactive as fresh
lime. The mass fraction of inert reaction products in the slurry had little effect on calci-
um utilization.

For short residence times on the order of the droplet drying time, the model was able to
adequately predict the effect of all the independent test variables except for inlet SOz
concentration. However, the limited amount of data at concentrations other than the
nominal 2000 ppm of the DITF tests introduces significant uncertainty about the effect
of inlet S02.

The model does not predict the 25% increase in S02 removal observed at longer resi-
dence times characteristic of the ESP (5 to 10 times the droplet drying time). This dis-
crepancy probably occurs because the reactivity of the unreacted Ca(OH)2 in the dry
product layer is completely ignored. S02 removal measured during dry injection of
Ca(OH)2 in a humidified flue gas indicates that significant S02 absorption occurs at
high relative humidity. The unreacted Ca(OH)2 in the dry product layer should also be
reactive, especially while evaporation is occurring or when the approach temperature is
low. This mechanism for S02 removal must be added to the model before it can be
used to predict S02 removal at conditions different than those used to optimize the
model parameters.

86-10


-------
DIAN1D —Sensitivity analyses performed with DIAN1D showed that poly-disperse
drop size distributions typical of the two-fluid atomizers used for slurry injection must
be represented by at least eight discrete size bins. Entrainment rate has a significant
effect on evaporation and calcium utilization predictions. One-dimensional calcula-
tions that assume instantaneous mixing can be significantly in error.

DIAN1D was evaluated by comparing predicted droplet evaporation rates with data
obtained from the 6x6 Humidification Duct and S02 absorption data from the DITF.
The predictions and data are in good agreement.

A comparison of DIAN1D predictions and DITF test results indicate that, up to some
point, large slurry droplets containing most of their original water content can impact
the duct walls without causing deposits to build up. It is likely that heat transfer from
the flue gas next to the walls is sufficient to evaporate the water in the drops striking
the wall until a critical impaction rate is reached. At this point, deposition increases
dramatically. Experimental data are required to define these conditions and a heat
transfer model should be developed that can represent this phenomena well enough to
provide a basis for design.

DIAN1D used in conjunction with its preprocessor, DIANUI, and the slurry droplet
model is an easy-to-use tool that can provide the designer of a duct injection system
with quantitative information about the effect of atomizer characteristics and duct inlet
conditions on slurry droplet evaporation and S02 absorption.

DIAN3D—DIAN3D was evaluated by comparing predictions of velocity, turbulent
intensity, and jet width to 6x6 Humidification Duct data20 obtained specifically for vali-
dating DIAN3D. The predicted velocity profiles and velocity fluctuations agree very
well with the data (Figure 4). The agreement between predicted and measured
unevaporated water at the duct exit is also acceptable. The overall agreement with the
data indicates that droplet transport to the duct walls and flue gas velocity and turbu-
lence can be predicted well enough by DIAN3D to evaluate the effect of atomizer ar-
rangement on droplet dispersion and deposition.

Droplet Dispersion Model—The dispersion of inert particles in a plane confined
jet^ was predicted to benchmark the model. The boundary conditions were based
on data collected during the dispersion experiments. The predicted and measured par-
ticle void fraction and gas velocity at several locations downstream of the jet were in

good agreement Including the measured initial fluctuating component of the trans-
verse velocity of the particles as a boundary condition was found to be essential to pre-
dict the observed dispersion in the plane jet.

86-11


-------
Average Velocity (ft/sec)

Axial Velocity Fluctuation (%)

12CT

as-

40 ^3s3^£C00£O599e9eeee©e^,,

10	30

duct width, indies

DIAN3D Predictions



loO°o

_ o

X=105"

X=3S*

X=9"

oooo

oo©oo

o

"o^b,



oo

<8

°o°

10	30	50	70

duct width, inches

O 6X8 Data

Figure 4. Comparison of DIAN3D Predictions With 6x6 Test Data

Comparison of 1D, 2D, and 3D Predictions

A one-dimensional model is often adequate for predicting SOz removal and droplet
evaporation. To illustrate this, ID, 2D, and 3D predictions were compared for DITF
Run 67SR01. Hie operating conditions for this run are:

*	Temperature - 307T

*	S02 - 2041 ppm dry

*	H20 - 8.44%

•	Approach Temperature - 39°F

•	Recycle Mass Fraction - 0.543

•	Ca/S Overall -1.48

86-12


-------
The ID predictions were made with DIAN1D, and the 2D and 3D predictions were
made with DIAN3D. Entrainment was calculated by DIAN1D based on the assump-
tion that the jet diameter and the entrained mass increase linearly with distance from
the atomizer. The jet was assumed to expand at a half-angle of 5.75" and the flue gas
was assumed to be completely entrained when the jet diameter equalled the duct diam-
eter. For both the ID and 2D predictions, the DITF duct was represented as a cylinder
with an area equivalent to one-sixth the total duct area (the area associated with one of
the six atomizers). For all three predictions, the initial diameter of the jet was based on
a jet velocity calculated from a correlation of atomizer thrust data23 and the mass flow
and density of the atomizing air. The duct walls were assumed to be adiabatic and
droplets were not permitted to "stick" to the walls.

Figure 5 compares the predicted average flue gas temperature for the ID, 2D, and 3D
models. The ID and 3D predictions agree quite well, while the 2D model predicts that

the initial rate of change in temperature downstream of the atomizers is slower than
predicted by the ID and 3D models. Even so, the overall agreement is good.

The predicted total evaporation and calcium utilization for all three models are in rea-
sonable agreement.

Figure 6 compares the flue gas H20 and S02 average mass fraction as a function of dis-
tance from the atomizers as predicted by the 2D and 3D models. Hie agreement Is
good with the 2D predictions lagging the 3D predictions just downstream of the atom-
izers. The results of this comparison indicate that DIAN1D is able to predict S02 re-
moval and droplet evaporation accurately enough for many applications.

Effect of Droplet Dispersion and Atomizer Spray Angle

DIAN1D adequately predicts S02 absorption in the duct injection process as long as the
flow in the duct can be represented by the expanding jet model assumed by DIAN1D.
However, droplet dispersion cannot be predicted with ID models. The relative impor-
tance of droplet dispersion was evaluated by comparing 2D predictions for Run
67SR01, with and without dispersion. As expected, dispersion causes the average
temperature to drop more rapidly and reduces the difference between the wall and
centerline temperatures (Figure 7). Since dispersion shortens the average droplet life-
time, S02 absorption is also reduced (Figure 8).

Run 67SR01

ID

99.4
31.9
169.6

2D

99.9
30.2
171.9

3D

99.6
29.2
171.6

% Evaporated
% Calcium Utilization
Exit Temperature, *F

86-13


-------
©

320
| 300
S 280
® 260
| 240
I- 220
8 200
g5 180
i 160
© 140
« 120

§ -20 0 20 40 60 80 100
<	Distance from atomizers, ft

Figure 5. Predicted Average Flue Gas Temperature (Run 67SR01)

c
JS

o
ta

li.

05

m
G3

0.04

0

CM

X
©

S

1

0.03

0.02

$02-









«/ ^	

mf





mf







mf







ml





H20-

Sf







	2D - 3D





.

.



0.002 c
o

Q

s

UL

0»
10

ro

0.001

-20

20 40 60
Distance from atomizers, ft

80

100

CM

o

UJ

CD

o>

s

>
<

Figure 6. Predicted Average HzO arid S02 Mass Fractions (Run 67SR01)

Dispersion — No Dispersion

•average
rwali

20 40 60
Distance from atomizers, ft

100

Figure 7. Predicted Effect of Dispersion on Average Flue Gas Temperature (Run 67SR01)

86-14


-------
c
o

B

(0

CO

to
CO

0.04

S02-

O

CJ
X

0
OJ

<5

©
>
<

0.03

H20-

0.02
-20

Dispersion

¦ No Dispersion

0.002 g
o

03

u.

m
to
OS

0.001

20

40

60

80

100

Distance from atomizers, ft

s

03
©
§>

o
>
<

Figure 8. Predicted Effect of Dispersion on H20 and S02 Concentration (Run 67SR01)

Atomizer spray pattern can affect dispersion and consequently S02 absorption and wall
deposition. The effect of spray pattern was evaluated by comparing 2D and 3D predic-
tions for DITF Run 77SL03. The operating conditions for this ran are:

•	Temperature - 309°F	• Approach Temperature - 24.5 °F

•	S02 - 2084 ppm dry	• Recycle Mass Fraction - 0.0

•	H20 - 939%	• Ca/S Overall - 2.22

The spray cone half angles for the 2D predictions were 1*, 3.5°, and 14°. The spray
cone half angle for the 3D prediction was 14°.

The 2D predictions of droplet volume fraction at the duct wall are presented in Fig-
ure 9. As expected, the drops reach the wall more quickly for the 14° spray angle than
for the 3.58 and 1 ° angles. Spray pattern is predicted to have only a small effect on flue
gas temperature at the wall for the 2D model (Figure 10).

The DITF duct was 39.875 inches high by 49.875 inches wide. The atomizers were ar-
ranged in two rows with three atomizers in each row. The top row was 8.5 inches from
the top of the duct and the bottom row was 19.875 inches from the bottom. As expect-
ed, the 3D temperature predictions showed that the flue gas temperature at the top of
the duct on the vertical centerline began to decrease before the temperature at the bot-
tom (Figure 11).

86-15


-------
Distance from atomizers, ft

- * 1 deg Cone Angle — 3.5 deg Cone Angle —— 14 deg Cone Angle

Figure 9. Predicted Effect of Atomizer Spray Pattern on Droplet Volume Fraction at the

Duct Wall (Run 77SL03)

40	60

Distance from atomizers, ft

100

1 deg Cone Angle

3£ deg Cone Angle —— 14 deg Cone Angle

Figure 10. Predicted Effect of Atomizer Spray Pattern on Fiue Gas Temperature at the

Duct Wall (Run 77SL03)

«~ 3D Bottom — 3D Top

2D 14 deg

40	60

Distance from atomizers, ft

100

Figure 11. Comparison of 2D and 3D Predictions of Flue Gas Temperature at the Duct

Wall (Run 77SL03)

86-16


-------
Duct Injection Reference Plant

S02 removal and slurry droplet evaporation were predicted for the DOE Duct Injection
Reference Plant with DIAN1D. The calculations were made at a 40 "F approach
temperature, a fresh lime stoichiometry of 1.18, and a ratio of recyded to total solids of
45%. The droplet residence time was 1.5 seconds resulting in a predicted S02 removal
across the duct of 47.7%, almost identical to the 47,9% calculated from the DITF data
correlation. Based on DITF test data, an additional 25% removal was assumed to occur
within the ESF resulting in a total removal of 56.6%. The net result is an overall calci-
um utilization of 50.5%. The moisture remaining in the drops at the exit of the duct
was calculated to be less than 0.5%, indicating that deposition should not be a limiting
factor at these operating conditions.

Summary

The purpose of the models described in this paper is to provide designers with meth-
ods lor making quantitative predictions of deposition, evaporation, and SOz absorption
for the duct injection process. This goal has been largely met, but additional develop-
ment and evaluation is required. The empirical correlation can be used to estimate S02
removal when the flue gas and sorbent conditions are in the same range as the data on
which the correlation is based. The one-dimensional model, DIAN1D, is an easy-to-
use tool that can provide the designer of a duct injection system with quantitative infor-
mation about the effect of drop size distribution and duct inlet conditions on slurry
droplet evaporation and S02 absorption, as long as the flow in the duct can be repre-
sented by the expanding jet model assumed by DIAN1D. However, droplet dispersion
and wall deposition cannot be predicted with ID models. A 2D model can provide a
reasonable estimate of how flue gas conditions at the duct walls are affected by droplet
dispersion. These conditions are important for evaluating the possibility of deposition.
If the atomizers are arranged in a non-symmetrical pattern or if the effect of gravity is
significant, a 3D model should be used to predict conditions at the duct wall.

The predictions presented in this paper are based on the assumption that wall heat loss
is negligible and that the droplets rebound from the duct wall without sticking. If heat
loss is significant or if the droplets stick to the wall, flue gas temperature at the wall
could be lowered to the point where deposition becomes a problem.

The empirical correlation and the S02 model should be compared to data obtained over
a wider range of conditions. In particular, comparisons should be made at inlet S02
concentrations significantly different than 2000 ppm and at approach temperatures less
than 25'°F. Limited DITF data indicate that the correlation may not be satisfactory at
these conditions. S02 absorption also appears to be underpredicted by the slurry drop-
let model at low approach temperatures. Hie reactivity of unreacted Ca(OH)2 in the
product layer should be added to the slurry droplet model to increase reactivity at low
approach temperatures.

86-17


-------
Acknowledgments

The authors would lite to acknowledge the important contribution of Professor Peter
Harriott, Cornell University, who defined the S02 slurry droplet model presented in
this paper, and Bill Mahaffey, Adaptive Research Corp., who assisted B&W with the
modifications to DIAN3D. We also acknowledge the cooperation of the Gilbert/Com-
monwealth and Southern Research team who provided B&W with the Dill1 test date.

References

1.	"Duct Injection Technology Approaches Commercialization," PETC Review,
Spring 1993.

2.	R. L. Claussen, C. E. Martin, P. V. Smith, W. J. Oberjohn, and G. E Weber, "Engi-
neering and Design Guidelines for Duct Injection Retrofits," 1993 S02 Control
Symposium, Boston, MA, August 1993.

3.	Felix, L. G., Gooch, J. P., Merritt, R. L., Klett, M. G., Demian, A. G., and Hunt, J. E.,
"Scaleup Tests and Supporting Research for the Development of Duct Injection
Technology," Draft Final Report, SRI-ENV-92-633-6715-F, July 27,1992.

4.	Harriott, P., "A Simple Model for S02 Removal in the Duct Injection Process," J.

Air Waste Management Association, July 1990.

5.	Harriott, P., personal communication, 1992.

6.	Harriott, P., and Kinzey, M., "Modeling the Gas and liquid Phase Resistances in
the Dry Scrubbing Process for S02 Removal," presented at the Pittsburgh Coal
Conference, September 1986.

7.	Kinzey, M., 'Modeling the Gas and Liquid-Phase Resistances in the Dry Scrubbing
Process for Sulfur Dioxide Removal," Ph.D. Thesis, Cornell University, 1988.

8.	S. V. Patankar, Numerical Heat Transfer and Fluid Flow, Washington, New York,
London: Hemisphere Publishing Corporation, 1980, p 15.

9.	Launder, B. E., and Spalding, D. B., "The Numerical Computation of Turbulent
Flows," Computer Methods in Applied Mechanics and Engineering, Vol. 3, pp. 269-289,
1974.

10.	Crowe, C. T., Sharma, M. P., and Stock, D. E., "The Particle-Source-in-Cell (PSI-
CELL) Model for Gas-Droplet Flows," Trans. ASME Journal of Fluids Engineering,
Vol. 99, pp. 325-332,1977.

86-18


-------
11.	S. V. Patankar, Numerical Heat Transfer and Fluid Flow, Washington, New York,
London; Hemisphere Publishing Corporation, 1980, p 118.

12.	Van Doormaal, J. P., and Raithby, G. D., "Enhancements of the SIMPLE Method for
Predicting Incompressible Fluid Flows," Numerical Heat Transfer, Vol. 7, pp. 147-
163,1984.

13.	Vlachos, N. S., Mahaffey, W. Av and Daley, P. L., fundamental Investigation of
Duct/ESP Phenomena, Topical Report No. 6, First Generation Duct Injection Mod-
el/' Adaptive Research Corporation Report 4770/12, Revised November 1992.

14.	Azevedo, J. L. T., and Pereira, J. C F, "Predictions of Gas-Particle Turbulent Free
or Confined Jet Flows," Part. Part. Syst. Charact., 7, pp. 171-180,1990.

15.	Milojevic, D., and Borner, T., "Third Workshop on Two-Phase Flow Predictions,"
Belgrade, June 1986.

16.	Milojevic, D., "Lagrangian Stochastic-Deterministic (LSD) Predictions of Particle
Dispersion in Turbulence," Part. Part. Syst. Charact., 7, pp. 181-190,1990.

17.	Ilan, R., and Bailey, R. T., "Duct Injection Technology Prototype Development,
Nozzle Development Subtask 4.1: Atomizer Specifications for Duct Injection
Technology, Topical Report 8," Alliance, Ohio: Babcock & Wilcox, Alliance Re-
search Center Report RDD:92:4572-57-01:01, February 1992.

18.	Ilan, R., and Bailey, R. T., ibid.

19.	Harriott, P., "Drying Times for Slurry Droplets in the Duct Injection Process,"
unpublished report, 1991.

20.	Holmes, M. J., Bailey, R. T., and Ilan, R., "Duct Injection Technology Prototype
Development, Subtask 4.3.2: Nozzle Array Tests, Topical Report 10B," Alliance,
Ohio: Babcock & Wilcox, Alliance Research Center Report RDD:92:4572-57-01:02,
February 1992.

21.	Borner, T., Durst, F., and Milojevic, D., "Second Workshop on Two-Phase Flow

Predictions/' Erlangen, May 1985.

22.	Milojevic, D., and Borner, T., 'Third Workshop on Two-Phase Flow Predictions,"
Belgrade, June 1986.

23.	Ilan, R., and Bailey, R. T., ibid.

86-19


-------
A


-------
OPTIMIZING ELECTRIC UTILITY AIR TOXICS COMPLIANCE
WITH OTHER TITLES OF THE CLEAN AIR ACT1

A.P. Loeb and D.W. South

The submitted manuscript has hem authored
by a contractor of the U.S. Government
under contract (Mo. W-31-109-ENG-38.
Accordingly, she U. S. Government retains a
nonexclusive, royalty-free license to publish
or reproduce the published form of this
contribution, or allow others to do so, for
U. S. Government purposes.

Technology and Environmental Policy

Argonne National Laboratory
9700 South Cass Avenue, EID/Bldg. 900
Argonne, Illinois 60439

Abstract

This paper provides an overview of regulatory issues under Title m of the Clean Air
Act Amendments that could affect electric utilities. Title HI contains provisions
relating to hazardous air pollutants (HAPs) and provides special treatment for electric
utilities. Generally, this discussion documents that if utility toxic emissions are
regulated, one of the chief difficulties confronting utilities will be the lack of
coordination between Title EI and other titles of the Act. The paper concludes that if
the U.S. Environmental Protection Agency (EPA) determines that regulation of utility
HAPs is warranted under Title HI, savings can be realized from flexible compliance
treatment

Introduction

The interaction of the Title III (air toxics) with Titles IV (acid rain) and Title I (criteria
pollutants) of the Clean Air Act presents a fundamental dilemma for the utility
industry. This dilemma arises when utilities' choice of compliance strategies to meet
the future requirements under Title III is influenced by their choices under Title IV
and I; in addition, their choice of strategies under Titles IV and I could be influenced
by their potential choices under Title ID. For the utility industry to devise an optimal
"least cost" strategy, it should consider the two sets of issues simultaneously, but as
the Act is currently written, that simultaneous consideration is made impracticable.
This paper examines whether the Act sets out the most efficient path for meeting its
clean air goals, and whether flexibility devices written into EPA regulations could
provide the legal opportunity for utilities to adopt lower-cost compliance options.

1 Work supported by the US. Department of Energy, Assistant Secretary for Fossil Energy, under
contract W-31-109-Eng-38. The authors wish to acknowledge the research assistance of T. Elliott, K.
Quinn, and L. Hedayat. All errors and omissions are the responsibility of the authors.

87-1


-------
Historical Background

Hazardous air pollutants (HAPs), often referred to as air toxics, have been regulated
at the federal level since the Clean Air Act of 1970 ("the 1970 Act").2 In Title HI of
the Clean Air Act Amendments of 1990 (CAAA),3 Congress replaced the prior
Section 112 entirely.4

Under Section 112 of the 1970 Act,5 the EPA was required to identify and list air
toxics and then apply standards to control them with an adequate margin of safety.
These were known as the National Emission Standards for Hazardous Air Pollutants
(NESHAPs).

By the 1980s, the failings of the NESHAPs program had come to be recognized.
(1) Because the deadlines for issuing controls for a pollutant once listed were so
stringent, the EPA simply avoided listing pollutants. (2) Because of margin-of-safety
requirements, substances without identifiable health thresholds (i.e., levels below
which no health detriments can be detected) could only have standards set at zero
emissions. Consequently, in the 20 years following the 1970 Act, the EPA listed only
eight pollutants and set controls for only seven.6 Some of the air toxics emitted by
electric utilities were listed by the EPA (i.e., beryllium, mercury, arsenic, and
radionuclides), but none of the controls applied to electric utilities.

With concern about the adequacy of the Section 112 program rising, a series of bills
was introduced into Congress to amend it.7 In May 1985, Congressmen Waxman,
Wirth, and Florio introduced a bill to replace Section 112 with a toxics release
disclosure system leading to promulgation of standards and compliance within four

2	Fub.L. 91-604,84 Stat. 1676 (1970), as amended and significantly tightened by Pub.L. 95-95,91 Stat.
685 (1977), codified at 42 U.S.C. § 7401 et seq.

3	Pub.L. 101-549, 104 Stat 2399 (Nov. 15, 1990). Title HI consists of a replacement to the old

Section 112 as well as provisions addressing other sections of the Clean Air Act and requirements of the
Occupational Safety and Health Act and the Solid Waste Disposal Act

4	The 1970 Act contained three titles, for stationary sources, moving sources, and general provisions.
The CAAA contained eleven titles, but only added three new titles to the codified law, for acid
precipitation, permits, and stratospheric ozone protection. Thus, the Act now contains six titles. Title III
of the CAAA is identical to Section 112 of the Act, as amended.

5	42 U.S.C. § 7412.

6	The substances are asbestos, beryllium, mercury, vinyl chloride, benzene, radionuclides, arsenic, and
coke oven emissions. All but coke oven emissions had been regulated at the time the CAAA were enacted
in 1990. See 40 CFR, Part 61.

7	In the early 1980s, the EPA decided to limit its own participation in air toxics regulation to the role
of acting as a clearinghouse for state programs. In the absence of federal action, the states took up the
call. Virtually all of the regulatory activity during the 1980s took place at the state level.

87-2


-------
years. The bill did not get very far, but the disclosure idea was enacted in the 1986
Superfund Amendments and Reauthorization Act (SARA),8 which established a list
of toxic substances and under Section 313 required the reporting of toxics releases
from industrial facilities. Electric utilities, however, were not required to report
under SARA.

In 1986, recognizing that the existing statutory framework was unworkable, the EPA
announced a comprehensive air toxics strategy that recognized two new areas.
(1) Area sources (i.e., the small but numerous sources that do not pose significant
risks individually but may pose risks collectively) were the source of as much as 75%
of the cancer risk. (2) The other air programs (e.g., those for mobile sources) had
actually made a greater contribution to the reduction of air toxics than had the
Title HI program itself.

By the end of the 1980s, momentum was building toward enactment of amendments
to the 1970 Act. The key to the amendment of the 1970 Act was solving the acid rain
problem. In 1989, certain understandings were reached that established the
fundamental premises of an acid rain program. With the logjam clearing, all other
issues came open for reexamination, including air toxics. But since acid rain was the
key, all other amendments had to be written in ways that conformed to the initial
acid rain understandings.

When President Bush proposed the Clean Air Act Amendments in the summer of
1989, his bill contained a separate title for air toxics that contained a two-phase
strategy: first, EPA would promulgate technology-based standards; second, it would
examine the residual risks. However, Edison Electric Institute (EEI) took the position
that utilities should be exempted from the air toxics program unless further study
showed that regulation was warranted. Its reasoning was that many of the controls
used to reduce sulfur dioxide and nitrogen oxides would co-control air toxics. The
White House later sided with the House version, which adopted EEI's position and
exempted utilities from the main channel of air toxics regulation.

8	Pub. L. 99-499, which amended the Comprehensive Environmental Response, Compensation, and
Liability Act (CERCLA or Superfund), codified at 42 US.C. 9600.

9	Title HI of SARA is also known as the Emergency Planning and Community Right-to-Know Act of
1986 (EPCRA), codified at 42 U.S.C. §§ 11001-11050.

87-3


-------
Air Toxics Provisions of the 1990 Amendments

The air toxics provisions of Title III enact an entirely new program containing three
elements:

1.	Program to Identify and Control Air Toxics

•	Phase 1: In the first phase, the EPA must issue technology-based standards
for each category of sources of listed pollutants, and the standards must
require the maximum achievable control technology (MACT) for major
sources.

•	Phase 2: In the second phase, the EPA must determine what residual risks
remain after application of the technology-based standards and report to
Congress by November 1996.

2.	Separate Research Programs

•	Section 112(n)(l) requires the EPA to study the hazards to the public health
from utilities, due by November 1993, and it also requires two studies of
mercury. In addition, several provisions in Title III initiate studies of risk
assessment methods and theory.

•	Section 112(m) requires the EPA and National Oceanic and Atmospheric
Administration to identify and assess HAP deposition in the Great Lakes and
Coastal waters and to report to Congress by November 1993. On the basis of
the findings, the EPA is required to issue such standards as necessary to
prevent adverse effects by November 1995.

•	Section 112(n)(2) requires the U.S. Department of Energy and EPA to study
coke oven technology and to report to Congress by November 1996.

3.	Program to Prevent Accidental Releases

•	Section 112(r) imposes regulatory risk assessment and management plans for
the 100 most hazardous substances and requires the EPA to establish a
program of long-term research on methods and techniques of hazard
assessment.

Title in preserves the rights of states to adopt their own control programs and to set
controls that may be more stringent than those adopted by the EPA. State programs,
however, must work through the federal permitting program established under

Title V of the CAAA. One study estimates that U.S. industry will spend more than

87-4


-------
$19.4 billion in this decade on control technologies and monitoring equipment needed
to comply with them.10

To implement Title III is in some respects quite specific. In others, much of the
substance of the law is delegated by Title HI to the EPA to develop and implement.
Thus, the impact of Title HI on most parties will be determined in EPA rulemaking
proceedings. Electric utilities are, for the most part, exempted from the current
proceedings. However, it would be a mistake to ignore them, as they will set
precedents that will affect possible future utility regulation. It is hoped that this
frame of reference will illuminate these potential precedents.

Program to Identify and Control Air Toxics

The bulk of Title III is devoted to the first of the three program elements — the

program to identify and control air toxics.

Identifying Hazardous Air Pollutants. The responsibility for identification changed

as follows:

•	Under the 1970 Act Section 112 of the 1970 Act established a definition of HAPs
and delegated to the EPA the task of identifying and listing those pollutants that
met the definition. By definition, no HAP could also be a criteria pollutant.

Prior to enactment of the CAAA, eight substances were listed as HAPs by the
EPA.11 The last HAP for which emission controls were issued was listed in
1980. The EPA promulgated standards for seven of the eight listed pollutants.
Under the savings provision of the 1990 CAAA, Section 112(q), any standard that
was previously promulgated remains in force under the new Act.

•	Under the 1990 CAAA: Frustrated with the slow pace of EPA's risk-based listing
process, Congress decided to take the listing function away from the EPA. Under
Section 112(b)(1), the CAAA established a list of 189 hazardous air pollutants for
the EPA to regulate.

Implications for Electric Utilities: As indicated in Snyder et al.,12 the Electric Power

Research Institute (EPRI) found about 35 of the substances on the (b)(1) list are

emitted by electric utilities. Three of the substances EPRI lists as utility HAP

10	Mcllvane Co., 1991, Air Toxics and VOCs: Markets and Technology for Compliance Analysis, Control
and Measurement, a market forecast. See Clean Air Report, Dec. 5, p. 18.

11	These were asbestos, beryllium, mercury, vinyl chloride, benzene, radionuclides, arsenic, and coke
oven emissions. The EPA published notice of intent to list an additional 25 substances as air toxics.

12	Snyder, T., et al., 1992, unpublished information, Argonne National Laboratory.

87-5


-------
emissions were listed by EPA under the old Section 112: beryllium, mercury, and

arsenic.

Controlling Hazardous Air Pollutants: A Two-Phase Strategy. Title III employs a

two-phase control strategy for sources of HAPs. During Phase 1, under authority of
Section 112(d), technology-based standards are to be set for specific source categories.
These must apply the maximum achievable control technology (MACT). Existing
sources must meet the standards within three years of their issuance. During
Phase 2, the EPA must evaluate the residual risk remaining after the installation of
MACT controls and report to Congress by November 1996.

Title III introduces the concept of rolling MACT standards. The EPA is required to
develop a list of source categories and rank the categories into four priority tiers.
Phase 2 follows this rolling schedule: within eight years of promulgation of the
original MACT standards (nine years for the first tier regulated), EPA must issue
residual risk standards for each of the categories. In sum, the principal strategy for
Phase 1 can be described as the three M's — mandatory MACT for major sources.
However, Title ID does not treat all parties equally. First, EPA must apply MACT to
major sources, but EPA's application of MACT to area sources is at their discretion.
Second, some other sources are subject to a totally different regulatory scheme.

• The Phase 1 Program (The Three Ms)

— Listing source categories: Section 112(c)(1) required that the EPA publish a
list of all categories and subcategories of major sources and area sources of
HAPs by November 15, 1991. Section 112(c)(2) required that the EPA issue
emission standards for each category and subcategory of sources. The object
of identifying source categories is to divide the emission sources into generic
groups, so that similarly situated parties can be given similar regulatory
treatment. The new Section 112(c)(1) reduced EPA's discretion by requiring it
to establish a generic list of source categories first and then to apply controls
to reduce emissions from those sources. The EPA's published list contained
far more source categories (appr oximately 750) than it originally envisioned
(approximately 250 to 400). The EPA assigned more categories to
industries for which extensive emission monitoring data was already
available.

Implications for electric utilities: Because electric utility steam generating units
(EUSGUs) are not expressly exempted from the three M's, the EPA was
placed in something of a dilemma in issuing the source category lists. In its
June 1991 Federal Register notice (56 Fed. Reg. 28,548), the EPA examined its

13 57 Fed. Reg. 31576 (July 16,1992).

87-6


-------
alternatives for categorization of EUSGUs. The EPA explained its choices as
follows.

Alternative 1 is to include EUSGUs on the list, but that would require them
to be regulated unless the utility study under section 112(n)(l) concludes that
regulation is not warranted; moreover, to delete the EUSGU category, the
EPA would have to go through risk-based findings under section 112(c)(9).

Alternative 2 is to exclude EUSGUs from the list, however, that choice might
appear inequitable because utility units smaller than 25 MW would have to
be on the list as well as nonutility generation units.

The EPA chose not to list EUSGUs as a source category.14

—	Ranking source categories:. The EPA is required by Section 112(e)(1) to rank
source categories to set the order for issuing MACT standards for them. The
source categories and dates for issuance of standards are ranked into four
tiers: (1) at least 40 source categories by November 1992, (2) at least 25% of
all listed source categories by November 1994, (3) at least 50% of all listed
source categories by November 1997, and (4) all listed source categories by
November 2000.

Pursuant to Section 211(e)(3), the EPA was required to list the categories that
will be classified in these tiers by November 15,1992. It released the
timetable on September 16, 1992.15 Once source categories and the rank
order for issuing standards have been established, the EPA must then
proceed to meet the schedule for promulgating MACT standards.

—	Distinct classes of regulated parties: While Sections 112(c)(2) and (d)(1)
require the EPA to establish emission standards for each source category that
is listed under Section 112(c)(1), Section 112(c) also gives the EPA some
discretion on whether to list some parties.

Section 112 establishes two classes of parties on the basis of the amount of
HAPs emitted. First, a major source is any stationary source or group of
sources within a contiguous area and under common control that emits, or
has the potential to emit, considering controls, 10 tons/yr (TPY) or more of
any HAP or 25 tons/yr of any combination of HAPs.16 Second, an area

14	The EPA did, however, list industrial electric generation turbines and cooling water chlorinarion-
steam electric generators but not electric utility small boilers.

15	"National Emission Standards for Hazardous Air Pollutants; Availability: Draft Schedule for the
Promulgation of Emission Standards/' 57 Fed. Reg. 44147 (Sept. 24,1992).

16	Section 112(a)(1). EPA also expresses TPY as 9.1 Mg/yr.

87-7


-------
source is any stationary source of HAPs that is not a major source but does
not include sources subject to regulation as mobile sources under Title II.17

The classification of individual sources by quantity of emissions determines
the severity of rules to be applied. Clearly, the burden on major sources is
going to be heavier. Under 112(c)(1), all categories of major sources must also
be listed; therefore, all major sources will be regulated under the three M's.
However, many individual major sources may avoid regulation as major
sources simply by reducing their emissions to be reclassified as area
sources.18

Implications for electric utilities: The possibility of aggregating emissions of
different HAPs from a source so that they exceed 25 tons/yr and so the
source thereby becomes a major source, creates a mechanism for controlling
small quantities of emissions. In addition, some utility emissions are
candidates for lesser quantity cutoffs, which could cause EUSGUs to be
classified as major sources.

- Early reductions program: Under Section 112(i)(5) of the CAAA, existing
sources can qualify for a six-year extension of the compliance date for MACT
requirements, if they achieve a 90% reduction in HAP emissions (95 percent
or more for particulates) from their 1987 baseline HAP emissions. Early
reductions either must be achieved before the MACT standard for the source
category is proposed or must be subject to an "enforceable commitment" to
achieve the early reduction by January 1, 1994, for sources subject to MACT
standards prior to that date. Sources can take credit for early reductions in
HAP emissions from voluntary actions as well as from reductions resulting
from compliance with federal, state, or local laws. Conversely, early
reductions can be used as offsets to meet New Source Review (NSR)
requirements. HAP emission reductions resulting from criteria pollutant
offsets could be credited toward the early reductions program. This last
result may be especially important for fossil-fuel-fired electric utilities that
may satisfy Title IV requirements before being subject to Title III
requirements.19

17	Section 112(a)(2).

18	Charles D. Malioch, Monsanto Co., cited in Environment Reporter, June 21,1991, page 489.

19	One must take care to distinguish the early reductions program under Title III from the EPA's other
current pollution prevention programs, such as its "33/50" toxics project, which encourages voluntary
reductions of 17 chemicals not only to the air but.also the ground and water. To qualify, a party must
reduce releases by 33% in 1992 and 50% by 1995, with an emphasis on pollution prevention techniques.
Note also that there are early reductions allowed in Title IV pursuant to Section 404(e). However, these
are only reductions before the Phase 1 deadline of 1995 and have strict conditions placed upon them.

87-8


-------
The early reductions provision allows a form of in-plant bubbling — a type
of emissions trading. However, the trading is limited by Section
112(i)(5)(e), which provides that the trading of certain high risk pollutants
(HRPs) should be limited in counting toward offsetting reduction of other
HAPs.

The early reductions program, by being associated with the permit process,
will be administered by the EPA until states have their own permit programs
up and running.

Implications for electric utilities: The EPA's notice does not address electric
utilities at all. Thus, no information currently exists about whether the early
reductions program applies to EUSGUs. If the EPA has discretion as to
whether to regulate in the entirety or not, then it has discretion to regulate in
part or to make certain distinctions.

Special rules for electric utilities: Title III makes two distinctions of EUSGUs
from other parties. The first distinction is in its treatment of EUSGUs
compared with its treatment of other source categories. Section 112(a)(8)
defines an "electric utility steam generating unit" as any fossil-fuel-fired
combustion unit of more than 25 MW that serves a generator that produces
electricity for sale, including regeneration units that supply more than
25 MW and more than one-third of their generation capacity to a utility
power distribution system for sale. It is clear from the EPA's actions so far
that a utility will not be able to immunize itself from regulation as a major
source by showing that its actual emissions are not more than 10 tons per
year.

The second distinction is in making emission controls contingent on the
results of studies the EPA will conduct Section 112(n)(l) requires the EPA to
undertake two studies of utility HAPs and mercury and requires the National
Institute of Environmental Health Sciences (NIEHS) to undertake a health
effects study of mercury exposure. The report must describe alternative
control strategies for emissions that warrant regulation. The EPA shall
regulate EUSGUs if the study shows that such regulation is appropriate and
necessary.

• Phase 2 Program (Residual Risk)

Section 112(f)(1) requires the EPA to report to Congress by November 15, 1996,
on the residual risk to public health remaining after application of the emission
standards, its risk to public health, and the positive as well as negative

20 Rules were proposed in 56 Fed. Reg. 27,338 0une 13,1991) and issued in final form in 57 Fed. Reg.
61,970 (Dec. 29, 1992).

87-9


-------
consequences of efforts to reduce such risks. If the EPA finds that the residual
risk requires regulation, it must promulgate rules within eight years from issuing
the MACT standard under Section. 112(d) —¦ nine years in the case of the first tier
of source categories. Residual risk standards must provide an ample margin of
safety to protect the public health unless the EPA determines a more stringent
standard is necessary to prevent an adverse environmental effect, taking into
consideration costs, energy, safety, and other factors. Two generalizations can be
made about Phase 2. First, Phase 2, like Phase 1, is a rolling process, since the
schedule is tied to the MACT schedule. Second, residual risk is the return to a
variant of EPA's original risk-based regulatory approach under the old Section
112.

Implications for electric utilities: When the residual risk element is added to the
mix, the treatment of utilities under Title III yields a distinct image. Hie utility
exemption under Section 112(n)(l) is designed to examine the hazards to public
health remaining "after imposition of the requirements of this Act" (with implied
reference to the acid rain program).

Mandated Title III Research Programs

The second element of the Title IH program is the establishment of research programs
designed, for the most part, to provide data for regulatory decision making.

Special Studies for Electric Utility Steam Generating Units. Section 112(n)(l) of
the CAAA requires that three reports related to emissions from EUSGUs be
submitted to Congress.

1.	By November 1993, the EPA must report on the hazards to public health posed
by EUSGU emissions and on alternative control strategies to reduce those

emissions.

2.	By November 1993, the NIEHS must report on its determination of the threshold
level of mercury exposure, below which human health is not adversely affected.

3.	By November 1994, the EPA must report on the level of mercury emissions from
EUSGUs, municipal waste combustion units, and other sources. The EPA must
also report on health and environmental effects and on the availability and cost
of control technologies.

The EPA plans to study (1) the rate and mass of emissions, (2) control technologies,
(3) health effects of emissions, and (4) costs of control.

Because two of the studies prescribed under Section 112(n)(l) specifically concern
mercury, it is of preeminent concern to electric utilities. In discussing the risk from

87-10


-------
mercury emissions by utilities, it is necessary to consider the models currently in use
and whether mercury has a safety threshold. One of the reported problems is that a
significant amount of mercury is unaccountable, so that a mass-balance study cannot
be completed. New sampling technologies are needed to provide the means to
collect total mercury samples. It is also important to track the various mercury
compounds individually, as each of the species differs in the pattern of its
environmental impact It is necessary to examine which of these species is most
problematic.21 It is not yet dear how the EPA plans to address these problems.

Studies of Risk Assessment Practices. Title HI establishes several research
programs to investigate risk assessment theory and methods, including these.22

•	Section 112(o) requires the National Academy of Sciences to conduct a review of
risk assessment methodology to determine the risks associated with HAP
exposure.

•	Section 303 establishes a Risk Assessment and Management Commission (RAMC)
to investigate the policy implications and appropriate uses of risk assessment and
risk management.

•	As part of the program for accidental releases, the EPA is required to conduct
long-term research on methods and techniques of hazard assessment.

Programs to Prevent Accidental Releases

The foregoing controls cover routine and predictable emissions. The third element of
the Title in program is the establishment of new organizations and procedures to
prevent and/or provide quick response to accidental or catastrophic releases of
HAPs.

Unlike other programs under the Act, the program for accidental releases is •
organized with the Office of Solid Waste and Emergency Response (OSWER) as the

21	Bloom, N.S., and B. Rand, 1991, Chemical Specialion of Mercury in Coal-Fired Pawcrplant Stack Gases:
Overcoming the Analytical Problems, Temarks at the EPRI Conference, Managing Hazardous Air Pollutants:
State of the Art, Nov. 4-6.

22	For more information on these, see Loeb, A.P., A.J. Policastro, R.G. Whitfield, and M.C. Wozny,
1992, unpublished information on methods and issues in air toxics risk assessment, Argonne National
Laboratory.

87-11


-------
lead office within the EPA.23 The rules for the program are being written by an
inter-office work group that includes representatives of the EPA's Office of Air.

Stale Air Toxics Programs

In the absence of federal action to control air toxics during the 1980s, a number of
state air toxics programs arose. While state programs were intended to fill the
federal vacuum, it is also true that the EPA actively encouraged states to take the
lead. Some of the general concepts contained in the state programs were
incorporated into Title HI of the CAAA. Beyond federal requirements contained in
Title III, state air toxics programs are independently important for the following
reasons:

•	States have the implicit ability to implement more stringent air toxics regulations.

•	States will administer the Title HI air toxics program through their state
permitting systems.

•	Regional control of air toxics emissions by of the CAAA.

Of the 33 state air toxics programs that presently operate in the United States,
12 states currently are active in revising existing programs or in developing new
programs. Four of these 12 active states are located in the Great Lakes region, which
is title region likely to be the most affected by EUSGU regulation.24

State air toxics programs vary in scope. States either operate regulatory programs
specifically designed to address air toxics, or they append, as a matter of policy,
consideration of air toxics emissions onto other existing programs, such as permit
programs. Few state air toxics programs operate as independent regulatory
programs. Most state programs cover new sources, and usually new and modified
sources. The following summary describes features of state air toxics programs that
either already conform to requirements of Title ID of the CAAA or that will need to
be adjusted to conform to Tide in requirements.

•	Most state programs will have to add coverage of existing sources to conform to
the coverage required in Title III.

23	A proposal rule was issued in 58 Fed. Reg. 5102 (Jan. 19, 1993). An extension of time for
comments was granted in 58 Fed. Reg. 13,174 (March 9, 1993). No final rule has been issued. The
Department of Labor has also issued rules pursuant to Section 112(r), including a list of substances, in 57
Fed. Reg. 6356 (Feb. 24,1992), codified at 29 CFR, Part 1910.

24	For more information on these programs see Loeb, A.P., et al., 1992, Air Toxics Provisions of the Clean
Air Act Amendments: Regulatory Issue Analysis, Argonne National Laboratory internal report

87-12


-------
•	Most state programs already operate as permit programs, thereby conforming to
the requirements of Titles III and V of the CAAA for air toxics programs.

•	Most state programs already impose control technology requirements that would
conform in principle to MACT in Title HI.

•	Only a few state programs evaluate maximum community exposure that may be
required as part of the residual risk analysis required in Title III after MACT is
applied.

Unlike Title I, which delegates to states the setting of emission controls for attainment
of ambient standards, Title III sets the standards at the federal level. The role of the
states under Section 112(1)(1) is to implement and enforce MACT standards and other
provisions under EPA-approved state programs.

The CAAA: Interaction between Title III and Titles I, IV, and V

Title HI should not be examined in isolation. Its interaction with the provisions of
Titles I, IV, and V of the CAAA will affect how electric utilities respond. The
possibility of co-control of air toxics resulting from the implementation of Title IV
provisions is the reason for the special treatment of electric utility studies in Title III.
The ability of state permit programs, authorized under Title V, to impose more
restrictive air toxics emissions limitations may affect how utilities respond to Title IE.
Table 1 provides a side-by-side comparison of the statutory schedule for
implementing the provisions of Titles I, III, IV, and V.

Interaction between Title III and Title I

Title I of the CAAA tightens the existing program under Sections 108 and 109 of the
Act. These provisions require the EPA to list criteria pollutants and set National
Ambient Air Quality Standards (NAAQS).25 "Die Act divides the country into local
air quality control regions and requires each one to meet the NAAQS. The Act
makes state regulatory agencies responsible for achieving attainment of NAAQS in
each region within their states. Areas that are in attainment are only subject to
standards for prevention of significant deterioration (PSD) of air quality, which are
designed to ensure that areas with dirty air are not cleaned up by moving their
industry to dean areas. In contrast, areas that fail to meet the NAAQS standard, for
any of the criteria pollutants are designated "nonattainment" and must take action to
come into attainment.

25 In addition, new stationary sources are subject to nationally uniform new source performance
standards, which require that sources install a level of control technology.

87-13


-------
TABLE 1

Time Lines for Various CAAA Provisions Relevant to Electric Utilities

Year

so2

NO„

Air Toxics

Permits

1991

3/31- Elections

6/30- Elections under §406

12/31- EPA to publish proposed list of

Phase II allowances



EPA to establish list of categories of
major sources and area sources

11/15- EPA to Issue min.
elements ol slate permit
programs

1992

5/15- EPA to issue rules to track
allowances

5/15- EPA to issue rules (or CEMS
12/31- EPA to publish final list of
Phase 2 allowances

5/15- EPA to issue rules for
tangentially fired boilers and dry-
bottom wall-fired boilers

MACT standards for first 41
categories; health risk threshold
quantities set for accidental releases
of 100 toxics

12/31- EPA to set standards for
coke oven batteries

5/15- EPA to Issue regulations
to Implement a federal permit
program; states must submit
programs to EPA for approval

1993

2/15- Phase 1 sources must submit
permit application and compliance
plan

11 /IS- Phase 2 units must Install
CEMS

1/1- EPA to propose revised new
source performance standards

EPA to issue rules for prevention of
accidental releases, emergency
response, risk management
11/15- Compliance date for coke
oven work practice regulations

11/15- States to submit
proposed permit programs
and legal opinions to EPA

1994



1/1- EPA to issue final rules for new
source performance standards
1/1- EPA to report to Congress
consequences of trading NQ„ for S02
allowances

MACT standards for 25% of listed
categories

11/15- EPA to approve or
disapprove timely state permit
programs

1995

1/1- Phase 1 begins and Phase 2 units
must install CEMS

1/1- Phase 1 begins; emission
limitations for first two types of
boilers take effect

11/15- EPA to list categories of area
sources sufficient to control 90% of
area source emissions of the 30
HAPs that present the greatest
health threat; EPA to list categories
to assure that sources accounting for
>90% of specifically listed
pollutants are subject to standards
12/31- Compliance date for coke
oven batteries

Facilities must submit permit
applications within 1 year of
approval of state programs

(Corit. )


-------
r fi jk tmr «-t	jr-*«	*

TABLE 1 Cont.

00

-j

1

01

Year

so2

NO*

Air Toxic#

Permits

1996





Operators of stationary sources
subject to accidental release
standards must submit risk
management plans; EPA to report to
Congress on methods of
determining health effects

7/1- States with approved
programs by this date shall
issue Phase 2 permits by
12/31/97; where states do not
have approved programs, EPA
will Issue permits

1997

After 1/1, if any Phase 1 extension
unit exceeds limitations, EPA shall
deduct equal amount of allowances
12/31- Last date to demonstrate
repoweriiig

1/1- EPA to issue emission limitations
for other types of boilers
1/1- EPA may make standards for
first two types of boilers more
stringent if such technology Is
available

MACT standards for aggregate 50%
of categories

12/31- Phase 2 permits must
be issued by states by this
date

1998







1/1- Units must submit NO„
permit application

1999









2000

1/1- Phase 2 begins (Phase 1 units
become subject to Phase 2)
1/1- Last date to submit
documentation of repowering

1/1- Phase 2 begins

EPA to issue MACT standards for
remaining categories
11/15- EPA to complete rules for
specifically listed pollutants; EPA to
complete rules for area sources
11/15 or 2 years after the date
listed, whichever is later- EPA to
issue emission standards for
additional categories EPA may list



2001





EPA to Issue residual risk standards



2002









2003

12/31- Extension for repowering ends








-------
Title I of the CAAA adds a number of specific provisions designed to force
nonattainment areas to come into attainment. In all nonattainment areas, the CAAA
require steady progress toward attainment until the federal standards are met.26

The Title I program is very decentralized. Beyond setting the NAAQS, most of the
activity takes place at the state level Because of this decentralized nature/ no single
generalization can be made on the interaction of Title I with Title III regarding
control technologies or schedules for compliance.

Interaction between Title III and Title IV

Title IV divides electric utility units into two categories: (1) high-emission units
(identified in Section 404, Table A of CAAA) that must meet S02 emission reductions
in Phase 1 by 1995; and (2) lower emission units, that along with the Phase 1 units,
must provide additional S02 emission reductions in Phase 2 by 2000 and 2010. A
two-phase strategy is also prescribed for NOx.27 The Title III study of HAP
emissions from electric utilities is not required to be completed until November 1993;
it probably will not be completed until 1994.28 Thus, it is unlikely that the EPA
win propose emission standards for electric utilities any sooner than 1994, and given
the history of MACT rulemakings, the agency will probably not publish a final rule
until one year hence (1995, at the earliest).

Phase 1 S02 Sources. By the time the EPA's Title III intentions are known, Title IV
Phase 1 affected sources will probably have been implemented. Although particular
Phase 1 control strategies could result in HAP emission reductions (via co-
control),29 Title IV Phase 1 compliance will not necessarily alleviate Title HI
concerns. Title IV compliance strategies may not succeed in reducing HAPs to EPA-
acceptable levels and may not ultimately be the most cost-effective strategies when
Title m and Title IV compliance strategies are viewed in combination. Indeed, some

26	See Sections 108-110 of the CAA, 42 US.C. §7408-7410.

27	Under Section 407(b)(1), the NOx controls become applicable to utility units at the same time that
they become affected units under the S02 provisions. The standards for Phase 1 for NOx, which were due
on May 15,1992, were proposed in 57 Fed. Reg. 55,632 (Nov. 25, 1992); corrections to the proposed rule
were published in 57 Fed. Reg. 61,489 (Dec. 24, 1992). NOx controls for Title IV are coordinated with
those required under Title I, which were proposed in 57 Fed. Reg. 55,620 (Nov. 25,1992) as a supplement
to EPA's General Preamble.

28	As indicated above, the Section 112(n)(l) studies of utility HAPs and the NIEHS study of the
mercury thresholds are not due until November 1993; the study of mercury emissions is not due until
November 1994.

29	See Molburg, J.C, 1992, unpublished information, Argonne National Laboratory.

87-16


-------
Phase 1 compliance strategies could actually increase HAP emissions (e.g., fuel
switching to particular low-sulfur coals).30

Phase 2 S02 Sources. In contrast to Phase 1 (Title IV) affected sources whose
compliance strategies will precede any Title HI actions, Phase 2 affected sources could
be confronted with a similar, but reversed, Title III and Title IV compliance timetable.
Given that the EPA will probably not issue a final rule on electric utility MACT until
1995 at the earliest, that rule would then become effective three years hence (1998).
The closer these two dates correspond, the greater will be the opportunity for utilities
to coordinate planning with both CAAA titles.

Title III could force utilities away from coal and toward other fuels that contain lower
levels of trace metal air toxics. However, utilities may still be able to use coal and
satisfy both Titles HI and IV, if integrated compliance choices are made. Thus,
careful calculations are required to determine the total toxics and acid rain emissions
before a utility commits to a certain fuel and compliance strategy to satisfy both
Titles IE and IV.

Interaction between Title HI and Title V

Nothing in Title III restricts state air toxics programs from imposing more stringent
requirements than the federal program. Title III will continue the state program
process of mediating air toxics control through the permit process as described in
Title V. Although the final rulemaking for state permit programs under Title V was
issued in July 1992,31 states will not submit their permit programs for approval
until 1993. The EPA does not need to take final action on those programs until 1994,
and states will not begin to issue the first phase of permits until 1996. State permit
programs will be in place prior to the likely effective date of MACT for utilities. So,
like the development of MACT for other source categories, experience with air toxics .
permits with other sources will establish precedents for the way utility air toxics .
permits are handled.

As with the general authority under Title III, states are empowered to make their
permit programs more stringent than the federal program. States can collect higher
fees per ton than the statutory $25 per ton. In the final permit rule, the EPA dropped
a proposed de minimis provision that would shield industries from lengthy permit

reviews if they experience only slight increases in emissions. Applied to air toxics
emissions, elimination of the de minimis provision could mean that an increase in any

30	Szpunar, C., 1992, Air Toxic Emissions from the Combustion of Coal: Identifying and Quantifying
Hazardous Air Pollutants from U.S. Coals, Argonne National Laboratory Report ANL/EAIS/TM-83.

31	57 Fed. Reg. 32,250 0uly 21,1992).

87-17


-------
of the myriad of air toxics emitted from an electric utility could trigger a 45-day
permit review.

Observations

There is a trade-off between Title IV and Title III in coal fuel-switching. To the extent
utilities reduce sulfur emissions by switching from high-sulfur coal to low-sulfur coal,
they increase the chlorine content of the coal.32 During combustion, the chlorine
becomes Cl2 or HC1, which either corrodes the steel pipes of the unit or is emitted.
HC1 is possibly already the largest single pollutant emitted from power plants.
Moreover, there is some indication that other toxic particulates may result, and that
these will be in the form of fine particles that enter deepest into the lungs.33

Potential Application of Flexible Regulation

In the last decade, the EPA has begun to consider the form of regulation as a separate
issue in regulatory decision making. Inspection of the list of 189 HAPs covered by
Title HI indicates that the pollutants themselves, as well as the sources from which
they are emitted, vary considerably. This diversity in nature and application suggests
that it is improbable that a single type of policy tool will be best suited for control of
all HAPs. Instead, HAP control can be achieved most efficiently by choosing from a
variety of policy instruments, so that a remedy can be specifically tailored to an
individual case. Within the legal constraints of the CAAA, the question before
policymakers becomes the matching of the HAP with the most appropriate
instrument.

Policy Instruments

For purposes of this paper, the alternative policy instruments for HAP control have
been grouped into three broad categories: command-and-control measures,
informational and subsidy measures, and market-based measures.

Command-and-Control Measures. Command-and-control measures are appropriate
when it is believed that certainty of outcome is of preeminent concern and when
market and the price system cannot be effectively used to control the risks posed by a
particular HAP.

32	Szpunar, C, 1992, Air Toxic Emissions from the Combustion of Coal: Identifying and Quantifying
Hazardous Air Pollutants from U.S. Coals, Argonne National Laboratory Report ANL/EAIS/TM-83.

33	Laudal, D.L., S.J. Miller, and R. Chang, 1991, Enhanced Fine Particulate Control for Reduced Air-Toxic
Emissions, remarks at the EPRI Conference, Managing Hazardous Air Pollutants: State of the Art, Nov. 4-6.

87-18


-------
•	Bans and limits: Bans and Emits are strict highly explicit prohibitions or ceilings
on HAP usage that sharply limit the compliance choices on the part of the
regulated party. They may be appropriate in cases where the environmental
damages associated with the emission of a given HAP are highly uncertain but
believed to be very large and in cases where measurement of emissions is
thought to be difficult and expensive.

•	Efficiency and technology-based standards: Efficiency standards are requirements
that products meet some minimum output per unit of emission. Technology-
based standards prescribe processes by which a good is produced. A distinction
between these two instruments is that efficiency standards tend to encourage
innovation in control technology while technology-based standards do not.

•	Work practice standards: Work practice standards are regulations that require
the implementation and use erf a particular procedure in specific types of work
places. Section 112(h) specifically provides for use of such requirements in lieu of
a MACT standard where it is not feasible to prescribe or enforce an emission
limitation. Work practice standards are appropriate in cases where it is believed
that worker habits or maintenance programs can achieve significant emission
reductions without expensive monitoring requirements.

Informational and Subsidy Measures. Informational and subsidy measures exhibit
aspects of both command-and-control and market-based measures. They attempt to
utilize market forces in some fashion without a complete reliance upon them.

« Informational standards: Informational standards require that manufacturers
properly label their products in some way to influence emission and exposure
behavior of workers or the public. The effective use of informational standards
presupposes that individuals and firms are unaware of the risks related to their
activities (due to the search costs of information) and that they would act to
reduce exposure or emissions if fully informed at no direct cost.

•	Subsidies: The government may subsidize particular types of behavior to achieve
reductions in HAP emissions. Subsidies might take the form of (1) funding for
research, development, and demonstration projects to speed commercialization of
substitute substances or abatement technologies; (2) enactment of emission
abatement investment tax credits; or (3) funding for deposit-refund programs.34
This instrument can effectively be used in conjunction with other measures to
lower the costs of compliance.

Market-Based Measures. Under certain conditions, market-based measures may
lead to more cost-effective HAP control than command-and-control or informational

Note that the actual payment to emitters per unit of reduction in greenhouse gas emissions is more
like an emissions tax (opportunity cost of foregone reductions).

87-19


-------
and subsidy measures.315 However, implementation, monitoring, and enforcement
are likely to be much more involved in market-based control regimes.36 The
ultimate determination of choice of method is rooted in (1) the balance between the
cost efficiencies provided by the market versus the additional costs of monitoring and
enforcement and (2) the difference in achievable environmental quality under each
type of regime.37

•	Emissions charges: An emissions charge is a fee levied on emission activities
assessed in proportion to the magnitude of the emission. An emissions charge
can lead to cost-efficient emission reduction by taking advantage of consumer
choice and profit motive of firms. HAP emission reductions are thus made
where they are least expensive.

•	Excise taxes: For some HAPs, there may be a large number of emission source,
some of which are difficult to monitor. In these instances, it may prove to be
unwieldy to tax emissions directly, although the economic efficiencies of a tax
system may still be desirable. Just as in the case of emission faxes, emission
activities would be influenced by prices. Note that excise tax systems allow the
regulatory body even less control of final emissions than do emission charge and
tend to spur innovation in emission-reducing technologies.

•	Ad valorem taxes: Another type of instrument is the ad valorem tax.

Application of this instrument would levy a tax on all HAPs on fee basis of their
current prices. There is no attempt to justify the resultant emission reductions or
the economic cost of compliance by a cost-benefit analysis. Monitoring and
enforcement, however, tend to be less involved under an ad valorem tax system
than under the other market-based instruments.

<2C

Note that the cost savings offered by market-based approaches depend heavily on the markets in
which they operate (Tietenberg, T.H., 1985, Emissions Trading', Hahn, R.W., and G.L. Hester, 1989, Where
Did All the Markets Go? An Analysis of EPA's Emissions Trading Program; and Atkinson, S.E., and T.H.
Tietenberg, 1991, Market Failure in Incentive-Based Regulation: The Case of Emissions Trading).

The costs of monitoring and enforcement need to be factored into the choice of tax rate or quantity
of allowable emissions (Downing, F.B., and W.D. Watson, 1974, The Economics of Enforcing Pollution
Controls).

rsn

See A. Loeb, Lead Phasedawn and Environmental Economics: Rethinking the Model of Emission Trading,
manuscript, forthcoming.

This conclusion assumes that the regulated parties operate in a business environment that would
reward cost-effective behavior. This situation is always true in well-functioning markets, but these parties
are often subjected to a myriad of other regulations (e.g., utilities). In these cases, it is up to the regulators
to provide the necessary incentives for cost effectiveness. The conclusion also assumes that other
nonfinancial considerations do not play a role in the decision process.

87-20


-------
• Tradeable emission rights: In the past decade or so, tradeable emission rights
have rapidly become a familiar environmental policy option. This technique
requires a party to possess an allowance or credit in order to legally emit some
quantity of a pollutant. Permits are denominated in discrete quantities and are
given as a quasi-property right to parties who can buy and sell them, either in
bilateral trades (i.e., between a single buyer and a single seller) or in brokered
multilateral markets (i.e., on existing commodities exchanges such as the Chicago
Board of Trade).

There are two basic types of tradeable emission rights: credits and allowances.
Credits are generated by regulated parties by reducing emissions below a
required standard or permit limitation; that is, they are in a sense earned.
Allowances are created by the regulatory body without reference to a particular
standard; that is, they are distributed or auctioned.

A well-designed tradeable emission rights system has several advantages over
command-and-control measures. Like fee and tax systems, emission rights
systems internalize the costs of polluting behavior, using overall prices to guide
emission behavior. However, emission rights systems allow policymakers to
retain more control over the resultant environmental quality standard than do fee
and tax systems.39 A properly designed emission rights system can also
promote efficient intertemporal allocation of resources while actually phasing out
emissions.40 Another advantage is that a permit system can be designed to
include several HAPs; these comprehensive systems can be even more efficient in
the overall reduction of environmental damage for a given cost.41 Also, as in
an emission fee system, innovations in emission mitigation technology are
stimulated more in a tradeable emission rights system than in a command-and-

*>g

The environmental quality standard depends on the total amount of emissions allowed, which,
in turn, is determined by the number of permits issued by the regulating body. Note that this does not
mean that the chosen environmental standard is necessarily the efficient one (Weitzman, M., 1974, "Prices
and Quantities," The Review of Economic Studies 61, pp. 499-77).

40 Kosobud, R.F., T.A. Daly, and KG. Quinn, 1991, Tradeable Permits for Global Warming Control:
Implications for Regional Economies and Public Utilities, Proceedings of the 53rd Annual Meeting, American
Power Conference, Chicago, Volume I, pp. 658-673 (April-May).

This is because it may be cheaper for a particular emitter to reduce emissions of one HAP than
to do so for another HAP. The result is that the same total reduction of environmental damage can be
achieved at a lower cost than if emissions of both HAPs were reduced individually in another way
mandated by the control authority. Note, however, that given the uncertainties associated with the
comparison of environmental damages between different HAPs at different concentrations, the prospect
for a multi-HAP trading system may be limited.

87-21


-------
control or tax regime, because it reduces an emitter's need for costly credits or
allowances.42

There are some limitations to emission rights schemes, however. Efficient trading
markets (without prohibitive transactions costs) may prove to be difficult to
design and administer, and practical procedures to create viable institutions for
trading have been largely unexplored.43 Also, monitoring and enforcement of a
credit-based system can become complex and subject to moral hazard.44 This
situation is particularly true for HAPs with diffuse or poorly understood sources.
Similar to fee and tax systems, emission rights systems do not prevent hotspot
problems from developing.

One other point should be noted: Reducing HAP emissions by using tradeable
credits or allowances implicitly assigns the right to pollute to their holders. Thus
the confiscation of credits or allowances (i.e., the reduction of the tradeable
rights) could be required. Traders who believe confiscation is a legitimate risk
will discount credit or allowance prices to allow for this possibility. Therefore,
complicated questions regarding compensation for any confiscation could affect
the efficiency of the system.

integration of Title Hi Requirements with Other CAAA
Provisions: implication for Instrument Choice

Because the slate of possible regulatory forms has now been discussed, it is next
appropriate to attempt to match them to the potential needs of EUSGUs under the air
toxics program. A full treatment of the issues requires a thorough scientific and
economic analysis of each HAP — with its sources and flows through the ecosystem,
risk mechanisms, and links with the economic system — as well as an analysis of the
policy tools available for its control. Such an analysis is beyond the scope of this
paper. What is possible instead is to examine the findings of this paper and match
possible inefficiencies with regulatory mechanisms that may afford more efficient
control policies.

This reduction is because command-and-control measures largely rely on the state of technology
available at the time of their promulgation. Once set, these requirements are often costly to change.
(South, D.WV R.F. Kosobud, and KG. Quinn, 1991, Greenhouse Cos Emissions Control by Economic Incentives:
Survey and Analysis, prepared for the Thirteenth Annual North American Conference of the International
Association for Energy Economics, Chicago, Nov.

Apt

South, D.W., R.F. Kosobud, and K.G. Quinn, 1991, Greenhouse Gas Emissions Control by Economic
Incentives: Survey and Analysis, prepared for the Thirteenth Annual North American Conference of the
International Association for Energy Economics, Chicago, Nov.

44 Nussbaum, B.D., 1991, Phasing Down Lead in Gasoline in the U.S.: Mandates, Incentives, Trading and
Banking, prepared for the OECD Workshop on the Use of Tradeable Permits to Reduce Greenhouse Gas
Emissions, Paris (June).

87-22


-------
The foregoing discussion has determined that the timing of control requirements
appears to be a significant issue for EUSGUs. One may think of the CAAA as setting
a stair-step approach to regulation of utilities. The regulations increase the severity
of controls with each successive step, adding reductions beyond those already
achieved:

•	Step 1; Pursuant to the 1970 and 1977 Acts, the EPA promulgated ambient air
quality standards for criteria pollutants. Utilities are most affected by ambient
standards for particulates and NOx. In CAAA Title I, Congress tightens
procedures for attainment of these standards, so that some states will force
additional controls upon local utilities.

•	Step 2: Pursuant to CAAA Title IV, the EPA is promulgating regulations for acid
rain control. Utilities may install control technologies as part of their strategies
for participating in the S02 allowance market; additional NOx reductions may be
mandated by regulation.

•	Step 3: Pursuant to CAAA Title III, the EPA may, given certain findings,
promulgate standards for air toxics from utilities. However, recognizing that
some of the technologies implemented for criteria pollutants (Title I) or acid rain
control (Title IV) will co-control HAPs, Congress commanded the EPA to
promulgate regulations for utilities only if, after examining whether the controls
already installed will reduce HAP emissions to acceptable levels, the remaining
risk warrants regulation (analogous to residual risk). Given that some of the air
toxics from utilities (e.g., mercury) are not controlled by control technologies in
the prior steps, at least some utility regulation is likely. In the meantime, the
issue results in much uncertainty for utility planning.

In principle, the object of utility compliance planning will be to coordinate the
programs so that no control investments are duplicated or wasted. However, the
stair-step approach creates both inefficiency and uncertainty problems. (1) Taken
together, the three titles promote a piecemeal approach to utility regulation, with
partial steps being taken one at a time, preventing coordination and planning. The
poor integration of the Act will likely result in incremental and isolated compliance
strategies and suboptimal decisions by utilities. (2) Moreover, Congress intended that
its wait-and-see approach to utility toxics regulation would benefit utilities. Taken by
itself, that outcome seems reasonable. However, in the context of the other titles, that
approach may only add uncertainty to an already piecemeal approach.

Coordination will be the key to efficiency, but the Act and state regulation provide
no incentive to do that. Indeed, the inefficiency and uncertainty problems present
planning dilemmas. For example, a utility that could realize long-term savings from
installing HAP controls at the same time it installs controls to meet Titles I or IV, in
advance of the deadline for compliance with Title III, may be prevented from doing
so by rate regulation. Because utility expenditures for controls that go beyond
existing requirements may not be recoverable in the rate base and may be challenged

87-23


-------
in prudency reviews, the utility may be limited to recovery of costs to meet currently
required emission standards. Typically, utility executives will be averse to taking the
risk that control costs would be unrecoverable. Thus, as long as the possibility
remains that utilities can avoid regulation, they will not plan nor undertake
coordination of Title III controls with the other titles. In consequence, utilities may be
forced to commit to controls to meet near-term regulatory requirements, where such
technologies would be poorly suited to an integrated approach to meeting all the
Act's requirements.

The trade-off between Title IV and Title EI in coal fuel switching places utilities in a
further dilemma: to meet the provisions of one title, they may actually increase
emissions of another substance that would later become subject to regulation.

Underlying these planning dilemmas is a temporal mismatch of the various titles.
The Act, as structured, places electric utilities in jeopardy of "creeping
incrementalism'' by regulation. What is absent from the Act is the opportunity for
electric utilities to act according to their needs within state rate-of-return regulation.
The absence of the opportunity to coordinate the various CAAA programs, in
combination with the statutory structure and the uncertainty of the utility exemption,
could make Title III far more costly than a coordinated program. Although the utility
exemption was intended to buy time for utilities, it is not assured that the time
bought will, in the long run, save utilities money.

What Congress might have done to carry out the intent of Section 112(n)(l) to its full
intent would have been to provide a mechanism to combine Titles I, III, and IV for
planning purposes. In many cases, the most efficient way for utilities to meet Title III
would be to install the level of control at the same time a unit would install controls
for the other titles.45 In the absence of such provisions in the Act, it will be
important for any regulations issued under Title III to allow utilities to integrate
planning for the three titles, so they may achieve the same efficiencies in Title III as
are hoped to be achieved in Title IV.

To correct these disincentives, incentives can be placed into the regulations to allow
utilities to be rewarded rather than penalized for such early reductions. A program
structured in this way would make it possible for utilities to make reductions early if
that strategy was the most cost effective. While only the contours of this idea have
emerged, it has the potential to resolve the temporal mismatch and provide efficiency
in control.

45 For example, in the lead phasedown program, the EPA was not sure of the level of control that was
necessary and so adopted a graduated reduction program {i.e., in phases). This approach turned out to
be the least efficient way of doing it, because it was not cost effective for a refinery to add octane capacity
gradually; rather, the most efficient way was to build the capacity all at once. However, by that point in
time, there was not yet a market for the octane. So there were refineries with excess octane they could
not sell, and others had a shortage but could not afford to build on their own. The solution, of course,
was to allow trading.

87-24


-------
There will be certain legal issues associated with examining this issue. For example;
Does Section 112(d) allow any market-based incentives? Would it be restricted to
bubbling of individual sources, or would trading among parties be permitted? What
other legal constraints would apply? How would such trading integrate with acid
rain allowance trading or with state programs under the EPA's General Preamble?46
How would this affect rate-making processes? Of course, any such approach must fit
within the legal constraints set out by Title III and other applicable law.

This paper concludes this problem requires additional research to evaluate the
flexibility mechanisms and potential savings to utilities (should the EPA determine
that regulation is warranted) and to suggest approaches that will be more efficient
when applied to this context

46 See proposed rule, 57 Fed. Reg. 13,498 (Apr. 16, 1992).

87-25


-------
MODIFIED HYDRATE PRODUCTION AT DRAVO LIME COMPANY

J. College, MX. Cupp, M. Babu, D. Stowe
Dravo lime Company

3600 Neville Road
Pittsburgh, PA 15225

Abstract

Dravo lime Company is currently involved in several projects testing modified
hydrated limes for SO2 control by furnace sorbent injection (FSI). An ongoing in-house
development project is focused on the production of modified hydrates. One of these
hydrates contains sugar added during the hydrating procedure. This specialty hydrate
used under FSI conditions at a 2.0 CaO/S stoichiometry shows a 10-15% increase in
SO2 removal over the commercial hydrate. This translates to 5-8% increased utilization
of the lime. Dravo lime operates a 1 million BTU/hr combustor to test the hydrate
samples that are made in a batch mode (30 lbs. of hydrate product per batch) or in a
continuous pilot scale hydrator (capable of producing up to 300 lbs. of hydrate per
hour).

Dravo lime sugar hydrate was tested in the recently completed B&W's SOx-NOx-Rox-
Box (SNRB) process and has demonstrated over 90% SO2 removal at a Ca/S
stoichiometry of 2.0. The SNRB process is an advanced air pollution control system for
the simultaneous removal of SOx, NOx and particulates emitted from fossil fuel fired
boilers. The project was funded by the DOE, OCDO, EPM and B&W as part of the
Clean Coal Technology Program.

Dravo lime has received a contract to supply enhanced hydrated lime to Virginia
Electric Power Company's (VEPCO) Station in Yorktown, Virginia. The 180 MWe
VEPCO unit is undergoing FSI sorbent testing under an EPA contract Dravo lime is
supplying several types of hydrated lime: base case commercial, lignosulfonate treated
and sugar hydrate.

88-1


-------
Dravo Lime has negotiated art exclusive North American licensee agreement to market
the Sorbalit® flue gas treatment technology developed by Marker Umwelttechnik
GmbH of Germany.

Sorbalit® is a patented specialty sorbent comprised of a combination of hydrated lime
and activated carbon. Sorbalit® substantially reduces eco-toxic pollutants such as
volatile heavy metals (mercury),, organics (dioxins and PCBs) and acidic gases (SO2 and
HQ). The Sorbalit® flue gas treatment technology is uniquely compatible with a
variety of air pollution control technologies.

This paper discusses the results from the continued testing of the specially hydrates
and provides information pertaining to the Sorbalit® flue gas treatment technology.

Background

Dravo lime's initial involvement in dry injection technology utilized a 5 million
BTU/hr furnace burning petroleum coke containing 7% sulfur. Pulverized lime was
injected into the burner for SO2 capture. Poor performance of the lime injection on
petroleum coke gas led to injection of dry lime hydrate and improved SO2 capture.
This testing was performed in 1979 and 1980.

In 1984 and 1985 Dravo Lime, as a sub-contractor to Southern Research Institute,
participated in an EPA program to produce and test "supersorbents." The
"supersorbents" were chemically modified hydrates that would enhance SC>2 capture
and maximize lime utilization.

Testing included production of hydrates which incorporated additives into the
hydration process. Additives investigated included CaC^ NaQ.2' KG, Na2CC>3,
Na(OH), sodium silicate, lignosulfonate, alcohols and sugar.

This test program necessitated the construction of al million BTU/hr coal fired
combustor. Furnace sorbent injection tests were performed to screen candidate
hydrates. Once screened, more detailed tests were conducted at the Southern Research
Institute, Birmingham, Alabama.

88-2


-------
Of the sorbents
tested, sugar
hydrate yielded
the highest SC>2
removals. Figure 1
shows data that
clearly indicated
more testing
should be
performed. Since
this early testing,
Dravo lime has
pursued the
development of
sugar hydrates.

Laboratory support of the EPA program included fix bed reactors, fluidized bed
reactors and thermogravimetric analysis. After the EPA program was completed, it
was demonstrated that the addition of moisture to the gas stream at the lower
temperature ranges (120-325°F) yielded a significant increase in SO2 capture. This
concept was tested in the combustor and produced promising results.

From 1985 through 1988, funded by a DOE-PETC sponsored program, Dravo lame
tested its HALT process. HALT is an acronym for the process using humidifieation
and Hydrate Addition at Low Temperatures. A test program was designed to optimize
the removal of SQ2. Bench scale testing was conducted at Dravo Lime's Research
Center utilizing the 1 million BTU/hr coal fired combustor.

A one year series of parametric tests were conducted on a 5 MWe pilot plant at Ohio
Edison's Toronto Station burning local Ohio coal (3.2 %S, 14% ash, 10,500 BTU/lb).
Program objectives included: determination of SO2 removal dependence on adiabatic
approach to saturation temperature, multi-nozzle humidifieation testing, determination
of removal dependence on Ca/S ratios, baghouse and ESP removal comparison, and
conceptual design and cost estimate for a full scale 180 MWe retrofit application.

A slip stream of flue gas (275-350°F) downstream of the boiler air preheaters, laden
with fly ash, was ducted to the HALT plant Injection of hydrate particles into the flue
gas stream occurred prior to cooling the gas stream. Approach temperatures of 10-40
°F above adiabatic saturation were tested. Multiple dual fluid (air-water) air atomized
nozzles sprayed water directly on the gas in the ducting.

Figuel

%9Q2 Rsrn^sfrariffAStLriy fcr FS Saterts

%SC2
Rarwate'

43-

30

20

to-

0

JSa^32%Sdfu-Co0l,i
22QO-296D Degrees F I











iilllllliiiiii

1





Jill It f ill«





M

M iMll lll 11 1





Sllllllll





¦illl





lilllltls!





WiiMiM







to
adiSves

G*S
Sach=1.5

Nj
actftwes

GsflS
SiucM.1

2

&pr Ca!S
Sfcach=l2

z.m
&prCa*S
Sh«Fl2

88-3


-------
The following reactions occur during the sulfur capture process:

S02+H20	>H2S03

Ca(OH)2 + H2S03	~ CaS03 + 2H20

Hie particulate removal of the gas leaving the HALT plant occurred in a single
compartment pulse jet baghouse or a three field mobile ESP. Maximum "once through"
hydrate utilization of 25% upstream of the baghouse was accomplished. Removal
efficiencies of 50-60% were obtained at 20 °F approach to adiabatic saturation with
Ca/S ratios of 2.0.

hi 1989 Dravo lime produced modified hydrates for Consolidation Coal Company
(CONSOL) for their Coolside process testing. Additives incorporated during hydration
included: Na2C03, Nad, CaCl2, sugar and others.

Current Test Program

Since 1991, Dravo has intensified independent research on sorbent injection technology.
Support of the current test program necessitated the reconstruction of the 1 million
BTU/hr coal combustor. Simulating at research scale the operation of a commercial
pulverized coal-fired furnace, the combustor is supplied with a steady flow of
pulverized coal from a storage bin. A controlled coal feeder system delivers the coal to
a down fired burner on top of the combustor. From the control room/ operators can
adjust combustor temperature and sorbent feed rate while monitoring gas temperature
and flow, SO2 content, baghouse draft, and other system conditions.

As shown in Figure 2, the coal combustor allows the operators to simulate three differ-
ent types of dry injection technology:

•	furnace injection (2,100 °F-2,300 °F)

•	economizer injection (1,000 °F)

•	duct injection (300 °F) with and without humidification

88-4


-------
Dravs Research Center

Caal Cornbuslor (Sorfcers? Injection Testing Capability
Q Venturi Row Meier
Q 300 °F ScfOem injector. Pert
gj Humttitication IftjgCiisn Pon
g| Duct Wan Wstting Detected
gj Gas Sampling % Analysis Probe
El Sample Corwftioreng and On-Ljie SCV NO*

Measurement
0 Dual-Filter Solids Ssmsimg Device
|3 Baghoyse

Q CaalFeedet
@ Pufceraed Coal Supply
0 Com6uS#on fclr
0 Sorben: Feeder
@ Ft.'nacc Sotbei! Injection Port
Q FWrsetoiy-UfteaDua
@ 20CO»F SosWrt Injecta Port
Q Economizer

U Eeoncmaer Scroent injection Port
H ShcD-arxS-Tu&e Heat Exchanger

Cr





1 ¦ "X











1 j





o~



|o]

a I 0 (



figure 2

Furnace Injection

The combustor is equipped with injection ports which allow operators to inject sorbent
directly into the furnace at three different temperature levels. Sorbent is conveyed
pneumatically at controlled rates, and injected at one of three levels to establish reliable
correlations between furnace temperature and sorbent utilization. Refractoiy-lined
ductwork equipped with injection ports allow for sorbent injection at downstream
temperatures.

Economizer Injection

An annular water-cooled heat exchanger downstream from the combustor lowers the
gas temperature and enables simulation of sorbent performance when injected at the
economizer temperature of 1000 °F.

88-5


-------
Duct Injection

Downstream of the annular cooler, fee flue gas alters a second heat exchanger, a water-
cooled vertical shell-and-tube device followed by a venturi flow meter which measures
gas velocity. Injection ports in this 300 °F gas treatment zone allow for both sorbent
addition and gas humidification. External thermocouple arrays are placed at several
locations along the duct surface to detect duct wall wetting. This technique has proven
to be very successful.

SO2 Monitoring

Gas sampling probes located downstream of the furnace, economizer, and duct
injection ports enable combustor operators to determine the oxygen and SO2 content of
the combustor flue gas foEowing various types of gas conditioning and sorbent
injection. A final monitoring point is located just ahead of the system's single
compartment fabric filter baghouse. A dual-filter solids sampling device in this same
location provides for collection, analysis, and evaluation of differential removal across
a bag filter element

Sorbent Preparation

Sorbent preparation capabilities include both a batch and continuous lime hydrator.
The batch hydrator is capable of producing approximately 30 pounds of hydrate per
batch, while the continuous hydrator can produce up to three tons per day of specially
formulated hydrate. Both hydrators allow for variable water and lime addition feed
rates, and both can produce hydrate with or without additives. The operation of the
continuous hydrator can be readily scaled up to commercial size.

Sorbent Analysis

Samples of the product solids from the injection technology tests and the specialty
hydrates produced are analyzed for chemical and physical characteristics at our in-
house laboratory. Analytic capabilities include ICP, X1F, AA, particle sizing, BIT
Specific Surface Area and conventional wet chemistries. We are awaiting shipment of a
SEM/EDS/Image Analysis system which will allow microscopic physical and chemical
characterization.

88-6


-------
Current Test Plan Objectives

•	Optimize hydrates to increase SO2 capture. Produce super-hydrates by:
modifying crystal structure, surface charge/ surface area and particle size.
Incorporate additives into the hydration process. Additives include sugars, clay,
alcohol, lignosulfonates and others.

•	Optimize SO2 capture at three different injection points and temperatures:
furnace (2,100 -2,300 °F), economizer (1,000 °F), in-duct (300 °F) with and
without humidification.

•	Reduce hydrate requirements by improving utilizations and using sorbent
recycle techniques.

Rguie3

Conprison of Operating Costs for PS Sabots
Tested at Dravo Lira Ffesearcii

z>

ISass SOMAfe 3&aitrCist

•naasiuc*

330
330
2B0

Ftaegarf &

Cos
(STaiSCE)

250

c

30
ZD
ZD

ill. J

rKBTO	fiKSJu

SOW	SOBS

«KU.	«KU.

Hgus4

Corrpaiison of %tercwals jixI	UBfeCons

ft* F3Sofberts Tested at Draw UrreResoach

Combustor Results

Extensive furnace
injection testing has
demonstrated that sugar
hydrates have the
highest SO2 removals,
best utilizations and are
the most economical of
the hydrates tested.
(See Figures 3 and. 4)
Hie cost basis in Figure
3 assumes $50/tonfor
commercial hydrate.
The cost of additives

were on a differential
basis. Disposal costs are
included in the
calculations. Alcohol
added hydrates costs
were based on an
assumed range of $60-
$80/ton (as the test and
worst possible costs for
the reagent). Also 40%
utilization was assumed
for the alcohol hydrates,
although measured
utilizations were much
lower.

88-7


-------
Sugar hydrate has demonstrated 10-15% increased SO2 removal efficiencies over
conventional hydrate during testing of coals ranging from 0,5-3.6% sulfur.

An example of the data generated during the furnace sorbent injection of black strap
sugar hydrate is given in Table 1:

Table 1 - Injection Point: Bottom
Injection Temperatures: 210Q-230Q°F
% Sulfur Coal: 2.8

Test

A

B

C

D

Actual Ca/S Stoich

2.00

2.17

1.99

2.17

SOj Initial (ppm)

1198

1057

1199

1198

SO2 Final (ppm)

226

180

234

183

SO2 Removal (%)

81.19

82.99

80.26

84.81

Gas Phase lime Util. (%)

40.58

38.22

43.70

39.06

Current Test Plan Target

90% SO2 Removal at 2.0 Ca/S Stoich

Other Current Programs

Sorbalit® is a commercially available product that is currently being used at 25 air
pollution control systems in Germany and other European countries. Applications
include: waste-to-energy facilities, medical waste incinerators, hazardous waste
incinerators, metal smelting plants, crematoriums, and wood and coal-fired boilers.

Sorbalit®, a blend of hydrated lime and activated carbon, can be tailored to meet
specific removal requirements and a wide variety of applications. Sorbalit® has
demonstrated significant emission reductions of eco-toxic pollutants such as:

•	Volatile heavy metals-mercury, cadmium, thallium, selenium, and arsenic

•	Organics-dioxins, furans, chlorinated hydrocarbons, PCBs and PAKs

•	Acidic gases-SO? and HC1

Mercury reductions ranging from 70-99.9% have been documented. Dioxin, furan,
HQ, and SO2 reductions in excess of 90% have also been documented. Test data
indicates that Sorbalit® process-based systems are capable of meeting federal standards
for air toxics and acidic gas emissions from both coal- and municipal waste-fired
boilers.

88-8


-------
The Sorbalit® flue gas treatment technology is less expensive than competing activate!
carbon-based technologies, can be implemented without extensive modifications to
existing emissions control systems and represents one of the most cost-effective
compliance alternatives currently available. The Sorbalit® process is uniquely
compatible with a variety of air pollution control technologies, including wet scrubbing
as well as dry-, conditioned dry-, and spray-adsorption.

Dravo Lime's exclusive rights under the agreement with Marker Umweltteehnik GmbH
of Germany allows Dravo Lime Company to provide North American clients with the
process design and otter engineering support services required to utilize Marker
Umweltteehnik1 s Sorbalit® process. The agreement also includes the rights to produce
the specialty sorbent for use in air pollution control systems.

SNRB-Dravo lime supplied sugar hydrate to B&W's recently completed SNRB process
testing at a 5 MWe equivalent demonstration facility located at Ohio Edison's R.E.
Burger Plant The SNRB process utilizes a high temperature baghouse in which the
combined removal of SOx, NOx and particulate takes place.

Sugar hydrate was injected upstream of the baghouse with a baghouse temperature of
850°F. SO2 removals of greater than 90% were achieved at a Ca/S stoichiometry of 2.0.
Sugar hydrate demonstrated the highest SO2 removals of the calcium based sorbents
tested*.

The sugar hydrate was produced at the Dravo Lime's Black liver facility located in
Kentucky. The 20 ton/hr. hydrator has been modified to accommodate the addition of
sugar (molasses), lignosulfonate and other additives that can be added as a solution.
Black strap molasses was added to the hydration water in quantities sufficient to yield
1-1.5% sugar hydrate, the hydrate was air classified at 325 mesh, had a mass mean
particle size of 3.5-5 microns and a BET specific surface area of 10-12 m2/ g.

YEP CO - Dravo Lime is currently contracted to produce base case commercial hydrate,
1% lignosulfonate added hydrate and sugar hydrate for testing at VEPCO. Hydrates
are produced at Dravo Lime's Black River facility.

CONSOL - Dravo Lime is a subcontractor to CONSOL for a DOE funded project
entitled "Advanced Jh-Duct Sorbent Injection for SO2 Control". Dravo Lime is carrying
out a continuous hydration experimental program to develop fundamental information
on the hydration, leading to the production of optimized hydrates for in-duct SO2
removal tests to be conducted by CONSOL.

88-9


-------
Zinc Corporation of America (ZCA)-Dravo Lime is currently conducting a sugar
hydrate test program at the George F. Weaton Power Station at ZCA in Monaca,
Pennsylvania. The power station was built to serve the electrical requirements of
ZCA's zinc smelter. The plant is a conventional, pulverized coal burning steam
operation with a boiler-turbine-generator arrangement

ZCA utilizes FSI for SO2 control in their two 60 MWe units. Both units usually share
the same power requirements and burn the same low sulfur (0.5%) coal To support the
ZCA test program, combustor testing was conducted at the Dravo lime Research
Center utilizing the coal being burned at ZCA. Testing of the sugar hydrate in Dravo's
combustor demonstrated a significant increase in SO2 removal efficiencies over the
conventional hydrate routinely injected at ZCA.

Conclusion

Dravo Lime is continuing to explore hydrate injection as a means of SO2 reduction for
industrial and utility application. Research efforts are focused upon chemical and
physical modification to hydrate aimed at reducing stoichiometry and/ or increasing
SC>2 capture. Dravo lime's additional capabilities of marketing the Sorbalit® flue gas
treatment technology for the removal of eco-toxic pollutants adds to our already
extensive array of technical services including sample analyses, system start-up and
optimization services, process design and evaluation. Developments over the last 14
years are encouraging and we remain confident that such technologies will be
technically and economically viable.

Acknowledgment

Significant contributions were made by the technical support staff and combustor
operators: E. Goetz, S. Chenault, W. Biles, D. McKinney, F. Beall, S. Tutokey.

Reference

1. A.R. Holmes, K.E. Redinger, G.T. Amrhein, SO2 Emission Control with the SOx-NOx-
Rox-Box, presented at the 1993 SO2 Control Symposium, Boston, MA (August 24-27)

88-10


-------
Comparison of Worldwide Emission Control Strategies and Their Effects on

Plant Availability and Costs

S. Stallard
Black & Veatch
8400 Ward Parkway
Kansas City, Missouri 64114

L. Salvaderi
ENEL

Via G. B» Martini, 3
Roma 00198
Italy

C. Reese
RWE Energie AG
Krappstrasse 5

RWE Energie AG
Krappstrasse 5
4200 Essen 1
Germany

H. Schlenker

4200 Essen 1

Germany

Abstract

Worldwide, utilities are addressing legislative and political pressures for reducing
emissions of air pollutants. Measures are being developed and implemented resulting in
either the installation of new emissions control equipment or in the significant
modification of past operating procedures. One of the most critical factors in the
selection of emissions control strategies is the required level of emissions reduction. The
demands of the legislation and the applicable emissions standards will unquestionably
affect the range of compliance options available. Such measures may alter the viability
of available, alternative primary fuel sources (coal, gas, oil) as well as have a
considerable impact on die production cost and, therefore, on the overall cost of
electricity.

Impacts of the evolving environmental pressures on technology selected, availability, and
costs are not easily understood nor are they necessarily consistent throughout the world.
Of particular interest are the two different ideological approaches, a "command and
control" approach adopted by the European Community (EC) (and with some variation
in Japan) and a "market-based" approach developed in the United States (US).

Became such issues could potentially have a broad impact on power production in the
future, the Joint UNIPEDE/WEC Committee on the Availability of Thermal Generating
Plant (ATGP) performed quantitative evaluations of the impact of emission control
measures on plant availability and generation. This paper summarizes this Committee
effort.

89-1


-------
This paper discusses, in brief, the three principal areas of concern:

•	Emerging emissions control regulations and compliance strategies of the US,
Europe, and Japan.

•	The role of emission control technologies, or more precisely flue gas desulfuriza-
tion (FGD) systems, on plant availability, based on review of FGD performance
in US, Germany, and Japan.

•	The economic impacts of measures such as FGD and selective catalytic reduc-
tion (SCR) on electricity production costs.

Regulatory Influences on Emissions Control Standards

Comparing legislation, emissions standards, and the associated "ideology" of Europe,
Japan, and the US is rather difficult because of the differences in the structure of the
utility industry, varying levels of reliance on oil/gas versus coal, and differing political
climates. However, even given these difficulties, such a comparison can be both useful
and interesting in better understanding how the environment will affect power plant
availability, the impact on the cost of electricity (COE) for future generating plants, and
the role of the power producer in cleaning up the environment

Legislation and the emission standards for different countries vary significantly. The EC,
Japan, and the US each possesses unique ideological and legislative stances. In partic-
ular, the US has recently legislated a unique, "market-oriented" approach to control S02
emissions. On the other hand, the EC relies on unit-specific emissions constraints,
although the form and method for determining such can vary widely depending on
location, age, etc. In addition, emissions requirements are indirectly tied to fuel quality.
Table 1 illustrates how emission control regulations relate to emission limits for coal-
fired power plants in the EC, Japan, and US. This information is shown on Figure 1.

European Community

The EC took actions to reduce the atmospheric pollution and establish homogeneous
rules for all the member states in the early 1980s. The bases for their actions were as
foEows;

•	To respect the 1979 Geneva Convention on long-range transboundary air
pollution.

•	To avoid unfair competition conditions in the EC.

•	To request authorization for construction and/or substantial modifications of
industrial plants.

•	To establish emission limits on the basis of the Best Available Technology
(BAT).

•	To leave the member states the responsibility of establishing supplementary
conditions for particularly polluted areas.

•	To allow a gradual application of the provision to the existing plants.

89-2


-------
2000-

S02 Emission Rates, 1% S Coal

SOs Emitted
S02 Removed

Uncontrolled Phase I Phase II New

US
Unit

Figure 1. Comparison of US and EC Emission Rates

89-3


-------
Table 1. Comparison of Emission Rates for EC, Japan, and US Utilities



No

Control

New US
Unit"5

Existing
US Unit1®

Existing

US Unit131

EC Unit,
New or
Existing'41

Japan,

New

Facility®1

Japan,

Existing

Facility'*

S02

(1% S Coal)

ib/MBtu'7'

1.76

0.18

1.2

2.5

0.42

0.18

0.45

mg/Nm3ISI

1670

166

1140

2380

400

170

430

Required
Removal
Efficiency,
percent

0

90

31*®*

0

76

90

74

SOj

(2.5% S
Coal)

Ib/MBtu

4.4

0.22

1.2

2.5

0.44

0.19

0.47

mg/Nm3

4170

208

1140

2380

400

170

430

Required
Removal
Efficiency,
percent

0

95

73

43

90

96

89

NO,

Ib/MBtu

0.6

0.30

0.45

0.45

0.21

0.13

0.39

mg/Nm3

568

284

426

426

200

120

360

Notes:

m New US Utility unfts are required by the local regulatory agencies to meet SOz removal efficiencies
from 90 to 95 percent. The removal efficiency that must be met is determined on a case-by-case
basis and is dependent on the sulphur content of the coal to be buried.

 At the beginning of Phase II, January 1, 2000; values based on "average" allocation of S02
allowance per unit.

M After the said transition period concerning increasing percentages of adjusted capacity, and uniform
application of the Directive standards to both new and existing facilities. Current situation remains
nonuniform, as shown in Table 1.

19 Based on assumed site in Tokyo, Japan, H, =300m, K=1.17.

"" Based on assumed site in Tokyo, Japan, H, =300m, K=3.0.

m Lb/MBtu emissions assume 6,300 KCal/kg (11,360 Btu/ib) coal.

181 Because of the large gap between coal and boiler design characteristics, the following typical values
are assumed:

•	Lower Heating Value = 6,300 kcal/kg.

•	Ratio of flue gas/fuel of 12 Nm3 flue-gas/kg-coal.

,S) Removal of S02 at the specific station in question may not be required; due to the CAA structure,
the utility could elect to cover offset from 0.5 g/MJ or 1.2 Lb/MBtu at another utility within its
system.

89-4


-------
These principles were first applied to large fossil-fired plants via Directive 88/609/EC, in
force since November 1988. This directive regulates plants with a capacity higher than
50 MWnnrBMAi .

The adopted framework provides latitude for each country to adopt the legislation to
meet their specific needs, particularly in actual level of emissions control prescribed—the
Directive sets minimum standards-and timing of installation of emission control equip-
ment. While some EC countries have not yet fully adopted the above standards, some
countries' standards, like those of Germany, are even more stringent.

EC S02 and NO^ limits for new plants and for thermal plants having capacity greater
than 500 MW^m, or, roughly greater than 200 MWELECTRIO are shown on Figure 2. The
EC limits are expressed in mg of pollutant/Nm3 flue gas. From the standpoint of the
construction of new facilities, it is clear that the EC is essentially in harmony with
respect to S02 emission rates; some differences can be noted with regard to plants rated
below 100	However, differences in regulations for NO, emissions do exist.

For large generating facilities, more severe limits are imposed by Belgium (from 1996),
Denmark (from 1992), Germany, Italy, and the Netherlands.

For existing facilities, the picture is more inconsistent. Germany, the Netherlands, and
Italy all follow policies which do not discriminate between new or existing facilities in
their application of BAT. The same limits will be progressively applied to each existing
plant. Conversely, other countries are more flexible and allow the utility to adopt less
expensive emission compliance strategies. For instance, UK does not adopt a plant-by-
plant limit, provided the overall emissions of each sector remain within values
determined on the basis of the overall emissions indicated by the Directive, with a
relaxation of such in the short term (1993). In some respects, this approach can be
considered a "bridge" to the US "National Cap" policy discussed later.

Became electric utilities are the most prominent business sector regarding control of
emissions, they have frequently prepared reduction programs more stringent than the
national ones. For example, extensive programs of desulfurization and denitrffication
systems in existing plants have been implemented by Denmark, Germany, Italy,
Netherlands and, in a few plants, by the UK,

Moreover, in all the countries, various other measures for reducing emissions are being
applied. These measures include fuel switching, installation of low NOx burners, and
construction of new, high efficiency plants. To date, to meet S02 emission standards by
design measures, nonregenerable processes such as FGD are the most widely used.

Japan

Japan's first regulations covering environmental concerns were placed into law in 1962.
Subsequent legislation, the Air Pollution Control Law passed in 1968, also provided the
basis for the stipulating emission limit values.

89-5


-------
so.

NO„

Solid-UqtildGaseous

Solid

Liquid

Gaseous

400
Solids
and
Liiguiils

DIR 88/606

DK.F.IH^I r.P.^UK.N* * ,FRO* *,S* * *

* Proposed Valua
** 400 + 8B% DesutphurUation

*** Derogation lor Spain:

BOO mg/Nm3 lor Imported solid lirels
60% desulphiirliallon rate lor
Indigenous solid fuels

650

B,DK,F,IR,S,UK

200

FRG,IT.B+,0K**

I

After 31.12.1395

| ** Allot Of.01.1992

I
I

DIR 68/606

4 BO

B.DK.F.P.S.UK.IR

300
N

226

200
IT

DK

**

160
FRG,B*

DIR B8/60S 4 50

3SO

B.DK.F.IR.P.S.UK

226

200
IT,N

DK

**

100
FRG.B*

Figure 2. Emission Limits in EC Countries (mg/Nm3) For New Plants >500 MWthermal


-------
With respect to S02 emissions, regulations passed in September 1976 established
nonuniform limits all over Japan. These limits do not set the highest permissible
concentration but, rather, set the highest permissible S02 flow (Nm3-S02/hour),
calculated on the basis of an index "k" different in various regions and the actual stack
and smoke ascent height H, [m], Sixteen "kj" ranks are established for existing plants and
three "kj" ranks for new facilities. Lower values of k correspond to more stringent
control standards.

For comparison to EC standards, consider a typical coal-fired installation with a natural
emission of 840 ppm1 or 1,510 Nm3 S02/hour. For this typical installation, the following
are the emission rates permitted in the Tokyo area for existing (k=3) and new (k=-1.17)
plants, respectively:

Qeytsttnr-tapan — 270 Nm3 S02/hour
Qnewjamn = 105 Nm3 S02/hour

For further comparison the allowable emissions rate for the EC would be 250 Nm3
S02/hour. Thus, for congested areas, Japan's existing plant regulations would essentially
parallel those of the EC; new facilities would face much more stringent regulations.

For NOx emissions, Japan controls emissions as a function of several factors.

•	Type of facility — gas-fired, solid-material fired, or liquid fired.

•	Stack gas volume.

•	Date of installation.

NOx emission rates range from 60 to 150 ppm for gas-fired facilities, 200 to 450 ppm for
solid material, and 130 to 250 ppm for liquid-fired facilities.

United States

The US has recently enacted new amendments (15.11.1990) to the US Clean Air Act
(CAA) which will dramatically alter the needs and strategies of US utilities in meeting
S02 and NOx reduction targets. Unlike prior CAA regulations which dealt strictly with
new power facilities, the most recent amendments target the existing, higher polluting,
fossil-fired, utility power plants.2

1 Based on the following assumptions: a facility burning 1 percent sulphur coal with a lower heating
value erf 6,300 kcal/kg, an assumed ratio of flue gas/fuel tor 12 Nm3 flue-gas/kg-eoal, an assumed
He=300 m (200 m stack + too m smoke), and for the Tokyo area, k=3 for existing plants and k=1.17
for new plants. For the EC emission standard for oil and coal fired plants, of 400 mg S02/Nm3 equates
to 140 ppm with the percent reduction required to meet the EC standard would be 100*(1 -140/840) or
83.3 percent

2Within the discussion of US regulations and emissions control strategies, both ISO tonnes (103 kg)
and English tons wfll be presented. For clarity, the symbols wDI be: 1 English ton = 0.91 t.

89-7


-------
Clean Air Act Requirements Prior to 1990. Recent amendments are a supplement to
existing CAA legislation covering new plants. Since 1971, allowable emission rates for
S02 have been governed by the CAA, to varying degrees. This Act established accept-
able National Ambient Air Quality Standards (NAAQS) and determined areas in which
air quality meets the standards (attainment areas) and which it does not (nonattainment
areas). In 1971, the CAA established New Source Performance Standards (NSPS) for
new facilities in attainment areas; these standards were designed to assure that the air
quality within the attainment area will remain acceptable. In 1979, NSPS standards
established the S02 removal as a function of uncontrolled S02 emissions potential, with
the minimum removal being 70 percent In addition, the licensing agency usually
requires the utility to better these standards by selecting the Best Available, economically
reasonable, emission Control Technologies (BACT). Today, under BACT, new facilities
would typically be required to reduce uncontrolled S02 emissions by at least 90 percent-
often by 95 pereent~and to comply with NO, emission limits between 0.09 to 0.17 g/MJ
(0.2 and 0.4 lb/MBtu). licensing of future installations is expected to require even
higher removal of S02 and NOx. Thus, for new facilities, highly efficient FGD systems
are required.

Recent Clean Air Act Amendments. Because of the large number of older units still in
service, the recent CAA amendments have targeted these older units. CAA amendments
target a reduction of approximately 9.1 Mt S02 (10 Mton) per year below 1980 emissions
levels of 17.3 Mt (19 Mton), beginning in the year 2000. TTie lowering of annual
emission to this national cap (53 percent of 1980 value) will take place in two stages,
2.7 Mt (3 Mton) in 1995 and 6.4 Mt (7 Mton) in 2000. In 2000, a national cap of 8.1 Mt
(8.9 Mton)/year will be enforced for the S02 emissions from all electric utilities and all
affected industrial cogenerators. Thus, the focus on the law is not one of "command and
control" for specific facilities, but the reduction of emissions from the utility sector as a
whole. To meet this objective, an elaborate system of emission credits or allowances
which can be freely traded has been developed. Each allowance corresponds to the right
to emit one ton of S02. Each utility has been allocated its proportionate share of S02
emission reduction obligations with respect to the 1985-1987 S02 and fuel burn rate (or
'baseline") levels, and, hence, Ms been allocated its fair share of the total allowance
pool Since its pool of allowances will not cover current emission rates, the utility must
decide how and where such reductions will be made. Therefore, the system provides the
flexibility necessary to allow utilities to determine flexible, highly utility-specific
compliance strategies, which minimizes the cost of compliance.

In addition to the permanent emissions caps for S02 commencing in the year 2000, the
legislation also targeted the highest polluters for SOa reductions by the year 1995. The
legislation has two distinct phases:

Phase I. Phase I takes effect January 1, 1995. Based on emissions during the

baseline yean, 110 of the largest, highest S02 emitting plants (corresponding to

261 units) were identified. Phase I legislation requires affected units to reduce

89-8


-------
emissions to roughly 1.08 g/MJ (2J lb-S02/MBtu), applicable to both coal and oil-
fired unite. (In terms of EC unite, this limit is roughly equal to 2,380 mg-S02/Nm3
for coal and 3,110 mg-S02/Nm3 for oil3).

The overall result of this legislation is to enforce ail average emission rate across
utility systems, varying by phase, on a lbs of S02/MBtu basis. However, by shifting
allowances between facilities, the utility can vary actual allowable emissions on a
unit-by-unit basis.

Special incentives (bonus allowances) for the installation of proven, efficient S02
emissions control equipment were also legislated for certain Phase I units which
apply appropriate technologies such as FGD that reduce S02 emissions by at least
90 percent, beginning no later than January 1997.

Phase H Starting January 1, 2000, all existing plants above 25 MW will be granted
allowances to emit roughly 0.5 g/MJ (1.2 Ib/MBtu). Most units currently emitting at
rates above 0 J g/MJ (1.2 lb-S02/MBtu) will be allocated allowances corresponding
to this limit Unite which historically emitted at rates below 03 g/MJ (12 lb-
S02/MBtu) will be panted allowances to emit approximately 120 percent of their
baseline emissions. The average emission rate would correspond to 1140 mg
SOj/Nm3 for coal and 1480 mg S02/Nm3 for oil4.

The new CAA amendment also calls for a 1.8 to 3.6 Mt (2-4 Mton) reduction of NOx
from 1980 baseline emission levels of 5.8 Mt (6.4 Mton) by 2000 to be obtained by
reducing emissions on an individual unit basis (a 30 percent reduction of the level
projected without controls). The legislation strongly implies the use of combustion
controls as the primary mode of reduction. Therefore, utilities are expected to employ
direct combustion control measures such as low NO, burners (or equivalent reduction
technology). However, the legislation is sufficiently vague enough to afford the US
Environmental Protection Agency (EPA) the opportunity to require more strenuous NOx
control measures during the course of time, although the act does not call for a cap on
NOr

Matching Standards to Emissions Control Alternatives

Response to emissions control regulations is the product of economic, technical and
political influences, "Command and control" philosophies in the EC undoubtedly provide
less latitude than the market-based approach in the US. However, even in the EC,
utilities must measure alternative approaches and technologies against cost of

3Caicuiated by assuming a ratio of air to flue gas of 12 and 14 for coal and oB, respectively, and by
assuming a Iowa* heating vaiue of 6,300 and 9,600 kcal/kg, respectively.

4Calculated by assuming a ratio of air to flue gas of 12 and 14 for coal and oil, respectively, and by
assuming a iower heating value of 6,300 and 9,600 kcal/kg, respectively.

89-9


-------
implementation, risk, and public opinion. In many cases, public opinion has been a
principal ingredient in utilities' adoption of more stringent compliance strategies than
required by law. Such conditions are likely to be present in the US as well. In fact, the
success of the allowance market will be undoubtedly be influenced by political (on both
state and load level) pressure and public opinion (e.g., from special interest groups such
as coal producers).

The Committee's review suggests that even though each country develops and enforces
its own regulations, and, in fact, its own ideology, the means for addressing environ-
mental concerns remains relatively consistent.

In general, new fossil-fired facilities will typically be fitted with a highly efficient FGD
system. Many commercial FGD technologies exist; the most popular appear to be the
calcium based systems employing either limestone or lime. TTiese systems often employ
additional facilities to develop salable byproducts, such as gypsum, which can also reduce
landfill requirements.

Fuel switching can be an attractive alternative in both the US and in EC countries which
do not enforce the Helsinki agreement for existing facilities. Similar^, it is likely that
fuel switching will be quite attractive for me in Eastern Europe or countries with
centrally planned economies, either as an intermediate or long-term S02 reduction
technique, as a result of the lack of capital and the presence of other, possibly more
pressing social issues.

The control of SOa and NOx emissions unquestionably affects both plant design and
operation. Fuel switching can improve emissions, though potentially at the expense of
lower efficiencies or higher operating costs. FGD systems increase both capital and
operating costs. Relative to a non-FGD unit, the presence of an FGD system typically
involves the need for additional staffing, additional maintenance, sorbent, additional
waste disposal needs, and potential for both increased efficiency losses and unavailability.

A brief discussion of EC and US compliance strategies follows. Of particular interest is
the impact of the evolving US allowance market on technologies and strategies selected.

European Community

Hie power requirements of 12 EC countries are supplied by European Electric Industry
(ESI). The electricity is produced by a wide variety of technologies employing a wide
variety of fuels: solids, 38.5 percent; oil, 10.4 percent; gas, 6.9 percent;
hydro+geo+renewables, 9.4 percent; and nuclear, 34.8 percent.

Current projections show that power demand will grow from the present value of 1,760
TWh to more than 2,470 TWh by the year 2010. On the supply side, strategic guidelines
have been developed which consider the following:

89-10


-------
•	Ability to pursue the freedom of access to energy sources.

•	Extension of the geographic integration of the electric systems in Europe.

Growth and expansion in the East represent two forces which, without proactive
response of the EC, could limit the success of emissions control regulations and
agreements, such as the Helsinki and Sophia Conventions. Even though the EC strategic
plan places emphasis on the ability to access varied power sources, in the near term it is
likely that the ability to simultaneously support national power markets and national
emission reduction obligations will rely principally on the utilities of that nation. In
other words, utilities within each EC country will generally develop viable strategies to
simultaneously support the power needs of its constituency and the emissions reductions
required by its national legislation (and EC agreements), notwithstanding the opportu-
nities for foreign competition to introduce subtle changes to the emissions reduction
strategies. To the extent that changes in emission regulations exist in the EC, coupled
with differences in national economies and resources, it is likely that the effective cost
and availability of power are issues that must continue to be explicitly considered from
country to country.

To further address this point, strategies of Germany and Italy are briefly discussed.

Germany. Since German regulations require any existing fossil-fired facility which will
continue to operate past 1993 to conform to new plant S02 emission regulations, use of
flue gas cleaning equipment or the development of alternative sources of clean power
are the only means for accommodating the legislative requirements. Currently, the
German utilities have chosen the first strategy.

For the future, one consequence of the present discussions on the greenhouse effect and
of die related efforts to reduce C02 emissions could be that, for future power plants, the
conventional concepts might be replaced by new technologies such as combined cycle
plants equipped with an integrated coal gasification (ICGCQ unit, with higher efficiency
and, thus, lower specific C02 emissions. Because of the gasification process, these future
plants would meet the limits of the German legislation without air pollution control
equipment such as FGD or SCR in the flue gas stream.

The situation in the new East German Federal States is different The main efforts will
be concentrated on retrofitting, by 1996, FGD/SCR systems on those power plants for
which it is reasonable, both from economical and ecological points of view. The other
power stations-mostly older ones—will be replaced by new ones. These replacements are
needed within die coming years. These new plants will be built based on conventional
concepts with emphasis on developments which will improve plant efficiency. Therefore,
they will be equipped with FGD, low NO, burners, and SCR systems to meet the limits
of die legislation.

ENEL-Haty. The utility industry in Italy is currently undergoing significant changes; how
such changes will affect compliance strategies, new plant construction requirements, etc.

89-11


-------
is still subject to speculation. However, it is deaf that emissions compliance strategies
will be closely tied to a mix of generation sources and technologies.

In November 1992, the previous ENEL generation plan was revised to consider the
following:

•	Slowdown of the economy and the need to reconsider the evolution of the
electric industiy demand.

•	Necessity to scrutinize the investments in the new "privatized" framework.

•	liberalization toward or boosting of the role of monutility generators (NUGs).

•	Technology evolution of the heavy duty combustion turbines.

Each of the above must be factored into the continued planning for the development of
the ENEL generation system, the associated ENEL fuel mix and, consequently, in the
overall emission control strategy as well. Based on ENEL's current view, the following
strategies will be employed:

•	S02 control, coal-fired plants. FGD will play a major role in S02 emissions
compliance for coal-fired units including its application in all 5,720 MW,,* of
multifuel plants forecasted for the year 2002, and 4,391 MW^ out of the total
5,578 MW of oil-fired units being converted to burn coal. As a result of these
projects, the total FGD capacity is expected to be 10.1 GW out of a total coal-
fired capacity of 113 GW. For the remaining 12 GW of smMl unite, fuel
switching (oil/gas) will be used.

•	S02 control, oil-fired plants. Existing plants, 16,555 MW at the year 2002,
should utilize a suitable mix of low sulfur oil and natural gas, to remain
compliant with the new plants' emission limit (400 mg/Nm3) for increasing
percentages of the total installed thermal capacity (gas turbine excluded), as
required by law, discussed previously.

•	NO, Control. To meet the NO, reduction targets, a significant recourse to
design measures will be needed. All of the multi-fuel plants, a considerable
portion of oil-to-coal plants, and almost all other oil/gas plants (for a total of 31
GW out of a total of 45 GW thermal capacity) should be equipped with SCR.
Low NO, burners have been forecasted few the combined cycles (6.2 GW) and
for the heavy duty gas turbines used for repowering (2.16 GW).

ENEL investments for its generation system up to 2000 are presently planned to be
about 42.2 blire (1.1.92 money), 24 percent of which is dedicated to environmental
protection measures.

Final decisions are still pending; the results of the examination of the NUG's proposal
could alter the present projections concerning the plant to be commissioned as well as
the related ENEL emissions.

89-12


-------
United States

As discussed above, new facilities are subject to minimum S02 reduction standards and
the need to meet BACT standards. Hence, new US coal-fired facilities will likely be
fitted with high efficiency FGD systems capable of 95 percent removal or greater. Also,
additional capacity will be constructed using gas turbines, combined cycle plants and,
ultimately, clean coal technologies, including pressurized fluidized bed combustion,
ICGCC, etc.

In terms of managing S02 reduction issues confronting other existing utility facilities, the
system of marketable allowances for S02 emissions will allow utilities to, in theory,
determine the best approach for complying with emission reduction targets to fit their
specific equipment, economic, or geographical considerations.

For Phase I of the CAA, from 1995 to 2000, the two prominent strategies most often
applied will be fuel switching and the installation of FGD. In many cases, fuel switching
is a feasible means for meeting Phase I (and often Phase II) emission reduction
requirements for US power plants capable of burning varied qualities of coals. As noted
previously, Phase I provides special incentives (bonus allowances) for the installation of
proven, efficient S02 emissions control equipment such as FGD systems. Based on the
incentives and the level of S02 emissions reduction required for the utility system, it is
anticipated that several major US utilities will elect to install FGD systems during Phase
I. It is not clear, however, if such FGD facilities will be designed in the same rigorous
fashion as those to be constructed for new facilities. In this case, the utility is not
required to meet strict hourly or daily emission limits, and hence is more free to explore
tradeoffs in efficiency (long-term) versus capital.

For Phase II, beginning January 1, 2000, it is generally accepted that the utility industry
will tend to use a number of alternative S02 reduction options. These include the
following;

•	Fuel Switching. Lower sulphur coal supplies are available in the US in the form
of both bituminous and subbituminous coal sources. For certain fuels, advanced
coal beneficiation processes may be considered.

•	Gas cofiring. Natural gas may be used to supplement coal firing.

•	Sorbent Injection. Sorbent injection offers, relative to FGD installations, less
capital-intensive and less effective S02 capture.

•	Installation of FGD systems. FGD systems most often use Mmestone or lime as
the scrubbing agent.

•	Repowering with clean coal technology. Under this option, the plant's existing
steam turbine would be retrofitted with a new gas-fired or coal gasification
circuit.

The flexibility shown above corresponds, in general terms, to the flexibility of the CAA
market treatment. True cost savings realized by a market approach will be subject to the
efficiency of the market. In theory, many of the affected units will receive substantially

89-13


-------
fewer allowances than their current S02 emissions and, therefore, could potentially be
ideal customers for other utilities which either have more demanding reduction targets as
a result of more restrictive state or local emission limits or are installing significant new
FGD capacity. Yet, intervention by regulators or state governments could significantly
hamper free operation of the market

To date, it is not clear how many utilities will actively participate in the allowance
market as a key element of its ongoing compliance strategies. This lack of "willingness"
to participate stems primarily from uncertainty and risk issues, including the following:

•	Regulatory agencies have not finalized decisions on how the allowances will be
valued and handled for ratemaMng purposes.

•	Utilities are also concerned about prudency reviews; some believe a trade can be
viewed as imprudent if the market price drops after the trade is complete.

•	Utility access to the free market set up by the CAA amendments may be limited
by state and local interests; states may enact laws to encourage specific
compliance strategies and maintain jobs.

•	State legislatures or public utility commissions may bow to pressure from public
or environmental groups that view allowance purchases as importing pollution.

According to common economic theory, lack of willingness to participate in the market,
or more precisely, intervention into the free market by either regulatory or political
influences may cause 'Inefficiencies" that ultimately increase the overall cost of
compliance for the US utility industry.

Although many utilities appear hesitant, other groups maintain optimism concerning the
potential for a viable allowance market. The EPA asserts that the ability to trade
allowances should be viewed as a means to reduce the cost of compliance, and that EPA
may try to block legislation or regulation that limits trading. The Chicago Board of
Trade, the New York Mercantile Exchange, brokerage firms, and Wholesale Power
Services are preparing to assist in allowance trades with services that range from futures
contracts and futures options to market-clearing auctions and electronic bulletin board
systems. These services target nonregulated traders such as NUGs, industrial owners,
and speculators, in addition to regulated utilities.

NOs emissions are also targeted by the CAA amendments, based, to a large extent, on
current performance expectations for low NOx burners. Although the legislation does
not specify the use of low NO, burners, it does not require higher performance, as
available, via either catalytic or noncatalytic reduction. For new facilities, expectations
surrounding the level of NO, emissions may be changing; recently, BACT provisions have
been enacted to support the installation of SCR for NOx control. Hence, it is possible
that, in the relatively near future, SCR may be required for the bulk of new fossil-fired
facilities.

89-14


-------
Impact of FGD on Plant Availability

The availability of a thermal plant is clearly a function of its design, the demands placed
on the equipment, its condition, and the effectiveness of operation. Prior discussions
have focused on other important issues and pressures, due principally to the legislative
environment of the country in question. Clearly, more facilities will soon be fitted with
FGD systems; it is important, therefore, to learn from the experience gathered to date.
The majority of FGD experience lies in Germany, Japan, and the US.

Evaluating and analyzing experience is a challenging activity. To assist, this paper
focuses on the more meaningful and potentially understandable issues confronting both
new and existing FGD installations:

•	Overall effects of FGD on unit availability, including impacts of various design
parameters.

•	Effects of familiarity with operation on availability. The German situation is
particularly well suited to consideration of this issue. In this case, many FGD
systems were installed and put in service without the luxury of developing a
"learning curve" on a few units. Alternatively, in the US, installation of FGD
systems has occurred over time; hence, the US industry has had an opportunity
to learn from prior experience.

•	Impacts of legislation on availability.

•	Analysis of the distribution of FGD Failure Data. Trends can potentially offer
important insights.

A complete discussion of these issues is available in the formal work of the WEC
committee;5 the following discussion highlights the key data and findings.

A review of the US and German availability data suggests that FGD systems have, after
a transitional period, had minimal impact on unit availability. limited data available
from Japan also confirms that FGD has a minimal impact on unit availability. The
additional total unavailability (planned + unplanned) in Germany/RWE (lignite fired
plants) appears to be about 2.8 percent on the average from 1986 to 1991, as opposed to
significantly lower values-0.7 percent for 1989 and 03 percent for 1990, at the end of
major transients related with the crew training and teething troubles of FGD technology
and flue duct repair work of 1990. This compares favorably with US results; for
prevailing coal-fired plants, the average total unavailability is 1.35 percent A
comparison of the availability results is shown in Table 2.

Quite low values for the unplanned/forced unavailabilities have been recorded. For
Germany/RWE and the US, average values of 0.5 percent and 0.45 percent, respectively,
can be assumed, based on 3 to 4 years of experience.

5WEC work is published in its entirety in "Thermal Generating Plant Availability and the Environment,"
by Joint UNIPEDE/World Energy Council Committee on the Availability of Thermal Generating Plant,
September 1992.

89-15


-------
Table 2. Comparison of US and German FGD Availability Trends

FGD Effect cm Total Unavailability

Source

Parameter

Type

Period

"Bonus"1

RWE

^An

lignite

1987: 4.5%
1988: 4.7%
1989: 0.7%
1990: 4.0%
1991: 0.3%
Average: 2.8%



US

EUF

111 Units,
90% Coal

Average, 1986-1988:
1.35%



FGD Effect on Unplanned Unavailability

RWE

^An A

Lignite

1987: 0.1%
1988: 0.9%
1989: 0.4%
1990: 0.1%
1991: 0.1%
Average: 0.3%

Yes
2.7%

US

EFOR

111 Units,
'90% Coal

Average, 1986-1988:
0.45%

Yes:

Beginning in
2000 for some
units which
install FGD to
meet 1990
CAA

Amendments

1 "Bonus" refers to time a utlity can legally continue to run the unit even if FGD is unavailable.

US and German findings also indicate that performance of FGD systems has improved
rapidly, in each respective country, since the first FGD systems were installed. It is likely
that the two primary reasons are that FGD technology matured through the cooperation
of manufacturers and utilities, and that utilities gained experience and knowledge in the
operation and maintenance of the FGD chemical cycle.

A closer look at the German and US experiences is warranted.

Germany

One consequence of the rapid implementation of German legislative stipulations has
been that hitherto unfamiliar process technology had to be integrated into existing power
plants. No information was available on the failure frequency of FGD facilities, nor was
there sufficient time to adequately train operating crews in the handling of this new
equipment Given this set of drcumstances, it was initially expected that the nonavail-
ability of power plants would increase, although neither the precise extent of the increase

89-16


-------
was clear, nor whether it could be reduced in the Long term. The most likely outcome
after startup of the new equipment was thought to be that unplanned nonavailability in
particular would rise initially, whereas no noticeable change in planned nonavailability
would occur. This expectation was founded on the premise that previously practiced
inspection cycles would also suffice for new flue gas cleaning equipment.

It was only with time that sufficient experience was gathered to establish precisely which
operating parameters influenced the breakdown frequency of environmental protection
components plus the power stations themselves, as well as exactly to what extent. At this
point, it would appear useful to refer to a "before and after" comparison such as that in
Tables 3 to 6 which show the nonavailability over time of selected coal-fired power
stations in West Germany.

Table 3 clearly charts the increase in planned nonavailability. The increase until 1989 is
largely a result of planned shutdowns in connection with the extensive construction
measures entailed by the installation of flue gas cleaning equipment. The increase in
unplanned nonavailability which commenced in 1988 is, on the other hand, the result of
the inexperience of operating crews in handling the new plant technology, as well as
various other "teething troubles." This becomes particularly clear if the unplanned
nonavailability level from 1988 to 1990 is analyzed cm the baste of technical cause.
Accordingly, the unplanned nonavailability caused by flue gas cleaning components is
decreasing from about a third (1.8 of 5.9 percentage points) in 1988 to less than one
tenth (0.3 of 5.2 percentage points) in 1990.

Table 3. Nonavailability Figures for West German Coal-fired Power Plants
Greater Than 100 MWa£CTf9C (Source: VGB)

Year

1984

1985

1986

1987

1988

1989

1990

Total Nonavailability

11.7

14.2

14.3

17.6

18.9

19.0

17.0

Planned Nonavailablity

7.9

9.2

9.9

13.1

13.0

13.8

11.8

Unplanned Nonavailability

3.8

5.1

4.4

4.5

5.9

5.2

5.2

Due to FGD









1.8

1.0

0.3

Unplanned Nonavailability for hard
coal-fired units









5.6

4.7

6.1

Due to FGD/SCR









0.6

0.6

0.2

Unplanned Nonavailability for
lignite-fired units









6.7

5.1

3.8

Due to FGD/SCR(LNi)









3.0

1.5

0.4

A slightly different picture emerges if lignite-fired power plants are examined in
isolation, as shown in Table 4. Here the drop in planned nonavailability began in 1988,
but increased once again in 1990 due to repair work at the flue gas ducts between the

89-17


-------
boilers and the FGD plants. The decrease of unplanned nonavailability began in 1989 as
in the general trend (compared to Table 1) but, on the contrary, the unplanned
nonavailability of the RWE plants again reached the level it was before the retrofit
campaign had started.

Table 4. Nonavailability Figures for Lignite-fired Power Plants Operated
by RWE Energie AG (Source: RWE's Own Calculations)

Year

1984

1985

1986

1987

1988

1989

1990

1991

Total Nonavailability

8.5

9.9

11.3

18.0

17.6

15.7

16.0

13.5

Planned Nonavailability

&3

6.6

8.8

14.9

12.9

11.7

12.5

10.5

Unplanned Nonavailability

2.2

3.3

2.5

3.1

4.7

4.0

3.5

3.0

As shown in Table 5, FGD system failures were a significant contributor to unplanned
nonavailability in 1988. However, for lignite-fired power plants operated by RWE
Energie AG in particular, construction-related problems with regard to corrosion
protection for flue gas cleaning equipment also had a bearing on the nonavailability of
the plants. These problems are reflected in the slow drop in planned nonavailability.
Table 6 shows the influence of flue gas cleaning equipment on planned nonavailability.

Table S. Planned Nonavailability Caused By FGD Used in Lignite-fired
Power Plants Operated by RWE Energie AG
(Source.* RWE's Own Calculations)

Year

1984

1985

1986

1987

1988

1989

1990

1991

Told Planned Nonavailability

6.3

6.6

8.8

14.9

12.9

11.7

12.5

10.5

Caused by FG Scrubbers







4.4

3.7

0.3

33

0.2

Caused by FG Reheat









0.1







Table 6. Unplanned Nonavailability Caused By FGD Used in Lignite-fired
Power Plants Operated by RWE Energie AG
(Source: RWE's Own Calculations)

Year

1984

1985

1986

1987

1988

1989

1990

1991

Total Unplanned Nonavailability

22.

3.3

2.5

3.1

4.7

4.0

3.5

3.0

Caused by FG Scrubbers







<0.1

0.8

0.3

<0.1

<0.1

Caused by FG Reheat









<0.1

<0.1





89-18


-------
It should be noted that as a result of special provisions in the German regulations,
known as the 72/240 hours rule, FGD plant failures do not (or at least do not fully)
govern the nonavailability of the power plants themselves. As a consequence, up to 2.75
percent annually of unplanned availability may be prevented by the rule.

Overall, it is apparent that when flue gas cleaning "plant construction has passed its peak,
availability losses because of FGD will also drop. It is, however, still too early to say
whether there will be an approximate return to prior nonavailability levels in the future.

United States

To assist utility planners in evaluating the potential implications of adding FGD systems
to existing coal-fired power plants, the North American Electric Reliability Council
(NERC) enlisted the Generating Availability Trend Evaluations (GATE) Working
Group to perform a detailed study evaluating the impacts of FGD systems on overall
unit availability and the sensitivity of unit availability to key FGD design considerations.®

GATE calculated mean and median distributions for unit Equivalent Unavailability
Factors (EUFs) and unit Equivalent Forced Outage Rates (EFORs). As shown in
Table 7, a wide discrepancy in average and median values calculated shows that a small
number of poor performers are inflating the average EUF and EFOR because of FGD.

Table 7. US Availability Statistics for EUF and EFOR as a Function
of Design Characteristics (Source: NERC, US, July 1991)

Category

Mean EUF, due
to FGD, Percent

Median EUF, due
to FGD, Percent

Mean EFOR, due
to FGD, Percent

Median EFOR,
due to FGD,
Percent

Overall

1.35

0.31

0.45

0.06

Spare vs.
No Spare

0.68 vs. 2.13

0.15 vs. 0.®

0.26 vs. 0.68

0.03 vs. 0.14

Bypass vs.
No Bypass

1.31 vs. 1.48

0J21 vs. 1.29

0.36 vs. 0.80

0.06 VS. 0,05

Retrofit vs.
Original

2.21 vs. 1.00

0-40 vs. 0.20

0.45 VS. 0.46

0.05 vs. 0.06

Reheat vs.
No Reheat

0.81 vs 3.28

0.24 vs. 1.25

0.47 vs. 0.38

0.06 vs. 0.045

6NERC/GATE Report titled "Impact of FGD Systems, AvaBabBity Losses Experienced by Hue Gas
Desiifurization Systems," July 1991, by Neath American Electric Reliability Council.

89-19


-------
Based on performance information from 1986 to 1988, it can be concluded that FGD
systems have had minimal effect on unit availability. In 70 percent of the FGD
installations, the FGD system contributed 1 percentage point or less to the total EUF
experienced by the unit At an additional 12 percent of the installations, FGD systems
contributed between 1 and 2 percent of the total EUF, -At only 7 percent of the
installations (eight units), unit EUF was increased by more than 4 percentage points.

Similarly, the FGD systems contributed 0.25 percentage points or less to unit EFOR in
over 67 percent of the FGD systems evaluated, and only 7 percent of the FGD systems
increased unit EFOR by more than 2 percentage points.

When evaluating FGD availability issues, it is critical to evaluate the distribution of the
data, not just the mean or median values. By analyzing distributions for EUFs and
EFORs for overall system performance and as a function of key design, GATE was able
to better understand impacts of such parameters. Additional conclusions drawn by the
NERC/GATE study are as follows:

•	The reduced performance that occurred in FGD systems was principally as a
result of damage to the stacks, plugging of mist eliminators, and repairs to duct
work and absorber towers.

•	No statistically significant performance difference was found in the unit EUFs or
EFORs between units where the FGD was part of the original design compared
to those where the FGD system was retrofit at a later date.

•	The inclusion of a spare scrubber module was the only design element that
proved to be statistically significant in reducing availability losses.

Approximately 30 percent of the units equipped with spare modules experienced
no change in the EFOR as a result of the FGD system. An additional 45
percent showed only slight changes (< 0.25 percent) in the EFOR.

•	In most cases, no difference was found in the current performance of the earliest
and latest designs. Some problems with the earlier designs still exist and
continue to degrade performance.

•	No conclusive difference was found between units equipped with or without flue
gas bypass systems. This impact may differ in the future, however, because units
with FGD systems installed under the CAA amendments may elect to continue
to operate (assuming bypass capability is present) if FGD fails.

•	No statistically significant performance difference was found in unit EUFs or
EFORs between FGD systems equipped or not equipped with flue gas reheat
capability. One anomaly was noted; the few FGD systems that had direct-
combustion reheaters did have significant impacts on unit EUFs and EFORs.

•	Date collected show that, irrespective of unit vintage, there have been "good"
units and "bad" units. However, through learning and recognition of design
problems present in the earlier vintage systems, the magnitude of problems in
newer vintage units is likely to be lower.

89-20


-------
Closer inspection of the data also shows that the gradual improvement in the mean
EFOR is also somewhat misleading in that the EFORs for the top 10 percentile were
essentially unchanged. The change in the mean value is due to a decrease in the
magnitude of problems evidenced in the worst 10 percent of the units (from 20 percent
to 10 percent).

Therefore, it is important to note that careful scrutiny may be required in the future
when analyzing statistical data for FGD systems. Historical data collected is based on
prior legislation; in the future, it may be necessary to consider the type of regulations in
estimating future performance. Because the recent CAA legislation does not impose
mandatory reductions of S02 on specific units, utilities electing to retrofit FGD systems
to reduce S02 emissions may have greater freedom in design than conventional, new-umt
applications. This would be a result of the following:

•	Reduction targets for the FGD system would be on an annual baas rather than
a daily or hourly average.

•	Most FGD installations would be designed to reduce S02 emissions beyond that
required for compliance to provide lower risk and to potentially free up
allowances for sale to others.

•	FGD system removal efficiency will be governed by the perceived value of
incremental S02 allowances liberated rather than by strict control requirements.

•	Component redundancy will be less critical since FGD system availability will
not be as closely tied to unit availability. Assuming adequate bypass capability,
the unit could continue to ran without the FGD system so long as the utility
possessed adequate allowances to cover its emissions.

In addition to the flexibility in design, flexibility in plant operations should be enhanced
as well. Historically, a significant portion of FGD outages for wet systems has been
attributed to problems with process chemistry. With the exception of economic penalties
associated with additional allowances required, the utility can elect to operate the FGD
system in a more conservative manner and use advanced diagnostic tools to minimize
chemistry problems and maximize availability.

Impacts of FGD on Cost

S02 and NO* emissions reductions are normally realized in one of three ways:

•	Design measures (FGD, SCR, and low NOx burners).

•	Fuel switch.

•	A combination of the above two measures.

Irrespective of the alternative chosen, the effect of emission controls on the production
cost will be quite significant. Comparison of such costs among various utilities or
countries is difficult-even at plant level. The bases for comparison (e.g., technical data
such as hours of operation, efficiencies, characteristics of the fuel used; economic data

89-21


-------
such as capital, O&M costs, cost of raw materials; and financial data such as rate of
return, period of the investments, yearly fixed charges) vary among countries, and
information is not readily available. In particular, confidentiality of economic data is
closely guarded, depending to a varying degree on the competition existing in various
countries, on constraints imposed by individual government, and on establishment of rate
structures. Other possible, potentially misleading factors are the currency conversion
rates and fuel costs used, present and projected, which differ from country to country
according to existing situations, markets, or the existence of government subsidies.

Yet, at the same time, it is virtually impossible to justify or interpret results from
different utilities without the benefit of the abovementioned figures. In view of this,
appropriate utilities and organizatons were contacted to gather data, as available to the
public, on costs associated with typical units.

Determination of costs at the generation system level would entail evaluation of the
overall system, through simulation with appropriate tools, for the present and future
operation policies. The final result, even in terms of percentage increase of the total
system production cost, would be greatly dependent on each system structure and, for the
fixture, on the energy policies chosen by each utility within each national strategy.

In reality, the "solution" selected for emission control is often a mixture of design and
fuel switch measures which are, in turn, affected by the standards and fuel sources
applicable to the various countries. Therefore, discussion will be limited to a more
general economic presentation according to the following:

•	Typical sizes of generating units of various utilities worldwide.

•	Percentage increases, in respect to the base case (without any control measure),
of the production cost of such typical units due to design measures such as FGD,
SCR, and low NOx burners.

•	Present cost of fuel in the various countries.

Consequently, cost estimates are to be taken with caution. Since the absolute production
costs of such typical units, and their variations, greatly depend on the fuel cost
assumptions, Figure 3 presents the results of a parametric analysis based on projected
fuel cost

The impact of FGD and SCR on the production cost will vary by country and fuel
source, as shown on Figure 3. On Figure 3, the 100 percent value corresponds to the
basic cost without any control measure, with percentage increases in the costs illustrated
for FGD only and for FGD+SCR. One exception lies in the German lignite unit; only
low NOx burners were required for NOj control. All results are displayed in US
$/MBtu.

Conclusions

Emissions control is a key issue for the electric supply industry worldwide. Various types
of legislation have been enacted. A "command and control" policy is used in Europe and
Japan, with a "market-based" approach being implemented in the US. In general, for the
EC countries, plant-specific constraints are, or will'be, progressively applied. In the US,

89-22


-------
r

140

00


-------
S02 emission reductions will be obtained by placing a national cap on total utility
emissions and a "bubble" over each utility's emissions. The US strategy allows flexible,
utility-shaped compliance strategies to be realized. The US allowance market, if
operated without significant intervention, has the potential to lower the overall cost of
emissions compliance to the utility industry by allowing each utility to determine its
most cost-effective compliance approach, while nevertheless constraining emissions from
the utility sector as a whole.

Even given the differences in legislations, the selection in technologies is similar, with
principal reliance on a combination of fuel switching and installation of FGD systems.
Implementation and technology selection will remain highly country-specific.

A review of the US and German availability data suggests that FGD systems have, after
a transitional period, had minimal impact on unit availability. For the US, in 70 percent
of the units evaluated, the overall unit EUF, as the result of an FGD system, which
includes at large both planned and unplanned outages, was less than or equal to 1
percent For Germany, less data is available for analysis; however, it appears that,
following a "learning" period, the influence of FGD equipment on unplanned outage
rates was less that 03 percent, as calculated for 1990. Limited data available from Japan
also shows that FGD has a minimal impact on unit availability.

The impact of design measures will have a considerable impact on the unit production
cost, roughly in the range of 8 to 23 percent for FGD with a farther 8 percent impact for
SCR, the accuracy of such estimates being limited by the varying financial, monetary, and
energy conditions present in each country.

As environmental pressures continue to be felt around the world, legislative, political,
economic factors, and technologies selected will undoubtedly change and evolve but,
given current experience to date, such changes are likely to affect costs, but not
reliability, of supply. Nevertheless, it is important for utilities worldwide as they install
FGD systems to employ appropriate data collection systems necessary to confirm the
role of FGD on plant availability.

References

1.	Salvaderi L., C. Reese, H. Schlenker, and S. StaUard, "Thermal Generating Plant
Availability and the Environment," Joint UNIPEDE/World Energy Council
Committee on the Availability of Thermal Generating Plant, September 1992.

2.	Salvaderi L, S. Stallard, and C. Reese, "Emissions Control: Comparison of the US
and Western European Approaches," presented at Power-Gen Europe '93, Paris,
France, May 1993.

3.	"Impact of FGD Systems, Availability Losses Experienced by Flue Gas
Desulfurization Systems," North American Reliability Council, July 1991.

89-24


-------
DEMONSTRATION OF PROMOTED DRY SORBENT INJECTION
IN A 75 MWe TANGENTIALLY-FIRED UTILITY FURNACE

D. Rodriguez
C. Gomez
Intevep SA
Caracas 1070A, Venezuela

R Payne
P.M.Maly

Energy and Environmental Research Corporation
18 Mason
Irvine, California

Abstract

The injection of dry calcium based sorbent materials into the furnace space is an S02
control technique which has been thoroughly studied and well documented in recent
years. Application of the technique has however been limited by the relatively low
utilization of typical sorbent materials, and by an ability to achieve only moderate levels
of SO2 emission control

In the study described here, a promoted Ca(OH)2 sorbent material has been developed,
in which small quantities of chemical additives are introduced into the water of
hydration. In pilot scale experimental testing, the resulting sorbent was found to yield a
reactivity towards SO2 which was 30 - 50% greater than conventional sorbent materials
under typical utility boiler conditions.

Injection of the promoted sorbent (a process called PromiSOx™) has recently been
demonstrated in a 75 MWe utility boiler firing a 3% sulfur bituminous coal. Sorbent
production was achieved in a commercial prototype hydration unit with a capacity of 10
tons/hour. Under optimum injection conditions, the promoted sorbent showed
significant improvement in performance, and recorded greater than 80% SO2 removal
at an injection rate of Ca/S = 2.6. Results from this demonstration program are
presented and discussed.

Introduction

The direct injection of dry, calcium based sorbent materials into the furnace space
(Furnace Sorbent Injection or FSI) is a well known SO2 control technique, which has
been frequently demonstrated in recent years. The sorbents most typically employed
are limestone (CaC03) or hydrated lime (Ca(OH)2), injected at flue gas temperatures
around 2200°F, and have been reported to achieve SO2 removal rates in the range of 50
- 60%. The process has been offered as a low capital cost alternative to scrubbers for

90-1


-------
some units, usually where title capacity factor is low, and only moderate levels of S02
removal are required. Major limitations of the basic FSI process have included: low
utilization of the sorbent material (typically 25% for Ca(OH2)); an ability to achieve only
moderate levels of S02 control; and difficulties associated with the collection of the
spent sorbent material in existing particulate collection devices; all of which have tended
to restrict the number of commercial installations.

Throughout the history of the development and evaluation of sorbent injection as an
S02 control technique, many studies have been conducted into the fundamentals of
those processes which dictate sorbent behavior, and into the development of new
sorbent materials to enhance reactivity. Such studies are weE documented in the
proceedings of recent 'Joint Symposia on Dry SO2 (and Simultaneous S02/N0x)

Control Technologies' (1986 -1991, published by EPRI), and are represented in papers
by authors such as: Slaughter et aL; Muzio et al.; Nolan et al.; and Gullet et aL. ITie
techniques employed have used the direct admixture of various additive materials with
the sorbent, or the use of additives combined with the water of hydration in the
preparation of enhanced Ca(OH)2. Mostly, these different studies met with moderate
success, and resulted in sorbent materials with small or no changes in reactivity towards
^2, particularly under conditions typical of utility boiler systems.

In the present study, a similar approach to sorbent enhancement has been adopted,
where small amounts of chemical promoters have been added to quicklime during the
hydration process, but in such a manner that sorbent reactivity has been significantly
improved. The development of this new approach has taken place over a three year
period. It was initiated with fundamental studies into sorbent sulfation, which lead to
the identification of specific promoter chemicals and processing techniques, followed by
studies in both bench- and pilot-scale facilities. Results from the experimental testing
provided sufficient confidence to proceed with the construction of a commercial
prototype hydrator with a capacity of 10 tons of promoted sorbent per hour. This
prototype unit allowed the program to proceed to a full-scale demonstration in a 75
MWe utility boiler firing a 3% sulfur bituminous coal.

Process Description

The promoted dry sorbent injection process will be marketed under the trade name of
PromiSOx™, and is a relatively simple variation on the well known Furnace Sorbent
Injection (FSI) process, where a dry calcium based sorbent material (limestone or
hydrated lime) is injected into the upper region of a furnace firing medium to high
sulfur fuels. A typical utility bote arrangement of this process is presented in Figure 1.

With PromiSOx™, a promoted sorbent material is used, which is based on a calcific
hydrate that is prepared in a patented process in which chemical additives are
introduced into the quicklime during the hydration process. A specially designed
hydrator has been developed for this purpose. The sorbent may be produced either in a
central hydration plant and shipped in bulk containers to the plant site, or it may be
manufactured on-site from quicklime in a dedicated hydration facility.

As with FSI, the promoted sorbent is injected at a location where flue gas temperatures

90-2


-------
are on the order of 2000°F. Under these conditions, the injected sorbent decomposes
in-situ to form CaO which subsequently reacts with SO2 to form a solid, dry CaS04. The
dry product, which consists of CaS04 and unreacted CaO, mixed with the fuel ash,
passes through the boiler to downstream particulate collection equipment, where it is
removed.

Hie sorbent is typically injected into the boiler through an array of wall mounted
nozzles arranged in the upper furnace region of the boiler, where six to eight
penetrations of no more than 4 inches (100 mm) in diameter are usually sufficient to
ensure uniform distribution of the sorbent throughout the furnace space. Injection of
sorbent from two elevations in the boiler may be necessary to ensure optimum S02
removal across the boiler load range. Sorbent feeding is usually from a small day-silo,
by means of a gravimetric feeding system, with pneumatic transport to the injection
locations. Bulk storage of the sorbent is provided in larger silos with a capacity sufficient
for several days' supply.

The application erf any furnace sorbent injection process can result in increased solids
deposition in the convective passes of the boiler, and in degradation of the performance
of existing electrostatic precipitators (ESPs). Available experimental data, and limited FSI
applications conducted to date, suggest that convective pass fouling should not present
a significant problem, and its occurrence can be managed by modifications to standard
sootblowing techniques. ESP performance may however be affected as a consequence
of the higher solids loading, as well as by the modified electrical properties. Mild
humidification of the flue gases, or conditioning by SC)3 injection, have been
demonstrated to be effective in restoring collection efficiency.

Experimental Studies

The promoted sorbent concept was developed initially as a result of fundamental
studies into the mechanisms of S02 reaction with calcium, oxide. These studies provided
insight into the reaction mechanisms, leading to the possibility of modifying the sorbent
particle morphology for enhanced reactivity, both in the initial 50 msec of reaction time
(the so-called 'prompt capture regime', where utilization occurs very rapidly), and at
longer times where utilization is limited by product layer diffusion. This lead in turn to
the identification of certain chemical additives, and to the development of processing
techniques for the production of highly reactive sorbent materials. Experimental studies
were carried out on: sorbent production; sorbent reactivity characterization; and
process evaluation under representative operating conditions.

Initially, sorbent production was carried out in small laboratory scale facilities where
CaO was hydrated under carefully controlled experimental conditions. Parameters
evaluated included: CaO source; additive combinations and quantities;
CaO/water/additive mixing conditions; and reaction time. The promoted sorbent
materials were subsequently characterized for reactivity in a small laboratory
drop-tube furnace, in order to define the optimum ranges of process conditions. The
manufacturing process was extended to a continuous hydration facility developed
specifically for this purpose, and producing 200-500 lb/hr. The production of larger
quantities of material allowed reactivity testing to be conducted in larger bench-scale

90-3


-------
facilities, and ultimately in a fully integrated pilot plant, more closely simulating the
conditions, geometry and auxiliary equipment of typical utility boiler applications.

Most recently, sorbent production has been successfully scaled to a 10 ton/hour
commercial prototype hydration facility.

Reactivity testing of the promoted sorbent has shown that its most important
characteristic is an ability to generate a CaO with, a high in-situ porosity and surface
area, and to retain these features through the temperature window most favorable to
the sulfation reactions. This is different from sorbent materials which begun with a high
initial surface area, but where this potential reactivity is lost to high temperature
processes, such as sintering, before significant reaction with SO2 can occur. In the bench
scale furnace facilities used in the initial phase erf this project, special sampling and
analysis techniques were developed to provide information on the morphology of the
sorbent particles, in-situ, where the sulfation reactions occur. Typical results are shown
in Figure 2, where surface area and porosity are compared for a promoted sorbent
material and a conventional calcific hydrate. Hie superiority of the promoted sorbent is
clearly evident from these data, as is its ability to maintain enhanced reactivity at
temperatures important to the sorbent sulfation process.

Prior to proceeding to a full-scale demonstration of the promoted sorbent process,
considerable proof-of-concept testing was carried out in a 10 x 106 Btu/hr pilot scale
combustion test facility. The facility was designed to simulate conditions of temperature
and time characteristic of utility boiler operation, and provided representative heat
transfer surfaces and downstream equipment for the evaluation of ash deposition and
particulate collection. Testing has been carried out with medium and high sulfur
bituminous coals, and with Orimulsion®, an emulsified natural bitumen from
Venezuela.

Testing has shown that the optimum location for the injection of the promoted sorbent
is slightly lower than that normally employed in PSI, and occurs where the flue gas
temperature is approximately 2000°F (1100°C). Under optimum conditions, the
promoted sorbent showed a reactivity which was between 30% and 50% greater than
that normally found with commercially available calcific hydrates. Experimental data
for the 10 x 106 Btu/hr pilot scale furnace are presented in Figures 4 and 5, for a 3.6%
sulfur bituminous coal, and for Orimulsion® fuel respectively. Hie significant
improvement in SO2 removal performance can be readily seen from these figures, as
can the tendency for the promoted sorbent to retain its reactivity at relatively high
overall S02 removal rates. SQ2 removal in excess of 80% was achieved with sorbent
addition rates corresponding to Ca/S molar ratios on the order of 2.5.

In addition to parametric testing for S02 removal performance, the promoted sorbent
has been subjected to several long term teste to evaluate potential impacts on furnace
ash deposition and heat transfer performance, and on the collectability of the fly ash in
downstream particulate collection devices. The results of the pilot-scale testing showed
that the promoted sorbent behaves in a manner consistent with that observed in boilers
and other experimental facilities operating with conventional FSL Compared to
operation without sorbent injection, there was an increase in ash deposition on
convective pass surfaces, although the ash deposits were found to remain friable and
were readily amenable to removal by sootblowing. The rates of ash deposition were

90-4


-------
comparable to those for conventional sorbent materials at similar mass addition rates.
Similar conclusions were also drawn with respect to collection of the fly ash particulate
in the facility baghouse and electrostatic precipitator.

Demonstration Testing

The host site for the promoted sorbent demonstration program was Illinois Power's
Hennepin Station Unit 1, which is a 71 MWe (net) tangentially-fired boiler burning a 3%
(as received) sulfur Illinois bituminous coal. In 1990, this unit was retrofitted with
systems for Gas Reburning (for NOx control) and Sorbent Injection (for S02 control), as
part of a demonstration program under Round 1 of the DOE Clean Coal Technology
program. Since that time, the unit has been operated to provide both parametric and
long-term evaluation erf the combined NOx and S02 control technologies, under the
joint sponsorship erf the U S Department of Energy (DOE), the Gas Research Institute
(GRI) and the State of Illinois Department of Energy and Natural Resources (ENR). A
general arrangement of the Hennepin unit is presented in Figure 5, showing the gas
reburn and sorbent injection systems, together with the flue gas humidification system
which provided conditioning in order to maintain performance of the ESP in the
presence of spent sorbent

Testing of the promoted sorbent was carried out over a nominal three-week period in
December of 1992. The promoted sorbent was produced in the 10 ton/hour prototype
hydration facility, trucked to the plant site in bulk carriers, and then off-loaded into the
main sorbent storage silo. The sorbent feeding and injection system developed for the
GR-SI demonstration program was retained un-modified for the promoted sorbent
testing. The feeding system employs a Fuller-Kinyon pump, which provides for dense-
phase transport of the sorbent in compressed air to the injection elevation, at which
point a booster fan provides additional air to support injection. The injection system
itself comprises an array of six three-inch diameter ports, located at the boiler nose-arch
elevation where flue gas temperatures are nominally in the range of 2250 - 2300°F. This
injection elevation was known to be non-optimum for the promoted sorbents, which
are known to prefer somewhat lower temperatures. A significant benefit to this
program however was that a considerable data base on sorbent injection and associated
boiler impacts had been developed during the course of the GR-SI demonstration. Since
the promoted sorbent was expected to behave similarly to the conventional,
commercially available, sorbents, this meant that the program could concentrate on
parametric testing.

During the course of the demonstration program, the effects of a large number of
operating variables were evaluated in order to identify optimum conditions for
promoted sorbent injection. These included parameters such as: burner tilt, excess
oxygen, boiler load, sorbent injection velocity, and whether gas reburning was
implemented or not Results from the testing are presented in Figure 6 for nominally
optimum injection conditions. The figure shows percentage S02 removal as a function
of the injected Ca/S molar ratio, where both parameters are related to the initial actual
S02 level to account for the reduced S02 concentration when gas reburning is
implemented. As can be seen from Figure 6, the promoted sorbent results in a
significant improvement in both sorbent utilization and S02 capture, relative to the
unpromoted material which was a commercial calcific hydrate from Linwood. At a

90-5


-------
sorbent injection rate of Ca/S = 2.6, more than 80% S02 removal was recorded,
resulting in a final SO2 concentration on the order of 500 ppm (at 3% O2).

Parameters which were found to most significantly influence sorbent utilization were
those impacting temperature at the injection elevation. Reduced burner tilt, increased
excess air, and reduced boiler load were all found to improve sorbent utilization to
some extent. The general impact of temperature on utilization is shown on Figure 7,
where temperatures have been estimated from available measurement data and from
boiler heat transfer modelling. Figure 7 suggests that the optimum injection
temperature for the promoted sorbent is on the order of 2050°F, at which point the
measured sorbent utilization ranges from 40% to about 70% higher than the normal,
unpromoted calcitic hydrate.

In addition to measurements of SO2 removal, a number of additional measurements
and observations were made concerning performance parameters such as sorbent
feeding, ash deposition in the boiler convective passes, and ESF performance. Such
observations were generally made relative to the available experience base with
conventional sorbent injection, and these are briefly summarized below.

In terms of sorbent feeding, no significant problems were encountered, and the sorbent
was found to flow freely through all stages of the feeding and injection system. Only
tihe maximum capacity of the system was somewhat reduced due to a slightly lower
bulk density of the promoted sorbent.

Sorbent injection into the furnace can add significantly to the particulate loading in the
boiler convective passes such that tube surface fouling and reduced heat transfer is of
potential concern. In general, although there is an increase in the rate of solids
deposition, experience has shown that the deposits remain friable and easily removed
by sootblowing. During the demonstration testing, heat balances were routinely
performed across individual convective tube banks in order to monitor fouling and to
calculate cleanliness factors. Results from such evaluations are presented in Figure 8 for
the secondary and primary superheaters and for the reheater sections, and for both
promoted and unpromoted sorbents. It can be seen from the rates of reduction in the
cleanliness factors that deposition is comparable for the two different sorbents, and that
application of the sootblowers tends to recover the initial starting values.

As part of the GR-SI demonstration program a humidification system was installed on
the Hennepin unit as a means of conditioning the flue gases in order to compensate for
the impacts of the sorbent modified fly ash cm ESP performance. The humidification
system comprises an array of twin-fluid atomizers which inject water into the flue gas
duct at a location just downstream of the air heater. Approximately 2 seconds of
residence time is available prior to the entrance to the ESP. The normal mode of
operation with sorbent injection is to add sufficient water to maintain the stack opacity
level comfortably below 20%, at which point measurements have shown the total solids
loading to be well below the permitted 0.10 lb /MMBtu. For both the promoted and
unpromoted sorbents the necessary humidification rate was found to be nominally the
same, and equivalent to that necessary to reduce the flue gas temperature to
approximately 250°F.

90-6


-------
Process Economics

Economic evaluations have indicated that promoted sorbent injection can be a cost
competitive S02 control technique for a wide range of applications, including utility and
industrial boilers, and incinerators. The process requires relatively small modifications
to the existing boiler system, and since additional equipment requirements are modest,
tiie initial capital investment necessary is low. The process is therefore ideally suited to
retrofit applications, and can be applied with most medium to high sulfur fuels (coal, oil,
Orimulsion®, etc) As with FSI, it may be most readily applied in coal fired units, or
plants designed for coal, where there is an existing particulate control device.

Economic analyses relative to wet-FGD and FSI have been carried out using the EPA
Integrated Air Pollution Control System (IAPCS) cost estimating model, and using data
obtained in the demonstration program. The results of selected economic comparisons
are presented in Figure 8 for a typical 100 MWe boiler application firing a 2.6% sulfur
coal, and where the retrofit difficulty for a wet-FGD system would be considered
moderate. The figure compares initial capital and operating costs (on a relative basis)
for wet-FGD, ESI, and PromiSOx™. The significant potential advantages of the
promoted sorbent process, in terms of both capital and operating costs, can be seen for
this particular application. Particularly attractive applications of the process are likely to
be found in smaller boilers (<300 MWe), where moderate levels of S02 control are
required, and in situations where the high capital costs of flue gas scrubbing cannot be
justified.

Acknowledgements

PromiSOx™ is a trade mark of Intevep SA (an affiliate of Petr61eos de Venezuela). The
authors wish to thank Intevep SA for their authorization to publish this paper.

The authors also wish to thank the staff at Hennepin Station for their cooperation and
assistance, and to acknowledge the US Department of Energy (DOE), the Gas Research
Institute (GRI), and the State of Illinois Department of Energy and Natural Resources
(ENR) for making the site and equipment available under the Clean Coal Technology
Program.

90-7


-------
Figure 1. Conceptual sorbent injection system.

1200	1400	1600

Injection Temperature (K)

0.4-

0.35-

«

o
»-
o

a

0.3-

0.25-

0J2-

Promoted
Sorbent

%

Unpromoted V
Sorbent	&

1200	1400	1600

Injection Temperature (K)

figure 2. Comparison of in-situ surface area and porosity data.

• 90-8


-------
Figure 3. SO2 removal performance - Coal. Figure 4. SO2 removal performance - Orimulsion®.

Sorbent Injection System

Transport Air Blower

Figure 5. Overview of the GR-SI and Humidification systems at Hennepin.

90-9


-------


a,

88

u

100
90
80
70
60
50 E-

« 40

o

m 30
20
10
0

PromiSOx

o

:x'

D • / •

^ ™ TTnnrrwnni

••

Unpromoted

V

/

/

A

111 • 1111111111



i.

.0 0.5 1.0 1.5 2.0 2.5 3.0
Ca/S

Figure 6. Effects of Ca/S on sulfure capture.

1800

PromiSOx

Unprompted

2000	2200

Injection Temperature (°F)

2400

Figure 7. Effects of temperature on utilization.

90-10


-------
— Unpromoted I,in wood : Ca/S=1.75
	 PromiSOx: Ca/S=1.60

Hennepin Unit 1

Secondary Superheater
Sootblowing

Rch eater

' 		



Primary Superheater

50

100
Minutes

150

200

Figure 8.

Impacts of sorbent injection and sootblowing on convective
pass cleanliness factors.

HI Capital Costs
(3 Operating Costs

90% DeSOx

50% DeSOx

80% DeSOx

Wet FGD	FSI	PromiSOx™

Figure 9. Economic comparison.


-------
DEVELOPMENT OF PROCESS TO SIMULTANEOUSLY SCRUB N02 AND S02

FROM COAL-FIRED FLUE GAS

V.M. Zamansky, R.K. Lyon, A.B. Evans, J.N. Pont, W.R. Seeker
Energy and Environmental Research
18 Mason, Irvine, CA 92718

C.E. Schmidt

Department of Energy, Pittsburgh Energy Technology Center
P.O. Box 10940, Pittsburgh, PA 15236

Abstract

Energy and Environmental Research Corporation (EER) has conducted extensive
research under contract to DOE to further develop and test a new process capable of
simultaneously scrubbing SO2 and NO2. This process involves injection of methanol
into combustion flue gas to promote the oxidation of NO to NO2. Nitrogen dioxide can
subsequently be removed, along with the sulfur dioxide, in a liquor-modified wet-
limestone scrubber. The focus of this study was to determine a liquor composition and
a scrubbing method capable of removing both compounds effectively. This paper
presents the results of kinetic modeling, and bench and pilot scale experiments of the
scrubbing study.

Introduction

The CombiNQx process is being developed to provide a low cost method of controlling
the NOx emissions of coal-fired utility boilers to very low levels. This process
incorporates a family of NOx reduction technologies including staged combustion,
Advanced Rebuming (i.e. reburning combined with selective non-catalytic reduction)
and methanol injection to convert NO to NO2 which can then be removed by wet
scrubbing. While individually, these technologies are limited in their NOx reducing
capabilities, in combination they are capable of reducing NOx emissions to extremely
low levels at a fraction of the cost of selective catalytic reduction.

The methanol injection step, however, is subject to the limitation that one must have a
scrubber that will remove NO2. Hie removal of SO2 and NO2 by sodium based wet
scrubbers is a well-established technology, but the majority of wet scrubbers currently
in use are calcium based. Accordingly, this study was undertaken to determine
whether or not the chemistry which occurs in calcium based scrubbers could be
modified to allow removal of NO2 as well as SO2.

Bench-scale experiments were performed in conjunction with chemical computational
modeling to evaluate the effect of scrubbing solution composition on SO2 and NO2
scrubbing efficiency. In addition, potential byproducts were identified. Finally, larger

91-1


-------
pilot scale tests were performed with a packed tower scrubber to address scale-up
issues and confirm the bench-scale results.

Bench scale studies

The bench scale NO2 scrubbing apparatus is displayed in Figure 1. A simulated flue gas
containing variable amounts of NO2 and SO2 was flowed through a constant
temperature, 80 cubic centimeter, bubbler containing the scrubbing solution to be
evaluated. After passing through the "scrubber", the gas was analyzed for O2, NOx,
N20, and SO2 to determine removal efficiencies.

Figure 2 graphically summarizes the bench scale results. The numbers symbolize
various cases that were performed while varying slurry composition, the lines show the
resulting performance as a function of time. A Ca(OH)2 solution (2 percent Ca(OH>2 by
weight) achieved 99+ percent SO2 reduction, indicating that the scrubber provides good
mass transfer. However, only 50 percent of the NO2 was removed. Reference 1
indicates that the crucial reaction for NO2 scrubbing is:

N02 + SO32- = N02- + SO3-	(1)

With pure Ca(OH)2 solution, the necessary sulfite ion (SO32-) tends to precipitate out as
calcium sulfite, instead of reacting, as desired, with NO2-

Since sodium sulfite (Na2$03) can provide the necessary sulfite ions for N02 absorption,
a 2 percent solution was evaluated. 99+ percent SO2 and 89 percent NO2 capture was
obtained initially, however, this performance decreased after approximately 2 minutes
of run time. After these 2 minutes, SO 2 reduction became negative and N02 reduction
dropped to 17%. SO2 is captured via the reaction

SO2 + SO32- + H20 = 2HSO3-	(2)

and NO2 via reaction (1). Unfortunately, the SO3- generated by NO2 removal is a chain
carrier in the oxidation of sulfite and bisulfite ions to sulfate and bisulfate ions.
Oxidation of the bisulfite ion to the bisulfate ion acidifies the solution, forcing SO2 back
into the gas phase.

Adding Ca(OH)2 to the Na2S03 scrubbing solution eliminates the problem of SO2
rejection by keeping the solution basic, however the NO2 capture remains poor and
short lived. Replacing the highly soluble Ca(OH)2 with very low solubility CaCOs
increases the concentration of sulfite ion that stays in solution. This improves N02
absorption, but the sulfite ion quickly oxidizes to sulfate, hampering N02 removal.
Experiments were conducted under conditions which inhibited the oxidation of sulfite
to sulfate (Le. decreasing reaction temperature and/or flue gas oxygen content). Even
though these conditions can not be applied to a real application, they did show that if
sulfite ion stays in solution, N02 capture improves and can be sustained for a longer
period of time.

91-2


-------
Sodium thiosulfate (Na2S20g) has been proposed by Gorin et al.2 as a method of
inhibiting the oxidation of sulfite to sulfate. We found that by adding this compound to
the scrubbing solution (1% by weight), 95% NO2 capture was achieved.and this capture
was sustained for the same length of time that the solution captured SO2.

The resulting recommended slurry solution, capable of scrubbing 99+ percent SO2 and
95 percent NO2 at bench scale level is:

9% Na2C03
9% CaCOs
1% Na2S203.

This scrubbing mixture was studied using computer modeling with the mechanism
described by Chang et al.l For the batch experiments, the reaction was found to
proceed in three stages. During the first stage, carbonate ion concentration falls while
the concentrations of sulfate, bisulfate and sulfite ion increase. This first stage ends
when the ratio of carbonate to sulfite ion becomes so low that calcium carbonate starts
to dissolve while calcium sulfite precipitates. During this second stage the ratio of the
concentrations of carbonate and sulfite ion are constant and the pH remains fairly
steady. The concentration of sulfate ion increases until calcium sulfate starts to
precipitate. The second stage ends when the calcium carbonate is exhausted. When this
happens the pH begins to rise until it reaches a level at which SO2 absorption fails and
scrubbing solution is spent.

Hie model predicts that NO2 capture will remain effective through these three stages of
the process, but the fate of absorbed N02 changes. During the first and second stages
nearly all of the absorbed N02 will be present as nitrite ion, with only trace amounts of
the complex nitrogen sulfur ions being formed. During the third stage, however, the
nitrite ion is converted to complex nitrogen sulfur ions, chiefly the aminetrisulfonate
ion. Near the end of the third stage the aminetrisulfonate ions are hydrolyzed to
sulfate ion and sulfamic acid.

While the model contains reactions which are capable of forming N20 and nitrate ions,
these reactions are only significant at very low pfi During the bench scale experiments
samples were taken of the bubbler's exhaust and, consistent with the models
predictions, no significant N20 production was observed. Measurements were also
made with EM Quant test strips on spent scrubbing solution. As one would expect
from the model, nitrite ion was found in fresh solution but not in solution which had
been allowed to age. Nitrate ions were not detected in the spent solution.

Pilot scale studies

Scale-up effects were investigated in two different pilot-scale facilities corresponding
nominally to heat inputs of 0.35 MBtu/hr and 10 MBtu/hr. The small pilot-scale
scrubber tests were performed by Research-Cottrell using the facility illustrated in

Figure 3. The small pilot-scale scrubber consists of a propane combustor, absorber
tower, absorber feed tank, analytical train, and solid disposal system. To simulate a
coal-fired flue gas, variable amounts of S02 and N02 were doped into the exhaust

91-3


-------
upstream of the absorber tower. At the absorber tower exit, NOw SO2, CO, and O2
were measured. The absorber tower is a vertical, stainless steel, 16 inch diameter tube,
approximately 20 feet in height. Hie simulated flue gas enters the . tower from the
bottom, travels through five sections to the top, and exits to the gas sample
conditioning systems and analyzers. The first two sections can be packed with a light
packing material to provide improved gas/liquid contact. Sections of the tower may
also be removed, if desired, to reduce absorber tower residence time. The scrubber
slurry is continually being mixed with dry limestone, sodium salts, and water in the
absorber feed tank. From the 200-gallon feed tank, the slurry is pumped to the top of
the absorber tower and dispensed in counter flow to the flue gas with a single slurry
nozzle. The slurry solution is drained by gravity from the bottom of the tower back to
the feed tank.

The first tests were performed to verify that SO2 removal was possible on the pilot scale

scrubber. For a 6 percent limestone slurry, flue gas flow rates were varied between 127
- 140 cfm, and slurry flow rates were maintained at 12 gpm. Up to 99 percent SO2
removal was obtained indicating satisfactory mass transfer.

Simultaneous scrubbing of NO2 and SO 2 was evaluated using a mixture of solid
scrubbing salts consisting of 49.5 percent CaCOs, 49.5 percent Na2C03, 1 percent
Na2&zQ3. Note that this is approximately 1/5 of the concentration of Na2S203 utilized
in the bench-scale tests. The mixture of salts or pure limestone were fed to the
absorber tank with different feed rates. During this test series, the following
parameters were varied: liquid/gas ratio (liquid flow and gas flow were independently
varied), concentration of sodium carbonate in slurry, concentration of sodium
thiosulfate in slurry, and initial NO2 concentration. NO2 removal efficiency ranged
between 65 and 90 percent while maintaining 97 - 99 percent SO2 removal.

The ratio of slurry flow rate to flue gas flow rate is defined as the liquid to gas ratio
(L/G), and is expressed here in unite of (gallons of slurry)/(1000 cubic feet of gas). Hie
slurry flow arid gas flow were varied independently of one another. Figure 4
summarizes the effects of L/G ratio on NO2 scrubbing efficiency. As would be
expected, a larger L/G ratio results in higher NO2 removal. Even though not depicted
in the figure, data indicate that gas flow rate has a larger affect on the NO2 scrubbing
efficiency than slurry flow rate. By decreasing the gas flow rate by a small fraction
(from 135 to 115 cfm), efficiency increased from 77 to 84 percent However, when the
slurry flow rate was nearly doubled (11.4 to 20 gpm), the efficiency only increased by
approximately the same amount, 77 to 85 percent.

Experiments were performed to determine the effect of initial NO2 concentration on
NO2 scrubbing efficiency. Slurry flow rate remained approximately constant as inlet
N02 concentration was varied by adjusting the doping gas flow rate. Results are
displayed in Figure 5. The general trend shows that scrubbing effectiveness drops as
initial NO2 concentrations increase.

The effect of Na2C03 concentration on NO2 and SO2 scrubbing efficiency was evaluated
by diluting the scrubbing solution by a factor of two, while continuing to add limestone
to maintain pH and ion concentration. Figure 6 shows that even though scrubbing

91-4


-------
efficiency was initially hampered by the dilution that occurs 200 minutes into the test,
with time the NO2 removal efficiency rose again to approximately the same level as
before the dilution. Even after diluting the slurry a second time, the NO2 scrubbing
efficiency returned to almost the original value These data indicate that NO2 removal
efficiency is not very sensitive to Na2CC% concentration in this range.

The effect of sodium thiosulfate concentration was tested by adding an additional 3.8
mmol of sodium thiosulfate per liter of solution. Figure 7 shows that NO2 removal
efficiency jumped from 65 percent to 89-90 percent within 15 minutes, while SO2
removal efficiency remained at 99+ percent The thiosulfate inhibits oxidation of sulfite
to sulfate, sustaining the presence of sufficient sulfite ions for NO2 capture. Summary
of bench-scale and pilot-scale scrubbing results is shown in Figure 8. From 90 to 95%
NO2 reduction was achieved at 99+% SO2 removal.

Throughout the experiments discussed above, scrubbing solution composition
measurements were periodically taken. In general, these observations were consistent
with the model and the assumption that the scrubber was operating in the second stage
(see previous discussion). The model is, however, limited in its ability to account for
sulfite ion oxidation, therefore, not surprisingly, sulfite to sulfate conversion was much
higher than predicted. The pilot scale studies also show nitrate ion as a major product,
contradicting both model's prediction and the bench scale results.

A single test was performed in lER's large pilot-scale facility. The large pilot-scale
scrubber facility consists of a simple spray tower with a pad-type demister. The spray
tower is 6 ft in diameter and 16 ft high, with an array of 16 spray nozzles. The
scrubbing solution consisted of an original composition of 9 percent CaC03,9 percent
Na2CC>3 and 1 percent ^28203. This slurry was circulated through the scrubber while
the furnace was firing a high sulfur coal for three consecutive days of testing. This
circulation allowed accumulation of sulfites within the slurry which is necessary to
achieve NO2 removal. The actual scrubbing experiment took place using natural gas
fired flue gas doped with 74 ppm of NO2- SO2 was not added. The test was conducted
at a L/G ratio of approximately 30 gal/103acf. The test result is shown in Figure 4.
Much higher NO2 removal was achieved than in the small pilot-scale tests due to the
much higher concentration of sodium thiosulfate used in the large pilot-scale test. The
results do indicate that high NO2 removal efficiencies can be achieved even with a
relatively primitive scrubbing system operating at L/G ratios similar to the of large
commercial scrubbers.

Discussion

Hie primary goal of this research, demonstration of efficient NO2 and SO2 scrubbing in
a calcium based wet limestone scrubber, has been achieved. Two important questions,
however, remain with respect to the disposability of the products produced by this
modified scrubber. First, there is the question of whether or not using sodium
compounds in the scrubbing solution will result in unacceptable sodium contamination
of the calcium- sulfate/sulfite product Since wet scrubbers require considerable
amounts of makeup water, it is theoretically possible to solve this problem by washing

91-5


-------
the calcium sulfate/sulfite with the makeup water, but this option would require an
engineering design study which has not been done.

The second question of disposability regards the formation of nitrate ion seen in the
pilot scale experiments. This ion formation was not detected during bench scale or
computer modeling studies. One possible explanation for this discrepancy is that as the
scrubbing liquid passes downward through the absorber tower, it readies a point at
which its ability to absorb SO2 is completely exhausted. This low pH condition was not
considered in the modeling study, yet it may responsible for the increase in nitrate
concentration. Mechanisms within the model do, in fact, show nitrate ion formation at
low pH levels.

While possible formation of nitrate ion will require further research, there are a number
of ways in which this problem might be solved. Adjustment of scrubbing conditions so
that the solution is prevented from over-reacting may prevent nitrate formation.
Alternatively, if nitrate forms by oxidation of nitrite ion, the removal of nitrite ion by
reaction with NH2SO3H may prevent nitrate formation, or, all else failing, nitrate ion
could be removed by selective reduction with scrap aluminums. In a brief study, the
authors found that this method works quite well with shredded soda cans as the source
of aluminum.

Acknowledgments

This work was funded under DOE Contract No. DE-AC22-90PC90363, "Development of
Advanced NOx Control Concepts for Coal-Fired Utility Boilers". EER would like to
acknowledge Mr. Charles E. Schmidt, Department of Energy Program Coordinator of
the CombiNOx program. The small pilot-scale experimental scrubbing studies were
conducted by Research-Cottrell under subcontract to EER.

References

1.	S.C. Chang, D. Littlejohn and N.H. Lin, Flue Gas Desulfitrization, ACS Symposium

Series 188, American Chemical Society, p. 128 -152, (1982).

2.	E. Gorin, MD. Kulik, R.T. Struck, U.S. Patent 3,937,788 (1976).

3.	A.P. Murphy, Chemical Removal of Nitrate from Water, Nature 350,223-225 (1991).

91-6


-------
Figure 1. Experimental set-up for NO2/SO2 scrubbing experiments.

91-7


-------
1	2% Ca(OH)2

2	2^%Na2»33

3	2.2% NajSOg, 9% Ca(OH)2

4	2.5% NagSO,, 10% CaCOg

5	9% Na^C03,9% CaCOa

6	9% N^COg, 9% CaC03,1% N^S203

_L

X

X



-10 0 10 20 30 40 50 60 70 80 SO 100 110 120 130 140 150 160 170 180 190 200

Time (min)

Figure 2. Bench scale NC^ scrubbing results.

91-8


-------
Demister

• Outlet Gas





Sampling

Datms

1

/l\

ooo
ooo
ooo
ooo
ooo
ooo
ooo
ooo
ooo
ooo
ooo
ooo

ooo
ooo
ooo
ooo
ooo
ooo
ooo
ooo
ooo
ooo
ooo
ooo
ooo
ooo
ooo

AT

"L

TC

no2-»no
Converter

Analytical
Train

Gas
Conditioner



NCVNO*

CO

o,

NO2—- NO
Converter

	TC

Air

I

1

OR

-L

H2°

I

1
1
J

I
I
I

~

* _

Absorber
Feed Tank

a r

TC

—<^FeederJ

Pump

o-

AT -Abso iter Tower
MFM - Magnetic Flow Meter
M - Manometer
N - Nozzle

TC-Thermocouple
R- Rotameter
P- Packing Materia!
OR - 2.5" Orifice

Figure 3. Schematic of the pilot-scale scrubber.

91-9


-------
90

80

o

m
>

70

©
cc

CM

o
z

60

50

~ 2 MMBtu/hr Pilot Scale Tests

O 10 MMBtu/hr Pilot Scale Tests
j	j	!	

20

40

60

80	100

->3,

120

140

160

L/G, apm/1(r acfm
Figure 4. Effect of liquid to gas ratio on NC^ removal.

90

80

to
>

70

®
cc

CM

O

60

50

50

100	150

[N02J, ppm

200

250

Figure 5. Effect of initial N0_ concentration, on NO^ removal

91-10


-------
100

Slurry diluted by
1/2.

200	300	400

Time (min)

600



100



90



80

5*

70





O



c



o

60

a

lD



?

50

£!



s

40

&



o~

30

2





20



10



0

Figure 6. Effect of N^COg dilution on N02 scrubbing performance.

UG ratio = 83 gal/1000 cf

Addition of 3.8
mmol/L of
Thiosulfate

10	20	30	40

Time (min)

50

60

Figure 7. Effect of thiosulfate concentration on N02 scrubbing performance.

91-11


-------
100-/

80-*-

N02 Removal

S02 Removal

Slurry Comoostion:
9% Limestone
9% Sodium Carbonate
0.5-1% Sodium Thiosulfate

Bench-Scale

Pilot-Scale

Figure 8. Bench-scale vs. pilot-scale scrubbing results.

91-12


-------
The NOXSO Clean Coal Technology Project:
Commercial Plant Design

James B. Black
Mark C. Woods
John J. Friedrich
Clay A. Leonard

NOXSO Corporation
P.O. Box 469
Library, Pennsylvania 15129

Abstract

The NOXSO process is a dry, post-combustion flue gas treatment technology which
uses a regenerable sorbent to simultaneously adsorb sulfur dioxide (SOj) and nitrogen oxides
(NOj) from die flue gas of a coal-fired utility boiler. In the process, the SOj is converted to
a sulfur by-product and the NOx is reduced to nitrogen and oxygen. Based on pilot plant
results, the process can economically remove 90% of the acid rain precursor gases from the
flue gas stream in a retrofit or new facility. The NOXSO Clean Coal Technology Project
will demonstrate the NOXSO process on a commercial-scale. The project is co-funded by
the U.S. Department of Energy (DOE) under round III of the Clean Coal Technology
program. The DOE manages the project through the Pittsburgh Energy Technology Center
(PETC). The NOXSO process, the plant layout, and economics for a commercial-scale plant
are described in this paper.

92-1


-------
Introduction

The NOXSO process is a dry, post-combustion flue gas treatment technology which
uses a regenerable sorbent to simultaneously adsorb sulfur dioxide (S(X) and nitrogen oxides
(NO,) from the flue gas of a coal-fired utility boiler. In the process, the SQ> is converted to
a sulfur by-product (elemental sulfur, sulfuric acid, or liquid SOj) and the NO* is reduced to
nitrogen and oxygen. Based on pilot plant results,the process can economically remove 90%
of the acid rain precursor gases from the flue gas stream in a retrofit or new facility.

Process development began in 1979 starting with laboratory scale tests and
progressing to pre-pilot scale tests (3/4-MW) and a life cycle test. Each of these test
programs have provided data necessary for the process design. Tests of the NO, recycle
concept which is inherent to the NOXSO process have been conducted on small boilers at
PETC and the Babcock & Wilcox (B&W) Research Center in Alliance, Ohio.4 A 5 MW
Proof-of-Concept (POC) pilot plant test at Ohio Edison's Toronto Plant in Toronto, Ohio was
recently completed.5 The Clean Coal Project is currently in the project definition phase
incorporating recently obtained pilot plant data into a commercial-scale design.

The commercial demonstration of the NOXSO process will be cost shared by the
Department of Energy through the third round of the Clean Coal Technology program
through a cooperative agreement between DOE and NOXSO. The cooperative agreement is
currently being assigned (novated) to NOXSO by Morrison Knudsen Corporation - Ferguson
Group and will be formally executed shortly. NOXSO will provide overall management for
the project while MK-Ferguson will provide engineering and construction services. NOXSO
will conduct the operation phase of the project. W.R. Grace & Co.-Conn will be the sorbent
supplier. DOE will manage the demonstration project through the Pittsburgh Energy
Technology Center (PETC).

NOXSO Process Description

Flue gas is drawn from the power plant duct work either upstream or downstream of
the particulate collection device through the flue gas booster fan. Figure 1 shows a process
flow diagram with flue gas drawn from the particulate collection device discharge. Figure 1
shows single pieces of equipment, however multiples will be used as required to provide the
necessary capacity. Tail gas from the sulfur by-product plant is mixed with the flue gas at
the booster fan suction. The flue gas then passes through a two-stage, fluidized bed adsorber
where S02 and NOx are simultaneously removed using a high surface area ^-alumina sorbent
impregnated with an alkali material. Water sprays into the fluid beds maintain a 250°F
temperature by evaporative cooling. The cleaned flue gas passes through a particulate
separator and is returned to the power plant before it exits through the chimney. Sorbent
fines removed by the separator are directed to the dense phase transport system.

Sorbent is removed from the adsorber and transported to the sorbent heater by a dense
phase pneumatic conveying system. Make-up sorbent to maintain the sorbent inventory is
added downstream of the adsorber. The sorbent heater is a variable area five-stage fluidized
bed where a hot air stream is used to raise the sorbent temperature to 1150°F. During the

92-2


-------
Figure 1. Process Flow Diagram


-------
heating process, NOx and loosely bound S02 are desorbed and transported away in the
heating gas (NOx recycle) stream. This hot air stream at SOOT can be used to heat a slip
stream of the power plant's main condensate before being injected into the combustion air
system upstream of the combustion air preheater. Hie NO% recycle stream provides
approximately 30% of the required combustion air. Upon entering the boiler, a portion of
the recycled NO, is converted to nitrogen (Nj) and either carbon dioxide (CO2) or water
(H20) by reaction with free radicals in the reducing atmosphere of the combustion chamber.
NOx recycle studies were performed during a previous NOXSO test program (a 3/4 MW pre-
pilot scale test). More recently, NOx recycle studies were conducted using a scaled model
cyclone boiler.4

Once the sorbent reaches a regeneration temperature of 1150°F, it is transported by
means of a J-valve to the moving bed regenerator. In the regenerator, sorbent is contacted
with natural gas in a countercurrent maimer. The natural gas reduces sulfur compounds on
the sorbent (mainly sodium sulfate) to primarily SOj and hydrogen sulfide (H2S) with some
earbonyl sulfide (COS) also formed. Some of the sodium sulfate (N%S04) is reduced to
sodium sulfide (Na2S) which is subsequently hydrolyzed in a moving bed steam treatment
reactor which follows the regenerator. A concentrated stream of H2S is obtained from the
reaction of steam with Na2S. The offgases from the regenerator and steam treater are
combined and sent to a sulfur by-product plant which produces elemental sulfur, sulfuric
acid, or liquid S02. The tail gas stream from the sulfur by-product plant is recycled to the
suction of the flue gas booster fan.

From the steam treatment vessel, the sorbent is transported by means of a J-valve to
the sort>ent cooler. The cooler is a five-stage variable area ftoidized bed using ambient air to
cool the sorbent. The warm air exiting the cooler is further heated by a natural gas fired in-
duct heater before being used to heat the sorbent in the fluidized bed sorbent heater. The
sorbent temperature is reduced in the sorbent cooler to the adsorber temperature of 250°F.
Sorbent from the sorbent cooler is transported by means of a J-valve to a surge tank located
above the adsorber. The surge tank is used as a source and sink for sorbent to maintain
constant bed levels in the other process vessels. From the surge tank, sorbent flow to the
adsorber is regulated using an L-valve, thus completing one full cycle.

General Arrangement

Figure 2 shows a general arrangement for a nominal 100 MW NOXSO plant. The
major components will be identified by tracing the flow paths of the flue gas, the
heater/cooler gas, and the sorbent. This arrangement shows two adsorber trains. Flue gas
enters the lower of the two large ducts in the foreground, splits and flows through the flue
gas booster fans, adsorbers, and particulate separators before recombining and exiting the
NOXSO tower.

Ambient air for cooling the sorbent enters through two of three 50% capacity
heater/cooler fans. The air is preheated by the sorbent in the sorbent cooler (hidden behind
right adsorber) before flowing through the air heater (located below the sorbent heater) where
it is heated with natural gas. The high temperature air enters the bottom of the tapered

92-4


-------
Adsorber(2

NOx recycle t
power plant-

Flue gas
outlet

Flue gas
inlet

Sorbent heater

Jiegenerafcoi

Steam
Treater

Sorbent
Cooler

Flue gas booster
fan (2)

Heater/cooler
fans (3-)

Figure 2. NOXSO Process Tower

* 3^ *

92-5


-------
sorbent heater and exits from the top. This exit gas is the NO, recycle stream which goes to
the combustion air system of the power plant.

Sorbent is transported from the adsorbers to the sorbent heater. After being heated in
the sorbent heater, the sorbent is transported to the moving bed regenerator and then to the
steam treater. From the steam treater, the sorbent flows to the sorbent cooler where it is
cooled before being transported back to the adsorber completing the cycle.

Pilot Plant Test Program

NOXSO pilot plant tests on a 5 MW slip stream from Ohio Edison's Toronto power
plant in Toronto, Ohio have recently been completed. Results of these tests are reported in a
paper by Haslbeck, et al.s Highlights of the test program are discussed below.

After the startup aid shakedown phases of the test program were completed, a
parametric test phase was begun. During the period of time from April through December
of 1992 while the parametric tests were being conducted, the pilot plant availability averaged
76%. This high availability for a pilot plant undergoing frequent operating changes is a
testimony to the simplicity and reliability of the process.

SO2 and NOx removal efficiencies were quantified as a function of flue gas flow rate,
solids circulation rate, adsorber solid inventory, adsorber temperature, inlet S02
concentration, and inlet NO, concentration. A modification to the original project scope
included installation of a second bed in the adsorber and adding in-bed water sprays to lower
the adsorber temperature. With these modification, S02 removals of 95% and NO* removals
of 90% were easily attained at economical operating conditions.

Sorbent attrition was determined by measuring the initial and final sorbent inventory
and the quantity of sorbent added to the system to maintain the sorbent inventory. After
3,232 hours of testing the attrition rate was 3.0 lb/hr. This represents 0.011% of the total
sorbent inventory per hour which is less than the 0.016% lb/hr previously estimated.

Economics

Data from the pilot plant have been incorporated into the design of a commercial size
NOXSO plant. Using this commercial plant design, an economic analysis was performed for
the NOXSO process. The basis for the analysis and cost information are included in Table
1. The analysis was conducted for a 500 MW power plant burning 3% sulfur coal and
emitting 0.6 lb NOx/MMBtu.

Since the NOXSO process is a combined S02/NO, removal process, it is not possible
to separate the cost of removing S02 from the cost of removing NO,. Consequently, an
assumption is made that the cost of removing NOx is 3.0 times higher than the cost of
removing SOz. The value of 3.0 represents a reasonable average for the relationship between
die cost of NOx and SO2 removal based on published economic studies of separate high

92-6


-------
Table 1.

NOXSO PROCESS ECONOMIC ANALYSIS (1)

POWER PLANT PARAMETERS

GROSS CAPACITY
CAPACITY FACTOR
HEAT RATE
COAL HEATING VALUE
COAL SULFUR
NOz EMISSIONS

500 MW

mo %

10,000 Btu/kWh
12,000 Btu/lb
3.0 %

0.6 tb/MMBm

ECONOMIC PARAMETERS

ELECTRICITY	S0.03 /kWh

NATURAL GAS	%USQMsd

SORBENT	$3.40/lb

NET SULFUR VALUE	$50 Aon

S02 ALLOWANCE VALUE	J30Q

FIXED CHARGE RATE (2)	10.6 %

REMOVAL COSTNOx/REMOVAL COSTS02	3.0

NOXSO PROCESS REMOVAL EFFICIENCIES

SOZ	95%

NO*	80%

EMISSIONS DATA

UNCONTROLLED S02
CONTROLLED S02
PHASE IS02 LIMIT

S02 ALLOWANCES GENERATED

UNCONTROLLED NOx
CONTROLLED NOx

POLLUTANT REMOVAL EFFICIENCY

76.550 toasfyear
3,833 tons/year
38325 umsfytat
34,493 tons/year

9.158 tons/year
1,840 tons/year

93,4 %

OPERATING AND MAINTENANCE COSTS

FIXED (3)
VARIABLE (4)
NATURAL GAS
SORBENT
ELECTRICITY
TOTAL

$5,714,000
$129,030
$5,131,000
$10 J 12,000
13,642,000
$24,728,000

CAPITAL COST

REVENUES

S12&5OO.OO0

$257 flcW

S02 ALLOWANCES
SULFUR VALUE
TOTAL

$10347,750
$1320,438
$12468.188

NET LEVEUZED COST

S26J80,813/year

8.5 railk/kWh
S276/ton~S02
$828/ton—NOx

(1)	1993 dollars.

(2)	Based on 30 year book life. 20 year tax life, 38% composite federal and state tax.
and 2.0% for property taxes and insurance.

(3)	Includes operating labor. fringes, and supervision; maintenance labor and equipment:
and general and administrative expenses.

(4)	Includes process water and Clans plant catalyst.

92-7


-------
efficiency technologies. This value does not affect the overall economics, however it does
affect the relative cost of S02 and NOx removal.

Emissions date are also listed in Table 1. The "Phase I S02 Limit" is calculated
based on allowable emissions of 2.5 lb S02/MMBtu. It is appropriate to consider over
compliance since the high removal efficiency of the NOXSO process will alow a utility to
generate S02 allowances which can be sold to partially offset the operating cost. A value of
$300 has been assumed for S02 allowances. Beginning in the year 2000, the number of
allowances generated will decrease, however it is also likely that the value of allowances will
be significantly higher offsetting to some degree the reduction in the number of allowances
generated.

The annual operating and maintenance cost is $24.7 million with the cost of sorbent at
$10.1 million representing 41% of the total. The capital cost of $257/kw is based on a
recent EPRI study.6

Revenues for the process will be generated by the sale of the sulfur by-product and
the S02 allowances. The sulfur by-product can be elemental sulfur, sulfuric acid, or liquid
S02. The choice of sulfur by-product will be influenced significantly by the local demand
for the specific product. Since the market for sulfur is larger than the other two, sulfur is
used in this analysis. If a local market exists for sulfuric acid or liquid S02, either would be
a more economical choice since the revenue from sulfuric acid would be approximately three
times more than sulfur and liquid S02 would be six to eight times more. Making sulfuric
acid or liquid S02 would also result in minor increases in capital and operating costs.

The net levelized cost for the process is presented from three points of view. The
cost of buying, operating, and maintaining the plant will be $26.2 million dollars per year.
This translates to 8.5 mills/kwh of electricity produced. On a pollutant removal basis, it cost
$276 to remove each ton of S02 and $828 to remove each ton of NOx.

Acknowledgement

NOXSO Corporation wishes to thank the following for their contributions to the
project: U.S. Department of Energy's Pittsburgh Energy Technology Center, Ohio Coal
Development Office, W.R. Grace Co-Conn., and Moirison-Knudsen Corporation - Ferguson
Group.

References

1.	Haslbeck, J.L., C.J. Wang, L.G. Neal, H.P. Tseng, and J.D. Tucker, "Evaluation of
the NOXSO Combined NO^/SO;, Flue Gas Treatment Process", U.S. Department of
Energy Contract NO. DE-AC22-FE60148, November 1984.

2.	Haslbeck, I.L., W.T. Ma, and L.G. Neal, "A Pilot-Scale Test of the NOXSO Flue
Gas Treatment Process", U.S. Department of Energy Contract No. DE-FC22-
85PC81503, June 1988.

92-8


-------
3.	Ma, W.T., J.L. Haslbeck, and L.G. Neal, "Life Cycle Test of the NOXSO Process",
U.S. Department of Energy Contract NO. DE-FC22-85PC81503, May 1990.

4.	Zhou, Q., J.L. Haslbeck, L.G. Neal, "An Experimental Study of NOx Recycle in the
NOXSO Flue Gas Cleanup Process", U.S. Department of Energy Contract No. DE-
AC22-91C91337, March 1993.

5.	Haslbeck, J.L., M.C. Woods, W.T. Ma, S.M. Harkins, J.B. Black, "NOXSO
SO^/NO* Flue Gas Treatment Process: Proof-of-Concept Test", presented at the S02
Symposium, Boston, MA (August 1993).

6.	J.E. Cichanowicz, C.E. Dene and W. DePriest, et al., "Engineering Evaluation of
Combined N0j/S02 Controls for Utility Application", Electric Power Research
Institute, Palo Alto, CA, (December 1991)

92-9


-------
INVESTIGATION OF SORBENT REGENERATION KINETICS
IN THE COPPER OXIDE PROCESS

Joanna M. Markussen
Henry W. Pennline
U.S. Department of Energy
Pittsburgh Energy Technology Center
P.O. Box 10940
Pittsburgh, PA 15236

Introduction

The Copper Oxide Process is a dry, regenerable process using a copper (Cu)-
impregnated alumina sorbent for the simultaneous removal of sulfur dioxide (S02)
and nitrogen oxides (NOx) from flue gas. Testing erf the Huidized-Bed Copper Oxide
Process has ranged from laboratory experiments to process-developmental-scale
integrated operation1"4. The Moving-Bed Copper Oxide Process is scheduled for
future testing in the life-Cycle Test System at the Pittsburgh Energy Technology
Center (PETC)5. The fluidized-bed and moving-bed processes use different gas-solid
contacting designs for absorption, but both processes use a moving-bed regenerator
for sorbent regeneration. Most of the kinetic information in the literature has focused
on the absorption process. Kinetic data dealing with the effect of changing gas
composition on the regeneration step (as would be found in a moving-bed
regenerator) are very limited. Such information is needed for a better understanding
of the regeneration step and for potential improvements in the design of the moving-
bed regenerator used in both copper oxide process configurations.

In the Copper Oxide Process, copper oxide (CuO) is converted to copper sulfate
(CuS04) by reaction with S02 and oxygen (O2) in the absorber at approximately
400°C.

CuO * S02 + 1/2 Oz - CuS04	(1)

Catalytic reduction of NOx to nitrogen (N2) is achieved with the injection of ammonia
(NH3) into the absorber. Both CuS04 and CuO act as catalysts in the reduction
reaction.

93-1


-------
4 NO + 4 NH2 + 02 - 4 Nz + 6 if20

(2)

2 N0Z * 4 iW3 + 02 ¦* 3 2\^ + 6 HzO

(3)

The sulfated sorbent is then regenerated at 450-500°C with methane (CH4) in a
moving-bed reactor, where the reducing gas flows countercurrent to the sorbent The
overall reaction is

Note that as the sulfated sorbent is reduced, the gas composition along the length of
the reactor changes due to the release of SOy C02, and H20. For every mole of CH4
consumed, five moles of gas are produced.

The regenerated sorbent is then cooled and returned to the absorber, where the Cu is
rapidly oxidized to CuO by 02 in the flue gas.

It is essential to get nearly complete regeneration of the spent sorbent so that 95-99%
S02 removal can be achieved in the absorber with only a slight excess in the Cu/S
stoichiometric ratio.

The Fluidized-Bed Copper Oxide Process was tested in an integrated mode in the
PETC Life-Cycle Test Unit using a fluidized-bed absorber and a continuous
moving-bed regenerator3'4. The regenerator had a solids residence time of 1-2 hours,
depending upon the solids flow rate, and the average regenerator temperature was
approximately 450°C. Only 20-50 percent of the CuS04 fed to the regenerator was
reduced, and this incomplete regeneration resulted in a decrease in the S02 removal
in the absorber. To get 90% S02 removal, solid circulation rates of 4-6 times the
theoretical minimum had to be used. In a commercial process, high solid circulation
rates would increase the operating costs through higher sorbent attrition losses and
higher process energy costs. Thus, for future process development, there is a strong
incentive to improve the regeneration efficiency.

The kinetics of S02 removal have been studied using fixed beds and different size
fluidized beds, but there is little information on the kinetics of regeneration. Early
tests by McCrea et al.6 in a packed-bed reactor showed slow regeneration with CH4 at
400°C, but iwarly complete sulfur removal was obtained in 30 minutes at 450°C or in
5 minutes at 500°C The sorbent retained about 1% sulfur even after 1 hour exposure
to CH4 at 500°C, but the sulfur was believed to be bound to the alumina support and
did not affect the capacity of the CuO for subsequent SO, removal. Tests by Yeh et
al.2 using a microbalance reactor showed 80% reduction of CuS04 with CH4 in 10
minutes at 450°C or in 25 minutes at 400°C. These rates of reaction were higher than

CuSOt + 1/2 CHi - Cu * SOz + Kp + l/2 COz

(4)

Cu + 1/2 Oz - CuO

(5)

93-2


-------
those reported by McCrea, probably because a larger excess of CH4 was used. Also,
since McCrea and coworkers used a 4-in.-deep sorbent bed, any inhibition of
regeneration by the product gases may have affected the observed reaction rate.

Harriott8 suggested that the incomplete reduction in the moving-bed regenerator may
have been due to low excess methane concentration, high product concentrations, or
a combination of these factors. In a commercial operation, a slight excess erf methane
will be used, such as 10% excess CH4. However, even when 10% excess CH4 is fed to
the regenerator, the exit gas concentration is reduced to 2% due to dilution by the
product gases (see equation 4). The sorbent near the gas exit of the regenerator is
actually exposed to a higher concentration of S02 than CH4. Thus, to aid in
regenerator scaleup and design studies, kinetic data are needed over a wide range of
gas concentrations.

As a result, a series of tests was conducted at PETC using a microbalance reactor
exploring the kinetics of regeneration9. Samples from a batch of sulfated sorbent
were regenerated at 450°, 480°, and 510°C using mixtures of CH^ SO^, COj, and N2.
Tests showed that the regeneration reaction is strongly inhibited by S02 in the
product gases. It is also postulated that unsulfated CuO entering the regenerator is
first sulfated before being reduced to Cu with CH4. The reactions of copper sulfate
and sulfite with CH4 do not go to completion when S02 is present, but not enough
data is available to dearly define the equilibrium limits. More laboratory tests are
needed to develop kinetic correlations that can be used to help interpret pilot-plant
data and determine the best conditions for sorbent regeneration.

Experimental

A series of regeneration tests is currently being carried out in a Cahn Electrobalance
(Model 1000) at PETC. A schematic of the microbalance system is shown in Figure 1.
Gases are blended from certified cylinders to generate the desired regeneration gas
mixtures. Methane concentrations range from 10-100%. During regeneration tests
with CH^SO-j-Nj mixtures, S02 concentration is varied from 10-40%. Total gas flow
rate to the microbalance reactor tube are maintained at about 2000 cm3/min (STP).
A series of regeneration tests in the microbalance at varying gas flow rates from 1000
to 3200 cmVmin (STP) showed that operation at 2000 em'/min results in negligible
gas mass-transfer resistance in the system. The temperature of the reactor is
measured with a thermocouple placed 0.5 cm below the sample basket Regeneration
kinetics are being studied at reactor temperatures of 450°C, 480°C, and 510°C.

The copper oxide sorbent, fabricated by Universal Oil Products, Inc. (UOP SOX-3), is
l/16-in.-diameter spheres of gamma alumina impregnated with 6.8 wt% Cu. A large
batch (15.0 g) of the sulfated sorbent being used for microbalance testing was
prepared in a packed-bed reactor unit. The CuO sorbent was exposed to two cycles
of S02 sorption and CH4 regeneration followed by a final sulfation step. Sulfation
was carried out at 400°C, with a simulated flue gas mixture for the final sulfation step

93-3


-------
of 3990 ppm SCX 2.72% Oy and a balance of N2. Regeneration was performed at
450°C using pure CH4. The total gas flow rate through the packed-bed reactor was
1600 cm3/min (STP).

In the microbalance regeneration tests, samples of sulfated sorbent weighing
approximately 50 mg are spread over a layer of quartz wool placed in a quartz
sample basket The basket is positioned inside the quartz reactor tube in the furnace
and suspended on a wire connected to the microbalance. The sample is then heated
to the reaction temperature using pure N2. The gas stream enters the microbalance at
tiie base of the reactor tube and passes through a section containing ceramic beads,
which are put in the tube to reduce the amount of dead gas volume. Steady-state
temperatures are reached in about 1-1.5 hours.

After the heat-up period, a regeneration gas mixture is introduced into the reactor by
switching the air-actuated four-way valves shown in Figure 1. The flow rates of the
two gas streams are matched to minimize shock buoyancy forces that can cause
inaccuracies in the sample weight measurements during the initial time period. In a
typical CH4-S02-N2 regeneration test, the sulfated sorbent is exposed to a CH4-S02-N2
gas mixture until no further change in weight is observed, which typically occurs
within 60 minutes. After the regeneration has reached completion, the S02 gas is shut
off and the N2 gas flow is increased. The sample is then exposed to a CH4-N2 gas
mixture or pure CH4 to determine if the sorbent can be further reduced. The reaction
system typically reaches steady-state conditions within an additional 30 minutes.
Upon completion of the regeneration, the sample is cooled to room temperature using
N2 gas. From the microbalance weight loss curves, fractional conversion data as a
function of time are obtained.

Discussion

A study of the kinetics erf sorbent regeneration occurring in the Copper Oxide Process
is being conducted using a laboratory microbalance unit This study is designed to
complement and expand upon the work of Harriott and Markussen9,10. The ultimate
goal of the research is to provide a greater understanding of the kinetics of sorbent
regeneration, such as the combined effect of temperature and gas composition on the
reduction reaction, in order to facilitate improvements in the design of the moving-
bed regenerator in the Copper Oxide Process.

The first series of tests in the microbalance unit involves the regeneration of fully
sulfated copper oxide sorbent at varying temperatures and gas compositions. Tests at
temperatures of 450°, 480°, and 510°C are being performed with CH4-S02-N2
compositions containing up to 40% S02. The tests with 40% S02 approximate the gas
concentration expected near the gas exit of a counterflow regenerator fed with a
mixture of CuO and CuS04 sorbent and a small excess of CH4. Previous laboratory
tests were mainly conducted at 450°C or 480°C9. However, because of the strong

93-4


-------
negative effects of SO, on the regeneration rate, thereby limiting conversion, some
additional tests at higher temperatures are being explored in this test program.

A second series of microbaknee tests is being conducted to measure the reaction
rates and limiting conversions for unsulfated sorbent, which is representative of
sorbent that is not sulfated in the absorber. The solid will be exposed to mixtures of
gases similar to those predicted at the gas exit of a counterflow moving-bed
regenerator. These tests will help determine if CuO is first sulfated prior to reduction
with CH4. If the sulfation reaction occurs before regeneration, the sorbent is expected
to gain weight rapidly as SO, combines with the solid and then slowly lose weight as
the reaction with CH4 takes place. Analytical determination of the sulfite species on
the solid will be attempted using X-ray diffraction, scanning electron microscopy, and
X-ray photoelectron spectroscopy. Sensitivity modeling studies of the process show
that the possible sulfation of CuO in the regenerator is potentially important to the
overall process performance and economics".

The laboratory tests with fully sulfated and unsulfated sorbent may not completely
simulate conditions in the moving-bed regenerator, because particles fed to the
regenerator may be partially sulfated and interactions between the solid phases
present may be possible. Therefore, a few regeneration tests with sorbent that is 50
to 70% sulfated are planned. Even though the individual reaction rates cannot be
determined, the overall weight loss can be compared with that predicted from the
kinetics of the separate reactions.

Most of the earlier work9 concentrated on the inhibition effect of SO^. Since C02 and
H20 are products of the regeneration reaction, COz and HzO are also expected to
have some effect on the regeneration conversion. A few tests conducted with CQ,
gave a slightly lower final conversion than when CO, was not present9. More tests

with C02 in the gas stream will be performed to better quantify its effect on the
regeneration reaction rate.

The possible inhibition effect of H20 on the regeneration reaction has not yet been
explored. To study this effect, tests will be conducted in a packed-bed reactor unit
designed for gases containing relatively high moisture content. A number of CuS04
regeneration tests using a thin sorbent bed are planned with 10-15% HjO in the feed
gas along with a slight excess of CH4. By following the S02 concentration in the
packed-bed unit, the amount of CuS04 reduced to Cu can be determined. After the
sorbent has reached steady state with the gas mixture, the sorbent will be exposed to
pure CH4 to determine the amount of sorbent not reduced by the mixture.

Combining the experimental data from this study with the results obtained in
previous microbalance tests will allow better correlation of the reduction kinetics
occurring in the moving-bed regenerator. Equations correlating the kinetic data will
be useful in interpreting pilot-plant data and will be instrumental in determining the
best conditions for sorbent regeneration.

93-5


-------
Summary

A study of the kinetics of sorbent regeneration occurring in the Copper Oxide Process
is being conducted using a laboratory microbalance unit at FETC. Regeneration tests
at varying reaction temperature and gas composition are being performed with fully
sulfated, partially sulfated, and unsulfated copper oxide sorbent. The laboratory tests
will provide data needed to develop kinetic correlations that will aid in the design of
the moving-bed regenerator, will help determine the best conditions for sorbent
regeneration, and will help interpret pilot-plant data.

References

1.	Demski, R.J., S.J. Gasior, E.R. Bauer, Jr., J.T. Yeh, JJP. Strakey, and J.I. Joubert,
"Simultaneous Removal of S02 and NOx from Hue Gas Employing a Fluidized-
Bed Copper Oxide Process," presented at the American Institute of Chemical
Engineers Summer National Meeting, Cleveland, OH (August 29-Sept. 1,1982).

2.	Yeh, J.T., R.J. Demski, J.P. Strakey, and JX Joubert "Combined SOz/NOx
Removal from Hue Gas." Environmental Progress. Vol. 4, No. 4, p. 223 (1985).

3.	Plantz, A.R., C.J. Drummond, S.W. Hedges, and FJST. Gromicko, "Performance

of the Huidized-Bed Copper Oxide Process in an Integrated Test Facility,"
presented at the 79th Annual Air Pollution Control Association Meeting,
Minneapolis, MN (June 22-27,1986).

4.	Williamson, R.R., J.A. Moriri, and T.L. LaCosse, "Phase I - Sorbent Life-Cycle
Testing Huidized-Bed Copper Oxide Process," UOP Project Report to C.J.
Drummond, Pittsburgh Energy Technology Center, Pittsburgh, PA, 1988. DOE
Contract DE-AC22-85PC81004.

5.	Yeh, J.T., J.S. Hoffman, and H.W. PennJine, "Design of a Moving-Bed Copper
Oxide Process for Simultaneous S02 and NOx Removal," Preprint 93-TA-
173.06P, presented at the 86th Annual Air and Waste Management Association
Meeting, Denver, CO (June 13-18,1993).

6.	McCrea, D.H., A.J. Forney, and J.G. Myers, "Recovery of Sulfur from Hue
Gases Using a Copper Oxide Absorbent." JAPCA. Vol. 20, p. 819 (1970).

7.	Yeh, J.T., J.P. Strakey, and J.I. Joubert, "SOz Absorption and Regeneration
Kinetics Employing Supported Copper Oxide," Unpublished Paper, Pittsburgh
Energy Technology Center, Pittsburgh, PA, 1987.

8.	Harriott, P., "Kinetics of Sorbent Regeneration in the Copper Oxide Process for
Hue Gas Cleanup," Report to C.J. Drummond, Pittsburgh Energy Technology
Center, Pittsburgh, PA, August 1989.

93-6


-------
9.	Harriott, P., and J.M. Markussen, "Kinetics of Sorbent Regeneration in the
Copper Oxide Process for Hue Gas Cleanup." Ind. Eng. Chem. Res. Vol. 31, p.
373 (1992).

10.	Harriott, P., "Studies of Sorbent Regeneration in the Copper Oxide Process for
SO2 Removal/' Report to CJ. Drummond, Pittsburgh Energy Technology
Center, Pittsburgh, PA, April 1990.

11.	Frey, H.C., "Performance Model of the Ruidized Bed Copper Oxide Process for
S02/NOx Control," Preprint 93-WA-79.01, presented at the 86th Annual Air
and Waste Management Association Meeting, Denver, CO (June 13-18,1993).

93-7


-------
M1CR0BALANCE

N,

Air

2-c^h^hxi— 3

-an

REGENERATION GASES

iXb

Sample
Basket
Thermocouple
Furnace

Ceramic Beads

Reactor
Tube

a



^ Regulating Valve

Micro Valve

Pressure Control
Valve

Four-Way Valve

Filter

Pleasure
Indicator

?

Balance
Mechanism

Nt Purge

» Vent

Exit
Qas

Water Trap

Indicator

Figure 1, Schematic of the mlcrobalance unit.


-------
Removal of Particulate Matters and Air Toxics

together with S02
in Chiyoda Thoroughbred 121 FGD Systems

Hiroshi Yanagioka
Minora U chid a

Chiyoda Corporation
2-12-1, Tsurumi-chuou, Tsurumi-ku, Y okohama, 230 Japan

Wet FGD systems remove particulate matters and air toxics together with
S02 from flue gas. Intimate contact between gas and liquid is prerequisite
for such ancillary capabilities. The Jet Bubbling mechanism is suited for
efficient gas/liquid contact and the use of the CT-121 FGD systems is
more advantageous for the total emission control of air pollutants than just
switching to low-sulfur fuel.

1. INTRODUCTION

In Japan, the use of sulfur-containing heavy petroleum fuel during the period of rapid
economic growth in the late 1960s caused serious acidic-gas pollution problems. Efforts
were made to develop flue gas desulfurization(FGD) processes, which have been widely
applied for flue gases generated from heavy oil, coal and even Orimulsion*.

It has become known that the FGD systems are also very effective for removing any
impurities in flue gas to provide more cleaner air in industrial areas than expected. This is
because of simultaneous removal of sub-micron particulate matters and other impurities

such as Cf, F", and Hg compounds.

Although electrostatic precipitators (ESP) are in general equipped upstream of FGD
systems for particulate matters removal, there are some industrial boilers without an ESP
or with an ESP de-energized for power saving.

1. Chiyoda Thoroughbred 121 FGD Systems

Fig. 1 illustrates the Chiyoda Thoroughbred 121 process, which is characterized by size
compactness and operation reliability. The Jet Bubbling Reactor(JBR), the heart of the
process effectively integrates all chemical reactions involving removal of S02 and

94-1


-------
formation of gypsum. Internal liquid circulatiorf which is induced by gas bubbling to
absorbent liquid through sparger pipes is utilized to eliminate external slurry circulation
using large pumps.

Fig. 2 shows a comparison of gas /liquid contacting devices in typical wet FGD
scrubbers. The JBR has a higher density of liquid phase even without external liquid
circulation by pumps. In this apparatus, the height of the absorption section involving gas
bubbling or the jet bubbling layer ranges only 1* to 3'. This low height of absorption
results in low energy consumption which is only about 1 % of the power output of a
power plant

The reliability of the system has been increased by suitable operation conditions and
careful design tailored to individual plant requirements. The absorbent having a low pH of
4 to 5 is advantageous for oxidation of sulfite to sulfate by injected air as well
neutralization by limestone with virtually 100 % utilization. Agitation by air bubbling and a
mechanical mixer maintains adequate amounts of suspended gypsum crystals in the lower
part of the JBR to allow crystals to grow to the desired size of about 100 micron. Efficient
washing systems are provided to remove gypsum deposits if any in such zones as a mist
eliminator through which flue gas passes.

The main body of JBR can be fabricated using 317 L type stainless steel to withhold
corrosion because of a thin protective passive film formed on the metal surface in an
oxidative environment. Flake glass reinforced plastic,whether solid or lined, is also
acceptable for this purpose.

A prescrubber, which is usually necessary when particulate matters are detrimental to
gypsum as a salable product, can be omitted if the amount of soot is small or the by-
product gypsum is left stacked at the site. A single length of ductwork with spraying
gypsum slurry from the lower part of JBR works for particulate removal as well as gas
cooling and humidification.

Table 1 shows a list of Chiyoda Thoroughbred 121 systems constructed to date in
different countries such as U.S.A.,Germany, Malaysia, Australia in addition to Japan,
The size of the system started from 23 MW and has recently reached an equivalent
capacity of 1,000 MW.

3. Particulate Removal Efficiency

When flue gas is contacted with scrubbing liquor, particulate matters collide with the
liquor by inertia and diffusion and are subsequently separated from the main stream of
gas. Thus, removal efficiency of particulate matters can be expressed as a function of their
diameters and properties.

94-2


-------
When particulate matters are composed mainly of solid spherical fine panicles that are
insoluble in water as encountered in coal-burned flue gas, collection efficiency is
expressed in terms of particle diameters only. Fig. 3 shows one of the collection
efficiency curves obtained from CT-121FGD systems. This curve indicates that a
collection efficiency of 90 % or more is expected for coal-fired flue gas where particle
diameters are generally greater than 1 micron meter. This has been confirmed in several
installations where outlet dust loadings are below 10 mg/Nm3 with an ESP upstream of
the FGD system.

On the other hand, when particulate matters are soluble in water or scrubbing liquor,
collection efficiency is largely dependent on their chemical properties rather than their
physical sizes or diameters. Particulate matters from oil-fired flue gas contain fine
panicles of ammonium sulfate((NH4)2$04, for example, which dissolves quickly into the
scrubbing liquor, resulting in virtually 100 % removal of this substance at the outlet of the
FGD system. However, carbon dust, which is rather sticky and water-repellent, tends
to slip from the contacting surface between gas and liquid, resulting in less collection
efficiency than expected basal on the particle diameter. Because of such differences in
chemical and physical properties, heavy oil or re-run oil burned flue gas shows collection
efficiencies ranging 2 to 43 mg/Nm from plants to plants.

This holds true in Orimulsion-fired flue gas, where 70% or more of the particulate matters
are composed of such chemically active substances as VOS04 and MgS 04,though finer
particles less than 0.6 micron meters are present. Thus, actual collection efficiency will be
much greater than predicted by the data obtained from coal-fired flue gas.

4. Mercury Removal Efficiency

Various coals contain toxic heavy metals such as Mn, Cr, Ni. As, Hg, and Be depending
on their mining sites though their amounts are considered negligible at present. Literatures
indicate that Hg is present in coal at concentrations of 0.02 to 0.09 ppm by weight.

Hg, when released from coal in boilers at high temperatures, can take various forms such
as Hg vapor, HgCl2, HgO and Hg2Cl2. These compounds significantly differ in
solubilitiy in water. While HgCl2 dissolves in water up to a concentration of 74,000 mg/1,
the solubility of Hg2Cl2 is only 2.1 mg/1 at a temperature of 250 C, indicating that Kg (I) is
difficult to be trapped even in wet FGD systems.

In a coal-fired boiler equipped with a CT-121 FGD plant, analyses were made to
determine whereabouts of Hg and the following results were obtained: 2 % in bottom ash,
12 % in fly ash, 3 % in gypsum, 41 % in waste water and balance in gas to the stack The
data suggest that Hg compounds generated by burning coal are relatively volatile and the

94-3


-------
large portion is water-soluble.

5. Conclusion

In addition to the intended SO2. the Chiyoda Thoroughbred 121 FGD system can
effectively remove particulate matters. Coal-derived particulate matters in particular
provide consistent removal efficiency as a function of particle diameter.

Particulate matters contained in heavy oil -fired flue gas are different from coal-derived fly
ash that can be assumed to be solid, insoluble spheres with a distribution of diameter.
Surface properties and solubility to the scrubbing liquor also play a role in removal
efficiency.

Hg, one of air toxics was found to be removed in the FGD systems, but more
investigations should be made. This added performance of wet FGD systems is certainly
more advantageous than switching to low-sulfur fuel in view point of total air quality.

Table 1 HISTORY OF CHIYQDA'S FGD PLANT

CT-101

14 plants started up during 1972 to 1975.

CT-121

21 plants awarded to date.

PLANT OWNER

LOCATION

PLANT CAPACITY

(MW)	FUEL

START UP

Gulf Power

Mitubishi Petrochemical
Nippon Mining
Toyama Kyodo Power
Kashima North Power
Hokuriku Power
Japan Oil Shale
State of Illinoi
Wieland Werke AG
Hokuriku Power
Tioxide Group PLC
Georgia Power
Chubu Power
Hokuriku Power
Chemical Co.

Okinawa Power
Western Mining
0. Power
T. Power
K. Power

Florida USA
Vokkaichi Japan
Chita Japan
Toyama Japan
Kashima Japan
Toyama Japan
Kitakyushyu Japan
Illinois USA
Voehringen Germany
Fukui Japan
Terer.gganu Malaysia
Georgia USA
Hekinan Japan
Ishikawa Japan
Japan

Gushikawa Japan

Australia

Japan

Japan

Japan

20x2	Cu Smelter gas

156	Coal

1000	Coal

900x2	Coal

110	Coal

700	Coal

500	Coal

85	Coal

156	Coal

40 Coal
12 Coal
350 Oil

225 Olimuision
250 Oil
2 Oil Shale

23	Coal

85	Oil

75	Oil

500	Coal

14 Calcine r Off Gas

1978

1982

1983

1984

1985
1987

1987

1988
1990

1990

1991

1992

1993
1995

1993

1994
1993

1995
1998
1997

Several plants expected in Japan before 2000.

94-4


-------
©	BOOSTER FAN

©	STACK


-------
Efficiency(I)







\ i I i i \ \ © \ i















































(























n \ii



















! j | | |





































't : : ¦





















! I





























*1	o. s 1	s 10 dp ( fi m)

Figure 3. Collection Efficiency of Particulate Matters

94-6


-------
A FINAL REPORT ON THE MOVING-BED LIMESTONE EMISSION CONTROL (LEC)

PROCESS PILOT PLANT PROGRAM

M. E. Prudich
Ohio Coal Research Center
College of Engineering and Technology
Ohio University
Athens, OH 45701

K. W. Appell
J. D. McKenna
ETS, Inc.
1401 Municipal Road
Roanoke, VA 24012

Abstract

ETS, Inc., a pollution consulting firm with headquarters in Roanoke, VA, has developed a
dry limestone-based flue gas desulfurization (FGD) system. This S02 removal system, called
Limestone Emission Control (LEC), can be designed for installation on either new or
existing coal-fired boilers. A nominal 5,000 acfm LEC plot plant has been designed,
fabricated and installed on the slip stream of a 70,000 pph stoker boiler providing steam to
Ohio University's Athens, Ohio campus. The SO2 in the flue gas reacts with wetted granular
limestone that is contained in a moving bed. A surface layer of principally calcium sulfate
(CaS04) is formed on the limestone. Periodic removal of this surface layer by mechanical
agitation allows high utilization of the limestone granules.

Hie primary goal of the current study is the demonstration of the techno/economic capability
of the LEC system as a post-combustion FGD process capable of use in both existing and
future coal-fired boiler facilities burning high-sulfur coal. Based on the results of the
moving-bed pilot program, an in-house economic study has shown the LEC to have definite
economic advantages (both in capital and operating costs) when compared to a wet limestone
scrubber.

A total of over 90 experimental trials have been performed using the pilot-scale moving-bed
LEC dry scrubber with run times ranging up to a high of 125 hours. The primary
processing variables studied were: superficial flue gas velocity (0.5 to 1.6 ft/s), limestone
bed velocity (4.4 to 21.6 ft/hr), inlet S02 concentration (400 to > 3000 ppmdv), and water
addition rate (0 to 2 gallons/min - includes both over-bed sprays and flue gas humidification
water).

SQz removal efficiencies as high as 99.9% were achievable for all experimental conditions
studied during which sufficient humidification was added to the LEC bed.

95-1


-------
Technical Discussion
LEC Basics

The limestone Emission Control (LEC) process is a unique system employing standard
quarry-sized limestone to remove SCX from coal-fired boiler flue gases. In the LEC process,
hot flue gases (<350°F) are contacted with a bed of 1/32 to 1/4 inch limestone granules
covered with a thin film of water. Sulfur dioxide is absorbed from the flue gas into the
water film where it subsequently reacts with dissolved limestone. A layer of reaction
products, primarily calcium sulfate and calcium sulfite, forms on the surface of the limestone
granules as the reaction proceeds.

History of LEC Development

In the early 1970s, the U.S. Department of Energy initiated bench-scale testing of dry
limestone for flue gas desulfurization. Shale and Stewart1 reported that SO2 removal
efficiencies of greater than 90% for extended periods of time were possible, and that the
limestone granules were regenerated easily with mild agitation. This early work resulted in
U.S. Patent No. 3,976,747s entitled "Modified Dry limestone Process for Control of Sulfur
Dioxide Emissions."

ETS, Inc. began development work on the present LEC system in mid-1982. In June 1983,
ETS and the U.S. EPA agreed that the LEC was potentially a simple, low cost FGD system.
It was agreed to proceed along three paths: (1) ETS performed a series of bench-scale slip
stream tests to verify the basic SOj emission reductions. Results showed greater than 90%
S02 removal; (2) EPA/IERL requested that TV A do a preliminary economic analysis of the
LEC system. The resulting analysis was in agreement with preliminary ETS economic
studies showing that the LEC process offers a significant economic advantage over both
spray drying and wet scrubbing systems; and (3) EPA/IERL initiated an in-house laboratory
bench-scale study of the LEC process aimed at verification of the ETS dip stream tests.
Although Step 3, EPA/IERL's in-house work, was not formally published, ETS arrived at
two significant conclusions based on the EPA bench-scale testing program. First, EPA's
bench-scale tests using simulated flue gases adequately substantiated ETS' slip stream test
results and Shale's early work at DOE. And second, although regeneration of the spent
limestone was demonstrated, the regeneration was not accomplished as easily as that reported
by Shale and Stewart, i.e., more vigorous grinding was required.

In the fall of 1986 the Ohio Coal Development Office awarded a research grant to Ohio
University to design, install and operate a small, semi-batch fixed-bed LEC pilot plant (400
acfm) on a slip stream of flue gas from the 70,(XX) lb/hr stoker boiler providing steam to the
Ohio University campus. Figure 1 shows a simplified flow schematic of the fixed-bed LEC
pilot unit. The flue gas first entered a spray chamber where it was conditioned to the desired
temperature and humidity. The conditioning was achieved by adding water via an atomizing
nozzle and/or by injecting live steam into the chamber. The conditioned gas then entered the
LEC reactor, passing downward through a fixed-bed of limestone. The reactor contained a
removable basket filled with limestone. The basket was 22 inches square and permitted the

95-2


-------
TO
STACr

FLUE
GASES

\	 ATOMIZING

NOZZLE

OVERHEAD
SPRAYS

±±±

STEAK
INJECTORS

ID FAN

.LIMESTONE
BED

REACTOR
DRAIN

Figure 1. Simplified schematic of the fixed-bed LEC process.

limestone bed depth (the bed dimension in the direction of flue gas flow) to be varied
between 0 and 24 inches. The scrubbed flue gases were then drawn out from the bottom of
the limestone bed by an induced draft fan and returned to the duct leading to the stack.

Over 100 experimental trials were conducted using the LEC fixed-bed pilot unit between
May 28, 1987 and December 18, 19873. Three different Ohio limestones covering the full
range of Ohio limestone compositions were used in these tests. The program studied the
effects of varying limestone bed depth (from 6 to 18 inches), superficial flue gas velocity
(from 1.0 to 2.0 feet per second), flue gas humidity, Met flue gas SOj concentration, and
inlet flue gas temperature. All three of the limestones used in the study were able to achieve
greater than 90% S02 removal in the LEC process. Many tests showed removals of up to
99%.

The fixed-bed LEC unit was shown to perform effectively for inlet S02 concentrations in the
flue gas ranging from 500 ppm to 3500 ppm. At an inlet SO2 concentration of 700 ppm, a 6
inch deep LEC bed was able to sustain SO2 removals of over 90% for 16 hours and removals
of over 50% for more than 24 hours (Figure 2). Even at 3500 ppm SO2, the 6 inch LEC

95-3


-------
100-

87121%

1 u *** * L

0 \ * *

wioja I



c 80-



Q.

60-

\

\ ^



I

Removal

O

. ! > . .

k

ft

\

^871217C

V
%



V

CM

0

01	20-















C

1 1 1 1 1 i 1 1 1 j 1 1 i 1 1 j i i 1 j 1 1 1 1 1

4 8 12 16 20 4 24

Elapsed Run Time,

1 1 1 1 1 1 1 i

28 32

hours

I 1 J i "1

36 40

Figure 2. Deactivation of an LEC bed due to surface blinding. [Run data from

Reference 3 with a limestone bed depth of 6 inches and a superficial flue gas
velocity of 1.0 ft/s.)

bed (the smallest bed tested) was able to sustain an SO2 removal efficiency of greater than
90% for about 1 hour and a removal efficiency of greater than 50% for more than 2 hours.

Two factors were identified which limit LEC operation. These were (1) drying of the
limestone surface and (2) build-up of the reaction product on the reactive limestone surface.
A dry limestone surface is effectively non-reactive under LEC processing conditions. The
build-up of reaction product on the limestone surface results in a decrease in the reaction rate
and eventually renders the limestone inactive.

Two operational modes were used to overcome deactivation by drying. Operating the
process with a saturated flue gas would halt evaporation and thereby prevent the limestone
bed from drying out. An over-bed water spray could also be used to keep the bed wet. In
order to keep the bed from drying out, the spray rate would have to be equal to or greater
than the rate of evaporation.

Deactivation due to surface blinding cannot be avoided but it can be overcome through the
removal of the reaction products. After the limestone became non-reactive due to the
blinding of the surface, the solids still retained a core of unreacted limestone. The layer of
reaction products could be removed by mild attrition. In the small plot plant studies the

95-4


-------
reaction products were removal from the reacted limestone through off-line attrition in a
stirred ball milL After attritting, the stone was returned to service. When the surface of the
stone was again blinded this process could be repeated until all of the limestone had been
utilized. Hie reactivated stone showed the same S02 removal activity as the virgin
limestone.

Three methods were used to supply water to the LEC process. One method was to prewet
the bed of limestone before starting the process. This was accomplished by pouring water
onto the bed before sealing the reactor. The life of the bed was determined by the amount of
conditioning of the incoming flue gas. The higher the relative humidity of the gas the longer
the time before the limestone bed became dry and therefore unreactive. The second method
was by condensation. Normally the limestone bed was cooler than the incoming flue gas.
As the gas contacted the cooler bed, water would condense from the gas phase onto the
limestone. The amount of water deposited by condensation would depend upon the
conditioning of the flue gas. As the relative humidity increased the amount of water
deposited onto the limestone would increase. Condensation would slow and eventually cease
as the bed was heated by the incoming flue gas. After condensation stopped the life of the
bed was once again determined by the condition of the incoming flue gas. The final method
of water introduction was by spraying water directly onto the limestone bed. As shown in
Figure 1, spray nozzles were positioned directly over the bed. The life of the bed was
determined by the rate of the water spray as well as the condition of the flue gas.

Figure 3 shows experimental data taken from the LEC pilot unit that illustrates both
deactivation due to drying and surface blinding. Runs 870714A', 870716A, 870716B,
870717A, 870720A and 870720B represent consecutive runs using the same 12 inch deep bed
of limestone. The superficial gas velocity remained constant from run to run at 1.5 ft/s but
the SO, concentrations ranged from 620 to 870 ppm. The inlet flue gas relative humidity
was approximately 50% but varied slightly from run to run. The variation of SOa
concentrations and gas humidities was due to the fluctuating conditions of the incoming flue
gas. Run 870714A' started with a prewet bed of limestone. After a little over 1.5 hours the
bed became unreactive. The same bed of limestone was then rewet before starting run
870716A. The bed regained its initial SOz removal efficiency before once again losing its
reactivity due to drying. The process of rewetting the bed before starting the next run
continued for the remaining four runs. Starting with run 870717A, rewetting the bed failed
to return its activity to its original level. This decrease in activity can be attributed to the
blinding of the surface due to the build-up of the reaction products on the reactive surface.

Deactivation due solely to surface blinding is illustrated in Figure 2. These data show the
performance of three different 6 inch deep beds of limestone. The inlet flue gas was
saturated and therefore drying of the bed was not a factor in the decline of the SOj removal
efficiency of the limestone. Run 871217C is a continuation of 871216B and uses the same
bed of limestone. Run 871105A continued using the same bed of limestone started by run
871103A. As can be seen from Figure 2 each bed lost activity at different times. The life
of the bed was inversely proportional to the product of the SQ, concentration of the incoming
flue gas and the flue gas flow rate. Run 871103A/871105A with an inlet S02 concentration
of 344 ppm lasted the longest. Next was run 871214B with an SQz concentration of 702

95-5


-------
ELAPSED RUN TIME (hr)

Figure 3. Deactivation of an LEC bed due to surface drying and surface blinding. [Run
data from Reference 3 with a limestone ted depth of 12 inches and a
superficial gas velocity of 1.5 ft/s.]

ppm. Run 871216B/871217C with an Met S02 concentration of 736 ppm had the shortest
life. The Met flue gas flow rate was held constant for all three sets of runs.

Besides the difference in operating conditions, the limestone used for the runs shown in
Figure 2 differs from the limestone used for the runs shown in Figure 3.

Moving Bed Process Description

A continuous moving-bed LEC pilot plant sized to handle 5000 ACFM has been installed on
the dip stream of the 70,(XX) Ib/hr stoker boiler providing steam to the Ohio University
campus. A amplified flow scheme of the moving-bed LEC process is shown in Figure 4. A
detailed schematic of the pilot plant is given in Figure 5. Limestone is added to the bed via
a 32 foot bucket elevator which is supplied with limestone from the feed hopper, the make-
up hopper, the vibrating screen, or the recycle chute by means of a feed screw. The make-
up hopper is used to compensate for limestone lost due to reaction with S02 and its
subsequent removal as reaction product. The limestone bed itself has dimensions of 14
inches X 36 inches X 128 inches (Figure 6). The depth of the bed in the direction of the

95-6


-------
flue gas flow is 14 inches. The LEC
reactor internals consist of inlet and outlet
louvers, overhead bed sprays and outlet
screening. The inlet and outlet louvers (as
originally designed) are four inches long
and are inclined at a 75° angle with a 3/8
inch overlap between louvers. The flue
gas inlet and outlet plenums on the sides
of the reactor are sized such that there is
uniform distribution of the gas across the
whole bed. The flue gas is drawn from
the duct leading to the stack from the
electrostatic precipitator. The superficial
velocity of the flue gas through the LEC
reactor can be varied from 0.3 to 1.6 feet
per second. The live hopper at the top of
the bed is provided so as to prevent the
leakage of ambient air into the outlet gas
plenum. An induced draft fen is used to

draw the processed flue gas from the	Figure 4. Simplified schematic of the LEC process,

outlet plenum. The processed flue gas is
directed to the stack.

WET
LIKESTONE
BED

rim

GA



DRY
LIMESTONE

BED

DESULFTOIZED
FLUE
GAS

Figure 6. Limestone bed geometry for
moving-bed LEC pilot plant.

limestone moves vertically downward
through the bed with a velocity sufficient to
prevent bridging of the limestone which
may occur when the flue gas dries up the
bed. The limestone removed from the
bottom of the bed is sent to the discharge
screw via a 19 foot bucket elevator. The
discharge screw has three outlets, any one
of which may be opened. The first outlet
directs the limestone to the recycle chute
which leads directly to the feed screw and
back to the reactor. The second outlet
leads to the reactivation system. The build-
up of reaction products on the limestone
surface tends to retard the rate of reaction
per unit granule surface area. To prevent
this surface blinding from rendering the
limestone inactive, the limestone granules
are passed through a reactivation device that
removes the reaction product layer by either
diy or wet abrasion. This abrasion exposes
a fresh layer of reactive limestone. This
mixture of reactivated limestone and

95-7


-------
LEGEND

1-Feed	Hopper

2-Waste	Hopper

3-Make-up	Hopper

4-Screen

5-Fluidizer

6-Bucket	Elevator

7-Bucket	Elevator

8-Inlet	Gas Plenum

9-Outlet	Gas Plenum

10-Reactor

11-ID	Fan

12-Bed	Twin Screws

13-Livo	Hopper

14-Inlet	Gas

15-Outlet	Gas

16-Reactor	Screw
17 Discharge Screw

18-Feed	Screw

19-¥aste	Screw


-------
abraded material is then passed through a vibrating screen were the abraded material is
removal as a waste. The reactivated limestone is then directed to the feed screw and
returned to the top of the LEC bed. The third discharge screw outlet is directed to the waste
hopper. Normal LEC operation directs the limestone through the reactivation system.

The S02 concentration in the inlet flue gas can be varied from 500 to 3000 ppm. Early
experimental runs using this pilot plant have shown that the LEC process has a capability of
removing more than 98% of the SOj contained in the inlet flue gas. The LEC process is
considered to be potentially attractive for enabling coal-fired power plants to meet stringent
S02 emission standards.

Discussion of Results

Initial Moving-Bed Experimental Results

A total of 30 experimental runs were conducted between May 29, 1991 and July 25, 1991
using the LEC moving-bed pilot unit. The length of each experimental run ranged from 30
minutes to 290 minutes. The primary variables studied were:

•	Superficial flue gas velocity - 0.5, 1.1, and 1.6 ft/s.

•	Limestone bed velocity - 4,4 ft/hr, 13.7 ft/hr, and 21.6 ft/hr
(these limestone bed velocities correspond to in-bed limestone
residence times of 2.4 hr, 0.8 hr, and 0.5 hr).

•	Inlet SQz concentration - 500 to 1100 ppm.

•	Water addition rate - 0 to 2.0 gallons/minute (includes both over-bed sprays
and flue gas humidification water).

As mentioned earlier, the SC^/CaCOs reaction is a liquid-phase reaction and therefore a thin
layer of water must be present on the limestone granules to allow significant SOb removal to
occur. The run philosophy to date has been to operate the LEC bed so that the limestone
exiting the bed is essentially dry. This is accomplished by controlling the position of the
drying front that exists in the limestone bed. As the limestone approaches the bottom of the
bed, the drying front moves closer to the outlet side of the reactor. This results in there
being less available wet limestone for S02 removal. Although this results in relatively poor
S02 removal values in the lower portion of the bed, the overall average bed removal remains
high because of the excellent performance of the top and middle portions of the bed.

The results obtained from the initial experimental trials have demonstrated SO2 removals in
excess of 90% and, often, SO, removals in excess of 99%. Performance plots for a
representative experimental trial (910723A) are shown in Figure 7. This figure also shows
the influence of die mechanical difficulties that inhibited long-term operation during the
initial period of operations. The outlet design of the LEC moving-bed pilot plant relied only

95-9


-------
~

/

/

s g*Ko 9

CM :

O 60-

tn :

Run No. 910723A

50

I | I I I I | I 1 1 I i ! I 1 1 J ! (l I I I O	*-0

50	100 150 200 250

Elapsed Run Time, min

Figure 7. LEC moving-bed experimental results. Trial number 910723A. AASHTO
No. 9 limestone. Met SQ,: 1100 - 1200 ppmdv.

on the outlet louver to retain toe Emestone in the moving bed. As it has turned out, the
original outlet louver design did not accomplish this purpose. During operations, limestone
was drawn from the bed into the outlet plenum and subsequently into the I.D. fen. In order
to prevent this limestone loss from the bed and to protect the I.D. fan, a screen was attached
to the inside (limestone bed side) of the outlet louvers. The screen was used to keep the
limestone in the bed. However, the presence of the screen introduced another operational
problem. Damp fines were deposited on the screen, plugging the screen and significantly
increasing the pressure drop across the limestone ted. The blinding of the screen occurred
most rapidly at the top of the bed (normally the most active part of the bed). Due to the
screen blinding at the top of the bed, the flue gas would begin to flow preferentially through
the bottom portions of the bed (normally the least active part of the bed). This action would
result in a simultaneous increase in pressure drop across the bed, decrease in SO2 removal
performance, and decrease in flue gas flow rate. This behavior can be observed in Figure 7.

After the initial run series, the outlet louvers and outlet planum wore redesigned. The
redesign modifications for these systems were made during a September/December 1991
shutdown. The new outlet louvers were eleven inches long with a 3-1/2 inch overlap
between louvers. The new and old outlet louver designs are shown in Figure 8. Results
from shakedown runs performed in February 1992 have indicated that limestone loss from
the reactor bed into the outlet plenum was reduced by several orders of magnitude.

It has been observed that significant limestone surface abrasion occurs within the LEC bed

95-10


-------
Figure 8. Old and new outlet louver designs used in the LEC moving-bed reactor.

itself. 3ii traveling downward through the LEC bed and in transport by the screw conveyors,
the limestone particles seem to provide most of the granule on granule abrasion necessary to
reactivate the limestone. In cases where this effect may be fully exploited, the external
reactivating attritor (equipment item #5 in Figure 5) may be eliminated from the process flow
sheet.

Enlarged Louver Design Experimental Results

A total of 60 experimental runs were conducted from May 5, 1992 through November 24,
1992. The duration of these experimental runs ranged up to 134 hours. The most important
variables studied, primarily with respect to overall SO2 removal efficiency, bed pressure drop
and stone carry-over, were:

•	mean granule diameter.

•	superficial gas velocity.

•	inlet S02 concentration.

Effect of Mean Granule Diameter. The four granule diameters that were studied in the
LEC moving-bed were:

•	AASHTO No.9 (Dp = 0.10 inches).

•	AASHTO N0.8 (Dp = 0.28 inches).

•	50/SO^otum- basis) No.8/No.9 mixture (Dp = 0.20 inches).

•	AASHTO No.9 pre-screened with a N0.8 mesh (Dp = 0.12 inches).

95-11


-------
100-

c
o

tv

75-

O

£

CO 50-

B

25-

O

0-
0.1

Particle Size, mm

10

Figure 9. Typical particle size distributions for the
AASHTO No. 8 and No. 9 limestones
used in LEC processing.

Typical particle size distributions for the two major stone sizes [AASHTO No. 8 and No. 9]
used in LEC processing are shown in Figure 9. The No.9 limestone has the smallest mean
diameter of the four types used and therefore has the largest surface area available for
reaction. Unfortunately, the diameter of a limestone granule is inversely related to its
minimum fluidization velocity. At smaller particle diameters this can result in a portion of
the limestone becoming fluidized by the flue gas stream and, subsequently, exiting the bed
through the outlet louvers. At superficial gas velocities greater than 0.75 ft/s, excessive
limestone carry-over was observed to occur. In order to reduce the fluidization potential of
the stone, limestone sorbent granules with increased mean diameters were studied.

The next larger commercially available size stone is AASHTO No.8 (Dp = 0.28 inches).
With the relatively large- particle diameter of the No.8 stone, the carry-over problem was
greatly reduced, even at superficial flue gas velocities in excess of 1.2 ft/s. The improved
material handling characteristics of the No.8 limestone enabled the first extended trial of the
moving- bed system (Figure 10). Run 920430 lasted for over 134 hours at an average
superficial gas velocity of 1.1 ft/s and achieved SO2 removal efficiencies of up to 85 percent.
However, subsequent experimental trials (LEC Runs 920601 and 920602), also using No. 8
limestone, that ware operated at Iowa1 superficial gas velocities (0.3 ft/s) with a 100 percent
bed moisture zone yielded S02 removals of only 40% (average). With die enlarged mean
diameter of the No.8 stone, the loss in available reaction surface area resulted in substantially
lower S02 removals when compared to No. 9 limestone. The low S02 removals resulting

95-12

A


-------
100

0

\> • i» \i
Run No. 920430

i—i—i—i—|—i—i—r

0	25

—j—r—r-T—i—j—

50	75

'I 1 1 1 1 I 1

100 125

.CM

o

CM
X

-oo

-to

Elapsed Run Time, hours

Q.

o

0

15

m

CD

-CN

o
,o

ro

.§
o

K>

o
.o
in

o

¦8

CM

o
.©
m

o

.o
o

o
¦o
m

"-o

U
D

i

o

Ll.

W
O
O

Q)

c

Figure 10. Long-term LEC operation using AASHTO No. 8 limestone. Run No. 920430.
Met S02: 800 - 1400 ppmdv.

from the use of No. 8 stone are unacceptable in terms of operating a viable commercial
system. This low removal more than negates the No. 8 stone carry-over and material
handling advantages. Increasing the LEC bed depth beyond 14 inches would add reactive
surface area and might compensate for the disadvantages of the No. 8 stone.

The next stone size used in an attempt to reduce stone carry-over and still achieve +90%
S02 removal was a 50/50 mixture by volume of No.8 and No.9 limestones. This mixture
had an average particle diameter of 0.20 inches (LEC Runs 920604 and 920605). The
resulting S02 removal was approximately 10% greater than with No.8 stone alone, but the
limestone carry-over was still excessive. Therefore, the combination of the two stones
yielded a mixture that exhibited the worst traits of both the No.9 (high stone carry-over) and
No. 8 (low S02 removal) stone sizes.

The last

method investigated in order to reduce the stone carry-over while maintaining high

95-13


-------
S02 removals consisted of scalping (removing the fines from) the No.9 stone with a No. 8
mesh screen (0.094 inches). This was done because the theoretical fluidization velocity of a
size distribution is weighted towards the higher individual fluidization velocities of the
smaller particles within the distribution. Thus, by scalping off the smallest 40% of the No.9
limestone granules, the fluidization velocity of the remaining stone would be increased while
at the same time maintaining a larger reaction surface area than that available from the No. 8
limestone. The resulting mean particle diameter of the screened stone was 0.12 inches. The
screened stone was used in LEC Runs 920810, 920811 and 920812. These runs yielded poor
results since there was no decrease in limestone carry-over when compared to No.9 stone
(Dp = 0.10 inches) although S02 removals were generally reduced to less than 90%.

The overall results of varying the mean particle diameter in order to achieve an optimum
stone size tot would limit the amount of limestone carry-over into the outlet plenum while
maintaining high SO2 removals was unsuccessful. The carry-over problem must be solved
through changes in the mechanical design.

Effect of Superficial Flue Gas Velocity. The superficial flue gas velocity affects three
dependent variables of the LEC system:

(1)	Reaction Theory.

(2)	Stone Carry-Over.

(3)	Bed Pressure Drop.

The reaction kinetics theory with respect to superficial flue gas velocity of the LEC system
has already been extensively dealt with by both Visneski4 and Reddy5 and their general
conclusions have been validated by the moving-bed project.

Stone Carry-Over, It has been empirically determined that flue gas superficial velocities of
greater than 0.75 ft/s result in excessive limestone carry-over when using No.9 stone (Dp -
0.10 inches). Fluidization of the limestone particles is believed to take place in fee 75° outlet
louver slot since the superficial flue gas velocity is more than doubled as it passes through
the outlet louvers. The excellent S02 capture performance of an experimental trial (920613)
operated processing No. 9 stone and at a superficial flue gas velocity of 0.31 ft/s is shown in
Figure 11. Hie use of No.8 stone with an average granule diameter of 0.28 inches allows
superficial flue gas velocities of up to 1.2 ft/s without excessive stone carry-over at the
expense of the loss of reactive surface area (at constant bed depth).

Bed Pressure Drop. Bed pressure drop is related to the square of the superficial flue gas
velocity and can be quantified through the Ergon equation. This is illustrated by LEC RUN
921020, during which the bed pressure drop was observed to be less than 2 in.H20 at a
superficial flue gas velocity of 0.31 ft/s. After approximately 26 hours into the trial, the
superficial flue gas velocity was increased to 0.47 ft/s which resulted in an increase in
pressure drop to approximately 7 in.H20.

Effect of Inlet S02 Concentration. All of the LEC experimental trials were conducted
using flue gas obtained from the slip stream of the Ohio University heating plant. The

95-14


-------
100

c


Q


O

E

Q)
CK

90

80

70'

CM

O 60-
CO

50-

S02 Removal

Inlet Gos Flow

Run No. 920613

Delta P

0

i—i—i—i—|—i—i—i—i—|—r

50	100

150

t—i—r

.CM

o

CN|

¦oo

•o

200

Elapsed Run Time, min

Q.

o

s—

0

L_

W
CD
95% of the SQ, entering the scrubber. However, whenever the
moisture (damp limestone) zone was reduced in order to allow the exiting stone to remain
dry enough for effective regeneration via the vibrating screen (between hours 4 to 5, 6.5 to
8.5, and 11.5 to 13.5), the SQa removal efficiency dropped dramatically down to
approximately 70%. In terms of sorption theory, this would indicate that at these high inlet
SO2 concentrations the SOj breakthrough curve resides within (but only barely) the fully
wetted 14 inch bed. This means that the S02 breakthrough curve resides close to the outlet
side of the bed even at low superficial gas velocities (0.3 ft/s). Therefore, careM water
addition is critical in order to ensure a maximum possible moisture zone within the bed when

95-15


-------
100-

c

a>
o

u_

a>

Cl

o
>
o

£
CD

cr

04

o

if)

80-

60'

40-

20-

S02 Remcrvol

O

¦coO
CM

-<£>



/ Inlet Gas Flow

Oelta P

Run Nos. 920716/920718/920719

"T	1	1	i	1	1	1	1	]	1	J	T

5	10

Eiapsed Run Time, hrs

-CM

15

Cl
O

k_

O


-------
The LEG process and conventional limestone scrubbing have been compared on an equatable
basis using flue gas conditions that would be expected at the outlet of the electrostatic
precipitator (ESP) on a 500 MW coal-fixed power plant. The equipment list for the LEC
case was obtained by conceptually scaling up the current LEC moving-bed pilot plant. The
equipment list for the wet limestone scrubber case was obtained from an EPRI technical
report®. Equipment costs for both the LEC and the wet limestone scrubber were generally
obtained using empirical estimating charts.

The LEC was found to have a definite economic advantage in both direct capital costs and
operating costs. The LEC equipment capital cost of $12,290,000 represented a 48% savings
when compared to the wet limestone scrubber process equipment cost of $23,534,000. On
an absorption tower basis, a capital cost advantage of 46% for the LEC occurs due to its
simpler mechanical design. Operating costs are also lower for the LEC process since a
slurry is not required, resulting in a lower water consumption. The solids material handling
associated with the LEC requires less electricity than that of the slurry handling system
required by the wet limestone scrubber.

The economic results quoted above are in general agreement with a 1989 study7 that
compared the LEC with wet scrubbing, a stand-alone dry scrubbing system, and retrofit
LIMB (with humidifieation). Hie results of this study are given in Table 1.

Table 1. Economic comparison of LEC and other scrubbing

technologies7.

Technology

SO2 removal,
%

Capital costs,
$/kW

SO2 removal
costs, $/ton

LEC

90+

168

406

Wet Scrubber

90+

220

680

Dry Scrubber

70

186

740

LIMB

50

74

608

Based on the success and findings of the present project, the next step will be a full-scale
commercial demonstration unit. Sales presentations to utility and industrial sites for a
demonstration unit have already been initiated. The commercial demonstration for the LEC
is currently under consideration in Taiwan.

95-17


-------
Licensing agreements with two major air pollution control suppliers have already been
negotiated. The first was with Procedaire Industries, which is the North American subsidiary
of the highly respected industrial furnace manufacturer, Stein Heurtey. Procedaire Industries
was granted the right to sell the LEG system to the North American industrial market. The
second agreement was granted to the U-Tah Industrial Company, Ltd., Taiwan, a major
supplier of air pollution control equipment, for the exclusive rights to sell the LEG process
throughout Taiwan, R.O.C. Taiwan's FGD market is expected to reach $2.8 billion U.S. by
1997.

Conclusions

The LEG moving-bed pilot plant project has demonstrated its capability as a high-
performance SO2 removal system. S02 removals as high as 99.9% have been achieved for
extended periods at inlet flue gas flow rate of up to 3200 acfm. Met flue gas SOj
concentrations ranging up to >3000 ppm have been successfully tested. This shows that the
chemistry behind this road-grade limestone-based technology is valid.

Mechanical difficulties, i.e., limestone blow-over and exit screen plugging initially made
long-term LEC operations difficult Outlet louver modifications made over the
September/December 1991 shutdown enabled long-term testing.

The next step in the commercialization of the LEC process is the demonstration of the
technology on an electric utility boiler. A conceptual schematic of a generic compact
moving-bed LEC is shown in Figure 13.

95-18


-------
wmmsmtm



(a.) Top View

/\:»:/\:::::
/Nh^x/NA

/N::::::::;XN:il

11

/N;::::::::><\

(b.) Face View

Figure 13. Schematic of conceptual compact LEC bed. (a) Top view, (b) Cross-
sectional face view. Arrows indicate flue gas flows. Shading indicates
limestone bed, Rue gas is distributed and collected through louvered
plenums.

Acknowledgements

This study was funded in part by the Ohio Coal Development Office under grant
GDO/D-88-49.

95-19


-------
References

(1)	C. C. Shale, U.S. Patent 3,976,747 (1976).

(2)	C. C. Shale and G. W. Stewart, "A New Technique for Dry Removal of SOj,"
presented at the Second Symposium on the Transfer and Utilization of Particulate
Control Technology, Denver, CO (1979).

(3)	M. E. Prudich, K. W. Appell, M. J. Visneski, J. D. McKenna, D. A. Furlong, I. C.
Mycock, J. F. Szalay, and J. E. Wright. Small Pilot Plant Demonstration ofETS'
Limestone Emission Control System, Vols. 1 & 2. Columbus, Ohio: Ohio Coal
Development Office, 1988. CDO/R-86-24.

(4)	M. J. Visneski. Modeling of the Low Temperature Reaction of Sulfur Dioxide and
Limestone Using a Three Resistance Film Theory Instantaneous Reaction Model,
Ph.D. Dissertation, Ohio University, 1991.

(5)	S. N. Reddy. A Mathematical Model ofETS' Limestone Emission Control Process
Using a Moving Bed Configuration, M.S. Thesis, Ohio University, 1991.

(6)	R. J. Keeth, M. J. KrajewsM, and P. A. Ireland. Economic Evaluation ofFGD
Systems, Volume 5; The NOKSO and SOXAL Sodiwn-Based Processes and Four
Additional Calcium-Based Processes. Palo Alto, Calif.: Electric Power Research
Institute, 1986. CS-3342-V5.

(7)	Ohio Coal Development Office. Proposal CDO/89-64, 1989.

95-20


-------
TUNG FGD PILOT PLANT PERFORMANCE

Peter Strangway
Niagara Mohawk Power Corporation
300 Erie Blvd. West
Syracuse, New York

Shao Tung
Raycon Research & Development
91 Blake Rd, Brookiine, MA

Robert J. Keeth
Raytheon Engineers & Constructors Inc.
5555 Greenwood Plaza Blvd.
Englewood, Colorado

Abstract

Raycon Research & Development Inc. (Raycon) contracted with Raytheon Engineers &
Constructors (Raytheon) to design, construct, and operate a nominal 2 MW pilot
facility. This test program will demonstrate the performance of the Tung FGD Process
when treating flue gas from a medium-to-high-sulfur, coal-fired utility boiler at Niagara
Mohawk Power Corporation's Dunkirk Station.

The Tung FGD Process pilot plant is designed to remove more than 90 percent of the
S02 from the flue gas in a conventional prescrubber/sodium sulfite packed tower
absorber system. The resulting bisulfite solution is extracted with an organic solvent,
thus regenerating the sodium sulfite. The organic solvent loaded with sulfite is then
steam stripped in a stripper column, producing a concentrated S02 gas stream. The
organic is returned to the extraction system for reuse.

This paper will present the project history, design considerations, and available
performance data generated during the parametric testing program, currently
scheduled for July-September, 1993. Integrated system operation will be evaluated to
determine the optimum operating conditions for the long-term reliability of the system.
Available data summarizing the parametric test performance to-date will be included in
the presentation.

96-1


-------
Introduction

Development of the Tung FGD Process began in the early 1980's and progressed
through a laboratory development phase to the point where a small pilot plant test
program was necessary to demonstrate the performance of the process on the flue
gas from a coal-fired utility boiler. Laboratory testing continues to enhance the
performance of various subsystems.

The program is funded by the Department of Energy (DOE), Niagara Mohawk Power
Corporation (Niagara Mohawk), the Empire State Electric Energy Research
Corporation (ESEERCO), and the New York State Energy Research and Development
Authority (NYSERDA). Raytheon Engineers & Constructors (Raytheon), Englewood,
Colorado, was contracted to provide engineering support to design, construct, and
operate a nominal 2 MW equivalent pilot plant. The pilot facility will demonstrate the
performance of the Tung Flue Gas Desulfurization (FGD) Process when treating flue
gas from a medium-to-high-suifur, coal-fired utility boiler. The pilot facility is installed at
Unit No. 4 of Niagara Mohawk Power Corporation's Dunkirk Station at Dunkirk, New
York.

Tung FGD Process Summary

The Tung Flue Gas Desulfurization Process utilizes an innovative liquid/liquid
extraction system to regenerate a sodium-based aqueous scrubbing liquor. Steam
stripping of the solvent recovers S02 as a byproduct, and requires only a fraction of
the energy needed to steam strip S02 from the aqueous solution itself.

A simplified process schematic developed for the Tung FGD Process is shown in
Figure 1. The process consists of three primary subsystems:

•	Scrubbing	S02 in the flue gas is scrubbed with a sodium sulfite

solution yielding a sodium bisulfite solution:

S02 + Na2S03 (aq) + H20 -~ 2NaHS03 (aq)

•	Extraction	S02 in the scrubbing bisulfite solution is transferred to an

organic solvent. The regenerated sulfite solution can be
reused for scrubbing:

2NaHS03 (aq) + Org. -* loaded Org. + Na2S03 (aq)

•	Steam Stripping The ioaded organic is steam stripped to separate S02

gas from the organic. The regenerated organic can be
reused for extraction.

loaded Org. -* Org. + S02

96-2


-------
The process also involves several minor steps such as sulfate formation and removal,
solvent regeneration, etc. They will not be discussed in this paper.

Major advantages of the Tung FGD Process include:

•	Relatively low steam consumption - about one pound of steam per pound of
S02 recovered is projected.

•	Relatively low scrubbing chemical consumption - the scrubbing chemical is
regenerated within the process. Only the sodium removed as sodium sulfate
waste or byproduct needs to be replaced.

•	Relatively low volume of waste products - sodium sulfate is produced in
moderate amounts and can potentially be marketed as a byproduct. The
prescrubber blowdown stream is neutralized and sent to disposal. There may
also be some small amount of organic bled off from the process. This organic
waste from a commercial plant can be incinerated off-site or perhaps mixed in
with the utility plant fuel and burned at the site.

Pilot Plant Description

The pilot plant is being installed at Niagara Mohawk's Dunkirk Steam Station located
on the south shore of Lake Erie about 45 miles southwest of Buffalo, New York. The
Dunkirk Steam Station Unit No. 4 (nominally 200 MW gross capacity) provides the flue
gas source for the pilot plant. The coal burned at the Dunkirk Steam Station is
bituminous coal from a variety of mines in western and northern Pennsylvania and
West Virginia. This coal has a sulfur content ranging from 2-3% by weight.
Approximately 70 percent of the coal used at the Station is washed and the remainder
is raw coal. The net heat rate for the Unit averages 9590 Btu/kWh. An ESP operates
upstream of the flue gas extraction point, providing an inlet particulate concentration of
about 0.04 Ib/MBtu.

The Tung FGD Process pilot plant is designed to remove more than 90 percent of the
S02 contained in a nominal 2 MW flue gas slipstream extracted from the Unit No. 4
ductwork downstream of the ESP and I.D. fan. The pilot plant is installed beneath the
existing Unit No. 4 precipitator. The following material describes the equipment
components currently incorporated into the system design. Figure 2 is a somewhat
simplified process flow diagram for the Tung system.

S02 Removal

The Niagara Mohawk power plant provides all utilities supplied to the pilot plant,
including fresh water, plant and instrument air, cooling water, steam, deionized water,
and flue gas. A 3 MW equivalent slipstream of the flue gas is taken from the ductwork
downstream from the existing plant I.D. fan. Two-thirds of this gas (approximately
6,500 ACFM) flows into a venturi scrubber/saturator constructed of rubber-lined
carbon steel with a Hastelloy venturi throat. The other third of the flue gas bypasses
the absorbers and serves as reheat for the scrubbed gases prior to their return to the
utility ductwork.

96-3


-------
In the venturi scrubber, the flue gas is contacted with recycled water. Fly ash,
hydrogen chloride/fluoride and a small amount of S02 are removed in this quench
operation. A purge stream from the venturi liquor recycle loop is pumped to the
quench biowdown/neutralization tank where lime is added for pH adjustment The
purge stream is then combined with organic-contaminated waste water and sent to the
local city sewer system.

Saturated flue gas exiting the venturi will flow through a cyclonic separator to remove
residual acidic droplets, then flow into and up through a packed tower absorber where
the flue gas is contacted with the recycled scrubber liquor to remove S02. The
packed tower absorber is constructed of 316L stainless steel. Two, 100% capacity,
recycle pumps are installed to supply both the absorber and the venturi/quencher,
allowing continuous operation of both units. A portion of the absorber recycle liquor is
diverted through a bleed filter for residual solids removal, prior to sending it to the
sodium bisulfite liquor tank. The bisulfite liquor leaves the scrubber at approximately
52°C (125°F). This solution is cooled to 24°C (75°F) by a heat exchanger and is
then pumped to the sulfite extraction system.

The scrubbed flue gas, with at least 90% of the S02 removed, exits the absorber
packing and then passes through a mesh pad mist eliminator before finally leaving the
scrubber. Scrubbed flue gas mixes with the by-passed flue gas for reheat, and then
passes through the pilot plant I.D. fan. All pilot plant ductwork is constructed of
rubber-lined carbon steel. The concentrated regenerated S02 from the stripper off-gas
condenser is combined with the reheated gas, and then returned to the Niagara
Mohawk Unit No. 4 flue gas duct system.

Sulfite Extraction

The bisulflte-rich spent scrubber liquor is pumped to a five-stage mixer-settler
extraction system where it mixes with an organic solvent. The solvent extracts sulfite
from the bisulfite solution, thus regenerating the sodium sulfite scrubbing liquid.
Following the sulfite extraction from the solvent, the regenerated sodium sulfite solution
is separated from the organic solvent by a centrifuge. This sulfite solution then serves
as the regenerated feed for the S02 scrubber. Any sodium losses are made up by
direct feed of soda ash solution.

The loaded organic stream exiting the mixer-settlers also passes through a centrifuge
to remove any entrained aqueous solution. The loaded organic flows to a storage
tank, and the solvent is then preheated to 205°F and pumped to the top of the steam
stripper. A nitrogen blanket is maintained in all solvent systems to prevent oxidation of
the organic reagent and sulfite solution.

96-4


-------
Steam Stripping

Steam is fed into the bottom of the stripping column to strip S02 from the suifite-
loaded solvent stream. The lean organic solvent stream exiting from the bottom of the
stripper is first separated from any entrained aqueous fraction in a horizontal settling
tank. The regenerated solvent is then cooled and returned to the lean solvent storage
tank. The stripping column is a 30-tray, 10-inch diameter column constructed of alloy
steel. Multiple inlets are provided in the column to allow evaluation of performance for
various sizes of columns.

The S02/steam stream exiting the stripper passes through the S02 off-gas heat
exchanger, which condenses the water in the S02 gas stream. The S02 gas is
returned downstream of the pilot plant l.D. fan and is mixed with the reheated flue gas.
In a commercial plant, the S02 would be processed to provide a saleable byproduct.

A general arrangement drawing is presented in Figure 3 showing the locations of
various subsystems on the pilot plant site.

Pilot Plant Design

While all major process steps use only standard chemical process equipment
(absorber, extractors, steam stripper), design of these equipment components for the
pilot plant was not always straight forward. There were specific problems that
emerged either due to the nature of the process or the size of the pilot plant. This will
be illustrated based on the design of the steam stripper.

Design Data

Vapor-liquid equilibrium data, chemical reaction rate data, and relevant hydraulic data
were gathered before starting the steam stripper design. These data are presented

below:

1. Vapor-liquid equilibrium data were assembled by Badger Engineers, Inc. (now
Raytheon Engineers and Constructors). The data, which is presented in Figure
4a, shows that the loaded organic is relatively easy to strip. At 70% sulfur
loading, the vapor pressure at 100°C is about 260 mm Hg.

96-5


-------
2. A Thermogravametric Analysis (TGA) rate study of the loaded organic

decomposition was conducted aid the results are summarized in Figure 4b.
This figure shows that at 100°C, complete decomposition of the 100% loaded
organic can be achieved in about 4.9 minutes. The variation of rate constant
with temperature is shown in Figure 5. From these data, a rate equation for
decomposition was developed:

JL» = k (i_a) = 0.019 (1 -a) at 100°C
at

Where « = fraction of decomposition

The significance of this data is that in the stripper column design, the residence
time of the organic liquid in the column must be longer than 4.9 minutes. To
provide a reasonable operating margin, 5.5 minutes was used as the residence
time in tine design specification.

3. Using a two-inch diameter Oldershaw column, a series of laboratory-scale

steam stripping tests with the loaded organic were completed. The results are
summarized in Figures 6a and 6b and Table 1. These data show the
importance of turbulence when stripping the organic liquid. In particular, Table
1 shows that to achieve effective stripping, the F value must be larger than 0.24.
The F value, defined as vVp, where v is the superficial steam velocity and p is
the vapor density, has been regarded by many as a mixing index. When it
drops below 0.24, the turbulence is not sufficiently intense, and the stripping
performance, as indicated in Table 1 as the percent stripped, decreased.

Design Considerations

1. Column Type - Three different stripper column designs were considered for
stripping the loaded organic: packed column, valve tray and sieve tray. The
type of column selected is a judgemental decision., In the test program
planning, a mathematical model was constructed to guide in the selection.
Unfortunately, for logistical reasons, the model computations were not
completed prior to the design selection; therefore, the decision was made
without the benefit of the model's guidance.

Although a packed column, under certain circumstances, can give a high F
value, it was decided that the F value in this instance was not a good criterion.
In a packed tower, mass transfer is predominantly the across-the-film mass
transfer, and that is not the type that would best meet the needs of the process.
Therefore, in spite of the fact that F value could be large, a packed column was
not suitable for this application.

96-6


-------
The decision of whether a vaive tray or a sieve tray should be used was a more
difficult one. It was concluded:

•	The valve tray column has an advantage over the sieve tray column in
providing larger turndown (because slots in the valves can be reduced
as the gas loading decreases).

•	For large industrial columns, the stripping efficiencies of these two types
of columns are comparable. However, for a 12 inch (or smaller)
diameter valve tray column, data were not available to confirm that
performance of the column would not suffer because only a small
number of valves can be positioned on a tray. Furthermore, most valve
fray vendors do not want to supply these trays if the column diameter is
less than 12 inches.

Therefore, selection of a valve tray versus a sieve tray depends primarily on the
column diameter. If the column diameter is larger than 14 inches, a valve tray
column should be favored. On the other hand, if the column diameter is less
than 12 inches, a sieve tray should be selected. The fact that the laboratory-
scale Oldershaw column, which used a sieve tray, worked well also favored the
selection of the sieve tray.

Column Diameter - Four pounds of steam per pound of S02 released, which
had been successfully demonstrated in Oldershaw column, was selected as the
design point. To be suitable for this application, the column must satisfy the
following hydraulic and kinetic conditions:

•	The F valve must be greater than 0.24

•	The liquid residence time must be greater than 5.5 minutes.

The F value calculation shows that for columns having 12, 10, 9, 8, and 7 inch
diameters, the F values are 0.32, 0.48, 0.57, 0.72, and 0.95, respectively. Thus,
a 12 inch column cannot demonstrate a steam use rate of two pounds of steam
per pound of S02 released, because at that steam rate, the F value will be
0.32/2 = 0.16, which is less than 0.24. A 10 inch column could possibly
demonstrate two pounds of steam/pound of S02, because for that column, the
F value at that steam rate is 0.48/2 = 0.24, which is barely sufficient to achieve
effective stripping. The column diameter would have to be reduced to 7 inches
in order to demonstrate one pound of steam/pound of S02 since 0.95/4 s
0.24.

For a 20 tray stripping column, the liquid residence time of a 10" column with a
weir height of 2 inches is estimated to be 5.7 minutes. This is sufficient time for
loaded organic to achieve 100% decomposition at 100°C according to the
kinetic data. The smaller the column selected, the higher the weir height should
be in order to keep the liquid residence time above 5.5 seconds.

96-7


-------
Since the column diameter is less than 12 inches, a sieve tray rather than a
valve tray was selected.

3.	Column Tray Design - The layout of a standard sieve tray consists of several
areas as shown in Figure 7 Area for down flow pipe positioning from the plate
above and to the plate below, calm zones between the downflow pipes and the
active areas, and active areas where the sieve holes are placed and the
liquid/gas interaction occurs must all be taken into consideration. As the
column gets smaller in size, the areas reserved to accommodate downflow
pipes as well as the calm zones cannot be reduced proportionally. As a
consequence, the active area where liquid/gas interaction (or stripping) actually
occurs becomes smaller and smaller. Actual layout design showed that for
column less than 10 inches in diameter, an effective design cannot be attained
for a standard sieve tray layout (downcomers on either side of the tray). In
addition, most vendors have little experience in making trays smaller than 12
inches in diameter. Some companies were willing to build a 10 inch column
tray, but they could not offer guarantees that the trays would maintain
satisfactory hydraulic performance.

It was decided that in order to build a smaller than 12 inch diameter column,
conventional tray design would have to be abandoned in favor of a special
design called "orbital flow." The orbital flow design is usually selected for
systems having a low liquid flow rate. Although the Dunkirk pilot plant will have
a very low liquid flow rate, the primary reason that the orbital flow design was
selected is that it has less wasted area. As shown in Figure 7, the orbital
design puts two down flow pipes on the same side. By so doing, the wasted
area is reduced and the active area is increased. Although a 7 inch column
was possible with orbital flow design, it was decided to remain conservative and
a 10 inch column was selected instead. By deciding to build a 10 inch column
rather than a 7 inch column, the possibility of demonstrating one pound of
steam/pound of S02 was eliminated. A demonstration of a steam feed rate of
two pounds of steam/pound of S02 will be adequate for the current pilot test.
Demonstration of one pound of steam/pound of S02 may occur in a larger-
scale demonstration.

4.	Column Height - Even though 20 trays were found to be sufficient in the
laboratory-scale test with the Oldershaw column, it was decided to build a 30
tray column in order to determine, among other things, whether higher
residence time would make any difference in column performance.

Project Schedule

Preliminary engineering and development of a definitive cost estimate by Raytheon
was started in November 1991 and was completed on January 31, 1992. Following
DOE approval of the definitive cost estimate, Raytheon proceeded with final equipment
and facility design and procurement.

96-8


-------
Based on a competitive bidding process, Quackenbush Inc. of Buffalo, New York was
selected to construct the pilot plant and to install the necessary equipment.
Quackenbush initiated site work in October 1992. Pilot plant construction started
shortly thereafter. Construction and equipment installation were completed in March
1993. Figures 8, 9,10, and 11 show various views of the complete pilot plant.

A Dunkirk Station Unit No. 4 annual outage occurred between March and April 1993.
This outage, along with DOE and Niagara Mohawk approval of the pilot plant Safety,
Failure Mode Analysis, and Operating Manuals, resulted in a delay in pilot plant startup
until July 1993. As is shown in Figure 12, the pilot plant test program schedule
consists of a one-month shakedown test during July 1993, followed by approximately
three months of parametric optimization testing. Following completion of the
parametric testing, a five-month long-term reliability test program will take place which
should be completed the end of March 1994. After completion of the test program,
the site will be restored to its original condition.

Test Program

The primary objectives of the pilot plant demonstration are to establish the
performance and costs of the Tung FGD Process when treating the flue gas from a
medium-to-high-sulfur coal-fired utility boiler. The test program is divided into two

phases:

•	Parametric testing - to establish the capacity, performance, and turndown
capability for each component in the system.

•	Reliability testing - to demonstrate the process performance after extended
periods of continuous operation.

Parametric Testing

The parametric testing immediately following the completion of the system startup
program. Testing consists of performance analyses for the following subsystems and
operating parameters as listed below:

•	S02 Absorption - establish the performance of the venturi/absorber system and
confirm that system removes more than 90% of the S02 from the inlet flue gas.
These tests are being done at 1 and 2 MW equivalent flue gas flow rates.

•	Extraction - evaluate performance of 1, 3, and 5 stages of extraction at 1 and 2
MW equivalent flue gas flow conditions. Study aqueous and organic
entrainment over a range of aqueous/organic phase ratios.

•	S02 Steam Stripping - evaluate stripper performance as it relates to the number
of trays in the stripping column and the steam feed rate.

•	Organic Regeneration - analyze performance of proprietary components at
various operating conditions.

The results of the parametric test program will be used to establish the optimum
operation conditions for the long-term reliability test program.

96-9


-------
Reliability Testing

The long-term reliability testing will begin as soon as the parametric testing has been
completed. The purpose of this test period is to demonstrate the performance
capability of the process during continuous, 24-hour/day operation. Equipment
reliability and chemical regeneration efficiency will be analyzed during this five-month
program. Test conditions will be changed periodically to gain additional information
regarding the impact of different equipment configuration impacts on performance over
extended operating periods. Various steam stripping and regeneration configurations
and flow rates will also be evaluated during this test period.

Laboratory Analyses

An on-site laboratory is providing analytical date on organic loading, process stream
compositions, organic viscosity, and stripping and extraction efficiency. Analytical
equipment includes a gas chromatograph, titration equipment, a viscosity meter, pH
meters, and other chemical testing equipment The date generated in the laboratory is
added to the data acquisition system data base to provide system performance results
for all operating conditions. A consistent sampling protocol is being used to allow the
laboratory data to be accurately combined with process operating conditions in order
to produce an overall performance evaluation for each set of test conditions.

Test Results

At the time when this paper had to be submitted for publication (July 2, 1993), the pilot
plant was in the initial stages of startup. Test data generated during July and August
of 1993 will be presented at the FGD Symposium.

Conclusion

To date, the Tung FGD Process has been successfully demonstrated in laboratory-
and bench-scale tests. It is currently being demonstrated at a nominal 2 MW pilot
plant-scale using a slipstream of actual flue gas from a coal-fired utility boiler.

Ultimately it is hoped that a full-scale facility will be built based on the anticipated
successful results from the pilot plant testing that is now in progress.

96-10


-------
Cleaned	Spent	Loaded

Flue Gas	Scrubber	Organic

Liquor	Solvent

Liquor	Solvent

Figure 1. Tung Gas Desulfurization Schematic

96-11


-------

-------
r

to

0>
CO

Figure 3. Tung Pilot Plant General Arrangement Drawing


-------
SULFUR LOADING-MOLES SULFITE PER MOLE ORGANIC

Figure 4a. Equilibrium Vapor Pressure of SO2 Over Loaded Organic

TEMPERATURE (deg C)

Figure 4b. Loaded Organic Decomposition Kinetics
Conversion Time Versus Temperature

96-14


-------
1M6

3J> 2.1 22 Z3 Z* £6 2.8 Z7 £8 2M 9X> 3.1 &2 3.3 &4
1/TEMPERATURE {lOOOMefl K)

Figure 5. Rate Constant Versus 1/Temperature

Figure 6a. Stripping Column SO2 Removal Versus L/G

Figure 6b. Steam Stripper Test - Steam Consumption

96-15


-------
ORBIT FLOW DESIGN
Figure 7. Sieve Tray Design Alternatives

98-16

A


-------
Figure 8. Process Storage Tanks

Figure 9. Absorber Recycle Pumps

96-17


-------
Figure 10. Stainless Steel Absorber, Bleed Filter, and Motor Control Center

Figure 11. Insulated Process Heat Exchangers

96-18


-------
Uffll "" "• • 			i KJOBlMWiro		""'•"X'-lBBmBBBWrWWm			

7UNBR3DTE5T FACILITY-PHASE il j

RATCON RESEARCH & DEVELOPMENT, INC. i
ti&AKEROAO ¦*- aawmaiatTt
SHOOKUNE.IMCK46 I

7" tLtMKT 1*

eoas i tuvor

« 0WT1M ^ I ,	 1

wtia ( fffi J f¥M

14 wwann
conpucm

1

II | 9 U j F j U A m 4 4 A i a j 0 | K 1 S J f m I A It j J | J -tiwjojujDijjF |m .*>

a. |t.
mm 1 4C*ttU.

tvommjm ^

¦*"** miwT

puu ryiw -7Tf«i

1



i







ntfj . hmm'

, ,UTUM

T7







j







:ST







1







TASK 4

watfer>ujuaB»Ai6orrf

HtlESTatE STATUS REPORT

ELBfEHT

CODE

PUNNED PLANT DESCRIPTION



ELEHEHT
CODE

PLANNED PLANT DESCRIPTION



Raveon Task 1 started



iC



_ _

Issue Phase I! Work Plan



40

Issue Contractor Soscifi cation for Bid

„„

Issue Phase II Mark Plan fRev. 1)



4E

Start Construction

1A

Task 1 - Raytheon Subcontract Hark Started



4F

Start Shakedown (Begin C/Q & S-U
Activity]

!3

Pre'pet. Coot 1 eteri



4S

Carolete Construction

ic

Issued Phase II Final Renort



4H

Finish Shakedown (Mechanical CcmnlRtionl

2A

laboratory Work Corns! eted



41

Test Facility Removed

3A

"¦"ask S - Paci J itv Oesfon Standard



SA



3E

Definitive Eaiiipment/Frograni Budget
Completed and Safin Engineering
(1st Ravcon Oesien Review)



53

Test Completed

3C

?nd Ra*con Desian ievipw




-------
TABLE 1

LABORATORY SCALE STRIPPING DATA

Steam Consumption

F Value = v vp

% of S02 Stripped

6.3 lb/lb S02

0.32

93%

4.0 lb/lb S02

0.23

97%

1.8 lb/lb S02

0.11

65%

1.0 lb/lb S02

0.06

42%

96-20


-------
INTEGRATED FLUE GAS TREATMENT BY WET FGD
OPERATING IN A WATER-CONDENSING MODE

J. P. Heaphy
J. C. Carbonara
Consolidated Edison Company
of New York
4 Irving Place
New York, New York 10003

W. Ellison
Ellison Consultants
4966 Tall Oaks Drive
Monrovia, Maryland 21770

Abstract

Results are presented for electric-utility, field testing of novel, heat-recovery type, wet
flue gas desulfurization facilities of ultra-high efficiency in removal of S02, S03 and
solid particulate matter. The focus of this technology development and
demonstration-scale, continuous performance testing is an upward-gas-flow, water-
cooled, condensing heat exchanger equipped with teflon-covered tubes and tubesheets
that also provides a basis for simultaneous NOx absorption in an oxidation/absorption
FGD process-mode while serving to reduce trace metal emissions in keeping with
objectives of new Clean Air Act, Title 1H. Hie mechanical design of this advanced
flue-gas cooling/scrubbing equipment is based on more than twelve years of
commercial application of such units for energy recovery by boiler make-up water
preheating in residual-oil-fired boiler operation. This Integrated Flue Gas Treatment
(IFGT) technology is additionally proposed for use in coal-fired boiler service,
typically positioned downstream of a high-efficiency electrostatic precipitator as in
conventional wet FGD practice.

97-1


-------
Introduction

The technical development by Con Edison described herein has the unique aim of
demonstrating a new technological approach fen* power plant flue gas heat recovery
and stack emissions reduction. This technology employs equipment known as a
condensing heat exchanger, and the first condensing heat exchanger demonstration at
Con Edison began in February, 1992. A condensing heat exchanger extracts generally
unavailable energy from the flue gas stream by lowering the temperature of flue gas
well below the sulfuric acid dewpoint temperature. Conventional flue gas heat
recovery equipment foregoes use of "waste teat" in the flue gas to avoid sulfuric acid
formation and corrosion that results when the flue gas temperature is lowered below
this acid dewpoint. These limitations do not exist with the condensing heat exchanger
design now being demonstrated because all gas-side tube and shell surfaces are
covered with a layer of teflon resin. Therefore, ultra-low flue gas exit temperature
can be achieved by utilizing such a condensing heat exchanger. Considering that the
heat lost through discharge from the stack is one of the major contributors to boiler
inefficiency, substantial stack temperature reduction and the attendant heat recoveiy
can reduce fuel consumption considerably. In addition, environmental benefits are
significant since a condensing heat exchanger system can be designed to serve as a
"scrubber" in that a substantial amount of water vapor in the flue gas may be
condensed to provide a unique counter-current interaction with the gas. By successful
demonstration during this Con Edison project, condensing heat exchangers can be
shown to have the potential to become standard boiler equipment for improving plant
efficiency while serving as the key component of a stack gas treatment system for
integrated flue gas cleanup.

Prior Use of Teflon-Covered

Condensing Heat Exchangers in Heat Rate Improvement

The mechanical design of this advanced flue-gas processing equipment is based on
more than twelve years of commercial application of single-stage units in downward-
gas-flow design/operation solely for enhanced energy recovery, e.g. in preheating of
boiler makeup water or combustion air in residual-oil and natural-gas fired service.
Con Edison has now pioneered commercial, electric-utility scale application of the
process.

industrial Boilers
Overview.

The Teflon-coated condensing heat exchanger, although new to electric utility service,
has been applied in the industrial sector for more than twelve years. There are
currently more than one hundred such condensing heat exchangers that have been

97-2


-------
installed in the U.S. recovering energy from boiler flue gases and other stack gases.
The use of corrosion resistant, Teflon-covered condensing heat exchangers allows
previously wasted energy to be utilized increasing plant efficiency. Companies such
as IBM, Exxon, Union Carbide, ADM, Bristol Meyers, and Anheuser Buseh have
installed Teflon-covered condensing heat exchangers to improve the efficiency of their
steam-generating operations. The corrosion proof design of the condensing heat
exchanger system is achieved by covering all gas side surfaces with DuPont Teflon.
Teflon has been used in chemical service for over forty years, its main application
being Teflon-lined pipe. Just as the Teflon lining protects chemical plant piping form
acid attack, the Teflon covering on the outside of the heat exchanger tubes protects
them from corrosion. The conventional criteria for minimum allowable metal
temperatures for common, exposed-metal, heat recovery equipment designs therefore
do not apply. Inside the shell and tube heat exchanger, the tubes are covered with a
0.015 inch (0.38mm) thick extrusion of DuPont Teflon on the outer surface. The
shell is protected with sheet Teflon 0.060 inch (1.52mm) thick. The tube to tube-
sheet seal is a dynamic Teflon-to-Teflon seal. The outlet plenum and discharge duct
are typically fabricated of fiberglass reinforced plastic (FRF), which, like Teflon, is
inert to add attack (1).

Tube Preparation.

The fabrication process for a Teflon covered heat exchanger tube is similar to that for
insulating a copper wire. Clean copper/nickel tubing is fed through an extruder
crosshead where molten Teflon flows evenly around the tube's circumference under
high temperature and high pressure. The thickness of the coverings, typically 0.015
inch (0.38mm) is precisely controlled by varying extruder speed relative to the tube
feed rate. After the Teflon is in place, the tubes are cooled in a water bath to set the
material. Next, a doughnut shaped, sine-wave spark tester is passed over the tube to
check for flaws in the covering. If flaws are found, the tube is rejected and the metal
tubing is recovered. If not, it is installed directly into a heat exchanger being
assembled to avoid possible damage during storage. The end result of this painstaking
manufacturing process is a covering that is far superior to conventional sprayed-on or
dipped coatings. The Teflon covering has a recommended upper temperature limit of
400F (204°C) but temperatures up to 500F (260°C) can be tolerated as long as there is
fluid flowing through the tubes to limit the Teflon temperature. Studies indicate that a
thin Teflon covering typically reduces the heat-transfer ratio of the tubing by less than
10%. Unlike exposed metal tubes, Teflon does not become fouled, nor does it retain
a continuous surface layer of water. Thus, the overall impact on heat transfer rate is
negligible. (2)

97-3


-------
Electric Utility Boilers

Heat Recovery.

Favorable heat recovery performance has been achieved for more than a year in an
initial electric utility application of this technology at Con Edison's 74th Street
Cogeneration Plant with 0.3% sulfur residual oil firing (3). Two, in-parallel, single-
stage exchangers with a total flue gas flow capacity of 320,000 lb/hour, one third of
average plant flow, reduce gas temperature to less than 100°F (38°C). By extracting
additional energy from the flue gas, boiler makeup water is preheated approximately
40°F (22°C), with the same fuel input. By doing so, an additional 23,OCX) lb/hour of
steam is made available for district supply as a result of reduction in the amount of
steam required in the boiler deaerator to heat the water to its boiling point. The
cooled gas is intermixed with the balance of boiler gas yielding a stack temperature of
250°F (121°C). Performance attained verifies the potential for integrating the
condensing heat exchanger into various power plant cycles. A realistic goal for heat
rate reduction (4) when boiler condensate is heated is:

•	Combined cycle plants operating at greater than 40% cycle efficiency:
approximately \xk% heat rate reduction

•	Reheat boilers at approximately 35% cycle efficiency: approximately 4%

•	Nonreheat boilers at approximately 30% cycle efficiency: approximately 6%

•	MSW (municipal solid waste) plants at 20% cycle efficiency: approximately
11%.

Flue Gas Cleaning Effects.

For the most efficient design for heat transfer, downward-gas-flow heat exchanger
arrangements, such as at the 74th Street Plant, have long been applied at industrial
plants. In all such applications in residual oil service significant removal of flue gas
pollutants has been observed through the substantial dissolved and suspended solids
content of condensate wastewater. However, experience with upward-gas-flow
installations indicates significantly greater flue gas cleaning effect with this latter
physical arrangement, the result of the effect of counter-currency of flow, gas vs.
descending liquid condensate, on the shell side. Such field observations have now
pointed to an improved integrated flue gas treatment (IFGT) technology in which;

•	A maximum degree of heat recovery and of flue gas cleanup may be achieved

•	A downward-gas-flow, first-stage heat exchanger, used primarily for flue-gas
sensible heat recovery and conditioning of flue-gas pollutants, is followed by an
upward-gas-flow, second-stage heat exchanger serving to recover latent heat
awl as a wet scrubber that treats water-saturated inlet flue gas.

97-4


-------
Innovative, Multi-Stage Use of Indirect

Heat Exchange for Simultaneous Flue Gas Cleaning

Process Rationale

The focus of this integrated flue gas treatment (IFGT) technology development is an
upward-gas-flow, indirectly water-cooled, condensing heat exchanger (the second of
two exchanger stages) fitted with acid-proof, teflon-covered tubes and tubesheets. It
provides a unique condensing (non-evaporative) wet-scrubbing mode to address air
toxics control objectives of new Clean Air Act, Title in. Moreover, it is also capable
of simultaneous S02 and S03 removal. Advantageous trace-metal
condensation/nucleation/agglomeration along with substantially enhance boiler
efficiency is accomplished in the IFGT system by use of boiler makeup water or an
alternative heat side. Boiler flue gas is indirectly cooled to a near-ambient-
temperature, low-absolute-humidity, water-saturated state. Applied in an optimal
manner via the two-stage arrangement, (EFGT), as per Figure 1, condensing heat
exchangers possess unique potential to remove S02 S03 HC1, sub-micron flyash and
trace heavy metals, volatile organic compounds and other toxics that may soon be
regulated under the new Clean Air Act (5). Thus, with beneficial use of the recovered
heat, this may be the first stand alone scrubbing process with simultaneous removal of
multiple flue-gas pollutants that actually improves the energy efficiency of a scrubbed
power plant.

System Design/Operation

The crux of this new technology (IFGT) is the inclusion of two stages of condensing
heat exchange designed to recover useful energy and to cool the flue gas to near-
ambient temperature. Flue gas is passed first through a downflow Teflon-covered
condensing heat exchanger that transfers energy to condensate from the hotwell pump
discharge in a closed cycle power system, to boiler makeup water, process water, or
some other sink for recoverable heat. The flue gas is cooled in the downflow heat
exchanger to a temperature slightly above the water vapor dewpoint, principally
removing sensible heat. The flue gas leaving the first heat exchanger is brought to
water saturation by sprays in a fiberglass transition chamber. These sprays cool the
flue gas to its wet bulb temperature, remove any coarse particulate matter and further
condition the gas for efficient removal of S02, S03, and other pollutants. The water
saturated flue gas then enters the upward-gas-flow Teflon-covered heat exchanger. As
the saturated flue gas contact the cold tube-surfaces of the heat exchanger, additional
energy is recovered and substantial condensation of water vapor and condensable trace
metals in the gas stream occurs.

97-5


-------
Flue-Gas Pretreatment.

In the case of applications with high ash-content fuel such as coal, the use of an
electrostatic precipitator, fabric filter, or mechanical collector is used to avoid a
massive solid particulate loading entering the EFGT system. In high-sulfur fuel
service, additional pretreatment of the flue gas prior to the JFGT may include flue-gas
NO to N02 conversion, e.g. in the rear boiler cavity, to accomplish IFGT operation
that achieves removal/absorption of NOx in conjunction with S02 removal.

Dry Bulb Temperature Reduction by Indirect Heat Transfer.

The first IFGT stage is a downflow Teflon-covered heat exchanger that cools the gas
by removing sensible heat. It also conditions the gas for subsequent effective removal
of raw-gas S03. The flue gas leaving the first stage is cooled and saturated by the
action of water sprays in a fiberglass transition chamber. These sprays not only cool
the flue gas to its wet bulb temperature, but also removes coarse particulate matter
and further conditions the gas for efficient removal of fine particulates, SQ2 and S03.

Spray Fogging and Water Saturation.

Hie water saturation of the flue gas in the transition zone from the first to second
stage provides an opportunity to apply fogging sprays serving to augment
nucleation/removal of small/sub-micron particulate.

Counter-Current Wet Scrubbing With Condensation.

As the water saturated inlet flue gas contacts the cold tube-surfaces
of the second stage heat exchanger, additional energy is recovered and substantial
condensation of water vapor occurs. This subcooling of the flue gas, accompanied by
water vapor condensation, creates a large amount of fine water droplets that form on
both pre-existing and newly-formed submicron solid particles and sulforic-acid
droplets. This supports a unique condensation-scrubbing action that makes possible
substantial collection of fine particulate matter. It is captured by impaction on large-
droplet and wetted-tube surfaces with nucleation of ultra-fine droplets by continuous
gas temperature reduction in the second stage heat exchanger. These sub-micron
droplets thereby coalesce and agglomerate to larger droplets that are removed by
gravity, by impaction on tube surfaces and by mist elimination. Hie addition of an
alkaline reagent with recirculation of the IFGT liquor allows the particulate removal to
be accompanied by simultaneous flue gas desulfurization. Soda ash provides a water-
soluble, clear-liquor, sulfite/bisulfite scrubbing medium compatible with the exchanger
construction and with capability for high S02 removal efficiency. When used in
conjunction with NO to N02 gas conversion technology, IFGT has the capability for
simultaneous absoiption of S02 and NOx. In both low and high-sulfur service,

97-6


-------
important potential exists for S02 removal without chemical reagent through the effect
of catalytic oxidation.

Mist Elimination.

The mist eliminator, which provides final cleanup of gasborne wetted particulate,
removes the small droplets (down to 10 micron size and smaller) from the gas exiting
the exchanger for return to the process liquid inventory. The inertial mist eliminators
used in large-capacity wet scrubbing systems for boiler flue gas operate on the basis of
gravitational (dynamic) forces generated through change in gas flow direction. For
maximum efficiency a horizontal gas flow arrangement, employed for the final
eliminator stage, uses elevated gas velocity through vertical, vaned elements. Phase
separation chambers on fee trailing edges of the vanes extending counter to gas flow
direction afford drainage paths external to the gas-flow regime for removal of the
collected liquid without its reentrainment by the high-velocity gas flow between the
vanes.

Stack Discharge.

It is foreseen that unique innocuous exit-gas properties will allow atmospheric
discharge, as is, without complication in wet stack system design and operation
encountered with conventional high-gas-humidity, wet scrubber operations. Exit-gas
humidity is expected to be very low, (approximately 0.03 lb water vapor per lb dry
gas at 85°F i.e. 29°C). This alleviates problems usually associated with condensation
of moisture on stack internals that often require costly gas reheat equipment. An
effective mist-eliminator system will further serve to minimize presence of liquid in
the stack and thereby preclude downpitching of the plume due to the thermodynamic
effect of internally-created, plume chilling (6). Additionally, because of the
anticipated efficient removal of principal flue gas pollutants, adverse impact on
ground-level air quality is expected to be avoided despite no significant plume rise.

Waste Management

In a low-sulfur applications, wet chemical treatment of the excess-condensate purge
stream (for outfall discharges) by lime generates waste gypsum particles that enhance
flocculation in the wastewater treatment operation and reduce outfall-liquid trace metal
concentrations to acceptable levels. Collected trace metals are rendered in a low-
volume, trace metal/gypsum sludge that would be disposed of in a secure landfill.

Based on extensive, favorable experience in the U. S., calcium sulfite cake from a
high sulfur application, intermixed with the dry fly-ash catch from upstream of the
heat exchanger system (and a small proportion of hydrated high-calcium time additive)
advantageously forms a pozzolanic fixation product that can be used to build a

97-7


-------
structural landfill. Such a disposal-mass structure has permeability so low that, in
perpetuity, it remains non-water-saturated and thereby cannot be the source of
significant outflow of leachate to any underlying groundwater body. This is of
particular significance and a major environmental benefit in exploiting the gas cleaning
performance of the heat exchanger system since:

•	Heavy metals, chlorides, etc., collected in the condensing heat exchanger and
contained in the solid and liquid phases of the sulfite cake waste are
permanently encapsulated/isolated in the structural landfill mass.

•	A substantial portion of excess condensate yield and its dissolved solids is
assimilated as surface moisture in the sulfite cake waste.

•	Sludges formed in treatment of the net excess condensate yield may be
conveniently discarded by blending into (interspersing with) the fixed landfill
mass during the cake-waste/fly-ash mixing step.

Alternatively, high-sulfur applications may use the Thioclear FGD process which
employs a recirculating, clear-liquor, magnesium hydroxide medium and yields
gypsum byproduct usable in large-volume gypsum wallboard manufacture. As in die
low-sulfur mode, above, FGD liquid effluent outfall is chemically treated to render
hazardous air pollutants in concentrated form as a low-volume, disposable sludge
solids and achieve a permitted discharge to an available receiving stream.

Performance Objectives

Simultaneous Pollutant Removal,

Ultra-low emission of diverse pollutants is foreseen via IFGT use, even with poorest
quality, residual fuel stocks. Although not generally recognized as such, residual fuel
oil is a waste product, the bottom-of-the barrel remnant of gasoline and distillate fuel
production from diverse petroleum resources. Much international residual production
is available only in medium/high sulfur form and is increasingly being restricted in use
by worldwide initiatives for S02 emission reduction. Depressed price of low-grade
heavy oil, both petroleum and bitumen based, makes it a uniquely attractive energy
source in operation of utility boilers equipped with advanced, state-of-the-art, flue gas
cleaning facilities. Thus, gas cleaning performance in IFGT use is aimed at gaining
futuristic, low, pollutant-emission levels in firing of available, minimum-cost, heavy
oil fuel.

Flue-Gas Heat Recovery.

The Teflon-covered condensing heat exchanger is an indirect-contact heat exchanger in
which both latent heat and sensible heat are transferred form the flue gas to a heat
sink. Typical heat sinks available at sufficiently low temperature are boiler makeup

97-8


-------
water, process water return, condenser condensate-water and combustion air. In a
typical application for the condensing heat exchanger in industrial or commercial
service, cold water is heated with flue gas to reduce steam consumption. For
example, preheating boiler makeup water prior to the deaerator decreases the steam
requirements associated with deaeration. By reducing this steam use, overall
efficiency of the steam generation operation is increased. Ideal use of IFGT exploits
available heat sinks to gain energy cost savings in the most cost-effective manner.
The most favorable heat-recovery application of a condensing heat exchanger system
is for a steam send-out (district steam supply) power system that uses a high rate of
boiler makeup water (e.g. raw water at 60°F to 70°F, i.e. 15.5 to 21°C), as well as
high heat-rate power systems such as those employing non-reheat boiler units. For
example, a condensing heat exchanger system installed at the Rochester District
Heating Co-op in Rochester, New York, heats 100 percent of the makeup water used
for steam sendout, increasing plant energy efficiency by 7 to 10 percent with an
investment payback within approximately 2 years.

Field Pilot Plant Testing to Date of

Gas Cleaning by Integrated Flue Gas Treatment

Field test work has been carried out in residual oil fired service at two plants, one an
industrial site, (Morgan Linen Company), the other a steam sendout facility of Con
Edison at its Ravenswood Generating Plant in Queens, New York.

Morgan Linen Company

Test Objectives.

Condensing Heat Company Coiporation has carried out a program for source emission
testing on a factory-pilot-scale. This Integrated Flue Gas Treatment facility has also
been installed at Morgan Linen Company, Menands, New York. Testing was
performed to determine emissions of solid particulate matter, sulfur dioxide (SOa) and
sulfur trioxide (S03). The primary purpose was to generate emissions data to provide
preliminary quantification of the gas cleaning efficiency of the unit in medium/high
sulfur fuel service.

Field Test Operations.

Morgan Linen Company has a 14,000 pph rated boiler producing an average of 7,000-
8,000 pph of steam at lOOpsig. The fuel burned in the boiler is 1.5% sulfur, No.6
residual fuel oil. This pilot IFGT System is of a capacity that treats a 5060 pph
slipsteam of this boiler exhaust flue gas at 320°F, i.e. 160°C (approximately 65% of
total boiler flue gas). The 5060 pph of flue gas was extracted from the existing stack
by the IFGT system's induced draft fan. The downflowing flue gas then passed

97-9


-------
through the first Teflon-covered condensing heat exchanger, transferring heat to the
water in the tubes. Hie flue gas was cooled from 320°F to 113°F, which is near the
water dewpoint temperature of the gas. Upon exiting the downflow heat exchanger,
the 113°F (45°C) flue gas entered the fiberglass reinforced plastic (HIP) spray
chamber/plenum where it was contacted by sprays of a reagent solution. These sprays
quench and water saturate the flue gas at 105T (40.5°C). The sprays promote liquid
to gas contact allowing dissolved reagent (sodium-base alkali) to begin the process of
S02 removal, resulting in a water-soluble reaction product. The saturated flue gas
enters the second (upflow) Teflon-covered condensing heat company and is cooled to
88°F (31°C) as sodium liquor recirculation completes the S02 removal operation. As
the saturated flue gas contacts the cold tube surfaces of the heat exchanger, water
vapor condenses. The condensation creates a mist of nucleating water droplets that
contribute to capture of particulate. Scrubbing liquid flow due to condensation in the
upflow teat company is augmented by the recirculating-liquor spray nozzles positioned
above the tubes. The gas passes through a mesh type mist eliminator of high
efficiency prior to discharge to the FRP stack.

Overview of Test Results.

Results of the test program for this particular boiler indicate that the IFGT system
removal efficiency for total particulates averaged 89%. Hie average system removal
efficiency for S02 and S03 during tests was 99.1 and 99.3%, respectively. The
efficiency of removal of various metals is believed to be very high and will be
quantified in follow-up testing at Ravenswood. A perfunctory test was performed with
methanol injection in the rear cavity of the furnace to convert NO to N02 for removal
by the alkaline IFGT liquor. Confirmation of simultaneous SO2/N0x removal was
achieved. Further technical development and field testing work will be carried out to
achieve higher particulate removal efficiency and to quantify metals removal
efficiency.

Ravenswood Steam Sendout Plant
Overview.

Preliminary Ravenswood test work was carried out in early 1993 in anticipation of a
late 1993 formal field program at the residual-oil-fired, Ravenswood *A"-House
Steam Sendout Plant. This program is to address the field operation and testing of a
recently installed Integrated Flue Gas Treatment (IFGT) system supplied by
Condensing Heat Exchanger Corporation with capacity to treat 25,000 lb/hour of
boiler flue gas. This system will ultimately serve as a proof-of-concept,
demonstration facility for heavy-oil fired service achieving substantial heat recovery as
well as integrated stack gas cleaning functions. Testing will establish environmental
pollution control performance with respect to principal toxic chemical (hazardous air

97-10


-------
pollutant) emissions, including nickel, in oil firing both so far as stack gas cleaning as
well management of collected wastes, liquid and solid. Other pollutant species
targeted for removal from flue gas include total particulates, (condensable in addition
to non-condensable), sulfur dioxide, sulfur trioxide (NO and NOJ, as well as
hydrogen chloride.

Focus of Long-Range Test Program.

This will demonstrate:

•	Heat recovery benefits and efficiency improvements significantly decreasing
plant rate

•	Potential to burn less costly, high-sulfur fuel (in lieu of low-sulfur) without
increase in SO* and S03, emissions

•	Marked reduction in air toxics (hazardous air pollutants) and S02 emission

•	Potential major reduction in emission of sulfuric acid mist (S03/H2S04)

•	Elimination of visible acid-mist discharge and, except for winter months,
absence of visible steam-plume discharge

•	Avoidance of visible plume occurrence in highly populated localities during
warm-weather months

•	Capability to use existing stacks for flue gas discharge, avoiding need to erect
new stacks for discharge of the low-humidity, water-saturated flue gas

•	Lessening of NOx emission.

Collaboration with Brookhaven National Laboratories.

Field test activity is being augmented by pilot plant testing at Brookhaven National
Laboratories in conjunction with a test combustor of capacity less than 1 million
Btu/hour. This collaboration provides a versatile means of assessing a broad range of
operating conditions including enhanced condensate flux generation per unit gas flow
volume with the aim of evaluating anticipated improvement in sub-micron particulate
collection.

Preliminary Field Test Results.

In low-sulfur oil fired operation flue gas testing has shown that:

•	Cooling, scrubbing and nucleation effects achieve, without benefit of
optimization of system design/operation, stack particulate loadings as low as
0,005 lb/MM Btu, one-sixth of the 1979-enacted New Source Performance
Standard (NSPS) for coal-fired boilers.

•	Mercury removal efficiency exceeds that for dry-bottom coal-fired boilers
served by a high-efficiency ESP and prescrubber-equipped wet FGD system,

97-11


-------
shown by field testing to be approximately 50%.

• Removal of S03 and sulfuric add mist averaged 66%, substantially greater than
the 30 to 50% removal level typically achieved in conventional, utility wet-
FGD scrubbing operations. IMs is expected to increase to over 90% with
alkaline reagent use as in the Morgan Linen Company tests reported above.

Influence of Key Variables in System Design /Operation

Pre-Coo!ing

The initial indirect cooling of the raw, low-dust, flue gas in the first exchanger stage
takes place in a protracted manner that provides advantageous conditioning of S03 gas
by its slow cooling down to and below the sulfuric acid dewpoint temperature. Such
gas its treatment provides an advantageous means of promoting growth of acid mist
particles formed in advance of small-micron particulate collection in the condensing
regime provided by the second stage exchanger.

Atomized Sprays

Use of fogging sprays generates ten micron water droplets immediately downstream of
the first-stage exchanger. It is believed to provide a significant mechanism for
improvement of vapor droplet ami particle agglomerating effects available within the
condensing regime of the second stage exchanger. Fogging sprays are augmented by
conventional coarse-particle sprays to ensure total water saturation of flue gas
upstream of the second-stage exchanger.

Flux Rate of Water Condensation From Flue Gas

Hie heart and essence of the technology as it is expected to be broadly/universally
applied is the ultra-low-temperature, final condensing heat exchanger stage. Favored
with water-saturated inlet flue gas to which to apply the condensation scrubbing, it
provides a mechanism for ultra-low emission of sub-micron particulate and of
condensing gaseous contaminants that produce the sub-micron particulate. Note that
sub-micron control means i.e. polishing superior in performance and/or attractiveness
to conventional scrubbing methods, is afforded even when the heat-sink (cooling water
supply) temperature is as high as 80 or 90°F, (27 or 32°C), or more. Subcooling of
the flue gas In the second-stage exchanger accompanied by water vapor condensation,
serving as a unique scrubbing "flux", creates a large amount of condensate that forms
on both pre-existing and newly formed submicron solid particulate matter. This
droplet formation and accompanying gravity flow of Equid in a counter direction to
the gas flow create a unique condensation-scrubbing action that utilizes intimate
dropwise and tube surface, gas-to-liquid contact, making possible substantial collection
of fine particulate matter. Capture is enhanced by impaction on droplet and tube

97-12


-------
surfaces with benefit of condensation caused by gas temperature reduction in the
second stage heat exchanger. After droplet formation on sub-micron particulate,
resultant matter coalesces and agglomerates to large droplets that are removed by
impact and through separation in the mist eliminator.

The principal means of increasing design condensation-flux rate is by upward
adjustment of the gas wet bulb temperature at the exit of the first-stage exchanger,
thereby providing increased absolute humidity in the water-saturated gas stream
entering the second stage exchanger.

Con Edison Late-1993 Test Program at Ravenswood Station
Test Objectives

The field work to be done at Ravenswood will specifically assess the degree to which
Con Edison's purposes ami objectives in implementing this technical development have
been fulfilled.

•	Verify that addition of this treatment system to utility-scale oil fired boilers will
significantly improve the economics of steam and electric power generation

•	Characterize the performance of Integrated Flue Gas Treatment equipment of
Condensing Heat Exchanger Corporation in improving plant energy efficiency
and in minimizing air quality impact of the boiler

•	Devise and demonstrate optimal design and operation of the process in cost-
effectively increasing energy recovery, and maximizing flue gas cleaning

•	Establish gross and net pollutant emission levels and determine the fate of
pollutants collected

•	Evaluate boiler fueling alternatives that include replacement of high-priced,
imported, 0.3% sulfur residual fuel oil, commonly fired by Con Edison, with
reduced cost, higher-sulfur, heavy oil fuelstock available from within the
Western Hemisphere

•	Collect data for use in utility and other commercial-scale system designs

•	Report the results of the testing and demonstration program to the electric
utility industry.

Test Program Overview

Con Edison is continuing to closely monitor and scrutinize the results of ongoing field
operations and testing of the IFGT system installed at Ravenswood. The pilot plant
scale demonstration unit will be fully evaluated for heat recovery and stack gas
cleaning functions in oil fired service. It will assess thermal heat recovery
characteristics, including parasitic energy and consumables use, as well as
environmental management aspects, including stack gas cleaning and management of

97-13


-------
collected wastes. Detailed test evaluations will be made utilizing EPRI PISCES air
toxics program test protocol. This includes commercially-representative system
operation reflecting performance in a high-sulfur oil firing mode with the objective of
simultaneous removal of particulates including sulfuric acid mist, trace metals and
condensables; as well as applicable acid gases: S02, N02» N0X) and HC1.

The field work will be conducted so as to optimize the results and benefits of the
IFGT technology for use in the electric utility industry.

Test results will be evaluated to ascertain:

•	Heat recovery benefits and system efficiency improvements as measured by the
extent of net overall plant heat rate reduction

•	Extent of achievable reduction in the emissions of air toxics, S02,N0X and
particulates

•	Ability of the equipment to suppress both sulfuric acid mist emission and stack
opacity

•	Potential for use of existing stacks for flue gas discharge

•	Degree of avoidance of a visible steam-plume in discharge of the water-
saturated exit gas.

Differentiation Between High and Low Sulfur Fuel Service

The test work will pursue two alternative IFGT process/design strategies preliminarily
defined as follows:

•	The high-sulfur mode, e.g. in firing of 2.5% sulfur heavy fuel oil or other
high-sulfur, low-grade residuum, with the purpose of:

—	Greatest fuel cost savings through use of IFGT

—	Utilizing high-sulfur flue-gas chemistry, i.e. moderate flue-gas (yS02
concentration ratio, permitting practical use of either of two "clear liquor"
wet scrubbing means (employing oxidation inhibition) for S02 removal: (a)
Thioclear, magnesia-buffered lime scrubbing process (Dravo Lime Co.)
yielding usable gypsum, or (b) sodium alkali scrubbing with conversion of
the S02-catch to usable byproduct, (Aquatech or Wellman Lord Process,
etc.)

—	Exploiting of high flue-gas SOj/NOx concentration ratio, permitting
simultaneous NOx removal (using augmental NO to N02 gas conversion).

•	The low-sulfur mode, e.g. in firing of 0.3% sulfur residual fuel oil:

—	Significant S02 removal possible in a zero-alkali-reagent type of scrubbing
(using recirculated scrubbing liquor, if justified) by virtue of trace metal
catalysis, accompanied by effective S03 gas removal in a low-pH, water-
condensing, process regime, forming dilute sulfuric acid waste

97-14


-------
—	Favorable trace metal removal at low (e.g. 1.0) scrubber pH

—	lime treatment of the excess-condensate purge stream (for outfall discharge)
generating suspended gypsum particles that enhance flocculation in
wastewater treatment operation for reduction of outfall trace metal
concentrations to permissible levels

—	NOx removal achieved by furnace injection of ammonia (using urea reagent)
with excess ammonia (ammonia slip) emission curbed by furnace exit
injection of methanol, or efficiently absorbed in the low-pH IFGT scrubbing
step

—	Minimal byproduct volume: A low-volume, toxic, tract-metal/gypsum
sludge, (directly comparable to that formed in treatment of excess-
condensate generated in the high-sulfur mode), is the only waste product.

Application to Coal-Fired Boilers

IFGT technology can be applied in coal-fired boiler service with or without provision
for recirculating an alkaline scrubbing medium in the second-stage heat exchanger.
Scrubbing liquor recirculation affords flue gas desulfurization (FGD) and deNO, (by
N02 absorption where applicable). The IFGT system is positioned downstream of a
high-efficiency electrostatic precipitator as in conventional wet scrubbing practice.

Overall System Integration

The most technically and economically attractive means of applying/adapting this
scrubbing/heat recovery equipment for use at coal-fired emission sources is by
combining FGD performance with its unique trace/heavy metal removal capability,
(particularly with regard to improved collection of mercury). Current or improved
means/technologies for control of diverse, targeted, acid-gas pollutants including S02
and NGX, can be designed into the second-stage heat exchanger operation. Note that
in the case of volatile particulate emissions such as mercury and selenium, the
pronounced reduction of flue gas temperature in the condensing exchanger step results
in substantial condensation of water contributing significantly to the anticipated high
efficiency of removal of these trace metals.

Plant Heat Rate Improvement

Successful achievement of energy cost savings objectives in condensing heat exchanger
use is tied to:

Nature of Available Heat Sinks.

Hie type of heat sink, the amount of energy that can be transferred to the sink, and
the variations of sink capacity with time are important elements in establishing both

97-15


-------
the technical and economical Justification for condensing heat exchanger use.
Examples of possible heat sinks include:

—	Makeup water, which offers the best choice for heat exchanger effectiveness,
with a combination of the lowest incoming temperature and the highest teat
capacity available

—	Condenser hot-well condensate that can be preheated upstream of the first LP
(low pressure) feedwater heater: This will decrease the extraction steam flow
to the LP feedwater heaters and increase the LP tuibine-generator output.

There will be an increase in heat loss in the condenser. However, the
recovered energy from the flue gas exceeds this and there is a net energy gain
in the cycle.

—	Combustion air, which of these three offers the poorest heat-transfer sink: In
general, utility-size boilers are equipped with air preheaters, and further
preheating is generally not as attractive as preheating boEer condensate or
makeup water. The disadvantages of gas-to-air condensing heat exchangers
include smaller heat-transfer coefficients resulting in less-effective heat transfer;
larger equipment to accommodate both the heat-transfer surface requirements
and the large volume of combustion air; larger and often lengthy duct work; or,
alternatively, paired heat exchangers, which lower ducting costs but decrease
heat-transfer effectiveness and raise equipment cost. Such combustion air
preheat increases the temperature of air entering the air preheater, increasing
the plate temperature and decreasing air preheater duty and its recovery of
energy.

Heat Availability.

The flue-gas flow rate, temperature and composition define the energy available for
recovery and flow, along with heat sink conditions, are the primary variables
influencing the condensing heat exchanger design.

Value of Recovered Heat.

The value of energy-use being avoided by heat recovery determines the value of the
heat recovery. The steam characteristics of the steam, use of which is being reduced
by a condensing heat exchanger installation, has a direct bearing on fuel savings. For
example, if 200 psig steam that could be "sent to the street" has been used to heat
makeup water in the deaerator, reduction in use of that steam is more valuable than if
water heating had hypothetically been by 5 psig exhaust steam.

Medium/High Sulfur Service
Adaptability for Use as FCD Means.

97-16


-------
The condensing heat exchanger facilities can be employed in medium/high sulfur,
bituminous-coal applications requiring enhanced collection of volatile and non-volatile
particulate together with simultaneous, high-efficiency S02 removal. SOj emission
control is achieved by augmental use of principal existing/emerging flue gas
desulfurization (FGD) process means incorporated into the second stage heat
exchanger system. Reduction of stack gas wet bulb temperature by unique indirect
gas cooling advantageously reduces the gas flow volume for FGD design process.

Use of Throwaway Waste Design.

Because of the substantial presence of coal-based toxic substances in the pollutant
catch of the heat exchanger system, FGD design, as ultimately influenced by CAA
Title HI, can be expected to have greatest overall environmental control effectiveness
if it is predicted on throwaway-waste system operation that assimilates all collected
wastes, solid and liquid, in an all-in-one solid waste used to construct a structural
landfill. This, of course, calls for calcium alkali, i.e. lime/limestone, use so as to
yield a comparatively insoluble chemical reaction product, calcium sulfite and/or
calcium sulfate. This is desirable because it can be landfilled as a solid waste as per
common electric utility practice in the U.S. Because of the very substantial internal
equipment surface-area present within the second-stage exchanger in which a major
portion of S02 mass transfer is accomplished, calcium sulfate (gypsum) surface-scale
control/prevention is highly critical. This greatly favors use of magnesia-buffered
lime (calcium oxide containing 2 to 4% active magnesium oxide) reagent with
elemental sulfur oxidation-inhibitor additive to ensure calcium-sulfate unsaturated
mode operation. (Common wet limestone scrubbing provides a calcium-sulfate
supersaturated mode of operation and is best applied through use of open spray tower
scrubbers devoid of internals that may accumulate gypsum scale without
foreshortening boiler operating campaigns). By this means, high, e.g. 98%, S02
removal efficiency can be foreseen with production of a predominantly calcium-sulfite,
fine-grained precipitate that will dewater by thickening and filtration to cake solids
containing a high proportion (approximately 50%) of surface moisture.

Low Sulfur Service

Focus on Air Toxics Removal.

As in the case of low-sulfur residual oil firing, Con Edison's present liquid fuel mode,
the IFGT system lends itself to meeting new Title III objectives in control of air toxics
emissions in low sulfur coal fired applications. This system would be installed (along
with an upstream-ESP), downstream of the air preheater with or without use of FGD
provisions. In either case, hydrogen chloride, also included in the new Title HI
listings as a toxic substance, is removal at very high efficiency.

97-17


-------
IFCT Operation Without Alkaline Reagent Feed.

Absent use of alkaline reagent supply, such as when formalized FGD operation is not
mandated (including some medium/high sulfur applications), substantial S02 removal
is nonetheless foreseen in IFGT operation. Particularly in an augmented oxidation,
wet-scrubbing mode with recirculation of condensate/liquor, trace-metal catalysis,
including oxidation of flue-gas S02 to S03 and uniquely high S03 removal efficiency,
even at low pH, by tihe water-condensing regime fixes a significant portion of the flue-
gas-S02 as dilute sulfuric acid byproduct-waste. Moreover, low-pH scrubbing
operation is expected to provide superior trace metal collection from flue gas (than
IFGT system operation in the alternative intermediate-pH mode, as in nominal
medium/high sulfur service.)

Management of Collected Wastes.

Lime treatment of the excess-condensate purge stream, (for outfall discharge),
generating waste gypsum particles form die waste add, will enhance fiocculation in
wastewater treatment operation designed to reduce outfall trace-metal concentrations to
permissible levels. But management of major volumes of sulfurous waste/byproduct
volumes is avoided. A low-volume toxic, traee-metal/gypsum sludge, (directly
comparable to that formed in the treatment of wastewater discharge form a
medium/high-sulfur operating mode,) is the only solid waste product.

References

1.	1. Heaphy, J. Caibonara, A. Kressner, J. F. Carrigan, and W. Ellison,

"Integrated Flue Gas Treatment System for Simultaneous Emission Control and
Heat Rate Improvement - Demonstration Project at Ravenswood", presented at the
"International Symposium on Improved Technology for Fossil Power Plants - New
and Retrofit Application", Washington, DC (March 1993).

2.	"Utility seeks to integrate heat recovery flue-gas treatment". POWER, May 1993,
pp. 65,66, and 68.

3.	"Boilers, Combustion Systems, and Their Auxiliaries". POWER, Special Report,
June 1992, p. 82.

4.	S. M. Cho, D. Dietz, M. Kandis, J. Carbonara, J. Heaphy, A. Kressner and J.
Carrigan. "Heat Rate Improvement with Condensing Heat Exchangers", presented
at the Fifth International Power Generation Exhibition and Conference, Power-
Gen '92, Orlando, FL, (November 1992), Figure 6.

5.	"Developments to Watch." POWER, June 1992, pp 103 -104.

6.	R. S. Scorer. Air Pollution. Oxford. U.K.: Pergamon Press Ltd., 1968, pp
86-106.

97-18


-------
RA /% t	I •	f IPH/^Nf11 A	|

gure 1. Schematic of IFGT System


-------
A


-------
10-MW DEMONSTRATION OF THE ADVACATE
FLUE GAS DESULFURIZATION PROCESS

L. R. Lepovitz
C, A. Brown
Radian Corporation
8501 North MoPac Blvd.
Austin, TX 78759

T. E. Pearson
J. F. Boyer
ABB Environmental Systems

1400 Centerpoint Blvd.
Knoxville, TN 37932-1966

T. A. Burnett
V. M. Norwood
E. J. Puschaver
Tennessee Valley Authority
17A Chestnut Street Tower
1101 Market Street
Chattanooga, TN 37402

C. B. Sedman
U.S. Environmental Protection Agency
Air and Energy Engineering Research Laboratory
Research Triangle Park, NC 27711

B. Toole-O'Neil
Electric Power Research Institute
3412 Hillview Avenue
Palo Alto, CA 94304

98-1


-------
This paper has been reviewed in accordance with the U.S. Environmental
Protection Agency's peer and administrative review policies and approved for
presentation and publication.

Abstract

A 10-MW ADVAnced siliCATE (ADVACATE) pilot plant was built and operated in
1992 under a cooperative agreement between the Tennessee Valley Authority (TVA)
and ABB Environmental Systems (ABBES), and supported by the U.S. Environmental
Protection Agency (EPA) and the Electric Power Research Institute (EPRI). The
ADVACATE pilot plant was incorporated into TVA's National Center for Emissions
Research (NCER) located at the Shawnee Fossil Plant in Paducah, KY. Flue gas was
diverted from the existing pilot spray dryer inlet into a new duct and then returned to
the inlet of the existing pilot electrostatic precipitator (ESP). The ADVACATE process
blended an enhanced sorbent with reaction products and additional water and fed this
solids mixture into the flue gas where cooling of the flue gas, reaction with sulfur dioxide
(SQz), pneumatic conveying, and flash drying of the solids occurred.

The ADVACATE process was studied at stoichiometric ratios of 1.0 to 1.6 and at an
approach-to-adiabatic-saturation temperature of 20° Pn while treating flue gas from the
combustion of a 3.0 % sulfur coal. The sorbent preparation results, SC^ removal
efficiencies, and particulate control performance for the process are presented and
discussed.

Introduction

ADVACATE, a sorbent for removing SQ, from flue gases, was first developed and
patented by the University of Texas, Acurex, and EPA in 1986 *, In 1987 ABBES
developed an *in-duct" flue gas desulfurization process followed by a pulse-jet baghouse
called Moist Dust Injection (MDI) at its 0.3 MW pilot plant located at the University of
Tennessee. The results from a parametric test program conducted in 1988 and 1989 led
to the conclusion that an enhancement to the sorbent was needed to improve calcium
utilization. ABBES tested the ADVACATE sorbent during the parametric test program
and found a substantial performance enhancement over the use of lime2. This finding
was consistent with the trends in performance reported by EPA and the University of
Texas at laboratory scale. Consequently, ABBES obtained an exclusive license for
ADVACATE, Subsequent ADVACATE patents3,4 were issued in 1991 and 1992 and
encompassed slurry preparation, humidification, and gas/solid contact including the MDI
process. Therefore, the remaining text will discuss ADVACATE as a process as well as
a sorbent

This paper focuses on the first pilot-scale, continuous process testing of the ADVACATE
process in an MDI configuration with the solids mixer located on the flue gas duct at the
NCER from May through September 1992. The NCER site was ideally suited for the
ADVACATE and MDI process research needs because the site was of sufficient scale-up
size; utilized an ESP; was currently conducting flue gas desulfurization (FGD) research;

(I) Conversion factors to convert to metric units are included at the end of this paper.

98-2


-------
and had all of the necessary analytical, numerical, and control capabilities. Previous
spray dryer and ESP research performance data provided an excellent data base for
comparison. The site was folly staffed to support continuous ADVACATE operations
and testing.

The ADVACATE process performance results for SO? removal and particulate control
performance with an ESP are presented for the 5 months of actual testing. The results
from ADVACATE slurry preparation tests are also presented.

ADVACATE Process Description

The ADVACATE process employs the reaction of lime and fly ash in water at elevated
temperatures to produce calcium silicate compounds. These compounds show increased
reactivity toward SOj and acid gas species over that of lime alone. ADVACATE solids
are dried by blending the ADVACATE slurry in a pug-mill-type mixer with additional
dry reaction products to create a mixture which is free-flowing with relatively high
moisture content In the pilot plant configuration, the mixer feeds this blended material
into the flue gas.

The flue gas conveys the moist solids material through the ductwork to the particulate
collection device. While the solids are being conveyed, flash drying occurs and the flue
gas temperature is decreased. The flue gas SOj is simultaneously absorbed by the
reaction with the moist ADVACATE sorbent

The reaction products and fresh fly ash are collected from the flue gas in a particulate
control device and stored in a silo. A fraction of this material serves as the fly ash
(silica) supply to prepare ADVACATE slurry. The major fraction of the stored solids is
blended with the ADVACATE slurry to form a free-flowing, high-moisture-carrymg
material. The remaining fraction of the reaction products is discharged as waste.

Pilot Plant Description

The 10-MW ADVACATE pilot plant is located at TVA's NCER near Paducah, KY, and
is adjacent to TVA's Shawnee Fossil Plant. A slipstream from the Shawnee Fossil
Plant's Unit 9 boiler provides the flue gas for testing at the NCER. Unit 9 is a front-
fired Babcock & Wilcox boiler with a nameplate rating of 175 MW. Unit 9 fired a
medium sulfur (4.0 to 5.0 lb SO^/MMBtu), low chloride (0.03%), eastern bituminous
coal during the ADVACATE test program. The coal was supplied by the Peabody Coal
Company from their Martwick mine in western Kentucky. Typical coal and coal ash
compositions are shown in Table 1.

The flue gas slipstream from Unit 9 is withdrawn downstream of the boiler air heater
and multidone system. The multiclones on Unit 9 remove approximately 70% of the fly
ash from the flue gas exiting the boiler such that the flue gas particulate concentration is

98-3


-------
approximately 03 grains per actual cubic foot (gr/acf). During most of the
AD VACATE tests, a fly ash injection system was used to compensate for the fly ash
removed in the multiclones. Typically, the pilot plant inlet particulate concentration
after the fly ash injection was approximately of 2.0 gr/acf.

The flue gas slipstream from Unit 9 pases through a 40-inch-diameter duct to the pilot
plant The slipstream is routed through a venturi located at the pilot plant inlet to
measure the flue gas flow rate. Downstream of the venturi is a preheater/precooler heat
exchanger, shown in the Figure 1 process flow diagram, which is used to control the flue
gas temperature. The flue gas then bypasses the existing spray dryer to the ESP via the
duct constructed for AD VAC ATE evaluation.

A mixer, located on top of the duct, blends ADVACATE slurry, additional dry reaction
products, and water before discharging the mixture into the duct The flue gas residence
time between the AD VACATE sorbent injection location and the inlet to the ESP is
approximately 22 seconds. This residence time is site specific and may exceed that
needed in a full-scale application.

The NCER ESP is designed with four electrical sections and eight gas flow passages. An
empty fifth field is not equipped with a hopper or ash collection system. The collecting
plates are 23 feet high by 92 feet long and are spaced 10 inches apart. This resulted in
an ESP specific collection area (SCA) of approximately 440 square feet per thousand
actual cubic feet of flue gas per minute (tf /kacfm), with all four fields energized at
humidified gas conditions during ADVACATE testing.

The corona electrodes are stainless steel spiral wires mounted in a rigid-frame
configuration. Both the collecting and discharge electrodes are rapped by tumbling
hammers mounted on a rotating shaft The ESP power supplies are rated at 55 kV and
200 mA in each field.

The pilot system is equipped with flue gas analyzers to monitor SOj and oxygen (00
concentrations at the pilot unit inlet, ESP inlet and ESP outlet. Flue gas static pressures
and temperatures are also measured at these locations. The ESP outlet flue gas is
monitored for opacity. The inlet flue gas wet bulb temperature is measured manually
using a wetted wick technique approximately once every 2 hours during testing. The
measured wet bulb temperature plus the desired approach-to-adiabatic-saturation
temperature are used to calculate the set point for the ESP inlet diy bulb temperature.

The solids collected in the ESP are transported via screw conveyors to a bucket elevator.
The bucket elevator feeds a solids storage silo. The silo is equipped with a fluidized
dual-pant-leg cone bottom. One of the lep feeds the dry reaction product material
through a rotaiy valve to a series of screw conveyors that transport the solids to the pug-
mill-type mixer. The other leg feeds the solids to either a reaction product slurry mix
tank or a waste slurry mix tank.

98-4


-------
The ADVACATE sorbent is prepared by mixing quicklime from a 100-ton capacity silo
with a reaction product shiny in the ADVACATE mix tank. This slurry is pumped to
the bottom of an attritor mill. The slurry overflows from the attritor to the
ADVACATE feed tank. Hie slurry then either overflows back into the ADVACATE
mix tank, completing a recirculation loop, or is pumped to the pug-mill-type mixer. A
progressive cavity-type pump is used to feed nozzles located on top of the mixer where
the slurry is sprayed and mixed with dry reaction product material and trim water prior
to feeding ADVACATE sorbent directly into the duct.

The ADVACATE feed and mix tanks are covered, insulated tanks with steam heat
exchangers located in the bottom to control the slurry temperature. Both tanks are
equipped with dual-blade agitators to keep the slurry well mixed. Another ADVACATE
feed tank was added later in the test program to provide additional slurry residence time.
Although this tank was insulated, it did not have a supplemental steam heating system
for slurry temperature control. The agitator in this tank was marginal in maintaining a
well-mixed slurry.

Results

The following sections present the SQ, removal results and the ESP particulate control
performance during the ADVACATE test program. Also included is a section
describing the results from ADVACATE sorbent preparation tests via batch and semi-
batch techniques at different slurry residence times.

S02 Removal Performance

The SOj removal results and operating conditions for all of the ADVACATE tests
conducted during the test program are presented in Table 2. Each data point represents
the average value for the duration of a test at specific operating conditions, which
includes 4 to 8 hours of test data following 12 to 36 hours of operation to achieve steady-
state conditions. Test data were collected in 15-minute average increments. The range
of 15-minute average SO^ removal data for each test is provided in Table 2.

The major operating variables were stoichiometric ratio (moles of Ca(OH)2 in the fresh
lime feed per mole of SOj in the inlet flue gas), ADVACATE slurry residence time,
ADVACATE slurry temperature, mixer solids moisture, and fly ash injection rate. All
tests were conducted with an inlet flue gas temperature of 320° F, an ESP inlet flue gas
approach-to-adiabatic-saturation temperature of 20° F, and a duct residence time of 2.1
to 23 seconds. Since only nine ADVACATE tests were completed during the test
program, statistical modelling of the data was not attempted. Therefore, these results
are compared and discussed in general terms as a function of the major operating
variables.

98-5


-------
The total system SOjj removal performance is a combination of the SO, removals across
the duct and the ESP. The data plotted in Figure 2 present the contributions from the
duct and from the ESP to total SOj removal performance. The ESP SC^ removal is
based oil the system Met SO? concentration, so the duct SO, removal and the ESP SO>
removal can be added together to obtain the total system SO* removal. In Figure 2, the
duct SOj removal ranges from 49% to 84%, while the ESP contribution to SOfe removal
is lower and ranges from 3% to 19%. Generally, the higher ESP removals correspond
with the lower duct removals. As removal in the duct increases, there is less SOj
available for removal in the ESP.

Stoichiometric Ratio. The total system SOjj removal performance results, as a function
of lime stoichiometric ratio, are presented in Figure 3. The average SO^ removal ranged
from 66% to 89% at stoichiometric ratios of 0.99 and 1.47, respectively.

ADVACATE Slurry Residence Time. Most test runs were made with an ADVACATE
slurry residence time of 2.8 to 3.5 hours. Two tests (l-AS-12 and l-AS-13) were
completed at higher residence times of 10 to 13 hours. Test l-AS-13 produced nearly
the same SQ> removal (80%) as test l-AS-3 (82% at the same stoichiometric ratio of
L25) and test 1-AS-l (80% at a stoichiometric ratio of 136). The comparison of test
l-AS-13 and l-AS-3 is confounded by a difference in solids moisture content, which
results from differing amounts of reaction product blending. Test l-AS-12 was run at a
stoichiometric ratio of 0.99, and there are no comparable tests at a low residence time at
this stoichiometric ratio. When plotted in Figure 3, the results show nearly the same or
lower SO* removal results for the two tests at long slurry residence times as other tests.
The Bnmauer, Emmett, and Teller (BET) surface areas measured for the solids in the
ADVACATE slurry feed at the longer residence times ranged from 13 to 19 meters
squared per gram (nf/g), whereas the slurry solids surface area ranged from 22 to 40
nf/g for the 2.8- to 35-hour residence time tests.

ADVACATE Slurry Temperature. Several tests were performed at different average
ADVACATE slurry temperatures with all other test conditions at the same level (20° F
approach-to-adiabatic-saturation temperature, 1.45 stoichiometric ratio, and 3-hour slurry
residence time). Hie results plotted in Figure 4 show that lower SQ> removal
performance is achieved at lower slurry temperatures. At an average slurry temperature
of 191° F, the average SOj removal performance was 89%. There was a significant
decrease in performance to 73% SOj removal at a slurry temperature of 169° F.

In the laboratory, elevated temperatures are required to promote the reactions that form
calcium silicate compoundsI'3*4. Therefore, the observed increase in SO2 removal
performance with increasing slurry temperature was expected.

Mixed Solids Moisture Content. The moisture content of the blended ADVACATE
solids from the mixer is controlled by the rate of reaction product solids addition to the
mixer. Increasing the solids moisture content might be expected to enhance SOj removal
performance, although the effect may be offset by decreasing the rate of solids injection
into the duct. Due to the limited number of tests, a definitive conclusion cannot be

98-6


-------
reached regarding the net effect Comparison of test l-AS-3 at a solids moisture content
of 7.0% and test l-AS-13 at a moisture content of 22.6% produced similar SOj removal
performance (82% and 79% total SGj removal, respectively).

Fiy Ash Injection. Fly ash is the source of silica in the reaction to form calcium silicate
compounds. Fly ash injection was used during most of the ADVACATE test runs,
because the source of the flue gas slipstream was downstream of multiclones that
removed about 70% of the fly ash from the flue gas. Initially, there was a concern that
low amounts of fly ash would limit silica dissolution and the production of calcium
silicate compounds, resulting in decreased SO^ removal performance. Therefore, during
most tests, fly ash was transferred from the multiclones to the pilot plant and
pneumatically injected into the flue gas at a rate that increased the fly ash loading to
approximately 2.0 gr/acf.

To determine if fly ash injection affected SO? removal performance, fly ash injection was
not used during two of the ADVACATE test runs. The test results plotted in Figure 3
show that fly ash injection did not have a significant effect on SOj removal performance.
More test data would aid in confirming this effect and develop an accurate explanation.

Comparison with Spray Dryer/ESP Pilot Plant. Prior to testing the ADVACATE
process at the NCER, a spray dryer/ESP test program was completed using much of the
same pilot plant equipment5. This provided a unique opportunity to compare the
performance of two processes using data from the same pilot plant while Unit 9 was
firing the same coal. The processes, however, are in different stages of commercial
development The spray dryer process is commercially installed on low- and medium-
sulfur coal FGD applications and has been tested on high-sulfur coals in several pilot
programs, while the ADVACATE process as tested at the NCER was the first pilot-scale
demonstration of this new FGD technology.

Results from the previous spray dryer tests and the ADVACATE tests are plotted in
Figure 5. Spray dryer test results at approach-to-adiabatic-saturation temperatures of
18° F and 23° F are plotted to bracket the ADVACATE results, which were obtained at
an approach temperature of 20° F. Comparison of the results shows nearly identical SQ
removal performance for the two processes at stoichiometric ratios above 1.2. The
results at a stoichiometric ratio of 1.0 show a greater difference, which could be due to a
number of factors which cannot be explained due to the limited amount of data.

ADVACA TE Sorbent Preparation

The ADVACATE sorbent preparation process evaluated at the NCER site was designed
to expand on prior laboratory and bench-scale research at the EPA's laboratories in
Research Triangle Park and at the University of Texas in Austin. ADVACATE sorbent
is prepared by reacting lime slurry and fly ash, which is collected with reaction products
in the particulate collection device, at elevated temperatures to produce highly reactive
calcium silicate compounds. During previous research efforts, the ADVACATE sorbent

98-7


-------
was prepared in a batch mode with ratios of fly-ash-to-lime of 3:1 without the presence
of reaction products. These laboratory tests showed that, under certain preparation
conditions, the slurry solids surface area could exceed 60 nf/g and the slurry and
reaction product mixture could remain relatively free flowing with a total moisture level
exceeding 30%6.

One of the main objectives of the ADVACATE test program at the NCER site was to
produce the ADVACATE sorbent on a continuous basis. ADVACATE was prepared
for the first time in the presence of reaction products and grit from the quicklime. The
fly-ash-to-lime ratios in the ADVACATE slurry mix tank were 0.18:1 without fly ash
injection and 0.58:1 at the highest fly ash injection rate.

During the test program, ADVACATE slurry solids surface area ranged from 13 to 40
nf /g. These values are encouraging since they indicate that, even though there were
significant differences between the pilot plant operations and laboratory experiments, the
trend in producing a higher surface area sorbent remained.

The BFT surface area analyses of ADVACATE sorbent solids early in the test program
produced results that, based on material produced in batches at the laboratory scale,
seemed to be low. There was concern that the 3-hour residence time in the
ADVACATE sorbent preparation system was inadequate to produce high-surface-area
solids. Therefore, separate tests were conducted to evaluate surface area as a function
of slurry residence time. The following discussions present the experiments performed
during the ADVACATE test program that were specifically designed to investigate the
ADVACATE sorbent preparation capabilities.

"Semi-Batch" ADVACATE Slurry Residence Time Tests. Two approaches were used
during the ADVACATE slurry residence time testing. In the "semi-batch" test mode,
normal operation of the ADVACATE process was discontinued by stopping the slurry
feed to the mixer along with the quicklime and reaction product slurry feeds to the
ADVACATE mix tank. The ADVACATE slurry at the beginning of the test was
prepared in the normal manner using two tanks in series with an average combined
residence time of approximately 3 hours. Three "semi-batch" tests were ran. Tempera-
ture control problems during the first two tests may have confounded the results; there-
fore, the results from only one of the "semi-batch" tests are presented.

The "semi-batch" test was started while operating at the test conditions for ADVACATE
Test l-AS-6. In this test, slurry was allowed to recirculate through the attritor, feed tank,
and mix tank throughout the entire test. A slurry sample was taken from the mix tank
each hour for 12 hours. Water was added to the mix tank several times during the test
to keep the solids concentration below 30%. Although the addition of water resulted in
a temporary reduction in slurry temperature, the temperature ranged between 188° F and
213° F and averaged 202° F during this test

BET surface area measurements for the slurry samples are plotted in Figure 6. The first
data point plotted was at 1.5 hours, because the average slurry residence time in the mix

98-8


-------
tank when the ADVACATE slurry, lime, and reaction product slurry feed streams were
turned off was approximately 1.5 hours. The first ADVACATE slurry sample was taken
as soon as the feed streams were turned off. Measured surface area increased steadily
from 28 to 65 nf /g between 3 and 12 hours. No further increase in surface area was
observed at 15 hours of hydration time. Therefore, it was concluded that approximately
12 hours of residence time above 190° F was necessary to produce greater than 60 nf /g
surface-area ADVACATE slurry solids.

"Batch" ADVACATE Slurry Residence Time Tests. In the "batch" mode for
ADVACATE slurry residence time testing, the tanks were empty at the beginning of the
test. lime was slaked in water, then reaction product slurry was added to start the
ADVACATE reactions. Although three "batch" tests were ran, temperature control
problems during the first two tests similar to those in the "semi-batch" tests may have
confounded the results. Therefore, results from the third "batch" test are presented.

Quicklime was added to water in the ADVACATE mix tank to produce a fresh lime
sluny containing approximately 25% solids, then allowed to slake for 1 hour. Reaction
product slurry containing 35% solids from the reaction product mix tank was heated to
195° F in the ADVACATE feed tank, then mixed with the lime slurry at a reaction
product-to-lime mass ratio of 15:1 to begin the test. The resulting slurry was circulated
through the ADVACATE sorbent preparation system for four hours while hourly slurry
samples were taken. After stopping the recirculation, hourly slurry samples continued to
be taken for 12 hours. The slurry temperature averaged 204° F and ranged between
194° F and 213° F during the test.

BET surface area measurements for the "batch" test slurry samples are also plotted in
Figure 6. For this test, the beginning of the hydration time period corresponds to the
time when reaction product slurry was added to the lime slurry. The surface area
increased steadily from 26 nf /g at 2 hours to 62 nf/g at 10 hours, with a negligible
further increase to 63 nr /g after 13 hours.

Based on surface area results from the two ADVACATE slurry residence time tests
shown in Figure 6, a hydration or residence time of approximately 9 to 10 hours was
required to obtain ADVACATE slurry solids surface areas of greater than 60 nf/g.
During operation of the pilot plant, the BET surface areas were actually lower with the
ADVACATE slurry residence time at 10-13 hours than at 3 hours. This phenomenon is
not fully understood.

ESP Performance

At many existing power plants, the typical fly ash ESP inlet particulate concentration is
about 2.0 gr/acf. The requirements to retrofit these plants with a FGD process will be
site specific. The particulate emissions following an ADVACATE process will be
required to: 1) meet the emission standard for the existing facility (typically 0.03 to 0.1
lb/MMBtu); 2) meet the New Source Performance Standard (NSPS) for particulate

98-9


-------
emissions (0.03 lb/MMBtu); or 3) not increase particulate emissions above the existing
fly-ash-only emission rate.

A secondary objective of the test program was to measure the effect an ADVACATE
process has on the particulate collection performance of an ESP. Very high particulate
collection efficiencies are necessary in a retrofit application using an ADVACATE
process because the ESP inlet particulate loading is typically 50 to 175 gr/acf, depending
on the inlet temperature and the design moisture content of the ADVACATE sorbent.
The higher particulate loadings of this range were used during most test runs because the
reaction product flow rate was high to reduce the moisture content of the mixed
ADVACATE sorbent

The humidifieation of the flue gas stream that occurs with the ADVACATE process
improves the ESP electrical characteristics. Since the spark over voltage, which limits
the maximum operating voltage of the ESP, is proportional to the gas density, the
decreased temperature of the flue gas can result in an increase in operating voltages.
Voltage/current characteristics for electrical fields 1 and 4 during baseline fly-ash-only
testing and ADVACATE testing are plotted in Figure 7. A comparison of the curves for
field 1 shows the effects of both increased mass loading and lower temperature. A
comparison for field 4 shows the effects of lower temperature only, since most of the
particles that would induce space charge effects are removed prior to entering field 4.
The curves presented in Figure 7 demonstrate that electrical operation was improved in
both fields 1 and 4.

The ESP particulate collection performance results for the ADVACATE tests are
summarized in Table 3. The results in Table 3 must be qualified, however, because no
attempts were made during this short test program to investigate methods to optimize
the ESP performance. At the end of the test program, two cracked insulators were
found, the collection plates were found to be misaligned, and there was significant
buildup of solids on the high voltage insulators. Although these problems could affect
the spark rate, limit the maximum achievable voltages, and reduce collection efficiency,
no reduction in the secondary power levels in each field was observed.

The average particulate emission rate for each day of testing is plotted in Figure 8. The
particulate emission rate for the ADVACATE tests ranged from 0.019 to 0.242
lb/MMBtu at SCAs of 412 to 475 ft2 /kacfm. The SCA is calculated at cooled and
humidified gas conditions, which lowers the gas volume to be treated and increases the
SCA by approximately 10% above typical untreated flue gas inlet conditions.

Figure 8 indicates a trend of increasing particulate emission rates with time during the
test program. During the first 2 days of testing, the NSPS limit of 0.03 lb/MMBtu for
particulate emissions was easily met. However, later in the program, particulate
emissions were significantly higher than the NSPS limit One data point, on September
3, 1992, was exceptionally high at 0.242 lb/MMBtu. This data point was investigated in
detail, but no reason could be found to explain why the result was so high or to indicate
that the data were not valid. On September 17, field 4 tripped off line but was re-

98-10


-------
energized. Upon start up for testing the next day, field 4 could not be energized, so
testing continued with only three fields.

The ESP performance results indicated that very high particulate collection efficiencies
(>99.95%) were obtained. There were some difficulties in obtaining representative
measurements of the ESP Inlet particulate concentration. The inlet sample probe often
plugged very quickly or could not be inserted into the lower traverse point on the
horizontal, rectangular duct. It appeared that the particulate concentration in the gas
stream at the bottom of the duct was higher than in the upper sections of the duct. Due
to the difficulty in sampling the bottom port, where the particulate concentration
appeared to be very high, the measured inlet mass loadings could be biased low. Thus,
the actual ESP collection efficiency may be higher than reported. The outlet emission
rate, however, is not affected by the measured particulate concentration at the ESP inlet.

Equipment Operation

A limited number of tests were conducted during the ADVACATE test program,
partially because of equipment and operational problems. Some of these problems were
due to using existing equipment that was incompatible with the ADVACATE process
design. However, some of the problems were specific to the ADVACATE process. The
main problems specific to the ADVACATE process were handling and conveying several
solid and liquid process streams. Corrective actions were taken to resolve most
equipment and operational problems to complete the test program.

Table 4 highlights the main problems encountered during ADVACATE testing that were
associated with the process and describes the on-site and future proposed corrective
actions. Pilot-plant-specific problems are not included in this table. The problems which
were not resolved during the test program are currently under study.

Conclusions

From the data generated during ADVACATE process testing at the NCER site, several
conclusions can be drawn:

•	The ADVACATE process SQ, removal performance achieved 89% at a
stoichiometric ratio of 1.47. SQ performance was similar to that of the NCER
pilot spray dryer/ESP under comparable conditions.

The addition of fly ash to the pilot unit flue gas did not appear to impact SO*
removal performance.

•	The continuously prepared ADVACATE slurry solids surface areas ranged from
13 to 40 nr/g.

98-11


-------
•	The ESP particulate removal performance with an inlet grain loading between 100
and 200 gr/acf was better than 99.9 %. Emission rates varied from 0.015 to 0.242
lb/MMBtu, but most were generally higher than the NSPS limit of 0.03
lb/MMBtu with an SCA of approximately 440 ff/kacfm. These results were
obtained without tuning the ESP for the high inlet mass loading. Effects of using
the ADVACATE process with an ESP are not fully known.

•	Hie effects of slurry residence time and temperature, solids moisture content, and
slurry solids surface area on SO, removal performance are not fully understood
and need further study.

Conversion Factors

1.0 lb/MMBtu = 430.5 Ng/J

1.0 tf/kacfm = 0.1968 irf/irf/s

1.0 gr/acf = 2373 g/irf

1.0 inch = 2.54 cm

1.0 ft = 0.3048 m

1.0 tf/min = 472 cnf/s

1.0° F = 0.55° C

°F = 1.8° C + 32

1.0 lb/min = 27.216 kg/hr

1.0 ton = 907 kg

References

1.	U. S. Patent No. 4,804,521. "Processes For Removing Sulfur From Sulfur
Containing Gases." February 14, 1989.

2.	B. W. Hall, C Singer, W. Jozewicz, C. B. Sedman, and M. A. Maxwell. "Current
Status of the ADVACATE Process for Flue Gas Desulfurization." Journal of the
Air and Waste Management Association. Vol. 42, No. 1, p. 103-110 (January
1992).

98-12


-------
3.	U. S. Patent No. 5,047,221. "Processes For Removing Sulfur From Sulfur-
Containing Gases." September 10, 1991.

4.	U. S. Patent No. 5,100,643. "Processes For Removing Sulfur From Sulfur-
Containing Gases." March 31, 1992.

5.	10-MW Spray Dryer/ESP Pilot Plant Test Program High-Sulfur Coal Test Phase
(Phase III) Find Report Tennessee Valley Authority, Ontario Hydro, Electric
Power Research Institute, and Kentucky Energy Cabinet Laboratory, July 1988.
TVA/OP/ED&T-88/35.

6.	C. B. Sedmau, M. A. Maxwell, W. Jozewicz, and J. C. S. Chang, "Commercial
Development of the ADVACATE Process for Flue Gas Desulfurization,"
presented at the 25th IECEC, Reno, NV (August 1990).

Table 1

Typical Peabody/Martwick Coal & Ash Composition

Coal Component % (dry basis*)	Ash Component	%

Carbon	7298

Hydrogen	4.92

Oxygen	7.652

Nitrogen	1-65

Sulfur	3.05

Chlorine	0.03

Ash	9.72

1	Typical moisture of 10.24%.

2	By difference.

SiOj

50.96

A1A

18.84

FeA

19.67

CaO

3.33

MgO

0.83

Na20

0.38

KjO

2.39

SO,

2.19

Titania

1.04

Other

0.37 2

98-13


-------
Table 2

S02 Removal Performance Test Results

Parameter

I-AS-1

1-AS-l

l-AS-2

I-AS-3

I-AS-4

l-AS-5

l-AS-12

l-AS-13

l-AS-20

Date

6/5/92

6/18/92

7/4-5/92

07/10-13/92

7/17/92

7/29/92

9/2-3/92

9/6/92

10/1-2/92

Inlet Gas Flow, scfm

20,100

20,200

20,000

19,200

20,100

20,500

21,600

20,700

21,300

Inlet Temperature, * F

322

320

320

320

320

320

319

320

319

Inlet SC% Concentration, ppm

1,833

1,767

1,587

1,669

1,741

1,714

1,711

1,718

1,677

Approach Temperature, * P

20.0

20.3

20.0

20.0

20.0

19.9

20.2

20.0

20.1

Ply Ash Injection, lb/min

0

0

9.3

9.2

9,6

9.9

10.2

9.9

10.2

Mixer Outlet Moisture, % (dried at
14(7 F)

5.7

10.4

8.3

7.0

9.5

7.6

8.9

22.6

10.7

ADVACATB Slurry Residence
Time, hrs

2.8

3.1

3.5

6.1

14

2,9

13.0

10.1

3.2

ADVACATB Slurry Temperature,
*F

183

192

166

191

131

191

203

198

188

Reaction Product Ratio,
lb Reaction Product/lb Ca(OH^

1.6

1.8

1.7

0.8

1.5

1.7

1.9

1.7

1.2

Stoichiometric Ratio, Ca/S

1.36

1.49

1.46

1.25

1.46

1.47

0.99

1.25

1.58

Total Reaction Product Ratio,
lb Reaction Product/lb Ca(OH^

76

72

77

82

70

74

60

26

40

Test Results



















Test Average SQ Removal, %
Duct
ESP

Total System

73

7
80

77
10
87

53
19
72

76
6

82

84
3

87

75
14
89

49
17
66

64
15
79

77
11

88

Range (15-min avg data pts)

76-84

84-90

66-81

75-86

83-89

86-92

59-71

76-81

84-93

Lime Utilization, %

59

59

50

66

59

61

67

63

56

Reaction Product Solids Ga Util,, %

53

59

52

58

59

59

66

61

56


-------
Table 3

ESP Particulate Control Performance for ADVACATE Tests

Test

ESP
Fields

SCA1
(ff/kacfm)

Inlet
Loading
(gr/acf)

Outlet
Loading
(gr/acf)

Collection
Efficiency

(%)

Emission
Rate2
(Ib/MMBtu)

1-AS-l, 6/5/92

4

419

152.6

0.0048

99.997

0.015

1-AS-l, 6/18/92

4

433

125.7

0.0076

99.994

0.023

l-AS-2

4

459

133.1

0.0216

99.984

0.075

l-AS-3

4

475

168.2

0.0160

99.990

0.054

l-AS-4

4

454

173.5

0.0104

99.994

0.033

l-AS-5

4

440

1293

0.0122

99.990

0.040

l-AS-12

4

419

150.8

0.0715

99.951

0.242

l-AS-13

4

414

585

0.0266

99.953

0.087

l-AS-19

3

312

963

0.0568

99.936

0.178

l-AS-20

3

310

125.1

0.0575

99.952

0.182

lSCA at cooled and humidified conditions. Before cooling and humidification, a four-field SCA would be
nominally 390 ft2 /kacfm, and a three-field SCA would be nominally 290 ff /kacfm.

2 Baffle installed in field 4 during all tests.

98-15


-------
Table 4

ADVACATE Pilot Plant Operational Problems and Solutions

Problem

Solution(s)

Solids dropout in the duct,
increasing duct pressure drop

Increase duct velocity

Add fluidization sootblowers or other devices
to promote flow

Ensure good mixing and reliable slurry flow
Add dropout hopper
Relocate mixer

Plugged slurry nozzles in
mixer

Add on-line maintenance capability and spare
nozzles1

Add monitoring capability

Non-uniform blending of
slurry and reaction product

Change mixer design1
Reliable slurry and solids flow

Erosion of expansion joints

Add shields inside duct1
Provide setback1

ESP emissions

Tune ESP operation for high inlet dust
concentration

Moist dust buildup on ESP
support insulators

Provide heated purge air to insulators

ADVACATE material
process control

Develop on-line technique to measure
ADVACATE material

1 Solution implemented during test program.

98-16


-------
Preeooler

100

as
>
o

£
m
cc
CM

O

CO

Figure 1. National Center for Emissions Research
10-MW ADVACATE Process Flow Diagram

1	12.	1.4	1.6

Fresh Lime Stoich. Ratio (mols Ca(OH)2/moi S02)

Duct S02 Removal ESP S02 Removal Total System S02 Removal

Notes: All tests conducted at 320F In, with
3% S/0.03% d coal and Mississippi lime, 20F

approach, siunytempentture >ia>F.

figure 2. Effect of Lime Stoichiometric Ratio on SO, Removal
Performance for the ADVACATE Process

98-17


-------
1	1.2	1.4	1.6

Fresh Lime Stoich. Ratio (mols Ca(OH)2/mol S02)

No Fly Ash Injection

~

3-6 hr Mix Tank Res. Time
•

Notes: An tests conducted at320F In, 20F
approach, 3% S/0.03% Q coal. Mississippi lime,
2.1 -2.3s duct l

Slurry Temp < 180F
10-13 hr Mix Tank Res. Time

figure 3. Effect of Major Variables on Overall
SO, Removal Performance

160	170	180	190	200

Average ADVACATE Mix Tank Slurry Temperature (F)

With Fly Ash injection

Notes All teste conducted at 20,000 sdm, 320F,
20F approach, 1,46-1.49 stoich, 1.5-1.8 reaction
product/lime rafo, 2,9^5 hr mix tank res time

Without Fly Ash Injection

Figure 4. Effect of Slurry Temperature on Overall
SQ; Removal Performance

98-18


-------
90

— 100
(0
>
o

E

0)

a:

CM

O 80
CO

E
©

(O
>.

03

© 60

."5

70

-





A

A

o

¦ 6 A

o

•#A

•

¦

I

t I



!

SD-23F Approach

o

1	1.2	1.4	1-6

Fresh Lime Stoichiometry (mols Ca(OH)2/moi S02)

ADVACATE - 3-6 hr Res. Time

SD - 18F Approach

A

ADVACATE -10-13 hr Res. Time ADVACATE - No Fly Ash Injection
¦ ~

Notes: A0 tests conducted at 320F in, with
394 S/0.03% Ct coal and Mississippi (me.

ADVACATE tasJs conducted at a 20F approach.

Figure 5. Comparison of Spray Dryer and ADVACATE
Pilot Plant SO, Removal Results

E

CO

6	8	10	12

Hydration Time (hours)

Residence Time Test #5 Residence Time Test #6
(Batch)	(Semi^Batch)

Notes: Tests conducted at a reaction product/
Erne ratio of approx. 15, and an average siurry
temperature between 200and 205F

Figure 6. Effect of Hydration Time on Slurry Solids Surface Area

98-19


-------


220



200



180

<



E 160

""W"







c

140

£



Jk-»

3

120

Q





100

CO



T3

80

c

o



Q

m

©



CO

40



20



0

Fly Ash, Field 1

	-E	

Fly Ash, Fiel

		

Field 4

ADVACATE, Field 1

—e—

ADVACATE Field 4
—*—

0	10	20	30

Secondary Voltage (kV)

Figure 7. Voltage/Current Characteristics for First and Fourth Electrical
Fields During Fly-Ash-Only Conditions and ADVACATE Testing

3

1

0.3

II 025

m
c
o

0.2

0.15

¦S 0.1

0.05

£
LU

©

3
O

¦c
a

CL
£L
£0
LU

-

-







¦



~

~

-



¦

%



¦







*

¦



NSPS Limit





i i

	! 	 1	

1 1

I 	 1 		

5/31/92 6/14/92 6/28/92 7/12/92 7/26/92 8/9/92 B/23/92 9/6/92 9/20/92 10/4/92

Date

4 ESP Fields 3 ESP Fields (Field 4 failed)

Notas: All tosts conduced with 3% SC.03% O
cool. WriN^lm, andtfwbaflSah ESP
fieU4ki place.

Figure 8. ESP Particulate Emissions During ADVACATE Test Program

98-20


-------
NOXSO S02/NOx Blue Gas Treatment Process:
Proof-of-Cancept Test

John L. Haslbeck
Mark C. Woods
Warren T. Ma
Scott M. Harkins
James B. Black

NOXSO Corporation

P.O. Box 469
Library, PA 15129

Project Funded By:

U.S. Department of Energy, Pittsburgh Energy Technology Center
Ohio Coal Development Office
NOXSO Corporation
W.R. Grace & Co.-Conn.

Morrison Knudsen Corporation - Ferguson Group

99-1


-------
Introduction

The NOXSO process is a dry, regenerable flue gas treatment system that
simultaneously removes 90% of the S02 and 70-90% of the NOx from flue gas generated
from the combustion of coal. Hie process was successfully tested at small scale (0.017 MW)
on high sulfur coal (2.5%) at the TVA Shawnee Steam Plant. The test results are contained
in two U.S. Department of Energy (DOE) reports.1*2

Tests of a NOXSO process Development Unit (PDU, 0.75 MW) were conducted at
the Pittsburgh Energy Technology Center (PETC) under a cooperative research agreement
between NOXSO and the DOE. Testing in the adsorber was done by continuously feeding
sorbent into a fluidized bed adsorber and collecting the spent sorbent from the adsorber
overflow. Regeneration took place in a separate batch reactor. The test results were
reported by Yeh et al in 1987,3 and by Haslbeek et al. in 1988.4

A Life-Cycle Test Unit (LCTU, 0.06 MW) was built at PETC in 1988 to test the
NOXSO process in an integrated, continuous-operation mode. The LCTU test program was
designed to determine long-term effects of the process on the sorbent reactivity and attrition
properties. The sorbent was successfully tested for over 2000 hours on flue gas. The test
results were published by Ma et al. in 1990,5 and by Yeh et al. in 1990.6

The Proof-of-Concept (POC) test is the last test prior to full-scale demonstration. The
POC test collected all of the information needed to design the full-scale NOXSO plant: e.g.,
data pertaining to materials of construction, process performance and cost, process safety,
process control, sorbent activity, sorbent attrition, heat recovery, etc. The POC plant (5
MW) is located at Ohio Edison's Toronto plant in Toronto, Ohio. Flue gas was first
introduced to the plant on November 23, 1991. The current test results and process
performance, along with a summary of process economics, are presented in this paper.

Pilot Plant Description

The NOXSO POC plant at Toronto, Ohio is shown schematically in Figure 1. A slip
stream of flue gas (equivalent to 5 MW coal-fired power) is extracted from either Boiler #10
or #11 of Ohio Edison's Toronto Power Plant. The flue gas first flows through a 250 hp,
F.D. fan, then is cooled to 160°C (320°F) or lower by spraying water into the flue gas
ducts. The cooled flue gas then enters a 3.2 m (126") diameter fluid led adsorber. The
NOXSO sorbent, 1.23 mm diameter 7-alumina beads impregnated with 5.0 wt% Na,
removes the S02 and NOx simultaneously from the flue gas as it passes through the fluid bed
adsorber. The cleaned flue gas mixes with the hot offgas from the sorbent heater, then
passes through the baghouse to remove all particulate before the gas vents to the atmosphere
through the power plant's stack. A cyclone is installed after the adsorber to recycle fines to
the bed (50% efficient on 50 micron diameter particles).

The spent sorbent in the adsorber overflows into the dense-phase conveying system
where 377 kPa (40 psig) air lifts the sorbent 24.4 m (80 ft) high to the top of the sorbent
heater, which is a 2.34 m (92") diameter, 3-stage, fluid bed vessel. A natural gas-fired air

99-2


-------
heater supplies the hot air to heat the sorbent in the sorbent heater to 621 °C (1150° F).
During the sorbent heating process, all the adsorbed NOx and a small portion of adsorbed
S02 (.01%) desorb from the sorbent. The hot sorbent heater offgas can either be vented to
the atmosphere through the power plant's stack or mixed with the cleaned flue gas entering
the baghouse. The hot sorbent in the bottom bed of the sorbent heater underflows into a J-
valve. Either nitrogen or steam can be used to carry the sorbent through the J-valve into a
1.22-m (48") diameter moving-bed regenerator. Natural gas is the first regenerant to treat
the hot sorbent in the regenerator at a temperature of 1130°F. The sulfate on the sorbent is
reduced to SOz, H2S, and sulfide on the sorbent. Steam is then used to hydrolyze the sulfide
to H2S which occurs at a temperature of 1080°F. The offgas from the natural gas treater
mixes with that from the steam treater before the combined stream enters the incinerator, in
which all the sulfur species are converted to S02.

Regenerated sorbent flows into a second J-valve which conveys it to 1.73 m (68")
diameter, 3-stage fluid ted cooler. A fan supplies ambient air to cool the sorbent. The teat
of the regenerated sorbent is recovered by the cooling air which is then used as the
combustion air in the air heater. The cool sorbent in the bottom bed of the cooler overflows
into a 1.83m (72") diameter surge tank. A third J-valve is used to transport the sorbent from
the surge tank to the adsorber.

The first two J-valves isolate the reducing environment (regenerator) of the NOXSO
plant from its oxidizing environment (heater and cooler train). The steam (for normal
operation) or nitrogen (for start-up) enters the two J-valves to carry the sorbent upward.

Since the steam is introduced at the lowest point of the J-valve, which is also the highest
pressure point between the two vessels, a steam barrier is created to prevent the mixing of
the reducing gas with air or vice versa. The third J-valve is operated using air to lift the
sorbent from the surge tack to the adsorber.

The NOXSO pilot plant differs from a commercial unit in two respects. First, since
the POC plant uses only a slip stream of flue gas from the power plant, the amount of NOx
evolved from the sorbent heater is too small to affect the NOs thermal equilibrium inside the
boiler furnace. Therefore, NO* is not recycled to the boiler as it would be in a commercial
unit. However, the ability of a coal combustor to reduce excess NO, introduced into the
combustion chamber was proven in simulated NOx recycle tests. The tests were carried out
using the Department of Energy, Pittsburgh Energy Technology Center's tunnel furnace and
227 kg/hr (500 Ib/hr) pulverized coal combustor.1,2 Additional NO, recycle tests were
conducted on a 227 kg/hr (500 lb/hr) small scale cyclone burner at the Babcoek & Wilcox
Research Center in Alliance, Ohio to obtain additional design date for the NOXSO
Demonstration Plant.3

The second difference between the POC and a commercial NOXSO installation is the
fate of the regeneration offgases. At the POC, the offgases are simply incinerated so that all
sulfur species are converted to S02 and then vented to the plant stack. In a commercial unit,
the regeneration offgases are sent to a sulfur recovery plant where the sulfur bearing gases
are converted to elemental sulfur. Depending on market conditions and regional demand,
either sulfuric acid or liquid S02 could be chosen as the end product rather than elemental

99-3


-------
suite. As the processes for making each of these end products are well established and
commercially available, they were not tested at the POC facility.

Reliability

From November 1991 through December 1992, the NOXSO pilot plant processed flue
gas for a total of 5213 hours. The first four months of operation were hampered by the
usual startup/shakedown problems encountered in a first-of-a-kind installation. For example,
the plant was shutdown the entire month of January to replace all seven grid plates in the
fluidized beds. Nevertheless, once the initial problems were solved, pilot plant availability
averaged 76% over the period of April through December 1992 (see Figure 2). This is an
excellent result for a pilot plant since the plant was subjected to frequent changes in
operating setpoints that stretched the limits of plant operability. The pilot plant operating
experience demonstrates that the NOXSO process is simple, reliable, and adjusts easily to
changing operating conditions.

Operation/Safety

The pilot plant was staffed with two operators. A minimem of two was required for
safety so that in the event of an injury to one man the other could assist the injured or call
for emergency response. In actuality, our experience is that less than one fiiU-time operator
is needed since process controls and safety systems are folly automated. Auxiliary personnel
for maintenance, repairs, Instrument calibration, etc., would be required as needed.

The NOXSO pilot plant operated for 17 months without a single lost man-hour due to
accidental injury. This safety record is a direct result of the design team's commitment to
safety from the start. Safety engineers from MK-Ferguson (design/construction general
contractor) and W.R. Grace & Co.-Conn, (sorbent manufacturer) conducted safety hazard
reviews of the plant as designed and as constructed. In addition, the plant was inspected and
approved by MK-Ferguson's Corporate Safety Officer prior to startup.

S02/NOx Adsorption

Figure 3 is a plot of S02/N0X removal efficiencies versus cumulative plant operating
hours. The data are averages computed over a minimum of four hours and a maximum of
twelve hours. The data are selected from periods in which the plant sulfur and nitrogen
oxides mass balance closures were 100 +15%. The removal efficiencies in Figure 3 vary
with time due to the fact that NOXSO process operating conditions were intentionally varied
to quantify their effect on process performance. He process operating conditions varied
were flue gas flow rate, sorbent circulation rate, adsorber sorbent inventory, adsorber bed
temperature, and adsorber inlet S02 and NOx concentrations. Also tested were two different
adsorber configurations: 1) a single-stage fluidized bed with flue gas cooling via water spray
into the ductwork approximately 90 feet upstream of the adsorber, and 2) two fluidized beds
in series with cooling via direct water spray into the beds. The vertical line in Figure 3
marks the time at which the second adsorber grid was installed. Note that both SOj and NO*
removal efficiencies improved with the installation of the second grid.

99-4


-------
Figure 4 is a plot of S02 removal efficiency versus adsorber gas residence time.

When the data are segregated into groups with essentially the same sorbent residence time,
an equation of the form, y=ax1/N, N>1, satisfactorily represents the data. This is true for
the entire database of 117 data points, although for clarity only a portion of the database is
shown in Figure 4. The correlation coefficients (i2) for the two curves shown in the figure
are 0.85 (53-59 min) and 0.89 (32-39 min).

The strictly empirical correlation is best for intermediate values of S02 removal and
short sorbent residence times when the relationship between S02 removal and gas residence
time is nearly linear. The correlation is worst for high values of S02 removal and gas
residence time, since the correlation gives no limiting value of removal efficiency, although
the actual limit is 100%.

In addition to gas ami sorbent residence time, SOj removal efficiency varies with the
concentration of S02 in flue gas inlet to the adsorber. Figure 5 shows that S(\ removal
efficiency is inversely proportional to inlet S02 concentration. The proportionality constant
(the slope of the lines in Figure 5) varies depending upon the ratio of flue gas flow to sorbent
circulation rate.

Figure 4 also shows that the two-stage adsorber consistently out-performed the single-
stage adsorber. This is seen more clearly in Figure 6 which shows the results of an identical
series of tests cm the one and the two-stage adsorber. For the one-stage adsorber, SOz
removal efficiency is shown to be inversely proportional to the flue gas to sorbent mass ratio,
all other operating variables constant as noted at the bottom of the figure. When the tests
were repeated with the two-stage adsorber, S02 removal efficiencies were higher by 5 to 10
absolute percentage points. This improvement is due to 1) better gas distribution with the
addition of the second grid plate and 2) counter-current flow of gas and sorbent so that in the
bottom bed of the adsorber partially sulfated sorbent is in contact with the highest
concentration of pollutants providing the driving force to put more sulfur on the sorbent. All
the data in Figure 6 were obtained at equal adsorber sorbent inventories, therefore the
pressure drop across the two-stage adsorber is only greater than the one stage by the pressure
drop across the second grid plate. (2-3" H20).

Figure 7 shows NO, removal efficiency as a function of flue gas to sorbent mass
ratio. As is the case with S02, NOx removal efficiency decreases in proportion to the
increase in mass ratio, all other operating variables constant. The line drawn in the figure
through the one-stage data has a correlation coefficient (i2) of 0.98. The two-stage data show
the same trend but removal efficiencies are 6 to 12 absolute percentage points higher than the
one stage. The best line through the two-sage data extrapolates to 86% NOx removal
efficiency at a flue gas to sorbent mass ratio of 4.6. The two-stage/bed spray data point
shown in Figure 7 is 93.5% NOx removal at a mass ratio of 4.6. This shows the effect of
adsorber bed temperature on NO* removal. Data obtained over an adsorber bed temperature
range of 250-356 °F show a definite trend of increasing removal efficiency with decreasing
bed temperature. Further improvement is probable at bed temperatures lower than 250°F.

99-5


-------
This trend was best illustrated in tests where the flue gas was spiked with SO* and
NOx from pressurized gas cylinders. Figure 8 shows NO* removal efficiency as a function
of inlet adsorber NOx concentration from 300-1065 ppm. This is the range of NOx
concentration that exists in flue gas from coal-fired utility boilers. All tests were run at flue
gas to sorbent mass ratio of 4.2 to 5 and total bed pressure drop of 19" H20 in the two-stage
adsorber. The date in Figure 8 clearly show that adsorber NOx removal efficiencies of 86-
88% are achievable at 917 to 1000 ppm inlet NOx using the two-stage adsorber with in-bed
water spray.

Figure 9 shows that S02 removal efficiency increases as the concentration of NOx in
the incoming flue gas goes up. This is because the S02 and NOx adsorption mechanisms do
not proceed independent of one another. In one-step in the mechanism, NO and Oj combine
with SOz on the sorbent's surface to form Na^O*, a stable compound.

Sorbent Regeneration

The NO, adsorbed in the adsorber desorbs as the sorbent is heated to the sulfur
regeneration temperature in the sorbent heater. The heater bottom bed temperature at the
pilot plant was typically controlled at 1150°F. Laboratory studies show that NOx desorption
is complete around 700°F. In the pilot test, the NOx gas mass balance between the adsorber
and the sorbent heater was consistently within 100±15% closure. At the pilot plant, the
sorbent heater offgas (700-900 ppm NOx in air) was vented to the stack. In a commercial-
scale plant, the NOx desorbed in the soibent heater is recycled to the coal burners where a
significant fraction of the NOx is destroyed. Pilot-scale tests of NOx recycle were performed
in 1992 at Babcock & Wilcox, Alliance, Ohio. The results are summarized in the fwiai
report "An Experimental Study of NOx Recycle in the NOXSO Flue Gas Cleanup Process*.3

Sorbent sulfur regeneration is achieved by contacting sorbent first with a reducing gas
and then with steam. The gas/sorbent contacting was accomplished in two sequential
countercurrent moving bed reactors. Although tests have shown that H2S, C jHg, Hj, CO,
H2/CO (synthesis gas), and CH4 are suitable regenerant gases, pipeline natural gas was used
exclusively at the pilot plant because of cost and availability.

The sulfur regeneration mechanism is a net endothermic reaction. The heat of
regeneration is equal to 17.7 kcal per gmole sulfur at (1112°F). Since no additional heat is
input to the regenerator, the sorbent's sensible heat must sustain the reactions. Tests have
shown that the threshold temperature for the reaction to proceed is approximately equal to
the auto ignition temperature of natural gas which is 1116°F. In order to maintain a
reasonable rate of reaction, the regeneration temperature must be about 1136°F. Allowing
for heat loss when transporting soibent from the sorbent heater to the regenerator, the
temperature in the bottom bed of the heater must be at least 1150°F.

The pilot plant regenerator was equipped with sorbent sampling ports at various
elevations. Sorbent samples were periodically withdrawn from these ports and tested for
sulfur content. These data were used to gauge the extent of the reaction with time and to
estimate the rate of sulfur regeneration.

99-6


-------
The regenerator may be modelled as ail integral reactor with a constant gas
concentration profile during a steady operation period. The sulfur molar balance gives the
rate of sulfur regeneration as a function of sorbent flowrate, inventory, etc. as shown below:

FsSadX = rg dw

where Fs is the sorbent flowrate, lb/hr,

rs is the sulfur regeneration rate, (lb sulfur)/hr/(lb sorbent),

Sa is the sulfur fraction of the spent sorbent,

S is the sulfur fraction of the regenerated sorbent,

X is the conversion of sulfur regeneration, X=l-S/Sa, and
W is the sorbent inventory, lb.

The slope of the plot of X versus W/(FsSa) will give the rate of sulfur regeneration.
Figure 10 shows the plot. The legend on Figure 10 indicates where the sample was
extracted: upReg at the top of the downward moving bed, gasReg at the end of natural gas
treatment and gas+stmReg at the end of steam treatment. The data were obtained during
two steady operating periods when the sorbent heater bottom bed temperature (1150 °F) and
the natural gas and steam feedrates to the regenerator were held constant, but the sorbent
circulation rate, residence time, and spent sorbent sulfur content varied.

The data in Figure 10 can be approximated by two lines, indicating that the
regeneration consists of two main reactions. The slope of the lines is the reaction rate. Hie
rate of the first reaction, the one that occurs in the upper section of the regenerator, is eight
times higher than that of the second reaction. Since the sorbent cools as it passes through the
regenerator, the change in rate might be due to the temperature drop. However, at two
different sorbent heater bottom ted temperatures, the lines associated with the second
reaction have approximately the same slope, as shown in Figure 11. Only the rate of the
first reaction is increased at the higher heater temperature. Also at X=0.6 in Figure 11, the
sulfur regeneration shifts from the first to the second reaction regardless of heater bottom bed
temperature.

The first reaction is associated with natural gas regeneration, the second with steam
treatment. The contribution of steam treatment to the sulfur regeneration can be seen in
Figure 10. Hie figure shows that steam treatment begins to work at X=0.6; however, steam
has little effect once the gas treatment proceeds to X=0.8. Furthermore, the value of X
does not change once W/FsSa exceeds 100.

The adsorbed sulfur oxides are reduced to H2S and S02 in the regenerator/steam
treater. Figure 12 shows H2S/S02 ratio plotted against die spent sorbent sulfur content.
H2S/S02 ratio increases when the spent sorbent sulfur content is less than 0.8 wt%. Once it
exceeds 0.8 wt%, the H2S/S02 ratio appears to be independent of the sulfur content. The
mean H2S/S02 ratio is approximately 0.4. For a sulfur plant, the H2S/S02 ratio is important
as it must be adjusted to 2.0 so as to have the correct reaction stoichiometry. When the ratio
is less than 2.0, it can be adjusted by reducing S02 to H2S. Alternately, sulfuric acid or

99-7


-------
liquid S02 can be produced as a by-product of the sulfur regeneration In the NOXSO
process.

Sorbent Attrition

Sorbent attrition is caused by physical and thermal stresses that come to bear on the
sorbent as it is transported through the processing loop and as it resides in the fluid beds.
These stresses can fracture sorbent beads and/or erode the surface of the beads. If the
sorbent bead becomes small enough, it can be entrained by the gas and exit the fluid bed.
Sorbent makeup is then required to maintain a constant sorbent inventory.

The rate of sorbent attrition equals the rate of sorbent makeup provided the starting
and ending sorbent inventories are equal. The sorbent makeup rate at the NOXSO pilot plant
for a 7 month period of operation is summarized in Table 1. The sorbent makeup rate is 3
PPH or 3/27,000 = 0.011% of total sorbent inventoiy per hour. This equates to replacing
the entire sorbent inventory approximately once a year. This makeup rate is slightly lower
than the makeup rate (0.016%/hr) used in previously published estimates of NOXSO process
operating costs.

Table 1
SORBENT MAKEUP RATE

Operation

Start date
End date
Flue gas, hrs

SorbentInventory

Total makeup, ibs	20,307

Sorbent lost, Ibs	- 6,415

Increase in sorbent inventory, lbs	+ 4,245

Net Sorbent Makeup, Ibs	9,647

Sorbent Makeup Rate, Ib/hr	3.00

Materials of Construction

Corrosion coupons were placed at six locations in the pilot plant. They were exposed
to 5500 hours of process operation. Physical and chemical analyses of the corrosion coupons
and of the materials used to construct the pilot plant were performed on shutdown.

The adsorber vessel and flue gas ductwork are exposed to the corrosive effects of flue
gas. The adsorber can be constructed of either carbon steel (with a corrosion allowance) or
specialty metals designed for flue gas exposure. The carbon steel corrosion coupons at the
adsorber inlet showed a corrosion rate of 3.8 mils/yr (1 mil = 0.001 inch). With a 30 year
lifetime a carbon steel duct would loose 0.114 inches of thickness. Carbon steel coupons in
the adsorber corroded at 2.1 mils/yr. Specialty metals for flue gas performed very well on

7/17/92
2/11/93
3,232

99-8


-------
the Met and showed 110 corrosion in the adsorber. 20CB3, C276, C22 INCO 625, and C4
all showed less than 0.15 mils/yr of loss. The stainless steels (316 and 304) showed
corrosion rates less than 0.5 mil/yr.

The adsorber at the pilot plant was made of carbon steel with acid-resistant linings on
the bottom shell ami the flue gas ductwork after the in-duct water spray. The adsorber
internals consisted of a Hastalloy bottom grid, a 304 SS top grid, and a carbon steel overflow
and downcomer. A 1" diameter hole in the ductwork developed 1' downstream of the water
spray nozzles after 16 months of operation. This was due to a failure of the acid-resistant
lining and a misalignment of die spray nozzle. Subsequently, spray nozzles were placed
above the fluidized bed and water was sprayed directly into the bed to cool the flue gas and
bed simultaneously. This has two advantages: 1) the acid dew point in the bed is lower than
in the flue gas; therefore the adsorber can be made of cheaper materials and 2) the flue gas
upstream of the adsorber can be maintained above the acid dew point so that specialty
materials are not required in flue gas ductwork.

Materials used in the regenerator of the NOXSO process are exposed to SOj, H2S,
H20, CH4 and trace amounts of COS and CS2 at temperatures near 1150°F. Although this is
a very aggressive environment for materials, appropriate materials have been identified for
use. HR160 showed no corrosion, and alonized 304 stainless steel showed only minor loss
(less than 0.33 mil/yr) when exposed in the regenerator environment. Haynes 556 and 446
stainless showed losses of 3 to 5 mils/yr and could be used with appropriate corrosion
allowances. In comparison, unalonized 304 and 316 SS corroded at 22 to 52 mils/yr while
carbon steel was destroyed.

The pilot plant regenerator and steam treater were made entirely of alonized 304 H
stainless steel. A vessel inspection after 5200 hours of operation showed that the alonized
material performed extremely well. The only problem was with 446 thermal overspray used
to cover the welds that had flaked off after 300 hours due to the difference in thermal
expansion between 446 and the alonized 304 H vessel. However, the base weld consisting of
a 312 root pass and 308 filler performed very well without the overspray.

The sorbent cooler was constructed entirely of carbon steel. The vessel internals
were made of types 304 or 316 stainless steel. Based on visual inspections, no corrosion or
erosion problems were encountered in the sorbent cooler.

The sorbent heater was made entirely out of type 304 stainless steel. The shell from
the top of the bottom head to the middle grid was alonized to protect against hot sulfate
corrosion. Over the life of the pilot plant, no measurable corrosion or erosion was observed
in the sorbent heater.

Economics

Data from the pilot plant have been incorporated into the design of a commercial size
NOXSO plant. Using this commercial plant design, an economic analysis was performed for
the NOXSO process. The basis for the analysis and cost information are included in Table

99-9


-------
2. The analysis was conducted for a 500 MW power plant burning 3% sulfur coal and
emitting 0.6 lb NO/MMBtu.

Since the NOXSO process is a combined S02/N0X removal process, it is not possible
to separate the cost of removing S02 from the cost of removing NO,. Consequently, an
assumption is matte that the cost of removing NO, is 3.0 times higher than the cost of
removing S02. The value of 3.0 represents a reasonable average for the relationship between
the cost of NOx and SO2 removal based on separate high efficiency technologies. This value
does not affect the overall economies, however, it does affect the relative cost of S02 and
NOx removal.

Emissions data is also listed in Table 2. Hie "Phase I S02 Limit" is calculated based
on allowable emissions of 2.5 lb S02/MMBtu. It is appropriate to consider over compliance
since fee high removal efficiency of the NOXSO process will allow a utility to generate SO*
allowances which can be sold to partially offset fee operating cost. A value of $300 has been
assumed for SO2 allowances. Beginning in the year 2000, the number of allowances
generated will decrease, however it is also likely that the value of allowances will be
significantly higher offsetting to some degree the reduction in the number of allowances
generated.

The annual operating and maintenance cost is $24.7 million with the cost of sorbent at
$10.1 million representing 41% of fee total. The capital cost of $257/kw is based on a
recent EPRI study.7

Revenues for the process will be generated by the sale of the sulfur by-product and
fee S02 allowances. The sulfur by-product can be elemental sulfur, sulfuric acid, or liquid
S02. The choice of sulfur by-product will be influenced significantly by the local demand
for the specific product. Since the market for sulfur is larger than the other two, sulfur is
used in this analysis. If fee market exists for sulfuric acid or liquid S02, either would be a
better choice since the revenue from sulfuric acid would be approximately three times more
than suite and liquid S02 would be six to eight times more. Making sulfuric acid or liquid
S02 would also result in minor increases in capital and operating costs.

The net levelized cost for the process is presented from three points of view. The
cost of buying, operating, and maintaining the plant will be $26,2 million dollars per year.
This translates to 8.5 mills/kwh of electricity produced. On a pollutant removal basis, it cost
$276 to remove each ton of SO* and $828 to remove each ton of NOx.

Conclusion

The 5 MW pilot plant has successfully shown that the NOXSO process can be scaled
up by a factor of 10. Maximum removal efficiencies of 99+% S02 and 95% NOs were
demonstrated over 6500 hours of process operation. Date from the pilot test may now be
used to design a commercial-scale plant wife an acceptable level of technical risk.

99-10


-------
Table 2

NOXSO PROCESS ECONOMIC ANALYSIS (1)

POWER PLANT PARAMETERS

GROSS CAPACITY
CAPACITY FACTOR
HEAT RATE
COAL HEATING VALUE
COALSULFUR
NOx EMISSIONS

ECONOMIC PARAMETERS

electricity

NATURAL GAS
SORBENT

NET SULFUR VALUE

S02 ALLOWANCE VALUE

FIXED CHARGE RATE (2)

REMOVAL COST NOr/REMOVAL COST S02

NOXSO PROCESS REMOVAL EFFICIENCIES
S02
NQr

EMISSIONS DATA

UNCONTROLLED S02
CONTROLLED S02
PHASE IS02 LIMIT
S02 ALLOWANCES GENERATED

UNCONTROLLED NOx
CONTROLLED NOx

POLLUTANT REMOVAL EFFICIENCY

OPERATING AND MAINTENANCE COSTS

FIXED (3)

VARIABLE (4)

NATURAL GAS
SORBENT
ELECTRICITY
TOTAL

CAPITAL COST

500 MW

mo«

10,000 Bta/kWh
12,000 Btu/lb
3.0%

0.6 Ib/MMBiu

S0.Q3 /kWh
$2J0/Mscf
$3.40®
ISO Aon
$300
10.6*

3.0

95%
80%

76.653 tans/year
3,833 ions/year
38.325 taas/yest
34,493 tonsfyear

9,198 toufyear
1.840 tons/year

93.4%

$5,714,000
5129.000
S5.131.000
510 J 12.000
$3,642,000
$24,728,000

5128400.000

$257 flcW

510347.750
5l.S20.435
$12,168,188

REVENUES

S02 ALLOWANCES
SULFUR VALUE
TOTAL

NET LEVELIZED COST

$26480.813/year

8-5 mills/kWh
S275Aon-S02
SS28Aon-NOx

(1) 1993 dollars.

(2> Based on 30 year book life, 20 year tax life, 38% composite federal and state tax.
and 2.0% for property taxes aad insurance,

(3)	Includes operating labor, fringes, aad supervision; maintenance labor and equipment;
and general add administrative expenses.

(4)	Includes process water and Clans plant catalyst.

99-11


-------
Acknowledgments

NOXSO Corporation wishes to thank the following for their commitment of personnel

and funds to the successful completion of the NOXSO POC test: The U.S. Department of

Energy's Pittsburgh Energy Technology Center, the Ohio Coal Development Office, Ohio

Edison Company, W.R. Grace & Co.-Conn, and MK-Ferguson.

REFERENCES

1.	J.L. Haslbeck, C.J. Wang, and L.G. Neal, et al., "Evaluation of the NOXSO
Combined N0,,/S02 Flue Gas Treatment Process," NOXSO Corporation Contract
Report submitted to U.S. DOE Report No. DOE/FE/60148-T5. November 1984.

2.	J.L. Haslbeck, L.G. Neal, and C.J. Wang, et al., "Evaluation of the NOXSO
Combined NOj/SC^ Flue Gas Treatment Process," NOXSO Corporation Contract
Report submitted to U.S. DOE Report No. DOE/PC/73225-T2. April 1985.

3.	J.T. Yeh, C.J. Drammond and J.L. Haslbeck, et al., "The NOXSO Process:
Simultaneous Removal of S02 and NO, from Flue Gas," Presented at the 1987
Spring National Meeting of the AIChE, Houston, Texas. March 29 - April 2, 1987.

4.	J.L. Haslbeck, W.T. Ma and L.G. Neal, "A Pilot-Scale Test of the NOXSO Hue Gas
Treatment Process," NOXSO Corporation Contract Report submitted to U.S. DOE
Contract No. DE-FC22-85PC81503. June 1988.

5.	W.T. Ma, J.L. Haslbeck and L.G. Neal, "life-Cycle Test of the NOXSO Process,"
NOXSO Corporation Contract Report submitted to U.S. DOE Contract No. DE-
FC22-85PC81503. May 1990.

6.	J.T. Yeh, W.T. Ma and H.W. Pennline, et al., "Integrated Testing of the NOXSO
Process: Simultaneous Removal of SO^ and NOx from Rue Gas," Presented at the
AIChE Spring National Meeting, Orlando, Florida. March 1990.

7.	J.E. Cichanowicz, C.E. Dene and W. DePriest, et al., "Engineering Evaluation of
Combined NO^/SC^ Controls for Utility Application," Electric Power Research
Institute, Palo Alto, CA. Dec 1991.

99-12


-------
NOXSO

PROCESS
TOWER

Figure 1. NOXSO POC at Toronto, Ohio

99-13


-------
100



Month

Figure 2. Pilot Rant Availability by Month
(December 1991-December 1992)

3.000

4,(MO	5,000

Cumulative Operating Hours

Figure 3. NOXSO Pilot Plant
S02/NOx Removal Efficiencies

6.000

7,000

99-14

A


-------
Adsorber Gas Residence Time (SEC)

Figure 4. S02 Removal VS Gas/Sorbent Residence Time
77 Data Writs Used In Non-Linear Regression

\



" \ iV



- \



\



o\

! I \

i i

1.000	1,500	2.000	2.500	3.000

S02 inlet To Adsorber, ppm

3.500

10000 pph.U- H20
[N0x)=350-<21 pfttl
—

7000t>Ctl.125'K20
(NOl]=18e-Z31 ppm
—©_

4,000

Figure 5. S02 Removal as a Function of Inlet S02
All Tests at 7000 scfm

99-15


-------
o

03

3	as

342-3W F «*» 3334M F <*» (ftp)
2£OF.250F(W6»*»9«ib«l»pn(f)

45

Rue Gas/Sorbent {lb/lb)

Figure 6. NOXSO Pilot Test
One-Stage VS Two-Stage Adsorber

100

ONE SJW3E ADSORBS?
—%r-

TWO STAGE ADSORBER

O

TWO STAGBBED SPfWY
~

Flue Gas/Sorbenl (LB/LB)

WH20 (on® stag*;. T & 7 *H20 (two Mags)
342-349 F (one stage), 333-344 F (two stage)
250 F. 250 f (MO siagtfMd Spray)

Figure 7. NOXSO Pilot Test
One-Stage VS Two-Stage Adsorber

99-16


-------
FUJ£ GAS COOLING
31M33F
#

(oo	aoo

Adsorber Met NOX (ppmw)

1,000

Rua en/Soiwnt» 4-5 ims
Mtortxr Prnsura Drop » 1PH20

Figure 8. NOXSO Pilot Test
Flue Gas Spiking with S02/NOx
{S02}=2300-2600 ppm Two-Stage Adsorber

FUJE GAS COOLING
319-333 F
*

BOTTOM BED COOLING
288-311 F

O

BOTH BEDS COOLING
251-25? f

o

200	400

F*» Gaj/Sort>«rt -4-5 KWIb
Adax&er Bad Pmssura Drop =19*M20

boo	aoo	1.000

Adsorber Iniet NOx (ppmw)

1.200

1,400

Figure 9. NOXSO Riot Test
Flue Gas Sinking with S02/NOx
[S02]=2300-2600 ppm Two-Stage Adsorber

9S-17


-------
88	tSO	ISO	200

W/(Fs*Sa), kg.sort>0nt/kg.S hr

Figure 10. Sorbent Sulfur Regeneration
(Open Symbol: 10-05-92d->10-18-92d)

(Solid Symbol: 11-2Q-92n->12-10-92n)

W/(Fs*Sa}, kg.sorbent/kg.S hr

Figure 11. Effect of Heater Bottom-bed
Temperature on Sulfur Regeneration

99-18


-------
J	I	I	I	I

0.4	0.6	0.8	1	1.2	1X	1.4

Spent Sorbent Sulfur Content, wt%

Figure 12. Regenerator Offgas H2S/S02 Ratio

99-19


-------
SOXAL™ DEMONSTRATION PROJECT AT
NIAGARA MOHAWK'S DUNKIRK STEAM STATION

David Hurwitz
Clifford Denker
Patrick Birdsall
Joseph Soltys
AlliedSignal, Inc.

7 Powder Horn Drive
Warren, NJ 07059

Peter Strangway
Niagara Mohawk Power Corporation
300 Erie Boulevard West
Syracuse, New York 13202

Abstract

Beginning in January 1993. AlliedSignal and Niagara Mohawk Power Corporation conducted a
six-month demonstration of the SOXAL™ Flue Gas Control Process at Niagara Mohawk's
Dunkirk coal-fired steam station. The Department of Energy's PETC office sponsored the
project, with additional funding provided by NYSERDA and ESEERCO. The SOXAL process is
an advanced regenerative system employing a sodium-based scrubber and AQUATECH™ bipolar
membranes to regenerate the scrubbing solution and to recover the sulfur values in a form suitable
for commercial use. Combination of the basic SOXAL process with urea/methanol NOx control
methods results in NO* removal efficiencies in excess of 90%. This paper discusses the successful
operation of a 3 MW pilot-scale demonstration over a six-month period, with continuous
operation and removal of over 98% of the SO2 and 90+% of the NO*. It also describes several
mechanical problems, not directly related to the process.

Introduction

The overall objective of the nominal 3 MW pilot plant was to demonstrate the technical and
economic feasibility of the SOXAL process for simultaneous removal of SOX/NOX contaminants
from the flue gas of a coal-fired boiler. The key demonstration was the integration of an
AQUATECH bipolar membrane system with proven sodium scrubbing and sulfur dioxide steam
stripping technology. Previously, the AQUATECH membrane system had been commercially
proven in applications unrelated to flue gas desulfurization. In addition, SOXAL was tested on a

100-1


-------
laboratory pilot unit for one year using a synthesized scrubbing solution of sodium sulfite.

The majority of the current project's funding was provided through a grant from the Department
of Energy's PETC office under a cost-sharing contract with AlliedSignal as the prime contractor.
Additional funding and the host ate were provided by Niagara Mohawk Power Corporation with
co-funding from the New York State Energy Research and Development Authority (NYSERDA)
and the Empire State Electric Energy Research Corporation (ESEERCO).

The Clean Air Act of 1990 made it necessary to accelerate the incorporation of acid rain control
systems into the planning process of some of those utilities and industrial facilities burning sulfur-
containing fuels, primarily coal. While many control systems east based on lime and limestone
scrubbing, these systems have severe drawbacks for incorporation into long-term emissions
control plans. Although the costs associated with disposal of large amounts of gypsum sludge
may be manageable today, the trend is towards more costly and less onerous disposal options. In
addition, these approaches simply take pollution out of the air and put it in the ground. Even if
economical today, limestone scrubbing does not provide an environmentally sound solution for
the long term.

Many SO2 control technologies are being pursued in the hope of developing an economical
regenerable system that recovers the SO2 as a useful commercial product, thus minimizing the
formation of waste. Some include the use of exotic chemical absorbents alien to the utility
environment and waste treatment facilities. These systems present new environmental issues to
the utility while attempting to solve old ones.

Sodium alkali scrubbing is an accepted, proven system for removing SO2 from gas streams. It is
the system of choice in industrial applications due to its lower capital requirements, high SO2
removal efficiencies (98+%) and low maintenance costs. A large number of sodium scrubbers
have been operated successfully at industrial as well as utility sites. The main drawback has

always been the higher cost of the sodium scrubbing solution versus calcium systems. SOXAL
was developed to minimize the cost of sodium scrubbing by regenerating the scrubbing solution
and recovering the sulfur in the form of commercial products such as liquid sulfur dioxide, sulfuric
acid, or elemental sulfur.

Another benefit of sodium scrubbing is its' ability to enhance existing NOx control technologies
based on urea or ammonia injection. Methanol injected as an adjunct to these systems converts
NO to NO2. The NO2 is then absorbed % the sodium scrubber providing overall NO* removal
efficiencies in excess of 90%. The SOXAL pilot unit at Dunkirk was adapted to simulate
incorporation of a urea/methanol NOx control process. NO2 will be injected into the flue gas to
approximate the composition resulting from urea/methanol injection.

100-2


-------
Process Description

The SOXAL process flow diagram (Figure 1) shows four major unit operations:

1 • Pre-scrubber - A water scrubber designed to remove chlorides and fluorides from the flue
gas as well as particulates escaping from the ESP. It also serves to quench the hot flue gas.

2.	Sodium sulfite scrubber - removes SO2 and NO2 from the flue gas.

3.	AQUATECH bipolar membrane cell stack system - regenerates the spent scrubber solution
(sodium bisulfite) back to sodium sulfite and produces sulfiirous acid for the stripper.

4.	Steam stripper - strips off" pure SO2 (98+%) from the suUurous acid feed.

The primary reactions in the sodium scrubber for SOx/NOx removal are shown below. SO2
removal is accomplished by reaction with the Na2S03 scrubbing solution to form sodium bisulfite
(NaHSCb). Note that a portion of the Na2S03 is oxidized to Na2S04 by reaction with oxygen
from the flue gas stream.

Na2S03 + S02 + H20 --> 2NaHS03

Na2S03 + 1/2 Qa — > Na2S04

The Na2SC>4 can be recovered in a salable crystalline form. Alternatively, a secondary
AQUATECH unit can be employed to convert the Na2S04 to NaOH and H2SO4. This NaOH is
recycled to the scrubber to reduce the alkali makeup requirements.

Removal of NO* by urea/methanol injection into the flue gas is accomplished in a two-stage
process. First, urea addition reduces approximately 70%-80% of the NOx in the flue gas to N2 as
shown in the reactions below. This reaction occurs at 1600-1900 °F.

NH2-CO-NH2 —> 2NH2 + CO

NH2 + NO —> N2 + H20

The remaining NO in the flue gas is oxidized to N02 by reaction with methanol. Addition of
methanol also reduces ammonia slip, thereby reducing deposition of salts in downstream ducting
and plume visibility problems. This reaction proceeds at 1000-1500 °F.

CH3OH + 302 — > 2HO2 + H20 + C02

H02 + NO —> N02 + OH

The N02 product from methanol injection is subsequently reduced to N2 by reaction with Na2S03
in the scrubbing solution. The Na2S03 is oxidized to Na2S04 in this reaction. The Na2S04
generated is eventually recovered as a solid crystalline product, or processed in the optional

100-3


-------
secondary AQUATECH unit.

2N02 + 4Na2S03 --> N2 + 4Na2S04

Regeneration of the spent scrubbing solution is achieved in the primary AQUATECH
regeneration unit using proprietary bipolar electrodialysis membranes. The bipolar membranes
separate water molecules into hydrogen (H4") and hydroxyl (OH") ions (Figure 2). NaHS03 in the
spent scrubbing solution is converted to Na2S03 as shown below.

2NaHS03 —> Na2S03 + H2S03

H2S03 —> H20 + S02 (gas)

In the primary regeneration unit (Figure 3), sodium cations (Na) migrate across the cation
selective membrane into the base compartment and become associated with OH" ions originating
from the bipolar membrane to form NaOH. Most of this NaOH reacts with NaHS03 to form the
regenerated Na2S03 for recycle to the scrubber. The HSCb" anions that remain in the acid
compartment associate with H"1" ions from the bipolar membrane to form H2SO3, The partially
saturated H2SO3 stream is continuously withdrawn from the primary regeneration unit and
subsequently decomposed into S02 gas and water in the steam stripper.

The above reactions describe what takes place in the pilot-plant system. However, in commercial
operations, the concentrated S02 gas would be passed through a cooler to condense most of the
water. The resulting gas stream would then be preheated, dried, compressed, cooled and stored
as liquid S02 product. An optional secondary AQUATECH unit comprising three compartments
can be employed to recover sodium alkali from the sodium chloride and sodium sulfate solution
from the stripper bottoms as shown below:

Na2S03 + H20 —> NaOH + H2S04

The amount of H2 S O4 production is essentially dependent on the degree ofNa2S03 oxidation in
the system. Alternatively, the sodium sulfate stream from the stripper bottoms can be evaporated
and crystallized to a salable Na2S04 crystal form.

Pilot Plant Operation

The layout of the pilot plant is shown in Figure 4. The operators of the demonstration facility
found the SOXAL process to be easy to operate and very responsive but were surprised by how
many unexpected mechanical and instrumentation problems occurred. Few issues surfaced
concerning the membranes; however, most of our time was spent fixing pumps, flow meters, weld
leaks, and assorted non-process related problems. Most of these failures were eventually
corrected. The majority probably would not have occurred if we had more experience in
choosing commercial scale components and instrumentation for scrubbing systems and if our
suppliers and contractors had conformed more closely to our specifications for materials of
construction.

100-4


-------
Instrumentation

Our greatest disappointment was the failure of the SC>2/NOx gas analyzer to operate throughout
the entire test period despite numerous attempts to fix the problems. We were forced to contract
for an outside CEM service beginning in April in order to complete the test program. Our PVC
flow sensors proved to be too fragile. When the PVC turbines were replaced by stainless steel
turbines, the problem was solved. We found our Vortek-type gas flow meters to be prone to
plugging. Pitot type devices would be recommended in the future. Initially, we also experienced
difficulties with some of the sensor electronics and various controllers. These problems, too,
were mostly solved with time.

Much of the instrumentation installed proved, however, to be extremely reliable and performed up
to expectations. These instruments included the variable area rotometers, differential pressure
level sensors, pressure transducers, pH sensors, pneumatic control valves, PLC, Data Acquisition
System and miscellaneous PID controllers.

Prescrubber/Scrubber

The prescrubber and scrubber were designed and supplied by Advanced Air Technology of
Arlington Heights, Illinois. They were chosen for their expertise in supplying sodium alkali
scrubbers to the industrial market and their ability to incorporate low oxidation into their designs.

The prescrubber measured 4.5 feet in diameter by 25 feet in height and was water based. The
sodium- based scrubber included two six-foot, packed (polypropylene-random) stages and
measured 4.5 feet in diameter by 40 feet in height.

Operation was easy and reliable. We observed no fouling although a lot of particulates were
removed by the prescrubber. The mist eliminators worked well and there appeared to be no
carry-over from the prescrubber to the scrubber. The only difficulties encountered were the
failure of one pump, some minor control problems and a few initial gas leaks.

Cell Stack/Membranes

The AQUATECH cell stack employed at the Dunkirk demonstration project employed 44 two-
membrane cells. Each cell was composed of a single bipolar and cation membrane. It required
only 176 ft2of cell area to handle the equivalent of 3 MW of flue gas.

The cell stack was operated for over three months, although not all continuously, and no
problems were experienced with the cell stack hardware. The standard commercial AQUATECH
bipolar membranes employed in the demonstration performed very well. While some individual
membranes were replaced, from time to time, the overall performance was well within

100-5


-------
expectations, A batch of experimental bipolar membranes which did not perform well early in the
test were replaced. At no time was the testing ever delayed due to membrane failure.

Stripper

The steam stripper, although representing conventional technology, presented us with numerous
problems. Most were related to materials of construction and inconsistencies that resulted in
leakage of SO2 and concerns about the durability of the column. Instrumentation Mures also
resulted in temperature control problems that gave the erroneous impression of low efficiency.
Once these problems were identified and corrected the column operated as designed.

Operation of the pilot plant has been achieved with minimal staffing. Two engineers and four
operators covered all the shifts required, including 7 days per week, 24 hour operation during
continuous testing in January and February.

Test Description

The original 3 MW pilot-plant test program at Dunkirk covered six months (Figure 5)
incorporating continuous operation and numerous parametric tests. Failure of the SOx/NOx
analyzer to operate severely limited the quantitative data collected during the first two months of
operation. During January and February of 1993 we were, however, able to demonstrate
continuous operation of the cell stack and its integration into the scrubber/stripper system. The
process was kept in balance providing regenerated scrubbing solution and concentrated SO2
which was sent to the stack in the case of the pilot-plant facility. This period of operation also
demonstrated the viability of the bipolar membranes in a sodium-based system. Unfortunately we
were unable to measure how much SO2 was being absorbed in the scrubber.

Due to a scheduled month-long shutdown of the Dunkirk Station boiler during March, we were
unable to resume testing until late April 1993. This allowed time to revise the parametric testing
to take place during a four-month time frame after boiler startup and to arrange for a CEM
service to set up to obtain accurate SO2 and NOx readings that were unavailable from the
defective S02/NOx measurement system.

Beginning in late April, we began parametric testing according to the following plan:

100-6


-------
Tablet

Test Plan for Parametric Studies

Test

•	Initial Baseline Studies

•	Absorber Parametric Studies

-	Lower Stage pH

-	Recycle Rate

-	Number of Beds

-	SO2 Concentration

•	Cell Stack Parametric Studies

-	Base pH

-	Recycle Rate
-Amps

-	Conversion

-	Cell Stack Temperature
Stripper Temperature
Optimized Baseline
Optimized Baseline
NO2 Concentration
Optimized Baseline - SO2/NO2

EDTA Addition

Objective
Establish Baseline

Maximize Absorption
Maximize Absorption
Minimize Oxidation

Minimize Oxidation/Maximize Absorption

Reduce Flush Cycle
Reduce Cost and Flush Cycle
Reduce Power Consumption
Reduce Power Consumption
Reduce Power Consumption
Optimize Efficiency
Maximize Absorption
Minimize Oxidation
Determine N02 Absorption
Continuous Operation

Reduce Oxidation

During the four months of parametric and continuous studies from April through July, the
Dunkirk boiler was shut down on weekends and was operated in a curtailed mode overnight due
to a lack of power demand. Parametric testing was run, therefore, on a decoupled basis, five days
per week In other words, when we ran studies on the absorber, the cell stack was shut down and
vice versa. We ran from a full storage tank of either spent or regenerated absorbent, depending
on the test. The part of the process not undergoing testing was ran overnight to replenish the
inventory for the next day's testing. Although the parametric studies were not appreciably
affected by the boiler cycling and shutdowns, continuous operation was affected. At no time were
we able to run the process for greater than five days due to a lack of flue gas.

Test Results

As of the time when this paper had to be written and submitted, the demonstration program was
still in progress and we have not had an opportunity to thoroughly analyze all the data collected.
The data collected to date is summarized in Table 2. During the first two months of the test, we
operated continuously, seven days per week. Unfortunately our S02/N0X analyzer was not
functioning properly and we have no definitive gas stream scrubber data. Since the testing was
restarted in April using an outside (trailer-mounted) testing service, we have consistently
demonstrated 98+% SO2 absorption (Figure 6). During this time, SO2 concentration in the flue

100-7


-------
gas has averaged between 1000 and 1500 ppm (Figure 7). We are confident that same high level
of absorption was achieved during our earlier continuous run. Preliminary data indicates that
higher SO2 levels (obtained by recycling recovered SO2 into the scrubber) enhances SO2
absorption and reduces oxidation. We will have an opportunity towards the end of the test period
to run a number of five-day continuous tests. Unfortunately, the curtailed operation of Niagara
Mohawk's boiler at the present time is preventing us from running a longer continuous test.

Oxidation in the scrubber is an important factor in the economics of the SOXAL process, when
there is no commercial use for the sodium sulfate formed. The SOXAL process allows for
recovery of the oxidized sodium values at a small additional capital and operating cost using a
secondary AQUATECH unit. The sulfur values, however, would be lost. Recent preliminary
data indicates that total oxidation, during the SO2 only (no added N02) part of the demonstration,
Ms well within the design parameters even without the use of additives or any attempt to
minimize oxidation through adjustment of the operating conditions.

The major parameters associated with the operation of the AQUATECH cell stack are the
membrane durability and power consumption. Power consumption for the cell stack has
consistently been in the range of 1100 to 1300 kWh/ton S02 (Figure 8), depending on the
parametric test performed. This corresponds with anticipated values and is consistent with the
consumption used in EPRI's 1990 evaluation of SOXAL process economics.

During over 1600 hours of pilot plant operation to date, the commercial AQUATECH bipolar
membranes have proved to be extremely durable. Some experimental bipolar membranes had to
be replaced due to delamination. A wash process is used to minimize fouling of the membranes.
Parametric testing has resulted in the identification of operating conditions that minimize the need
to acid wash the membranes.

The majority of the testing remaining to be completed at this time involves 1) additional tests at
increased SO2 levels, 2) NO2 injection, 3) continuous operation at optimized conditions, and 4)

minimization of oxidation through addition of EDTA.

Conclusions

Although testing was incomplete at the time when this paper had to be written and submitted,
some very positive conclusions had already been achieved.

1.	Continuous integrated operation of the total absorption and regeneration system has been
demonstrated.

2.	The ease of operating the absorption and regeneration systems independently has been
demonstrated. This is important in dealing with variable loading, use of off-peak power and
general maintenance, where it may be desirable to operate parts of the system separately for
a period of time.

100-8


-------
3.	Stable bipolar membrane performance has been proven,

4.	The ability to operate the process with minimal staffing has been demonstrated. AlliedSignal
1ms operated the test with a staff of two engineers and four operators covering either a three
or four shift schedule. Except during parametric testing the demonstration was run by an
operator and engineer or an operator working alone.

5.	Consistent 98+% SO2 removal. SO2 removal has been demonstrated up to almost 100%
during parametric testing.

6.	Power consumption and oxidation consistent with design expectations for SO2 removal only
operation has been proven.

The demonstration has proven to be successful to date although instrumentation and mechanical
issues have limited the amount of quantitative data available from the first two months of
continuous testing.

100-9


-------
Dirty Am Gas
S02.HC1

Purge Stream

HCI

XI

Clean
FhieSas

NaOH
Makeup

Recovered
SO,

Flue Gas
SO.

i

J

Scrubber Faad
NajSO,

Stripper Feed
HzS0>lNaIS04

Base

Acid





vy

Spent Absorbent
NaHSOj, NajSO,

Steam



Met so2

Assoesfft AifWPfP

Cm

Smac

I - Purge Stream
NajSO,

Smw
Srmm

Figure 1
SOXAL Process Flow Sheet

o

h2o

Anode

OH-

pr

W, v.;. .vVggSI

C^ivAviwS®

OifSSsiSli

h2o

OH" H+

o

Cathode

Figure 2
Bipolar Membrane Operation

100-10


-------
Sodium Sulfite
Na2SOs

--V-

T

©T© ©

T

T

^	

Sulfurous Acid
h2so3

oh-

H*

OH"

©

— SO."

H*

SO,*

Anode

Na1



Na»-

OH-

Na*

Sodium Bisulfite 4 4 4 4
N3HSO3 	'	—Cell Unit	L

1

©

Cathode

Figure 3

Primary Regeneration Unit Operation

Flue
Gas

Figure 4
Equipment Layout

100-11


-------
Figure 5
Pilot Plant Operations Schedule

100

98

¦ ¦

i ¦

c
o

o
«
¦Q
<

-------
<<
-------
Table 2



Summary of Test Data - Daily Averages









S02

Flue Gas

SO?

Power

Test



Concentration

Flow Rate Absorption

Consumed

Code

Test Description

fppm)

©SCFM)

(%)

(kWh/ton SCW

A

Simultaneous* Baseline







A

Simultaneous Baseline









A

Simultaneous Baseline









A

Simultaneous Baseline

916

7,9922

98.8



A

Simultaneous Baseline

1202



99.3



A/H1

Simultaneous with 1.25x Feed to Base

1062

5,205

96.3



A/H1

Simultaneous with 1.25x Feed to Base

1115

4,798

97.5



A/H2

Simultaneous with 1.5x Feed to Base

1149

7,562*

98.9



A/H2

Simultaneous with 1.5x Feed to Base

1189

7,3662

97.2



A/H2

Simultaneous with 1.5x Feed to Base

988

4,758

96.6



B

Absorber Baseline

861

5,201

96.8



B

Absorber Baseline

1385

5,609

98.7



B

Absorber Baseline

1365

5,201

98.8



B

Absorber Baseline

1316

5,709

98.4



C

Cellstack Baseline







1122

C

Cellstack Baseline







1251

C

Cellstack Baseline







1082

D1

Absorber Bottoms at 5:1 Ratio

1361

5,309

99.6



D1

Absorber Bottoms at 5:1 Ratio

1278

5,175

98.9



D1

Absorber Bottoms at 5:1 Ratio

1085

5,711

91.9



D1

Absorber Bottoms at 5:1 Ratio

1182

5,402

97.7



D2

Absorber Bottoms at 1:1 Ratio

1086

5,744

100.0



E2

Absorber Recycle at 60 gpm

1146

5,880

97.9



E3

Absorber Recycle at 45 gpm

1123

5,771

95.9



F1

Absorber with One Stage

1211

5,865

82.7



G

Absorber with Recycled S02

2141

5,407

99.6



G

Absorber with Recycled S02

2167

5,394

100.0



G2

Absorber w/ Recycled S02, 5:1 Ratio

2228

5,377

99.4



H2

Cell stack with 1.25x Feed to Base







1131

11

Cell stack at 80 amps/sq. ft.







1049

12

Cell stack at 125 amps/sq. ft.







1331

J1

Cell stack at 80% Add Conversion







1334

J2

Cell stack at 120% Add Conversion







1292

Notes:

1.	"Simultaneous" indicates continuous operation of both absorption and regeneration
processes. All other tests conducted are decoupled.

2.	These flow rates in ACFM. All others in Dry SCFM.

100-14


-------
OPTIMIZATION OF ADVANCED COOLSIDE DESULFURIZATION PROCESS

M. R. Stouffer, W. A. Rosenhoover, J. A. Withum, J. T. Maskew
CONSOL Inc., Research & Development
40£X) Brownsville Road, Library, PA 15129

Abstract

This paper describes the development status of a high efficiency sorbent injection desul-
furization process (Advanced Coolside). The Advanced Coolside process involves the
use of a contacting device to humidify flue gas to near saturation while removing fly ash.
Hydrated lime, injected into the highly humid flue gas downstream of the contactor, cap-
tures S02 before being removed in the existing particulate collector. The high humidity
allows high S02 removal. High sorbent utilization is achieved by sorbent recycle, the
potential for which is increased with upstream fly ash removal. The objective of this
project is to develop a low-capital-cost process for Clean Air Act compliance, capable of
over 90% S02 removal and of sufficiently high sorbent utilization to be cost competitive
with wet limestone scrubbing over a wide range of compliance situations. The original
performance targets of 90% S02 removal and 60% sorbent utilization were exceeded in
1000 acfm pilot plant operations. The 90% S02 removal target was achieved with com-
mercial hydrated lime at sorbent utilizations of 70-75%. Up to 99% S02 removal was
attained at sorbent utilizations of at least 60% A simplified equipment design was
tested and operability was confirmed in pilot plant tests. An interim economic evaluation
was completed based on these results. This economic analysis compared Advanced
Coolside with wet limestone FGD on the basis of an "ntil" plant design, i.e., as if it were a
demonstrated technology. Projected capital costs are approximately 40% lower than wet
limestone scrubbing costs over the range of coal sulfur contents (1.5-3.5%) and plant
sizes (160-500 MWe) evaluated. Levelized S02 control costs are lower than wet scrub-
bing over the range examined. The economic study identified several design improve-
ments for cost reduction. Ongoing development work will focus on these design
improvements and on sorbent improvement. The goal is to establish at least a 20%
levelized cost advantage over wet FGD to provide a sufficient incentive for
commercialization of this lower capital cost process.

Background

In-duct dry sorbent injection technology has been actively developed in the U.S. since
the early 1980s. The performance of these processes has been well-established through
the development of the Coolside process (CONSOL)1"6 and the HALT process
(Dravo)7,8 and through the DOE duct injection technology development program.9"11
These development efforts have included pilot-scale tests, proof-of-concept tests, and a

101-1


-------
full-scale utility demonstration. Established performance is in the range of 40-50% S02
removal at 2.0/1 Ca/S molar ratio and 20~25°F approach to adiabatic saturation tempera-
ture using hydrated lime as the sorbent. Additionally, the 105 MWe demonstration of
the Coolside process at the Ohio Edison Edgewater Station4*6 showed that an S02
removal of 70% can be attained by improving calcium hydroxide sorbent activity with
sodium-based additive injection at a 0.2 Na/Ca molar ratio (—32% sorbent utilization).

Process performance data and economic analyses support the attractiveness of duct sor-
bent injection for site-specific applications.12 However, the applicability as a compliance
option for the Clean Air Act or other regulations can be expanded by increasing S02
removals and sorbent utilizations. The performance targets for developing an advanced
process (90% S02 removal and 60% sorbent utilization) represent a substantial improve-
ment over previous technology.

The Advanced Coolside process is being developed using a 1000 acfin pilot plant13 The
pilot plant was used in previous development and scale-up of the Coolside process;2 it
was modified to include all elements of the Advanced Coolside process. Process devel-
opment has focused on improving the design of the contactor and on improving sorbent
utilization by optimizing sorbent recycle. A test program to investigate sorbent
improvement was recently initiated. This report will discuss progress in these areas,
results of the interim economic study and approaches for future process improvement.

Description of Advanced Coolside Process

Figure 1 show a schematic of the Advanced Coolside process. The process achieves
greater S02 removal and sorbent utilization than previous duct sorbent injection pro-
cesses by operating at a higher flue gas humidity and by more fully exploiting the
potential of sorbent recycle. The key to the process is a gas/liquid contacting device
downstream of the air preheater. The contactor serves two purposes: to nearly saturate
the flue gas with water, and to remove most of the coal fly ash from the flue gas. The
sorbent is injected downstream of the contactor into the highly humid flue gas. Hydrated
lime is very active for S02 capture near the saturation point, even in the absence of
liquid water droplets. Because the flue gas is already humidified prior to sorbent injec-
tion, there is no strict residence time requirement for droplet evaporation. S02 is
removed in the duct and by the sorbent collected in the existing electrostatic precipitator
(ESP) or baghouse. The heat of reaction between S02 and hydrated lime raises the
temperature of the flue gas by roughly 8-10 °F for each 1000 ppm of S02 removed.
Therefore, the particulate collector can be operated at an elevated approach to
saturation without flue gas reheat. However, because hydrated lime activity is highly
sensitive to the approach to saturation, this reaction heat effect also acts as a limiting
mechanism for S02 capture.

The spent sorbent is captured by the existing particulate collector as a dry powder. It
can be disposed of with the fly ash or separately. Sorbent recycle is an integral
component of the Advanced Coolside process. Laboratory and pilot plant tests have

101-2


-------
shown that recycle sorbent is quite active for S02 capture at high humidity. The
potential for recycle is increased because fly ash is removed separately before sorbent
injection. Furthermore, recycle sorbent performance can be improved by a simple physi-
cal pre-treatment step prior to re-injection; the nature of this pre-treatment step is
currently a proprietary feature of the process.

Design optimization has focused on the flue gas/water contactor. For the initial pilot
plant testing the contacting device was a Waterloo scrubber.

14,15 This js

a commercially

available device, marketed by Turbotak Technologies Inc. (Turbotak), and used primarily
for removal of submicron particles. The Waterloo scrubber consists of a conditioning
zone, a centrifugal fan and a mist eliminator, and uses two-fluid nozzles to finely atomize
water sprays at a liquid/gas ratio of about 1 gal/1000 acf.

Experimental

Hie Advanced Coolside process is being developed using a nominal 1000 acfm (0.3 MWe
equivalent) pilot plant. In addition, exploratory work is being conducted in fixed-bed
laboratory reactors.

Figure 2 is a schematic of the pilot plant. It was used in the development of the
Coolside process and in the scale up to a 105 MWe demonstration. In this develop-
ment the plant was shown to adequately simulate large-scale process performance. The
pilot plant was modified to fully simulate integrated operation of the Advanced Coolside
process, including combined flue gas saturation and fly ash removal by a contactor,
sorbent injection into nearly saturated flue gas, and semi-continuous sorbent recycle. A
baghouse is used for particulate collection. As described below, the baghouse could be
operated to simulate S02 reduction expected in an ESP. The pilot plant is described in
more detail in Reference 13.

Discussion
Recycle Optimization

The improvement in desulfurization performance which allowed project performance
targets to be exceeded resulted primarily from recycle optimization. By more folly
exploiting recycle, sorbent utilization efficiencies of 70-75% were attained, while
maintaining S02 removal around 90%. Also, high S02 removals ranging from 90% to
over 99% were attained, while maintaining sorbent utilization of at least 60%.

Recycle optimization tests were conducted in the 1000 acfm pilot plant in a semi-
continuous manner. Spent sorbent was removed from the baghouse hopper on a regular
basis; a portion of the material was discarded and the remainder, after pretreatment, was
returned in a batch to the recycle feeder hopper. Test durations were sufficiently long to
assure that steady-state continuous recycle was simulated closely (typically 20-70 hr).

101-3


-------
Tables 1 and 2 list process conditions and results for pilot recycle optimization tests.

Teste 1 through 4 (Table 1) were conducted with reheat before the baghouse (to a 25 °F
approach) and with frequent pulse cleaning to minimize baghouse S02 removal. The
purpose was to simulate conditions in a retrofit application with an existing ESP. In this
case, S02 removal in the ESP would be limited by bulk gas mass transfer. Based on
literature information and on theoretical calculations, an ESP removal of 30% of the
S02 remaining in the ESP inlet gas is a reasonable assumption. As shown in Table 1,
S02 removal in the baghouse with reheat averaged 5% (absolute), with most removal
occurring in the duct. Tests 5-9 (Table 2) were conducted with no baghouse reheat. The
9 to 12 °F baghouse approach temperature was a result of the flue gas temperature rise
from the heat of reaction. In these tests S02 removal in the baghouse was greater than
with reheat, although the large majority of S02 was still removed in the duct.

The recycle test results indicate that for systems with an existing ESP, 90% S02 removal
can be achieved at sorbent utilizations of 70-75%, substantially higher than the original
target of 60% utilization. For example, with a fresh Ca/S mol ratio of 1.2, duct and
system S02 removals were 87% and 90%, respectively (Test 2, Table 1). The increase in
sorbent utilization over the initial target has a substantial impact on economics, because
sorbent cost is a major component of the total S02 control cost.

The recycle results also indicate that very high efficiency S02 removal can be attained in
systems with a baghouse operated at a close approach to adiabatic saturation. For
example, 99% S02 removal was attained at 61% sorbent utilization (Test 9, Table 2). In
this test most of the S02 removal (88%, absolute) occurred in the duct. This capability
to achieve very high S02 removal may be attractive to new units using a baghouse for fly
ash collection. As discussed below, high removal efficiencies have been achieved at
higher sorbent utilizations (ca. 75%) by using small amounts of additives for promotion.

In the recycle tests in Tables 1 and 2, recycle ratios ranged from 3.3 to 6.9 lb/lb fresh
lime. Relatively high recycle ratios are possible because fly ash is removed upstream of
sorbent injection. Total dust loading ranged from 9.5 to 14J gr/scf. Pilot testing
indicated that recycle sorbent particles tended to aggregate during handling, pretreatment
and reaction; this is expected to improve the ability of an existing ESP to handle higher
dust loadings. For example, the mean particle size of sorbent entrained in the flue gas
upstream of the baghouse was 45 jwn, compared with a typical value of around 15 fim for
fly ash from firing of pulverized bituminous coal.

As shown in Tables 1 and 2, the recycle optimization tests were relatively long-term tests.
With one exception, operating durations ranged from 21 to 115 hr. This allowed process
operability to be evaluated. It also provides added confidence in data reliability.

Data reliability was confirmed by comparing utilizations based on gas analysis with those
based on solids analysis. Figure 3 shows a parity plot of utilizations by the two
independent methods for the tests in Tables 1 and 2 and for more recent tests with addi-
tive promotion. The good agreement of sorbent utilization data confirms the accuracy of
process performance data. It also confirms that steady-state continuous recycle

101-4


-------
conditions were closely simulated. Establishing near steady-state operation is critical in
assessing the sorbent usage of a process with a large inventoiy of recycle material. In-
duct S02 removal data accuracy was confirmed by manual flue gas sampling using EPA
Method 6 for selected ran periods. Continuous gas analyzers agreed with the manual
analyses within the accuracy of the EPA method. Duct exit gas was sampled using a
probe to minimize sorbent entrainment; the probe was also heated to minimize sorbent
reactivity.

Design Optimization

A major portion of the process development is devoted to contactor simplification. The
contactor is a key capital cost component, and the contacting device initially tested (the
Waterloo Scrubber) was designed for more stringent applications (i.e., submicron
particulate control) than required for Advanced Coolside. Because the process is applied
upstream of an existing particulate collector, some fly ash slippage through the contactor
is acceptable. It is only necessary to remove a large portion of the particulate mass (ca.
90%) to avoid recycling much of the inert fly ash. Approaches to reduce the capital and
operating cost of the contactor included eliminating the fan, an integral component of the
original Waterloo scrubber system, and redesigning the contactor to reduce water and
atomization air requirements.

Preliminary pilot plant studies indicated that these approaches are feasible. Tests were
conducted using the original Waterloo scrubber system with and without its centrifugal
fan over a wide range of atomizing air pressures and water flow rates. The test results
(Figure 4) indicated that high flue gas relative humidities can be achieved with or without
the fan, as long as sufficient water droplet surface area is generated in the contactor.
The test results also showed that the droplet surface area and, thus, atomization energy
can be reduced to below typical operating conditions with a relatively minor effect on
flue gas humidity. For example, reducing atomization air pressure from 45 psig to
25-30 psig resulted in a slight decrease in humidity, from ca. 99% to ca. 97%. Particulate
removal tests indicated that removal efficiency was not sensitive to the nozzle operating
conditions over the ranges tested and that the scrubber fan was not needed to achieve fly
ash removal greater than 90% (by wt). These results indicated that there is flexibility for
design and operating modifications.

Based on the pilot plant tests using the original contactor, a mechanically simpler con-
tactor was designed by Turbotak, marketer of the Waterloo scrubber. The new contactor
(Figure 5) consists of a spray chamber and a downstream mist eliminator. The spray
chamber uses two-fluid nozzles and an internal baffle design to promote droplet/gas/
particle interactions and a reduced velocity zone to allow droplet settling. Most of the
particles and water droplets are removed in the spray chamber. The mist eliminator
removes remaining droplets from the flue gas. The Waterloo scrubber fan was elimi-
nated in the new contactor, significantly reducing the cost.

Tests were performed which verified the humidification performance, particulate collec-
tion efficiency, and operability of the simplified contactor. Over 100 optimization tests

101-5


-------
were conducted to reduce atomization air pressure and flow and water flow relative to
the design conditions of Turbotak. Figure 6 shorn the results of these tests, plotted as
percent relative humidity vs. the relative droplet surface area. Relative droplet surface
area is defined as the total droplet surface area per unit flue gas volume produced at the
test operating conditions divided by the total droplet surface area per flue gas volume
produced at the original design conditions. Efficient humidification requires sufficient
droplet surface area together with good dispersion of the droplets in the flue gas; both
are affected by variations in nozzle operating conditions. Based on the results of S02
removal tests performed earlier, high S02 removal can be attained at relative humidities
of 95% and higher. The results in Figure 6 show that the droplet surface area can be
significantly lower than the area produced by the design conditions without significantly
affecting the humidification efficiency, provided good dispersion is maintained. Many of
the operating configurations that were tested still achieved >95% relative humidity even
though they produced droplets with as much as 50% less surface area than the design
conditions. To produce sprays with less droplet surface area, lower atomizing air
pressure and/or water flows rate is required.

Particulate removal tests showed that >90 wt % fly ash removal (measured upstream of
the mist eliminators) can be maintained with the nozzles operating at lower air pressure
and water flow rate than the design conditions. Figure 7 shows that most all of the
conditions tested gave better than 90 wt % fly ash removal and better than 95% relative
humidity.

Based on these results, alternative operating conditions for the contactor were identified
that reduce the water flow rate and atomizing pressure (Table 3). The pilot recycle tests
discussed above used these alternative operating conditions. As shown in the table, water
and air flow requirements were reduced by about half, and the air pressure requirement
was reduced from 45-50 to —30 psig, while maintaining humidification (>95% relative
humidity) and fly ash removal efficiency (> ca. 90%). These alternative operating condi-
tions will result in lower operating and capital costs (e.g., for air compressors).

Operability Observations

Pilot plant operating experience in tests up to 115 hr in duration is a positive indication
of the operability and retrofit potential of the Advanced Coolside process. Although the
pilot plant is not of sufficient scale to make a complete assessment of process operability,
observations of pilot plant operation provide initial information on key operability issues.

The contactor operability was simplified by the elimination of the fen. With the modified
design, no significant operability problems have been observed in over 1500 hours of
operation. The mist eliminator was washed periodically to maintain contactor pressure
drop at about 1.5 inches of H20.

Accumulation of solids on the duct walls was not an operating problem, even at very

close approach to saturation and with different duct configurations having short straight-
run residence times (<0.5 sec) and numerous changes in flow direction. There was

101-6


-------
generally a light surface coating of dry solids. At bends, there was somewhat more
accumulation. The amount of solids on the duct surface tended to reach a steady value
after 10 to 30 hr of operation, after which the rate of accumulation approached zero.
The solids were loose and easily removed. Solids accumulation only became a significant
operating problem if the solids injection system was oriented in such a way that the
sorbent impacted a wall directly after injection.

No major problems were encountered in preparing, handling and feeding the recycle sor-
bent. Operability of the pneumatic transport system was similar to that with hydrated
lime. A minor feeder modification improved feeding operation by reducing bridging in
the hopper. Operability of the recycle handling system was observed to deteriorate at
very high sorbent utilization (>70%). Operability was improved by adding the fresh lime
to the recycle material during pre-treatment and co-injecting the sorbents.

Baghouse operability was good at the close approach to adiabatic saturation (down to
10 °F) investigated in this program. The material did have a tendency for compaction
under compression and for hopper bridging at the lowest baghouse approach tempera-
tures. This would be an important consideration for a large-scale design.

Sorbent Optimization

Sorbent improvement can increase the attractiveness of the Advanced Coolside process
in several ways. Increasing sorbent utilization reduces sorbent usage and waste disposal
requirements. Increasing sorbent activity can reduce the required level of sorbent recycle
and could increase the applicability of the process for very high S02 removal levels.
Finally, the results of sorbent studies could allow use of lower cost sorbents by reducing
process sensitivity to sorbent source.

Pilot plant tests reported previously in this paper were all conducted with a single com-
mercial hydrated lime. A sorbent optimization test program is currently under way. The
program includes work in three areas: a lime hydration study, evaluation of alternate
sorbents, and evaluation of additive enhancement.

The objectives of the lime hydration study are to determine the effect of hydration vari-
ables on the properties of hydrated lime and to determine the effect of lime properties
on desulfurization performance. The hydration study is being conducted in cooperation
with Dravo Lime Co. using their continuous pilot (120 lb/hr) hydrator. Hydration vari-
ables being investigated in a statistical experimental design include the following:
quicklime source, quicklime grind size, hydration water temperature, residual H20 in the
product, and hydrator residence time. Hydrated limes will be characterized for chemical
composition and physical properties such as particle size distribution, BET surface area,
and pore size distribution. Desulfurization performance will be measured in laboratory
reactors and in the pilot plant.

Evaluation of alternate sorbents includes testing of different commercial hydrated limes
and testing of other sorbents, for example, specially prepared hydrated limes. Recycle

101-7


-------
tests have been conducted for three commercial hydrated limes. These hydrated limes
were obtained from different geographic regions and from lime plants among the ten
largest in the U.S. In tests at 1.2 Ca/S mol ratio, system SOz removals (with baghouse
reheat to simulate ESP removal) ranged from 86 to 90% for the hydrated limes with
surface areas ranging from 14 to 22 m2/g. Also, once-through (no recycle) screening tests
of different commercial hydrated limes from different sources and with varying surface
areas showed only small differences in S02 removals. These results suggest that process
performance is relatively insensitive to surface area and to commercial lime source. This
is an economic advantage, allowing use of the lowest cost sorbent available. Preliminary
screening tests have also been conducted on two specially prepared high surface area
hydrated limes, each with a BET surface area of about 35 m2/g. For one of these
sorbents, once-through S02 removals were higher than for any of the commercial
hydrated limes tested (by about 10-15%, relative). For the other, S02 removals were
lower than any commercial hydrated lime tested. Process performance must be verified
in recycle tests; however, these results indicate that a systematic investigation of the
effect of lime properties on desulfurization is warranted.

Previous laboratory studies7 simulating Advanced Coolside process conditions indicated
that sodium-based additives can substantially increase the utilization of hydrated lime (by
over 20% absolute). In the current test program, different approaches for additive
promotion will be investigated, including addition to lime during hydration. Based on
previous lab studies and literature information, additives to be evaluated include
Na2C03, NaCl, and CaQ2. Chloride additives are of interest because they could be
generated by neutralization of contactor recycle water. One pilot plant test was
conducted with NaCl promotion. Results are encouraging, indicating that sorbent
utilization can be increased to 80-85% without increasing recycle by using very small
amounts of additive — 0.025 Na/S mol ratio, about 1/15 of that employed in the conven-
tional Coolside process.1"3 In another pilot plant test conducted with 0.03 mol CaCl2/mol
fresh Ca, system S02 removal without baghouse reheat was 97% at 1.2 Ca/S. This result
indicates the potential of the process for high efficiency S02 removal in a new plant with
a baghouse. Further testing is under way using different additives and additive dosages
and varying process conditions.

Process Economics

An interim process economic study was completed based on current process performance
data with a commercial hydrated lime and a conceptual process design. The objectives
for this study were to confirm the potential economic advantages of the Advanced Cool-
side process and to identify priorities for further process development. A final economic
study will be conducted at the conclusion of the pilot-scale development program.

The economic study compared costs of the Advanced Coolside process with limestone
wet scrubbing. Figure 8 shows a schematic of the Advanced Coolside process conceptual
design used in the study; many of the process systems, particularly ancillary equipment,
have not yet been optimized. The limestone wet FGD design used in the study is shown
in Figure 9; the design includes forced oxidation and use of a single absorber module

101-8


-------
with no spare. Economic assumptions (Table 4) were selected to assure comparison on
an equivalent basis. Both processes were evaluated for 90% S02 removal, an assumed
capital life of 30 years, 65% capacity factor, and using the same retrofit factors. The
analysis was based on an "nth" plant design, using an 18% contingency for each process.
Assumed delivered reagent costs were $60/ton for hydrated lime with 1% inerts and
$15/ton for limestone. All costs are reported in 1992 dollars.

The economic study confirmed a substantial capital cost advantage for the Advanced
Coolside process. Figure 10 shows that for a 2.5% sulfur coal the capital cost was about
40% less than forced oxidation limestone scrubbing, over the 150-500 MWe range of
plant sizes studied. The relative difference in capital cost was about the same for 1.5 and
3.5% sulfur coals. The lower capital cost can be important to utilities in making compli-
ance decisions because it reduces financial and regulatory risk.

The economic study quantified the potential S02 control cost advantages of the
Advanced Coolside process. Figure 11 shows that the process has a lower levelized cost
($/ton S02 removed) than limestone wet FGD over a wide range of coal sulfur contents
and plant sizes. The cost differential ranged from 21% for 1.5% sulfur coal and a
150 MWe plant, to 11% for 2.5% sulfur coal and a 250 MWe plant, to break even for
3.5% sulfur coal and a 500 MWe plant

The interim study also indicated that there is potential for further improvement of the
Advanced Coolside process and identified areas for improvement with the greatest
potential impact on economics. Figure 12 shows a breakdown of the levelized S02
control costs for a medium plant size and medium sulfur coal. Areas identified for cost
reduction include sorbent cost and equipment capital cost for certain process systems.
Approaches for equipment cost reduction include further contactor optimization and
improvement in other systems on which optimization studies have not yet focused (e.g.,
recycle handling, waste disposal and handling, and flue gas handling).

The goal of further development is to establish at least a 20% levelized cost advantage
over wet FGD over a wide range of compliance situations. This would make it more
attractive for utilities to employ a newer, less established technology.

Future Work

Based on the process economic study, the focus of future process development will be to
increase the cost advantage of Advanced Coolside over commercial technology through
equipment design optimization and sorbent improvement. For the economic study,
Turbotak developed preliminary full-scale designs for the simplified contactor based on
the test results with the original contactor. The results of pilot tests using the new
contactor will be used by Turbotak to develop a commercial design to further reduce
costs. In addition, other commercially available gas/liquid contacting devices will be
evaluated for use in the process. Equipment design optimization efforts will be expanded
to look at other systems with potential impact on process capital cost, as identified in the

101-9


-------
economic study. The sorbent improvement work under way will continue as described
above. The goals are to reduce sorbent usage or the required recycle ratio and to allow
use of lower cost sorbent sources. Another area for future investigation is air toxics
control, particularly that of mercury. A literature analysis under way suggests that the
Advanced Coolside process has potential for substantial reduction in Hg emissions. The
capability for air toxics control would provide an additional incentive to use this
technology for S02 compliance.

Acknowledgment

This work was conducted under partial sponsorship of the U.S. Department of Energy
Contract DE-AC22-91PC90360. The authors are grateful to C J. Drammond, T. D.
Brown, R. E. Tischer, and others at DOE Pittsburgh Energy Technology Center for their
support and encouragement of this project.

References

1.	Yoon, H., M. R, Stouffer, W. A. Rosenhoover, and R. M. Statnick. "Laboratory and
Field Development of Coolside S02 Abatement Technology." Proceedings, Second
Pittsburgh Coal Conference, Pittsburgh, PA (September 1985).

2.	Yoon, H., M. R. Stouffer, W. A. Rosenhoover, 3. A. Withum, and F. P. Burke.
"Pilot Process Variable Study of Coolside Desulfurization." Environ. Progress. Vol. 7,
No. 2, p. 104-11 (1988).

3.	Stouffer, M. R., H. Yoon, and F. P. Burke. "An Investigation of the Mechanisms of
Flue Gas Desulfurization by In-Duct Dry Sorbent Injection." I&EC Research. Vol.
28, No. 1, p. 20 (1989).

4.	Stouffer, M. R., W. A. Rosenhoover, and H. Yoon. "Pilot Plant Process and Sor-
bent Evaluation Studies for 105 MWe Coolside Desulfurization Process Demonstra-
tion." Proceedings, Third International Conference on Processing and Utilization of
High-Sulfur Coals, Ames, IA (November 1989).

5.	Withum, J. A., H. Yoon, F. P. Burke, and R. M. Statnick. "Coolside Desulfurization
Demonstration at Ohio Edison Edgewater Power Station." ACS Div. Fuel Chem.
Preprints. Vol. 35, No. 4, p. 1463-72 (1990).

6.	Yoon, H., R. M. Statnick, J. A. Withum, and D. C. McCoy. "Coolside Desulfuriza-
tion Process Demonstration at Ohio Edison Edgewater Power Station." Proc.

Annual Meet. - Air Waste Manage. Assoc. Vol. 9A, Paper No. 91/102.8 (1991).

101-10


-------
7.	Babu, M., R. C Forsythe, C. F. Runyon, D. A. Kanaiy, H. W. Pennline, T. Sarkus,
and J. L. Thompson. "Results of 1.0 MMBtu/Hour Testing and Plans for a 5 MW
Pilot HALT Program for S02 Control." Proceedings of the Third Annual Pittsburgh
Coal Conference, Pittsburgh, PA (September 1986).

8.	Babu, M., I. College, R. C. Forsythe, R. Herbert, D. A. Kanary, D. Kerivan, and
K. Lee. 5 MW Toronto HALT Pilot Plant." Proceedings, 4th DOE-PETC Contrac-
tors' Meeting, Pittsburgh, PA (August 1988).

9.	OT)owd, W. "Duct Injection Experiments at DOE-PETC," Technical Update

No. 18, Duct Injection Technology Development Program. (January 1991).

10.	Brown, C. A., M. Maibodi, and L. M. McGuire. "1.7 MW Pilot Results for the Duct
Injection FGD Process Using Hydrated lime Upstream of an ESP," Presented at
the 1991 S02 Control Symposium, Washington, DC (December 1991).

11.	Felix, L. G., J. P. Gooch, R. L. Merritt, M. G. Klett, J. E, Hunt, and A G. Demlan.
"Scale-Up Tests and Supporting Research for the Development of Duct Injection
Technology," Presented at the 1991 S02 Control Symposium, Washington, DC
(December 1991).

12.	Nolan, P. S., D. C. McCoy, R. M. Statnick, M. R. Stouffer, and H. Yoon.

"Economic Comparison of Coolside Sorbent Injection and Wet limestone FGD
Processes," Presented at the 1991 S02 Control Symposium, Washington, DC
(December 1991).

13.	Stouffer, M. R., W. A. Rosenhoover, and J. A Withum. "Advanced Coolside
Desulfurization Process," Presented at the 1992 Summer National Meeting of the
American Institute of Chemical Engineers, Minneapolis, MN (August 1992).

14.	Spink, D. R. "Gas Solid Separation and Mass Transfer Using a Unique Scrubbing
Concept," presented at the AIChE National Meeting, Houston, TX (March 29-
April 2,1987).

15.	Spink, D. R. "Handling Mists and Dusts," CHEMTECH, p. 364-8 (June 1988).

101-11


-------
Table 1. Advanced Coolside Pilot Plant Recycle Tests with Flue Gas Reheat Prior
to the Baghouse and 1500 ppm S02.

Test

1

2

3

4

Test Duration, hr

36

115

13

73

Process Conditions









Fresh Ca/S, mol

1.4

1.2

1.5

1.2

Recycle Ratio, lb/lb fresh lime

4.5

6.9

4.4

6.7

Recycle Pretreatment

Yes

Yes

Yes

Yes

Baghouse Approach Temp., °F

23

23

24

22

Hydrated Lime

A

A

A

B

Process Performance









S02 Removal, %, In-Duct

83

87

84

80

System

90

90

90

86

Baghouse

7

3

6

6

Sorbent Utilization, %, By Gas Analysis

63

75

60

70

By Solid Analysis

63

70

59

71

Table Z Advanced Coolside Pilot Plant Recycle Tests with No Flue Gas Reheat
Prior to the Baghouse and 1500 ppm S02.

Test

5

6

7

8

9

Test Duration, hr

40

28

21

25

23

Process Condfttons











Fresh Ca/S, mol

1.2

1.5

1.2

1.6

1.6

Recycle Ratio, lb/lb fresh lime

3.3

3.5

4.9

3.9

3.8

Recycle Pretreatment

Yes

Yes

Yes

Yes

Yes

Baghouse Approach Temp., °F

9

12

9

11

12

Hydrated Lime

A

A

A

A

A

Process Performance











S02 Removal, %, In-Duct

60

70

81

91

88

System

84

90

88

97

99

Baghouse

24

20

7

6

11

Sorbent Utilization, %, By Gas Analysis

67

61

71

60

61

By Solid Analysis

68

63

68

58

61

Table 3. Optimization of Contactor Operating Conditions.

Base Design

Conditions
Nozzle Water Row
Nozzle Air Pressure
Nozzle Air Flow
Performance
Exit Humidity
Fly Ash Collection

1.13 gpm/1000 acfm

45-50 psig
17 scfm/1000 acfm

>98%

>95%

Alternative

0.6 gpm/1000 acfm

30 psig
9 scfm/1000 acfm

>98%

-95%

101-12


-------
Tabie 4. Key Assumptions of interim Process Economic Study.

Advanced Coolside

Forced Oxidation Wet FGD

Delivered Sorbent Cost
Waste Disposal Cost
S02 Removal
Capacity Factor
Capital Life
Retrofit Factor
Location Factor
Design Philosophy
Sparing
Indirect Costs
Construction

$60/ton, 7% inerts (hydrated lime)
$6.50/wet ton
90%

65%

30 years
Medium (1.22-1.34)

1.06

¦nth" plant, 18% capital contingency
Auxiliary equip, only, no major equip.
37.2% of direct
2 years

$1S/ton (limestone)
$6.50/wet ton

90%

65%

30 years
Medium
1.06

"nth" plant 18% capital contingency
Auxiliary equip, only, no major equip.
37.2% of direct
3 years

HYDRATED
LIME

Figure 1. Conceptual Diagram of Advanced Coolside Desulfurization Process.

101-13


-------
Figure 2. Schematic of Advanced Coolside Pilot Plant.

SORBENT UTILIZATION, %

65	75

BY GAS ANALYSIS

Figure 3. Parity Plot of % Sorbent Utilizations by Gas Analysis
and by Solids Analysis for Recycle Tests.

101-14

A


-------
Figure 4. Initial Pilot Test Data For Contactor Simplification. Droplet Surface
Area is Relative to that at Design Conditions.

GAS
FLOW

ii

—a

'M*

/ < . v

MIST
ELIMINATOR

Figure 5. Schematic of Simplified Contactor Design.

101-15


-------
. -i?i¦£$£'

"	fc

¦

•v .... . .

¦ i!##l
¦ ¦ - .

J*"

»»

RELATIVE SPECIFIC DROPLET SURFACE AREA

Figure 6. Optimization Test Results Showing Relative Humidity vs.
Relative Droplet Surface Area.

96	99	190

PERCENT RELATIVE HUMIDITY

Figure 7. Optimization Test Results Showing Fly Ash Capture Results for
Tests with >95% Relative Humidity.

(Fly ash capture was measured upstream of the mist eliminator.)

101-16


-------
Figure 8. Schematic of Advanced Cooiside Conceptual Design
Used in Interim Economic Study.

Figure 9. Schematic of Limestone Forced Oxidation Wet FGD Process
Design Used in Economic Study.

101-17


-------
WET
FGD

ADVANCED
COOLSIDE

300

5

m

250

CO
O

o

g

CL
<
o

200

150

100

50











i

* • • m











* m „

• a «

* • • •

»









2.5 % S
COAL











100

200

300

400

500

600

PLANT GROSS MW

Figure 10. Comparison of Capital Costs for Advanced Coolside and Wet Limestone
Forced Oxidation FGD at 2.5% Coal Sulfur Content and Varying Plant Sizes.

WET
FGD

ADVANCED
COOLSiDE

250

100 150 200 250 300 350 400 450 500 550 600
PLANT SIZE, GROSS MW

Figure 11. Comparison of Levelized S02 Control Costs for Advanced Coolside and
Wet Limestone Forced Oxidation FGD as a Function of
Coal Sulfur Content and Plant Size.

101-18


-------
Capital Charges = 53.7%	O&M Charges = 46,3%

Figure 12. Breakdown of Levelled S02 Control Costs for
262 MW/2.5% S Coal Case.

(Capital costs are effective costs, including capital maintenance.)

101-19


-------
ECONOMIC ANALYSIS OF FGD BY-PRODUCT DISPOSAL ALTERNATIVES

D, Lynn Forster
Department of Agricultural Economics and Rural Sociology
Hie Ohio State University

2120 Fyffe Road
Columbus, Ohio 43210

Jonathan Rausch
Cornell Extension Service
0949 Larsen Road
Geneva, NY 14456

Abstract

A mathematical programming model of electric utility by-product disposal alternatives
is developed. Least cost disposal includes widespread use of the by-product on
cropland; however, the vast majority of the material is landfilled. Agricultural use
represents substantial cost savings to utilities.

Introduction

Acid rain has long been suspected as a cause of the deterioration of streams, lakes,
forests, soils, and various fabricated structures. These resources are adversely effected
by acidic precipitation linked to increased sulfur dioxide and nitrogen oxides emissions.
Acid rain is formed when sulfur dioxide and nitrogen oxide react with other chemicals
in the atmosphere (Helme and Neme, 1991). The primary source of sulfur dioxide and
nitrogen oxide, as identified by the United States Environmental Protection Agency, is
associated with the combustion of coal used in the production of electricity (Helme
and Neme, 1991). The amount of sulfur dioxide produced depends upon the sulfur
content of the coal being combusted. Coal higher in sulfur inevitably produces higher
concentrations of sulfur dioxide and precipitation that is more acidic.

102-1


-------
The actual or potential degradation of resources by add rain are vast. For example, in
the Adirondack Mountains up to 15 percent of the medium to large lakes, those lakes
greater than ten acres, are chronically acidic due primarily to add rain, and 25 percent
of small late are likewise effected (Helme and Neme, 1991). Hie National Add
Predpitation Assessment Program (NAPAP) has estimated that nearly 20 percent of
the nation's lakes and streams have little or no add-buffering capacity, thus are
susceptible to current and future addification,

Sulfiir dioxide (SQ,) emissions are concentrated primarify along the OMo River Valley.
Forty-four percent of the U.S. S02 emissions are produced in this region by Ohio,
Indiana, Pennsylvania, Illinois, West Virginia, with the indusion of Missouri and
Tennessee, in addition, four of the five highest S02 producers are also among the top
ten nitrogen oxide (NOJ producing states (EPA, 1986). Clearly, the OMo River
Valley is a major producer of emissions associated with acid rain, and sipiificantly
impacted by legislation mandating emission standards.

Emission Abatement Policy

Title IV of the 1990 Clean Air Act addresses sulfiir dioxide, nitrogen oxide, and
particulate matter emissions assodated with the burning of fossil fuels. This legislation
mandates a 10 million ton (40 percent) reduction in the nation's sulfiir dioxide
emissions (based upon 1980 emission levels) by the year 2000, and a two million ton
reduction in nitrogen oxide (Claussen, 1991).

The add rain program developed by the EPA under this title sets a ceiling on sulfur
dioxide emissions from electric power plants and allows individual utility companies to
determine the most cost effective means of achieving these new mandates.

Compliance is expected to be achieved through conservation efforts, using fuels lower
in sulfur, purchasing of emission allowances, retrofitting existing plants with pollution
control devices, and/or a combination of the above.

Currently the only pollution reduction technology which can be used on existing power
plants to reduce S02 emissions to mandated levels is flue gas desulfurization (FGD).
Through the use of a sorbent, such as limestone, exhaust gases are "scrubbed" of S02.
One such process is referred to as a dry scrubber process. These FGD technologies
are capable of reducing S02 emissions by as much as 95 percent from current power
plant emissions (EPA, 1986). However, this process of "scrubbing" creates another
environmental concern-disposal of the used sorbent.

Not all power plants are expected to convert to dry injection technology. Of the 23
power plants identified in Ohio, potential converters to dry injection FGD technology
by the 2000 were estimated. In this analysis, a least cost transportation model was
developed depicting the least cost distribution of dry FGD by-product material from

102-2


-------
these potential electric power generating sites to numerous end use alternatives. The
scenario assumes six power plants retrofit with this dry injection technology. This
analysis is only preliminary since options regarding S02 emissions reduction at the
plants are still being weighted.

Proposed Uses of Dry FGD By-Product

LandfTlling of this by-product material is the current means of disposal. However, it
has been suggested that dry FGD by-product material has chemical properties which
makes it valuable as an agricultural lime substitute, soil amendment in coal surface
mine reclamation, and highway embankment construction material This analysis
focuses on dry FGD by-product use in agriculture land application, current coal
surface mine reclamation, and landfilling.

Agricultural Land Application

FGD by-product has many characteristics similar to agricultural lime, and it is
expected that dry FGD by-product materials could be a close substitute for agricultural
lime. Annual usage of agricultural lime was estimated for each Ohio county. These
estimates were based on annual sales of agricultural lime in 1987-92. Since FGD by-
product has 60 percent the neutralizing potential of agricultural lime, agricultural lime
sales were adjusted to estimate FGD by-product demand. That is, FGD by-product
has a lower total neutralizing potential (TOP), thus higher application rates are
necessary to achieve the same results as agricultural lime. The estimated quantity of
FGD by-product demanded reflects these adjustments in TNP.

Transportation of the FGD by-product is expected to be similar to that of agricultural
lime. Only slight modification of existing equipment is necessary to transport the by-
product from the source (power plant) to the destination (farm, coal mine or landfill),
and to apply the by-product to agricultural land. Under these assumptions, FGD by-
product is expected to be hauled from the power plant to various locations throughout
the state by tracks. Once the by-product has reached the farm it is expected that
conventional lime spreaders will apply this by-product to agricultural lands.

Strip/Surface Coal Mine Reclamation

Coal surface mine operations are required to reclaim lands which have been mined.
During the reclamation phase lime is often used to return the mined spoils back to a
pH level conductive to plant growth. It is expected that FGD by-product materials can
also be used for this same purpose.

102-3


-------
The estimated quantity of FGD by-product material demanded to meet current surface
mine reclamation work was based on the estimated area of surface mining in each
county. These estimates are derived from data reporting tons of coal sold in each
OMo coal mining county (1989). From these estimates an average number of tons per
acre of coal extraction were used to estimate the number of acres displaced by surface
coal mining in a given year. Based upon the estimated number of surface acres mined
and an application rate of 20 tons per acre, an estimate for the quantity of FGD by-
produce demanded can be determined. Note, the application rate is that which is
expected to be used in experimental field work.

The final end use alternative identified is that of landfffling. It is expected that FGD
by-product will be landfilled in the event that the available quantity of the product is
larger than its economical use in agricultural and/or coal surface mine reclamation. It
is farther assumed that landfiUmg is not constrained in the quantity of the by-product
that ran be accepted, and that landfills are in close proximity to the power plants or
sources of FGD by-products.

Objectives of this Research

The objectives of this research are (1) to develop a model to identify the least cost
disposal methods of FGD by-product among the three stated alternative end uses from
the producer or power plants perspective, (2) estimate the quantity of by-product used
in each alternative, and (3) estimate the shadow price associated with each alternative
end use. This particular component of the research does not address the social
amenities/disamenities associated with FGD by-product disposal. Examples of these
omitted off-site impacts include: deterioration of roads and bridges from increased
traffic; property value pin/loss from landfill activities, surface mine reclamation, and
abandon mine land reclamation; and increased/decreased ground and/or surface water
quality from surface mine reclamation and abandon mine land reclamation, or
landfilling activities. However, work is in progress to quantify these
amenities/disamenities. Ones quantified the least cost disposal model can be re-
estimated with the appropriate increase/decrease in cost associated with the given end
use alternative. That is, the model can be modified to consider the social gain/loss
associated with each end use alternative, providing a least cost disposal model which
captures both a private and social accounting stance.

Model Development

Minimizing the total cost of transporting a product from some production point to
various destination or demand points can be done through a series of linear equations.
This type of least cost mathematical modeling, or transportation modeling, has been
applied to the distribution of FGD by-product material to various destinations

102-4


-------
throughout Ohio. This least cost transportation model is based upon six source nodes
(power plants) associated with the production of dry FGD by-product materials and its
use as a substitute for agricultural lime in 88 Ohio counties, as a soil amendment in 21
Ohio coal surface mining counties, and disposal in six landfill sites located in proximity
of the power plant

To formulate this model mathematically, the following terms are defined.

Zi = the number of tons of FGD by-product material available at the power
plant or source i, i=l,2,...,m;

bj = maximum number of tons of by-product required at each destination or
alternative use (e.g., county for agricultural lime, reclamation site, or landfill), j
= 1,2,...n;

Cjj = unit transportation and application cost from each source i to each
destination j, (i « l,2,...,m; j ~ l,2,...,n).

E aj = E b^ the quantity of FGD by-product available at source i must be equal
to the demand for FGD by-product material in alternative end uses (including
landfilling options).

aj > 0; supply of by-product at each source node is positive.

bj > 0; demand for by-product at each demand node is positive.

The problem then becomes determining the amount of FGD by-product material
shipped to each of these alternative end uses, given that the cost of distribution and
application of the by-product is known or can be estimated. Thus, the decision
variable, x^, equals the number of tons of by-product material shipped from each
source i to each destination j annually given some cost per unit shipped.

Hie transportation model estimated is:

Assuming:

N N

Minimum Cost = ^ J) c.x~

(1)

i /

subject to:

E Xjj = ai (i = 1,2,...,m);

£ Xg < bj (j = l,2,...,n);

Xjj > 0 (i = l,2,...,m; j = l,2,...,n).

(2)

(3)

(4)

102-5


-------
Equation (1) represents the minimization of total distribution costs, assuming a linear
cost structure for shipping, processing, and application of the dry FGD by-product
material. Equation (2) shows that the quantity of by-product shipped from each source
to all alternative end use destinations most be equal to the quantity of by-product
materials available. Equation (3) states that the quantity of by-product shipped from
all sources to each destination must be less than or equal to the maximum quantity of
by-product demanded at that destination. Finally, equation (4) indicates that the
quantity of by-product shipped from each source to each destination must be greater
than or equal to zero.

Data Used for Estimating the Transportation Model Using Six Power Plants

All estimates pertaining to the quantity of dry FGD by-product demanded at various
demand nodes for the two end use alternatives (agricultural and surface coal mine
reclamation uses) have been adjusted for a 25 percent rate of adoption. It is expected
that not all individuals using agricultural lime or reclaiming surface coal mines will
completely adopt this new technology. Thus, the model uses a conservative or lower
rate of adoption. However, it is important to note that the model can be re-run at
various levels of adoption.

linear distances from each power plant or source of diy FGD by-product to the center
of each county were estimated. There were a total of 528 possible agriculture routes
(88 counties supplied from six plants), and 126 possible coal reclamation routes (21
counties supplied from six plants).

Transportation costs associated with moving the specified distance were estimated. In
the case of agricultural land application, cost estimates were derived from the
agricultural lime industry. It is expected that the dry FGD by-product will be
transported in much the same manner as current agricultural lime. Thus, an estimate
of $0.10 per ton per mile was used. In addition to moving the product from the source
to the destination an application expense is incurred. Again, the application of the diy
FGD by-product is expected to be similar to agricultural lime, which 1ms an estimated
application charge of $3.50 per ton. This expense includes the spreader and
mechanism to load the product from a pile onto the spreader. Therefore,
transportation costs were calculated at $0.10 per ton per milk, then an additional $3 JO
per ton was added to each for ejected application costs.

Transportation of the diy FGD by-product to coal surface mine reclamation sites is
also expected to cost $0.10 per ton per mile. Application of the dry FGD by-product
is expected to be at significantly higher levels, potentially 20 tons per acre, than
application rates associated with agricultural use. Thus, different equipment is
expected to be used in distributing the by-product (e.g., a bulldozer or equivalent type

102-6


-------
of reclamation machinery), but application costs would not vary significantly and were
estimated at $3.50 per ton.

Cost estimates for landfilling were obtained from interviews with representatives of
electric utilities. All landfilling activities are regulated, and industry representtives
expected stringent landfilling requirements. Estimated costs were about $27 JO per ton
of material to landfill dry FGD waste and meet expected EPA guidelines.

Results

Quantities of dry FGD by-product shipped from the six power plants to each county
for agricultural uses, surface mine reclamation, and landfilling are shown in Table 1.

Figure 1 shows the percentage distribution of dry FGD by-product produced at six
Ohio power plants among the three end use alternatives. Notice that at five of the
power plants (Mountainer [Mount], Muskingum [Musk], Sammis, and Burger) at least
85 percent of the dry FGD by-product is landfilled. At two of the six power plants
(Edgewater [Edge] and Niles) nearly all by-product material is applied to agricultural
land. Surface coal mine reclamation accounts for negligible quantities of total FGD
by-product produced (1.03%) from all six power plants.

Figure 2 shows the quantity of dry FGD by-product distributed among the three end
use alternatives from each source. Of the total quantity of dry FGD by-product
produced, assuming a 25% adoption rate, little dry FGD by-product is used in
agriculture (8.87%) or surface coal mine reclamation (1.03%), and most is landfilled
(90.1%). This suggests that (a) a higher adoption rate among farmers and surface
mine operators would reduce the quantity of landfilled, (b) the amount of dry FGD
by-product produced greatly exceeds current quantities demanded or (c) additional end
uses for the by-product must be found. Based upon the estimated total output of dry
FGD by-product from these six power plants to be 4.1 million tons annually, current
agricultural land application and coal surface mine reclamation use at 100% adoption
is estimated to be about 1.7 million tons per year. This suggests that 2.4 million tons
of dry FGD by-product must be either landfilled or some alternative use must be
found assuming complete substitution of dry FGD by-product for agricultural and
surface coal mine reclamation.

Figure 3 shows the distribution of dry FGD by-product as a substitute for agricultural
lime from these six power plants. Quartiles were developed based upon the quantity
of dry FGD by-product estimated to be shipped to each Ohio county for agricultural
land application. The first quartile, or counties using the lowest 25% of dry FGD by-
product, are not shaded. Counties with shading indicate the second lowest 25% of
agricultural use through the largest 25%.

102-7


-------
Figure 4 shows a quartiled distribution, of dry FGD by-product for coal surface mine
reclamation. Since only 21 counties have coal surface mine activity, all quartiles are
shaded, and represent the lowest to highest quantities of dry FGD by-product used.
Again, the quantity of dry FGD by-product associated with a given quartile are in
hundreds of tons, and represent one-fourth of the total dry FGD by-product shipped
for coal surface mine reclamation from the six power plants selected.

The final geographic representation of dry FGD by-product distribution shows both
agricultural land application and coal surface mine reclamation, or a quartiled total
distribution of dry FGD by-product (Figure 5). Each quartile represents one-fourth of
the dry FGD by-product used as an amendment to land (both agricultural and coal
surface mine declamation lands). For example, four counties (Belmont, Coshocton,
Noble, and Tuscarawas) within the fourth quartile receive one-fourth of the total dry
FGD by-product applied to land.

The final objective of this research was to estimate the shadow price associated with
each demand or destination node. Shadow prices are imputed prices which reflect the
increase/decrease in total costs if one additional unit of product were available. In
this case, the additional unit is the use of one additional ton of dry FGD by-product on
agricultural land or in coal surface mine reclamation. Shadow prices for diy FGD by-
product are calculated as the difference between landfilling (non-binding constraint)
and agricultural land application or coal surface mine reclamation options (both
binding constraints). It would be expected that as the distance from the power plant
increases, the cost to move this by-product also increases, therefore the imputed value
of shadow price for binding end use options father from the source or power plant
would be lower. That is, the difference between landfilling and shipping dry FGD by-
product material greater distances would be smaller. Thus, counties located father
from dry FGD by-product sources would have lower shadow prices or lower cost
savings to the utility companies than would land application sites closer in proximity to
the power plant Figure 6 shows the shadow price associated with the distribution of
dry FGD by-product to each Ohio county.

Another interpretation of these shadow prices is the amount the power plant would be
willing to pay for the disposal of an additional ton of by-product in each end use
alternative. For example, Figure 6 shows Williams County in the first quartile or
having a shadow price between $3.29-15.01 per unit. Hie calculated shadow price for
Williams County is $3.29 per ton, suggesting the power plant would be willing to pay
the fanner up to $3.29 to use an additional ton of dry FGD by-product as opposed to
landfilling it at a cost of $27.50 per ton.

102-8


-------
Conclusion

Under Title IV of the 1990 Clean Air Act, electric power generating plants will be
required to reduce S02 and NOx emissions by about 40 percent no later than the year
2000. These emission standards are most significant for power plants which burn coal,
particularly high sulfur coal, as an energy source. In order to achieve compliance with
these mandates, power plants will have to reduce the quantity of coal burned, use fuels
lower in sulfur, purchase emission allowances, retrofit existing power plants with dean
air technology, or a combination of the above. Currently the only clean air technology
available to existing power plants is Hue Gas Desulferization (FGD) technology. EPA
has estimated that this technology can reduce S02 by as much as 95 percent.

However, dry FGD technology creates another environmental concern—disposal of the
used sorbent. Based upon current coal consumption estimates, Ohio could potentially
produce nearly 4 million tons of diy FGD by-product (used sorbent) annually from six
Ohio power plants. The objective of this research was to estimate a least cost disposal
model for the movement of this by-product to various geographic locations throughout
Ohio and for use as a soil amendment for agricultural land, coal surface mine
reclamation, and landfilling. In doing so, total disposal costs and quantities were
derived as well as shadow prices for each county (demand node) identified.

This analysis used one scenario for future dry FGD by-product distribution and uses.
Others are plausible and easily analyzed by this model. The linear transportation
model estimated suggests that the least cost disposal of diy FGD by-product would
cost $107 million to dispose of 4.1 million tons of diy FGD by-product material
annually. This is equivalent to average cost of $26.10 per ton of dry FGD produced.
In addition to landfilling, the two end use alternatives currently identified are dry FGD
by-product use as soil amendments on agricultural land and coal surface mine
reclamation. Of these three options, landfilling accounts for about 90 percent of the
total by-product produced, while agricultural and surface coal mine reclamation
account for about 8.87% and 1.03% respectively. Therefore, on average land
application of dry FGD by-product represents a savings of $1.40 ($2750 [landfilling]-
$26 .10 [average cost of disposal per ton]). However, at the margin (one additional
ton), land application represents a cost savings of about $14.81 per ton. That is,
electric power plants can reduce total cost of dry FGD by-product disposal by $14.81
per ton through land application rather than landfilling.

In this analysis, land spreading dry FGD by-product is dispersed widely over the state.
Agricultural use is concentrated in the western two-thirds of Ohio, while coal surface
mine reclamation use is important in the eastern one-third of the state. Yet, of the
alternative disposal options identified, the vast majority (3.8 million tons) of dry FGD
by-product is landfilled, followed by agricultural land application (374,276 tons) and
then coal surface mine reclamation (42,507 tons). Given these end use alternatives,
more FGD by-product will be buried in landfills than will be used in other alternatives
combined. However, electric utilities have an enormous economic incentive to supply

102-9


-------
dry FGD by-product to land. They could pay farmers substantial amounts and save
lanrlfflTing costs.

Acknowledgements

This research was conducted as part of the "Land Application Uses for Dry FGD By-
products" project which is a cooperative project of the Ohio Agricultural Research and
Development Center, Hie Ohio State University, The U.S. Geological Survey and the
Dravo lime Company. Funding support for this project was obtained from the Ohio
Coal Development Office (Columbus, OH) Grant No. CDO/D-89-35, The U.S.
Department of Energy (Morgantown Energy Technology Center, Morgantown, WV)
Award No. DE-FC21-91MC2806G, Dravo Lime Company (Pittsburgh, PA) Grant No.
RF768342, Electric Power Research Institute (Palo Alto, CA) Grant No. RP2796-02,
Ohio Edison Company (Akron, OH), American Electric Power (Columbus, OH), and
The Ohio State University (Columbus and Wooster, OH).

References

EPA Journal. "Acid Rain: An EPA Journal Special Supplement" Office of Public
Affairs, EPA Journal, Washington, DC, June/July 1986, Vol 12, No. 5.

Ned Helme and Chris Neme. "Acid Rain: The Problem." Office of Communication
and Public Affairs, EPA Journal, Washington, DC, Jan./Feb. 1991, Vol. 17, No. 1.

Eileen Claussexi. "Acid Rain: The Strategy." Office of Communication and Public
Affairs, EPA Journal, Washington, DC, Jan./Feb. 1991, Vol. 17, No. 1.

102-10


-------
Table 1

Estimated Tons of Dry FGD By-Product Shipped to Each End Use Alternative From
Each. Source or Power Plant



Mountain-

Edge-

Musk-







Total



eer

water

ingum

Niles

Sammis

Burger

Ship-

County

S25

Sll

S16

S9

S8

S12

ped

Agricultural Land Application













Adams

6,718

0

0

0

0

0

6,718

Allen

5,091

0

0

0

0

0

5,091

Ashland

0

0

0

1,821

978

0

2,799

Ashtabula

0

0

0

5,599

0

0

5,599

Athens

0

0

1,258

0

0

0

1,258

Auglaize

8,726

0

0

0

0

0

8,726

Belmont

0

0

0

0

0

4,921

4,921

Brown

6,523

0

0

0

0

0

6,523

Butler

2,021

0

0

0

0

0

2,021

Carrol

0

0

0

0

2,041

0

2,041

Champaign

4,048

0

0

0

0

0

4,048

Clark

3,194

0

0

0

0

0

3,194

Clermont

4,998

0

0

0

0

0

4,998

Clinton

7,002

0

0

0

0

0

7,002

Columbiana

0

0

0

0

6,853

0

6,853

Coshocton

0

0

0

0

0

11,259

11,259

Crawford

0

0

0

0

0

6,857

6,857

Cuyahoga

0

3,989

0

0

0

0

3,989

Darke

6,290

0

0

0

0

0

6,290

Defiance

0

11,016

0

0

0

0

11,016

Delaware

0

0

9,199

0

0

0

9,199

Erie

0

6,586

0

0

0

0

6,586

Fairfield

0

0

9,724

0

0

0

9,724

Fayette

5,612

0

0

0

0

0

5,612

Franklin

2378

0

0

0

0

0

2,378

Fulton

0

9,404

0

0

0

0

9,404

Gallia

3,917

0

0

0

0

0

3,917

Geauga

0

0

0

313

0

0

313

Greene

4,418

0

0

0

0

0

4,418

Guernsey

0

0

0

0

0

2,827

2,827

Hamilton

638

0

0

0

0

0

638

Hancock

0

0

3,577

0

0

0

3,577

Hardin

0

0

7,457

0

0

0

7,457

102-11


-------
Table 1 (eont.)



Mountain-

Edge-

Musk-







Total



eer

water

ingum

Mies

Sammis

Burger

Ship-

County

S25

Sll

S16

S9

SS

S12

ped

Harrison

0

0

0

0

456

0

456

Henry

0

5,745

0

0

0

0

5,745

Highland

14,993

0

0

0

0

0

14,993

Hocking

499

0

0

0

0

0

499

Holmes

0

0

0

0

0

3,237

3,237

Huron

0

0

0

16,983

0

0

16,983

Jacikson

109

0

0

0

0

0

109

Jefferson

0

0

0

0

1,373

0

1,373

Knox

0

0

8,513

0

0

0

8^13

Lake

0

0

0

1,539

0

0

1^539

Lawrence

148

0

0

0

0

0

148

licking

0

0

5,395

0

0

0

5395

Logan

1,009

0

0

0

0

0

1,009

Lorain

0

431

0

3,812

0

0

4,243

Lucas

0

4,770

0

0

0

0

4,770

Madison

712

0

0

0

0

0

712

Mahoning

0

0

0

2,438

0

0

2,438

Marion

0

0

1,903

0

0

0

1,903

Medina

0

0

0

4,879

0

0

4,879

Meip

116

0

0

0

0

0

116

Mercer

2,534

0

0

0

0

0

2£34

Miami

2,268

0

0

0

0

0

2,268

Monroe

0

0

0

0

0

370

370

Montgomery

611

0

0

0

0

0

611

Morgan

0

0

628

0

0

0

628

Morrow

0

0

1,068

0

0

0

1,068

Muskingum

0

0

5,665

0

0

0

5,665

Noble

0

0

398

0

0

0

398

Ottawa

0

6,070

0

0

0

0

6,070

Paulding

6,170

0

0

0

0

0

6,170

Perry

0

0

1,241

0

0

0

1^241

Pickaway

2,946

0

0

0

0

0

2,946

Pike

5,062

0

0

0

0

0

5,062

Portage

0

0

0

3,483

0

0

3,483

Preble

2,660

0

0

0

0

0

2,660

Putnam

0

0

2,501

0

0

0

2^01

Richland

0

0

0

0

0

4,119

4,119

Ross

1,630

0

0

0

0

0

1,630

102-12


-------
Table 1 (eont.)



Mountain-

Edge-

Musk-







Total



eer

water

ingum

Niles

Sammis

Burger

Ship-

County

S25

Sll

S16

S9

S8

S12

ped

Sandusky

0

3,019

0

0

0

0

3,019

Scioto

454

0

0

0

0

0

454

Seneca

0

5,542

0

0

0

0

5,542

Shelby

1,962

0

0

0

0

0

1,962

Stark

0

0

0

0

2,381

0

2^81

Summit

0

0

0

1,543

0

0

1,543

Trumble

0

0

0

3,846

0

0

3,846

Tuscarawas

0

0

0

0

2,695

0

2,695

Union

1,948

0

0

0

0

0

1,948

Van Wert

1,115

0

0

0

0

0

1,115

Vinton

153

0

0

0

0

0

153

Warren

1,720

0

0

0

0

0

1,720

Washington

0

0

1,963

0

0

0

1,963

Wayne

0

0

0

0

11,462

0

11,462

Williams

0

6,405

0

0

0

0

6,405

Wood

0

7,023

0

0

0

0

7,023

Wyandot

0

0

0

0

0

5,724

5,724

102-13


-------
Table 1 (cant.)



Mountain

Edge-

Musk-







Total



-eer

water

ingum

Niles

Sammk

Burger

Ship-

County

S25

Sll

S16 "

S9

S8

S12

ped

Coal Surface Mine Reclamation and Landfilld Alternative





Athens

0

0

312

0

0

0

312

Belmont

0

0

0

0

0

5,566

5,566

Carroll

0

0

0

0

1,637

0

1,637

Columbiana

0

0

0

0

1^68

0

1,268

Coshocton

0

0

0

0

0

3,904

3,904

Guernsey

0

0

0

0

0

287

287

Harrison

0

0

0

0

4,262

0

4,262

Hancock

262

0

0

0

0

0

262

Holmes

0

0

0

0

0

933

934

Jackson

2,049

0

0

0

0

0

2,049

Jefferson

0

0

0

0

3,954

0

3,954

Lawrence

55

0

0

0

0

0

55

Mahoning

0

0

0

409

0

0

409

Muskingum

0

0

1,858

0

0

0

1,858

Noble

0

0

6,917

0

0

0

6,917

Peny

0

0

841

0



0

841

Stark

0

0

0

0

404

0

404

Tuscarawas

0

0

0

0

4,343

0

4,343

Vinton

3,047

0

0

0

0

0

3,047

Washington

0

00

135

0

0

0

135

Wayne

0

0

0

0

66

o

0

66

landfill

757,524

0

869,447

0

1,689,128

386,656

3,702,755

Summary of Shipment Estimates











Total Ag.

120,393

70,000

60,490

46,256

37,823

39,314

374,276

Total Reclaim

5,413

0

10,063

409

15,933

10,690

42^07

Total Landfill

757,524

0

869,447

0

1,679,544

386,656

3,693,172

Total

883,330

70,000

940,000

46,665

1,733300

436,660

4,109,955

% Ag.

13.63

100.00

6.44

99.12

2.18

9.00

8.87

% Reclaim

0.61

0

1.07

0.88

0.92

2.45

1.03

% T-a-nrtfjH

85.76

0

92.49

0

96.90

88.55

89.86

% Total

100.00

100.00

100.00

100.00

100.00

100.00

100.00

102-14


-------
Percentage Distribution of FGD
By-Product By Source

1.2-T

Mount

S25

Edge

S11

Musk Nifes Sammis Burger
Si 6 SB	SB	S12

m % Ag. Use p//j % Reclainn Use ESS % Lancfi

Figure I. Percentage of Total FGD By-Product Produced and Distributed Among
the Three End Use Alternatives Selected by Plant

102-15


-------
Quantity Distribution Of FGD By-Product
By Source

-i.su	

Mount: Edge Musk NUes Sammis Burger
S25	S11	S16	S9	SB	S12

B51 to Use W7k Reclaim Use g^LsncfB

Figure 2, Quantity of FGD By-Product Produced and Distributed Among the Three
Stated End Use Alternatives % Selected Ohio Power Plants

102-16


-------
Figure 3, Quartiled Distribution of Agricultural Land Application of FGD By-
Product Assuming a 25% Rate of Adoption

102-17


-------
OinKlied distribution ©f
FCD hy product %m*4 m
curfKi eeal xnlna rKliiBition

Fini Qwilla
(S47-2JMB tons)

StcofMi Qt2»rill«

C3^W7-3^04 ton*)

Third Quuiflc
(3JiS3*4,34Z ions)

Fourth fiunifla
C5.S85-M1? tons)

Figure 4. Coal Surface Mine Reclamation Use of Dry FGD By-Product Assuming a
25% Adoption Level

102-18


-------
Figure 5. Agricultural Land Application and Coal Surface Mine Reclamation Use
of Dry FGD By-Product Assuming a 25% Level of Adoption

102-19


-------
Figure 6. Shadow Prices for Land Application or Electric Utilities Estimated Cost
Savings From Land application Versus Landfflling Dry FGD By-Product

102-20

A


-------
CONVERSION FROM DISPOSAL TO
COMMERCIAL-GRADE GYPSUM

An Alternative Approach for Disposal of Scrubber Wastes

Stanley K. Conn

Owensboro Municipal Utilities

4301 U.S. Highway 60 E
Owensboro, Kentucky 42303

Michael G. Vacek
Sargent & Lundy
55 East Monroe
Chicago, IL 60603

John T. Morris, Jr.
Wheelabrator Air Pollution Control
441 Smithfield Street
Pittsburgh, Pennsylvania 15222

Introduction

Owensboro Municipal Utilities' (OMU) Elmer Smith Units 1 and 2 were identified as
Phase 1 units in the 1990 Clean Air Act Amendments. To comply with the
amendment requirements, OMU opted to install a wet flue gas desulfurization (FGD)
system producing disposal-grade gypsum. The FGD system is common to both units.
Project engineering, procurement, and construction activities were proceeding on
schedule when difficulties were encountered with the development of a suitable
disposal site. These difficulties caused OMU to reevaluate its overall station waste
disposal plan. OMU's evaluation considered continuing with the development of an
OMU landfill, contracting for a third party to dispose of the waste in an engineered
landfill and modifying the process to produce a commercial-grade gypsum.
Ultimately, the FGD process equipment was modified such that gypsum could be
used for wallboard production.

103-1


-------
This paper summarizes events leading to this decision and OMU's economic
evaluation of the disposal alternatives- Also discussed are the technical
considerations for modifying the process and equipment design to produce a
commercial-grade gypsum. The revised process equipment design and arrangement
is also presented in the paper.

Background

The OMU Elmer Smith Station consists of two coal-fired generating unite with a total
capacity of 441 MW, Unit 1 is a cyclone-fired unit with a gross capacity of 151 MYV.
Unit 2 is a tangentially-fired pulverized coal unit with a gross capacity of 290 MW.
Both units fire coal from various sources in the Illinois basin. The sulfur content of
the coal ranges to a maximum of 6.0 lb/MBtu.

The work at the Elmer Smith Station to comply with the Clean Air Amendments is
identified as the Clean Air Project The Clean Air Project includes replacing the
existing electrostatic dust precipitators on each unit and installing a wet FGD system
that is common to both, unite. In addition, a low-NOx burner system will be installed
on Unit 2 to comply with the NOx provisions of the amendments.

Design activities were initiated on the project in March 1991 when Sargent & Lundy
(S&L) was selected as the architect-engineer (AE) for the project In November 1991,
Wheelabrator Air Pollution Control Corporation (WAPC) was awarded three
contracts to provide the FGD system, electrostatic precipitators and an acid brick-
lined chimney. WAPC's scope of work on these contracts is to design, furnish, and
install all "process-related" equipment for the project. Sargent & Lundy, as the AE, is
to provide balance-of-plant equipment facilities design such as foundations, structural
steel, building enclosures, all electrical facilities, control system, limestone handling
system, ash handling system modifications, booster fans, and station services (water,
air, etc.). A summary of the design features of major equipment associated with the
project is shown on Table 1.

In parallel with the S&L and WAPC efforts to design the on-site facilities, OMU also
contracted with Fuller, Mossberger, Scott and May (FMSM) of Lexington, Kentucky,
to locate a suitable disposal site for the waste gypsum and develop the site. After
several site selection screenings, OMU secured options on a 400-acre site that is
located 15 miles from the Elmer Smith Station. The design of the landfill was based
on a state-of-the-art lined landfill facility with a leachate collection system. Design
and permit preparation activities for the facility had been initiated when, in April
1992, pressure from a citizens group, local to the landfill site, caused OMU to
consider disposal alternatives other than an OMU-owned landfill.

To access the potential "market" for plant by-products, OMU and FMSM developed a

request for proposal (KFP) that was issued to landfill operators, bulk material
contractors and wallboard manufacture's. The RFP emphasized maximum utilization

103-2


-------
Table 1

Summary of Project

Electrostatic Precipitators

•	Unit 1

Type
SCA

Outlet emissions

•	Unit 2

Type
SCA

Outlet emissions
Reactant Preparation

•	Type

•	Number/capacity

•	% capacity each

•	Fineness
Absorbers

•	Number/capacity
Type/additives
S02 removal efficiency
L/G ratio

Material of construction
Primary Dewatering
Type

Solids in/out
Secondary Dewatering
Vacuum filter type
Number/capacity
Size

Filter area

Vacuum rapacity/hp
Solids at discharge

Design Features

Rigid electrode
238 tf/IOOO acfm
0.2 Ib/MBtu

Rigid electrode
266 ft2/"! 000 acfm
0.2 Ib/MBtu

Wet ball mills
2/16t/h
75%

95% through 325 mesh

2/67%

Limestone spray tower/none

96%

132 gpm/1000 acfm
317 LMN stainless

Thickener
12%/30%

Rotary drum
3/50%

6'0 x 8'L
151 ft2 each
530 acfm/30
75%

103-3


-------
of all plant by-products including bottom ash and fly ash. The RFP addressed both
disposal-grade gypsum that the process was currently designed to produce and
wallboard grades of gypsum that could be made in the system with modifications to
equipment. Two purity levels of gypsum were outlined in the RFP based on the
expected range of impurities in the potential limestone sources.

Bids were received in response to the RFP in July 1992. The bids encompassed total
by-product management (ash and gypsum marketing with disposal of balance) and
smaller scopes such as gypsum utilization with transportation by others. A total of
14 bids were received, and 7 were short-listed: 2 for total management, 3 for disposal
only, and 2 for gypsum utilization. An addendum was issued to the short-listed
bidders to obtain final pricing.

To support the OMU and FMSM effort, S&L and WAPC were requested to evaluate
technical requirements for wallboard quality gypsum based on technical information
provided by wallboard manufacturers in their bids. A summary of the wallboard
manufacturers' technical requirements are summarized on Tables 2 and 3. The
following sections discuss the technical evaluation of wallboard quality gypsum, the
design modification required to plant equipment to produce wallboard quality
gypsum, the results of the economic evaluation of disposal, and utilization
evaluation.

Table 2

Wallboard Manufacturer Technical Requirements

Mfq. A

Mfq. B

pH
Purity

Size criteria:

Free water

Min.

6

Max.

<10%
8

Desired

0% to 3%
6

97+%

Refected
>10%

8

<94%

Mean part
Surface area
Part area (X x Y)

20 Jim
3500 cm2/g

30 to 50 jim >70, <20p.m

>2000 jim >4000, <1000 Jim

Aspect ratio:

X to Y
Z to Y

Average particle
size (50 percentile)

<0.5, >0.1 <0.05, >0.75
30 to 50 jim >70 <20 Jim

2 to 5

>10

103-4


-------
Table 3

Specific Chemical Constituent Limits

Constituent

Fly ash

Mfg. A
Allowed
MaxJMin.
Limit

1.0%

Mfg. B

Allowed

MaxJMin.

Limit

Calculated
at

Maxiumum
Conditions

0.13% Max. 0.76%
(1250 ppm)

SiO,

1.1%

0.5%

0.41%

Fe.,0.

2W3

1.5%

Not given 0.45%

RA

3.5%

Not given 2.52%

Na

Not given 70 ppm

9 ppm

Mg

Not given 70 ppm

135 ppm

CaS03*%H20 1% max. 0.1%	0.72%

CI	120 ppm 100 ppm 100 ppm

Water soluble 600 ppm Not given Expected

<600 ppm

Worst Operating Conditions

0.2 Ib/MBtu ash loading and
60% fly ash collection in the
absorbers

60% FA collection, 51% Si02
in FA and 1.5% Si02 in LS.

60% FA collection, 43% Fe203
in FA and 0.2% Fe203 in LS.

60% FA collection, 25% Al203
and 43% Fe203 in FA, and
0.2% Al203 and 0.2% Fe203 in
LS.

25-250 ppm from Power
Magazine. Feb., 1988. 0.7%
Na20 in fly ash and 90%
solids and 90% wash
efficiency. 232 ppm Na in
46-gpm blowdown.

50-250 ppm from Power
Magazine. Feb., 1988. Based
on 90% solids, 90% wash
efficiency. 1.5% MgCOs in
limestone and 1 % Mg in ash.
Assumed that 50% of MgCOa
will dissolve.

At 99% oxidation, 97%
limestone utilization and 97%
CaC03

103-5


-------
Technical Feasibility Evaluation
Free Water

Free water is defined as the weight of the noncombined water divided by the total
weight of the wet sample. Excess moisture in the delivered product will require the
wallboard manufacturers to install special handling and drying equipment for the
synthetic gypsum that is not required for natural gypsum. As part of the wallboard
production process, the free water and a portion of the combined water must be
removed. Therefore, wallboard manufacturers prefer the lowest free water levels
possible with rejection levels ranging from 10%-15%. This point is the moisture level
where drying the gypsum is no longer economically feasible. Moisture will be
subject to penalties directly related to the costs of drying the gypsum.

Various vacuum filter options and centrifuges were considered to improve the solids
content of the final product with minimum changes to layout and equipment.
Utilizing rotary drum vacuum filters would provide 85% solids. Indexing, or carrier
belt-type horizontal filters and centrifuges would achieve in excess of 90% solids.
Horizontal belt vacuum filters have outperformed the rotary drum vacuum filters in
producing drier material and can achieve higher wash efficiencies due to counter-
current wash capabilities. A major consideration in the change from disposal-grade
to commercial-grade gypsum was minimizing the cost and schedule impact on the
project. The selection of vacuum belt filters for secondary dewatering equipment was
influenced by the fact that thickeners had been purchased and thickener vendor
engineering was complete. In addition, the water balance and distribution system
was designed without provision for a second high solids reclaim water system.
Technical evaluations indicated that thickeners in combination with vacuum belt
filters would produce an acceptable product. Financial and schedule evaluations
indicated that the thickener filter combination was the most cost-effective. The
indexing belt-type filter was selected over the carrier belt type after analyzing the
projected maintenance costs over a 5-year period. Material that is drier than the 90%
target will have a higher value. However, dusting of the gypsum during handling
and transportation may become a problem if the free water is less than 6%. A
centrifuge could also be used to produce drier than 90% solids. Centrifuges would
be expected to require a higher capital cost than both the rotary drum and belt
vacuum filters. The evaluation concluded that the horizontal indexing belt filter was
the most economic choice since guarantees of 90% solids were provided with
expected production between 91%-92%.

Sargent & Lundy recommended that a 4-day working stack be maintained to allow
for production and shipping flexibility. A storage building would be needed to
control moisture penalties, reduce potential contamination, arid minimize dusting and
product loss. Silo options were also investigated but proved to be more expensive
and less reliable than a more conventional pre-engineered building approach.

103-6


-------
Gypsum Purity

Gypsum purity is defined as the weight of CaS04'2H20 divided by the weight of the
dry sample. The minimum gypsum purity required by wallboard manufacturers is
between 94% and 95%. The purity at this level is primarily a market driven function.
Natural gypsum reserves can run in the 80%-90% range. The major concern for
wallboard production is not the level of gypsum, but is the makeup of the impurities.
The gypsum by-product purity can be optimized by maximizing oxidation, increasing
the available CaC03 in limestone, improving limestone utilization, and reducing fly
ash loading to and removal by the FGD system. Each of these items is discussed
separately.

Oxidation. Oxidation is defined as the total moles of CaS04*2H20 divided by the
sum of the total moles of CaS04*2H20 and CaS03*l/2H20. The higher the oxidation,
the higher the production of CaS04«2H20, which increases the gypsum contribution
to the total sample resulting in a higher purity.

A minimum of 99% oxidation is required to produce 95% pure gypsum. The original
sparger design was based on 98% oxidation of calcium sulfite to calcium sulfate.
WAPC's original design was to force oxidation to occur with one absorber operating
at 67% load. The design air stoichiometry was 2.4. Three 50% compressors were
provided in the original design. In changing the design to commercial-grade
gypsum, a 99.3% oxidation rate is required. In order to achieve the required
oxidation, the air stoichiometry was increased to 3.0. All three compressors must be
operated during MCR conditions to produce the required oxidation required for
commercial-grade gypsum.

The original sparger consisted of 4-inch pipes that were selected to minimize pressure
losses through the pipe to ensure adequate pressure and distribution of air to the
absorber. In changing to commercial-grade gypsum design, some of the sparger
pipes were increased to 6 inches in order to accommodate the higher air flow rate at
the original pipe pressure losses and acceptable distribution.

The current FGD design has a large solids residence time in the reaction tank in
addition to 32 feet of liquid level above the air sparger. OMU's anticipated 80% load
factor will further increase the average residence time. This design may achieve 99%
oxidation without requiring operation of the third compressor. However, provisions
were incorporated to allow three compressor operation.

Available CaC03 in Limestone. The available CaC03in the limestone is defined as
the percent of CaC03 dissolving in a laboratory experiment. A higher CaC03 content
in the limestone will reduce the quantity of inerts and other impurities such as SiOz,
Fe2Os, and R203, which will increase the gypsum purity. Table 4 presents the

103-7


-------
Table 4
Limestone Utilization

Gypsum
Purity. %

94

Available CaC03 in
Limestone, %

96

97

95

Required LS
Utilization, %

97.4

96.4

95.4

95

95

99.2

96

98.1

97

97.0

required limestone utilization at 95% through 97% available CaC03 content to
produce gypsum purities of 94% and 95%. The required limestone utilization
numbers are calculated based on 99% oxidation and 60% fly ash collection by the
absorbers at a loading of 0.2 lb/MBtu. The original limestone utilization guarantee
was based on a minimum 90% CaC03 content in limestone.

Table 4 indicates that the lower the limestone utilization value, the higher the CaC03
content required. For example, to produce a 95% pure gypsum, available CaC03
values of 95% and 97% would require very high utilizations of 99.2% and 97.1%
respectively. The table shows that a maximum gypsum purity value of 94% should
be contemplated to allow a realistic variation in limestone quality.

Limestone Utilization. Limestone utilization is a measure of the available CaC03
that reacts with the collected S02. The higher the limestone utilization, the lower the
unreacted CaC03 and MgC03 in the final gypsum product leading to higher gypsum
purity. Some amount of MgC03 in the limestone may react. The MgC03 credit is not

considered since the amount reacted depends on the reactivity of the specific

Table 5 presents the expected gypsum purity for available CaC03 between 94% and
97% at the design basis limestone utilization. A 96% limestone utilization can be
achieved for limestone with available CaC03 values between 95% and 97%. The
gypsum purity numbers are calculated based on 99% oxidation and 60% fly ash

collection by the absorbers at a loading of 0.2 lb/MBtu. The data indicate that the

limestone.

103-8


-------
Table 5

Expected Gypsum Purity

Available CaC03 in
Limestone Utilization, % Limestone, % Gypsum Purity, %

96

95

93.2

96

96

93.8

96

97

94.4

higher the CaC03 content, the higher the gypsum purity. For example, a limestone
with 97% CaC03 at 96% utilization will produce 94.4% gypsum purity.

A sulfuric acid addition system at the filter feed tank or upstream of the thickeners
could dissolve some of the unreacted carbonates (CaC03 and MgCOs) resulting in a
higher gypsum purity. For example, if 98% CaCOs utilization is achieved by acid
addition then 96% CaC03 limestone can be used to produce 95% pure gypsum. The
handling of acid adds to the complexity of the process. Also, if acid addition is not
monitored properly, corrosion problems can be created. Furthermore, acid addition
will add significant operating costs since 93% H2S04 costs approximately 4
-------
Fly Ashm In general, fly ash from properly operating precipitators will not impact the
wallboard quality of gypsum. Wallboard manufacturers' specified limits are between
0.13% and 1.0% If the fly ash loading to the FGD system is at the precipitator outlet
guarantee levels of 0.2 lb/MBtu and 60% of the ash is collected by "the scrubbers then,
the fly ash in gypsum would be 0.76%. In such a case, the manufacturers'
requirements would be impossible to meet. Fly ash loading is a function of the
precipitator design that could not be modified at that point in the project.

Excessive fly ash may cause whiteness problems; however, color consistency seems to
be a larger concern for U.S. wallboard manufacturers.

Chloride

High chlorides can cause paper adhesion problems in the manufacture of wallboard.
Wallboard manufacturers' requirements on chlorides in the gypsum are less than 100
to 120 ppm. This requirement can be met by washing the gypsum cake. The
recovered chloride will have to be purged to maintain the required chloride level
(<8000 ppm) appropriate for the current materials of construction. A continuous
blowdown from the filtrate pump (adjusted based on laboratory analysis) is required.
The indexing belt filter offers an 88% wash efficiency. A higher than 88% efficiency
can be achieved by adding an additional wash box to the belt filters at a later date if
needed.

The blowdown rate is inversely proportional to the wash efficiency. Therefore, the
higher the wash efficiency the lower the blowdown rate. Figure 1 is a graph of
blowdown versus chloride to sulfur ratio at 88% wash efficiency for various weight
percent solids. At 90% solids and 88% wash efficiency a blowdown flow rate of 50 to
280 gpm would be required (for Cl/S ratios of 0.02 to 0.1 respectively) to keep the
chloride level below 8000 ppm. A permit for blowdown up to 300 gpm is currently
being sought by OMU.

Chemical Composition of Gypsum

Combined Water, CaO and SOT The chemical formula of gypsum is CaS04-2H20.
Wallboard manufacturers have expressed the total gypsum purity of 95% into three
individual chemical species such as combined water, CaO and SC)3. The minimum
quantity of 19.9% combined water, 30.9% CaO and 44.2% S03 add up to total
minimum 95% CaS04«2H20 purity specified.

103-10


-------
Gypsum Cake @120 ppm d- with wash efficiency at 88%
Chloride Equilibrium Concentration - 8000 ppm

Slowdown, GPM

Chloride/Sulfur Ratio

Figure 1. Blowdown Versus Cl/S Ratio

Specific Constituent Limits. Table 3 is the list of constituents' maximum/minimum
allowable limits from wallboard manufacturers and the expected level at worst
operating condition at the Elmer Smith FGD installation. Specific items on the table
are addressed as follows:

•	As discussed earlier, expected fly ash level in the solids would be 0.76% and
0.7%, below a manufacturer's 1% requirement, at 0.2 and 0.185 lb/MBtu fly ash
loading respectively based on 60% collection by the scrubbers. The lower
requirement of 0.13% fly ash would be practically impossible.

•	Wallboard manufacturers' maximum Si02 limit of 0.5% and 1.1% will not be met
if the limestone contains greater than 2.5% Si02. High level at Si02 are a concern
to the wallboard manufacturers due to equipment wear. The wallboard
manufacturers have indicated that the Si02 limit can be exceeded if a cost
penalty is levied.

103-11


-------
•	R203 is a total amount of inerts including all metal and silica oxides but
excluding other inerts in fly ash and carbonates (unreacted CaC03 and MgC03).

•	One manufacturer's Na requirements of 70 ppm will be easily met but the 70-
ppm Mg level may be difficult to achieve. Other manufacturers have not given
any requirements for either Na or Mg. Instead of individual constituents they
look at total dissolved salts, which is discussed below.

•	One manufacturer's 1% maximum CaS034£H20 requirement will be met. The
expected level will be 0.72% at 99% oxidation. Other requirements of 0.1% are
very stringent and would require 99.86% oxidation.

•	The total water soluble salts should be substantially lower than the maximum
600-ppm requirement.

All the constituents listed in Table 3 are projected to be lower than the maximum
allowed limit by one manufacturer. The maximum limit by other manufacturers is
very stringent and will be difficult to achieve.

Other requirements (such as, aspect ratio, particle size, NH4, and ppm) are difficult to
monitor and, therefore, are not included in all manufacturer's specifications. A
minimum particle size requirement of 20 pm seems reasonable and should be
achievable. All other requirements, such as aspect ratio and x-y axis dimensions,
should be excluded from manufacturers' specifications.

Technical Feasibility Results

The technical feasibility evaluation concludes that wallboard quality gypsum is
technically feasible on the basis of the following:

•	Horizontal belt filters will be used for secondary dewatering in lieu of rotary
drum filters. Solids contents of 90% can be provided with a horizontal indexing-
type filter. The filter will be equipped with a two-stage wash system allowing
for a chloride level in the gypsum product that will not be greater than 120 ppm.

•	A continuous blowdown from the filtrate pump of the vacuum filter based on
chloride analyses is incorporated to control chloride level in solids and to
maintain the equilibrium level appropriate for the current materials of
construction. A blowdown stream of up to 300 gpm should be assumed. This
blowdown stream will also allow flexibility in coal procurement.

•	Limestone reagent with as high a CaC03 content as possible should be
purchased (a minimum of 96% CaC03 content may be required to meet a 94%
purity requirements). A higher CaCO, content in the limestone will increase
limestone utilization.

103-12


-------
• A minimum 99% oxidation can be achieved while operating the three 50% air
compressors that had been previously purchased. Sufficient conservatism exists
in the oxidation process design that may allow operation of only two
compressors.

Economic Results
Economics Analysis

Table 6 summarizes the evaluation of capital and operating costs associated with the
alternatives considered for disposing or utilizing the FGD system by-product. To
focus on the cost for alternate FGD disposal methods, the evaluation results
presented in Table 6 exclude any of the costs associated with reuse or disposal of fly
and bottom ash. The results shown in this table represent the optimum combination
of loading, hauling, and disposal and reuse bid costs based on the 14 bids received
byOMU.

The capital costs shown for the commercial-grade gypsum alternative represents the
incremental investment required for process and balance-of-plant supporting
equipment over the base cost for the rotary drum vacuum filter process equipment.
The costs considered in the process equipment category shown in Table 7 include the
incremental belt filter investment, air sparger modifications, and deletion of the radial
stacker and associated conveyors. In addition, building size and electrical power
supply system costs were impacted by the larger size of the belt filters and the larger
power requirement of the vacuum pump and auxiliaries. The total cost impact to the
process equipment is approximately $5.4 million. The $5.4 million incremental
investment compares to the $20 million investment required to develop an
environmentally acceptable landfill. No additional capital investment increment
would be required for a third-party landfill. The cost for a third-party landfill
operator to develop the landfill are included in the operating costs associated with
this alternative.

Operating costs included in the economic evaluation include costs for loading,
transporting, disposal, and/or utilization of final gypsum product for each
alternative. The commercial-grade gypsum evaluation assumes that 95% of the
gypsum will be sold for wallboard production. The remaining 5% is assumed to be
material that does not meet specification that would have to be disposed of in a
third-party landfill. For the commercial-grade gypsum alternative, an on-site barge
loading facility, owned and operated by a third-party contractor, was determined to
be the lowest risk economic choice. The overall cost of an OMU owned and operated
barge loading facility was comparable to the third-party operator costs, however,
OMU favored a lower initial capital investment approach to minimize the risk
associated with possible changes in the long term market for gypsum. Note, to avoid
compromising negotiations that are still in progress with potential gypsum
contractors, a breakdown of the operating cost components cannot be provided.

103-13


-------
Table 6

Economic Comparison of Solid Waste Disposal Alternatives

Commercial-
OMU-Owned Third-Party	Grade

Landfill	Landfill	Gypsum

Capital costs (1995$)

A. Landfill dev.

20,000,000

N/A

N/A

B. Sec. dewatering equip.

Base

Base

2,200,000

C. Gypsum stockout build.

N/A

N/A

2,108,000

D. Conveyors

Base

Base

434,000

E. Contingency

N/A

N/A

662.000

Subtotal

20,000,000

0

5,404,000

O&M Costs (1995 $)







A. Loading, trans., utilization







or disposal

1,906,000

2,763,000

1,245,000

B. Power

11.000

11.000

76.000

Subtotal

1,917,000

2,774,000

1,321,000

Limestone (1995 $)

1,016,000

1,016,000

1,632,000

IV.	20-Year PVRR	58,731,000 50,048,000 44,619,000

V.	20-Year PVRR Over Base	14,112,000	5,424,000	Base

The limestone cost evaluation presented in Table 6 represents the expected difference
between the higher quality stone required to meet gypsum purity requirements and
the lower quality, higher inert content percentage limestones.

OMU is currently in the process of procuring limestone for the project. Bids have
been obtained from suppliers in Kentucky and neighboring states. Preliminary
indications are that the premium for the higher quality stone can range to 100% of
the cost of the lower quality stone.

103-14


-------
Table 7

Development of Commercial-Grade Gypsum Process-Related Costs

Landfill Options Commercial-Grade Gypsum

A.	Belt filter

B.	Air sparger

C.	Delete rad. stacker

D.	Sec. dewatering building

E.	Electrical equipment

Total

Base

Base

Base

Base

Base

Base

50,000
(1,000,000)
400,000
250,000
2,200,000

2,500,000

The economic analysis shows a payback period of approximately 5 years on the
investment to convert to commercial-grade gypsum production. This payback was
sufficient for OMU to conclude that the commercial-grade gypsum conversion could
be justified. WAPC and S&L were authorized to proceed with the process and
balance-of-plant equipment changes in September 1992. OMU is currently in the
process of finalizing negotiations for gypsum purchase, barge loading services, and a
backup third-party disposal site.

Revised Equipment Arrangement

The arrangement of the revised secondary dewatering facility and gypsum stockout
building are shown on Figures 2 and 3. The three 50% capacity rotary drum vacuum
filter in the original design are replaced with two 100% capacity horizontal belt-type
vacuum filters. Each belt filter is designed for a capacity of 73,200 lb/hr and consists
of a filter assembly approximately 12-feet wide and 79-feet long weighing 12 tons, a
vacuum pump /washwater system skid approximately 12-feet wide x 16-feet long x
10-feet tall and a filtrate pump/receiver skid approximately 8-feet wide x 24-feet long
x 12.5-feet tall. The filters use a seamless cloth 2.5 meters wide with a filtration
length of 21 meters. The process flow for one belt filter is shown on Figure 4. The
filters will be continuously fed slurry by a recirculation loop from the filter feed
pumps to the filter feed tank. Modulating slurry control valves will be used to
control slurry feed to the filters. The length of the filter is broken down into a
number of process steps including a pool section, cake forming, first cake wash, first
cake dry, second cake wash, and final cake drying sections. "ITie filter cloth and cake
are moved from one processing step to the other as the filter indexes in increments of
approximately 30 inches. A countercurrent wash design in incorporated into the
filter system that uses wash boxes to distribute the washwater across the filter cake.

103-15


-------
Figure 2. Revised Secondary Dewatering Facility—Second Floor Plan

Spent vacuum pump seal water and spent cloth washwater are used for second-stage
cake washing. Filtrate water from the second stage wash is used for first-stage
washing.

Two 72-inch-diameter x 96-inch-high filtrate receivers are provided for each filter.
The primary receiver collects slurry filtrate and first-stage washwater for
transportation to the thickeners and for chloride blowdown. The secondary receiver
collects filtrate from the second wash stage for use in first-stage washing. Both
vacuum filtrate receivers are fabricated from rubber-lined carbon steel.

One filtrate pump is provided with each receiver. Filtrate pumps are constructed of
rubber lined carbon steel construction with gear box drives.

103-16


-------


| tew*. • * srocHour suiidinc

Figure 3.

Gypsum Conveyor System and Stockout Building


-------
w

00

Figure 4. Revised Secondary Dewaterlng Process Flow


-------
Each filter system includes one liquid ring vacuum pump designed to deliver 9040
acfm at 20 inHg and 450 bhp. The pumps come complete with a direct drive gear
box and 500-hp drive motor. Seal water requirements of the pump are minimized by
a 50% recycle of seal water. The seal water system also provides spent seal water
with an elevated temperature beneficial for use in second-stage cake washing.

Vacuum pump seal water and spent cloth washwater are collected in a washwater
tank of fire resistant FRP construction. A supplemental electric heater is provided in
the tank for use during system startup and when city water supply temperatures
drop below 40 °F. One washwater pump is provided to pump washwater to the
second-stage wash box.

Each vacuum filter includes a local control panel with a PLC for control of vacuum
filter indexing operations. Supervision of the PLC system will be by the owner's
distributed control and monitoring system (DCMS) via various alarm and status
signals transmitted from the PLC. All other belt filter system equipment will be
controlled through the DCMS.

The conveyor system will be designed to permit the simultaneous operation of both
vacuum filters. Each vacuum filter discharges gypsum product cake through a chute
and onto the reversing transfer conveyor for transportation to the forwarding
conveyor and finally the stacker conveyor system. The conveyor system is capable of
handling both vacuum filters simultaneous operation. The reversing transfer
conveyor can also discharge onto the emergency discharge conveyor with a final
emergency cake discharge elevation approximately 12 feet above grade. The gypsum
is conveyed from the secondary dewatering building to the gypsum stockout
building. A reversing shuttle conveyor in the stockout building allows material to be
stacked at any point in the building. The gypsum is removed from the pile and with
a front-end loader and moved by truck to an on-site barge loading facility.

Conclusions

There are several advantages to the production of a commercial-grade gypsum versus
a disposal-grade gypsum, but the advantages are not without some risks. In
addition, a decision to produce commercial-grade gypsum requires a site-specific
study. Some general conclusions from the OMU study are as follows:

•	The production of waliboard-grade gypsum is technically feasible. The

dewatering system is the main area requiring major capital investment.

•	The liability associated with a waste product are minimized with the wallboard
approach versus disposal in a third-party landfill.

•	The major economic factors are dewatering system and gypsum storage capital
investment, high-grade limestone costs, transportation costs to the wallboard

103-19


-------
plant, and gypsum utilization fees versus alternative costs (e.g., disposal,
alternate dewatering system, and limestone).

•	The economics are very sensitive to the long-term market for synthetic gypsum.
Length of contract and price adjustment clauses are very important factors to be
considered. The gypsum utilization fee in a pure economy should be a function
of competing gypsum mine costs and competing utilities. Only very close
power plants could compete economically with a gypsum mine located at the
wallboard plant Competition by utilities closer to wallboard plants could also
defeat the favorable economics of this modification.

•	Optimum economics occur when scrubber production matches wallboard plant
capacity to allow 100% synthetic gypsum wallboard production.

•	Maintenance of gypsum quality must be a prime consideration in addition to
removal efficiency when operating a FGD system.

•	A landfill should still be permitted to accept off-spec gypsum or in case of force
majure or contract termination.

Reference

1. W. Ellison and E. Hammer. "FGD-Gypsum Use Penetrates U.S. Wallboard
Industry." Power, pp 29-33. February 1988.

Bibliography

Ransom, J. M., Torstriek, R. L., and Tomlinson, S. V. "Feasibility of Producing and

Marketing Byproduct Gypsum from S02 Emission Control at Fossil-Fuel-Fired Power
Plants." EPA-600/7-78-192, October 1978.

Armour, D. W., Conn, S. K., and Frizzell, K. "By-Product Management Philosophy at
the Elmer Smith Station." Proceedings of the American Power Conference. 1993.

Collins, S. "Managing Powerplant Wastes." Power, pp 15-27. August 1992.

103-20


-------
PLANT GROWTH AND SOIL PROPERTIES RESPONSES
TO ADDITIONS OF DRY FLUE GAS DESULFURIZATION BY-PRODUCTS

W. A. Dick, R. C. Stehouwer, P. Sutton, J. M. Bigham, R. Lai,

S. J. Train a, E.L. McCoy and R.. Fowler
Department of Agronomy
The Ohio State University
Wooster, OH 44691

Abstract

Substitution of alkaline flue gas desulfurizatlon (FGD) by-products for conventional
liming materials in agriculture, mine reclamation and for soil stabilization are
potential uses for these by-products. Greenhouse and field studies demonstrated
improved plant growth when add soil or surface coal mine spoil were treated with
FGD by-products from three different sources. For agricultural soils, optimum
growth occurred when application rates were equal to the lime requirement of the
soil. For the acid spoil, optimum growth occurred when application rates were
between 6 and 12% (by weight). Leachate composition indicated that when FGD by-
products were applied at rates of 12% or lower, concentrations of elements of
environmental or regulatory concern remained vety low. Boron was an exception
when LIMB by-product was used to amend spoil and a period of time for the boron
to be leached from the amended spoil prior to planting of vegetation cm* crops may be
required when high application rates of LIMB by-product are used. There seems to
be little potential for adverse effects on water, soil and plant quality when FGD by-
products application rates are based on the amount required to neutralize the acidity
in soil or spoil. The improved plant growth observed in this study, when FGD by-
products ware applied to acid soil or spoil, represent a beneficial reuse of these by-
products.

Introduction

A number of diy FGD technologies have been developed in the past decade. The US
Department of Energy's Clean Coal Technology program and the 1990 amendments
to the Clean Air Act, mandating a 2-stage 10 million ton reduction in SO2 emissions
in the United States, have especially encouraged the development of these
technologies. Dry FGD technologies are generally smaller in scale and require a
lower capital investment than do the wet FGD processes. Dry FGD technologies are
also generally designed for retrofit on existing coal-fired power plants and, therefore,
represent an option for bringing older plants into compliance with clean air
legislation.

The classification of the diy FGD by-product as a solid, or even hazardous, waste by
most states has inhibited the development of beneficial uses of these materials.

Thus, in addition to installation and operating expenses, plants burning high sulfur
coal and using dry FGD technologies must also bear increasingly expensive landfill
disposal costs. We have recently completed a comprehensive study of the chemical,
physical, mineralogical and engineering properties of 58 dry FGD by-product
samples in Ohiol. With this information serving as a base, a greenhouse study was
conducted to demonstrate potential reuses of diy FGD by-products. These reuse

104-1


-------
applications, if environmentally benign or even beneficial, would significantly
reduce the costs of SO2 emission control.

Dry FGD by-products are composed of a mixture of conventional coal combustion
ash (either bed or fly ash), the SO2 reaction product (primarily anhydrite, CaSOa),
and unspent sorbent (generally lime, limestone, or dolomite). Due to the presence
of unspent sorbent, dry FGD by-products are usually highly alkaline with significant
neutralization potentiaP-5, Anhydrite does not neutralize acidity but may
ameliorate problems of A1 toxicity through the formation of Al(SC>4)x complexes. It
is also a source of both Ca and S for plant nutrition. Fly ash may supply other plant
nutrients (e.g. B, Mo, Zn, P, K)2. These FGD by-products may also be high in soluble
salts and contain some trace elements of environmental concerns.

Of the several dry FGD technologies tested in Ohio, four were selected for more
intensive by-product characterization: duct injection, lime injection multistage
burner (LIMB), pressurized fluidized bed combustion (PFBC), and spray dryer. Duct
injection involves injecting the sorbent (hydrated lime) into the flue gas as it enters
a humidification chamber in the ductwork downstream of the boiler and air heater.
Injection of fine water mist lowers the gas temperature and raises the humidity of
the flue gas to increase the reaction of the SO2 with the sorbent. Dry FGD by-product
is removed from the flue gas and contains the reaction products, fly ash, and
unspent sorbent. LIMB is a process whereby calcium based sorbent is injected
directly into the boiler where it calcines to CaO and reacts with SO2 and O2 in the
combustion gases to produce CaS04- The reaction product and unspent sorbent are
collected with the fly ash. In PFBC systems, a cataum-based sorbent (usually
limestone or dolomite) and crushed coal are introduced together into the boiler bed
where they are "fluidized" or suspended by jets of air. This mixes the coal and
sorbent and allows for reaction of SO2 and sorbent. Two by-product streams are
created - one being the heavier, granular bed ash material and the other the finer
materials which are removed by the particulate emission control equipment The
spray dryer scrubber includes a separate vessel located downstream of the boiler and
air heater where sorbent, usually a slurry of hydrated lime, is injected. Reaction
products and unspent sortsent are removed with the fly ash in the particulate
emission control system.

Based on the results of laboratory investigations of the properties of FGD by-
products, we proposed several greenhouse or tightly controlled field studies to
further develop beneficial reuses of these FGD by-product materials. The
applications tested and reported in this paper include using FGD by-products as an
alkaline amendment for acid agricultural soils and acid spoil from surface coal
mines. Growth responses and environmental parameters were measured.

Materials and Methods
Soil and Spoil

The soil used was a Wooster silt loam (fine-loamy, mixed, mesic, Typic Fragiudalf).
It had a pH of 4.6 and a lime requirement of 215 Mg/hectare (9.6 tons/acre) to reach
a pH of 7. The cation exchange capacity was 15.1 cmolc /kg (15.1 milliequivalents/100
g soil) and this soil was noted for Mn toxicity.

Spoil was obtained from the Fleming abandoned mined land (AML) site near New
Philadelphia in Eastern Ohio. This site contains approximately 10 hectares (25 acres)

104-2


-------
of exposed, highly erodible underclay bordered on two sides by approximately 18
hectares (45 acres) of unreclaimed spoil and 2 hectares (5 acres) of coal refuse. The

fH of the underclay ranges between 2.5 and 35 and that of the spoil between 32 and
0. Both the underclay and spoil were collected for use in the greenhouse study, but
only the results related to the spoil are reported.

Dry FGD By-Products Used

By-products from the LIMB, PFBC and spray diyer FGD technologies were used. The
LIMB by-product was collected from the full-scale commercial LIMB demonstration
at Ohio Edison's Edgewater plant in Lorain, Ohio. The PFBC by-product was
obtained from the demonstration facility operated by American Electric Power at the
Udd plant located near Brilliant, Ohio. The McCracken Power Plant at the Ohio
State University was the source of the spray dryer FGD by-product.

Detailed characterization data for the LIMB and PFBC by-products are summarized
in Table 1.

Experimental Procedures

After passing through a 10 mm screen, 48 leg erf air-dry Wooster soil was mixed with
FGD materials at rates of 0, 0.35, 0.7,1.4 and 2.8% by weight and placed into a FVC
pot (15 an diameter, 30 can deep). These rates approximate 0, 7.8,15.6, 31.2 and 624
Mg/hectare (3.5, 7.0,14, and 28 tons/acre) or 0, 0.25, 0.5,1 and 2 times the soil liming
requirement. Since the FGD by-products contained only 60% calcium carbonate
equivalency (Table 1), the rates required for neutralization of acid in the soil, as
compared to agricultural lime, were approximately double. Separate pots of each
mixture were planted with fescue and alfalfa. The treatments were replicated three
times.

The spoil from the Fleming AML site was air-dried, passed through a 10 mm screen,
and mixed with FGD by-product at rates of 0, 3,6,12 and 24% by weight
(approximately 0, 30, 60,120 and 240 tons/acre) using a complete factorial
experimental design. Sewage sludge (Table 1) was mixed with all treatments at a
constant rate of 6% by weight The sewage sludge was included because it has teen
our experience that addition of an organic material with alkaline material creates an
excellent amendment for reclamation of acid spoil and growth of vegetation. In
addition, four replications of spoil alone or mixed with 12% FGD by-product, but
without sewage sludge, were also included in the experiment. Mixtures were placed
in FVC pots (15 cm diameter and 30 cm deep) and the plots planted to Kentucky 31
fescue. The treatments were replicated four times and arranged in randomized
complete blocks.

The field experiment at Coshocton, Ohio involved inserting 21 PVC columns (30 an
diameter and 90 am deep) into the ground. The columns were filled (torn the
bottom to the top) with a layer of sand (5 cm), untreated Fleming spoil, and then
Fleming spoil treated to approximately 20 cm depth with either LIMB by-product (10,
20, 40% by weight) or spray diver FGD (10, 20, and 40%, by weight). An untreated
control was also included in the experimental design. An alfalfa and fescue
mixture was planted in the columns.

104-3


-------
Data Collection and Analyses

Similar data collection procedures and laboratory analyses were applied to materials
from pots containing treated Wooster soil or Fleming AML spoil,

Leachate was collected from pots containing treated soil and spoil at various times
(including before planting) by adding enough water to collect from 100 to 200 ml of
volume. Leachates were analyzed for pH, electrical conductivity, metals (AJ, As, B,
Ba, Ca, Cd, Co, Cr, Cu, Fe, Hg, K, Li, Mg, Mn, Mo, Na, P, Pb, S, Sb, Se, Si, Sr, V, and Zn
by taductivelv coupled plasma spectrophotometry (ICP)), and anions (F, CI"' NO3*,
SO3 and SO4 by ion chromatography (IC)).

Immediately following the first leaching the pots were weighed to determine water
content at field edacity. Soil water content in the pots was maintained at
approximately 25% below field capacity throughout the experiments to insure that
plants were not under moisture stress, that reducing conditions did not occur in the
pots, and that no leachate drained from the pots between intentional leaching^.
Artificial lighting was used to provide 14 hours of light, and the temperature was
maintained on a 25/20*C day/night cycle.

Hants were harvested eveiy 30 to 32 days with the first harvest occurring 94 days
after planting and continuing for six cuttings (approximately nine months after the
mixes were prepared. Cuttings were dried at 65°C, weighed, ground and digested
with perchloric add. The digests were analyzed for the same metals, using ICP, as
listed for the soils.

After nine months, the pots were split using a saw and the treated soil or spoil
removed. Various chemical, physical and mineralogical properties were measured
and the results compared across treatments. Only some of the data collected at the
conclusion of the experiment will be reported in this paper.

Rant growth at the Coshocton field site was determined 125 days after seeding. The
amount of dry matter production as alfalfa or fescue was determined.

Analyses of variance (ANOVA) was conducted on the soil, spoil and leachate data
collected.

Results and Discussion
pH, Alt and Mn Changes

The primary benefit of liming an add plant growth medium is increasing the pH to
near neutrality and redudng the potentially toxic concentrations of A1 and Km. The
FGD by-products used in our tests were effective in rapidly raising the pH of the
Wooster soil and the Fleming AML spoil (data not shown). For example, untreated
spoil maintained a pH of approximately 3.0 and spoil treated with 6% sewage sludge
obtained a pH of 40. Spoil treated with 6% sewage sludge and 12% FGD, however,
resulted in a pH of 8.3 (LIMB by-product) or a pH of 8.0 (PFBC by-product). The pH
of all treated soil and spoil materials decreased over time, with the largest decreases
occurring near the column surfaces. After nine months, pH in the surface 5 cm
layer of the pots containing spoil never exceeded 8.0, even when FGD by-product
was applied at the highest rate (24% by weight). This pH would not restrict plant
growth, as many agricultural soils in the western states of the United States have

104-4


-------
natural pH values of 8.0. Also a soil pH of 8.0 would be much more conducive to
increasing overall plant growth than the pH originally observed for the soils and
spoil materials used in this experiment. Thus, the addition of FGD by-product to
acid soil or spoil would represent a beneficial reuse.

The leachate chemistry is a reflection of the chemistry that occurs in the pots. pH in
the first leachate from treated soil and PFBC treated spoil was ^proximately 8.u at
the highest application rates (Figures 1 and 2). The leachate pH from the LIMB
treated spoil, however, varied from 7.0 (3% application rate) to 12.0 (24% application
rate) in the first leachate. With time, leachate pH remained either unchanged or
tended to decrease (especially for the LIMB by-product treated spoil). After nine
months, the final pH or leachate at the highest PFBC and LIMB by-product
application rates (24%) were 77 and 85, respectively. This decrease in pH is
apparently due to carbonation of lime, portlandite, and peridase since the pH was
tending toward that of free carbonates in equilibrium with atmospheric carbon
dioxide. Carbonation was somewhat inhibited in pots with high FGD application
rates, however, because high pH and soluble salt concentrations inhibited plant root
growth and microbial activity in these pots. Plant root and microbial activity act as a
source of carbon dioxide and leads to increased carbonation of the FGD by-products.

Conditions of high pH and adequate moisture also promoted the formation of
cementitious secondary minerals. Extensive enttringite (Ca6Al2(SQ|)3(OH)i2'
26H2O) formation occurred with LIMB by-product above the 12% application rate.
Enttringite is highly expansive and cementititious and thus severely restricted root
growth and water movement below the depth at which it formed. At the 24% LIMB
by-product application rate, extensive enttringite formation occurred below 15 cm
depth (Figure 3). The PFBC by-product did not promote enttringite formation to the
same extent as did the LIMB Dy-product. This very different behavior of the two
FGD by-products with respect to secondaiy mineral formation is due to the differing
mineralogies and solubilities of their constituent components.

As previously mentioned, one of the primaty benefits of adding an alkaline
amendment to acid soil or spoil is to decrease the potentially toxic concentrations of
A1 and Mn. Concentrations of A1 and Mn in the leachates collected from the pots
were determined. The first leachates were collected from previously air-dry
materials and are not representative of equilibrium conditions in the soil or spoil
plus FGD by-product mixes. The final leachates, however, provide a much better
taction of equilibrium conditions and long-term impacts of using FGD by-
products.

Leachate concentrations of A1 and Mn decreased with increasing application rates of
FGD by-products (Figures 1 and 2). The highest concentrations of Al and Mn were
found in the untreated Wooster soil and Fleming AML spoil. FGD by-product
additions thus created an improved plant growth medium. The solubility of Al and
Mn is greatest under acid conditions® and decreases rapidly with increasing pH. The
beneficial effect of FGD by-products is due to increasing the pH of the system.
Application of sewage sludge at the 6% rate resulted in large decreases in Al and Fe
(data not shown) even though the corresponding pH increase was slight (from 3.0 to
4.0). This is due to the strong complexes that organic matter forms with these
elements?. An inverse relationship between soil organic matter and AP+ solubility
and plant growth toxicity has been reported^. These results support our previous
recommendations where we maintain that best results occur when a mixture of
organic and inorganic amendments are usedlO.

104-5


-------
Sulfur Concentrations in the Leachate

Increasing application rates of FGD by-products increased soluble salt concentrations
(as determined by electrical conductivity; data not shown). Similar patterns were
seen for S, Ca, and Mg which were the major elements in the leachate. Only S
results are reported (Figure 4) but torn these results we can also gain information
about the fate of other elements, such as Ca and Mg, in FGD by-products applied to
soil and spoil.

Concentrations of S (and Ca) in the soil and spoils treated with LIMB by-product,
where enttringite did not form, appeared to be controlled by the solubility of gypsum
(CaSO^ ¦ 2H2O). Because enttringite is much less soluble than gypsum, S
concentrations decreased in leachates where this mineral formed, even though the
amount of LIMB by-product added was greater. With PFBC by-product, leachate
concentrations of S (and Mg) would be controlled by epsomite (MgS04 • H2O). The
concentrations of S would, therefore, be greater where PFBC is applied because
epsomite is approximately 300 times more soluble than gypsum. The
concentrations of Ca, also increased in the PFBC as compared to the LIMB by-product
treated soil and spoil, due to the solubility of caldta The results in Figure 4 indicate
that the presence of Mg in the PFBC by-product gives this by-product a greater
potential for excessive salt loading than the LIMB by-product.

Concentrations of Regulated Elements in the Leachate

Concentrations of elements regulated with respect to land application of sewage
sludge are shown in Table 2. These regulatory levels for sewage sludge were chosen
as a basis for evaluating our leachate results because we had included in our
experimental design the addition of 6% sewage sludge.

Not listed in Table 2 are Hg and Fb. Mercury was below the detection limit (<0.04
mg/kg) in all leachates ana also was not detected (<0.0002 mg/kg) in Toxicity
Characteristic Leaching Procedure (TCLP) extracts of LIMB or PFBC by-products.

Lead also was not detected (<0.04 mg/kg) and the TCLP levels for Fb in the LIMB and
PFBC by-products were 0.005 and <0.001mg/kg, respectively. Cadmium
concentrations in the leachate showed either no effect or a decrease in concentration
with increasing rates of FGD application (Table 2). Thus, with respect to these
metals, there appears to be little potential for environmental contamination as a
result of FGD by-product used as a soil or spoil amendment.

Concentrations of Cu, Ni and Zn generally decreased with increasing rate of FGD
application. These results are consistent with most studies of trace metal behavior
in soils, which show that metal solubility and mobility decreases as pH increases
above the acid range to neutralitylU2. Copper, Ni and Zn were much higher in the
sewage sludge than in FGD by-products (Table 1). Since the same amount of sewage
sludge was applied to all pots, regardless of FGD rate applied, the amount of these
elements added to the pots would be approximately the same. However, at very
high application rates of FGD by-product (24% by weight), concentrations of these
metals were found to be increased in the leachate. This is due to high pH
solubilizing the organic matter in the sewage sludge. Soluble organic matter can
complex with metals bringing them into solution' and these complexes are then
leached from the soil or spoil at elevated levels.

Leachate concentrations of the oxyanion, B, tended to increase with increasing FGD
by-product application rate (Table 2). At the highest FGD application rate applied to

104-6


-------
the Fleming AML spoil (24%), however, the concentration of B decreased (data not
shown) due to formation of enttringfte in the pots. Concentrations of the other
oxyanions (As, Mo, and Se) seemed to be unaffected by FGD application (Table 2) at
rates of 12% or lower.

Both the FGD by-products and the sewage sludge materials contained As, B, Mo, and
Se (Table 1). Increased concentrations observed at the highest application rate (24%)
(data not shown) is due to both the increasing amounts of oxyanions added to the
pots as well as increased competition of sulfate and organic Mgands for anion
absorption sites in the soil or spoil. This competition would force some of the
oxyarrions, that would normally be retained in the soil or spoil, to enter the solution
phase and be leached from the pots. In contrast, enttrmgjte formation would limit
contact between the leachate water and As, B, Mo and Se in the pots, thereby
decreasing their concentrations in the leachate The increases in B concentrations
were much greater in leachates from spoil treated with LIMB by-product than the
FFBC by-product. This is not surprising since total B in the LIMB by-product was
much higher than in the PFBC by-product Most species of B that are phytotoxic to
plants are water soluble and will rapidly leach from the root zone. However,
growth of plants may be initially inhibited by B when high rates of some FGD by-
products are applied

Plant Growth Responses to FGD By-Product Applications

Plant (alfalfa) growth was increased in the acid Wooster soil when treated with
LIMB by-product or the bed and cyclone PFBC by-product (Figure 5). In the first
harvest the optimum growth generally occurred at application rates of from 0.35 to
1.4%. At the highest rates, equivalent to two times the lime requirement for this
soil, growth during the first harvest period was inhibited. This is attributed to
excessively high pH values that initially developed in the soil when these high rates
were applied. However, by the third harvest the inhibitory effect was no longer
evident and, in fact, the 2.8% application rate yielded significantly more alfalfa than
did the untreated control. By this time, carbonation had occurred and decreased the
soil pH and the inherent buffering capacity of the soil had time to react with the
excess alkalinity. Our results (Figure 5) indicate that dry FGD by-products may be
used as agricultural lime substitutes, provided the rates applied are approximately
equivalent to the lime requirement of the soil If higher rates are applied, then fall
applications would be recommended to provide sufficient time for chemical
reactions to occur in the soil, during the winter months, prior to seeding of crops.

Similar results were observed when the acid spoil from the Fleming AML site was
treated with LIMB and PFBC (cyclone) by-products (Figure 6). This experiment used
much higher rates, however, and the high pH inhibited fescue growth until the
fourth or fifth harvest. The growth of fescue on spoil treated with 12% PFBC by-
product during the first harvest period seemed poor (Figure 6), but this was due to
poor growth in one of the four replications. However, subsequent growth periods
showed good growth response to applications of the FGD by-products. On the basis
of our greenhouse experiments, we would recommend application rates of FGD by-
products onto acid spoil of approximately 6 to 12% by weight. This rate supplies
sufficient alkalinity to neutralize the acidity in the spoil but still avoids inhibition of
plant growth because of excess soluble salts or high pH.

The field columns at Coshocton compared alfalfa and fescue growth, under natural
field conditions, in the add Fleming AML spoil treated with FGD by-products that
differed in the sulfur forms. The LIMB by-product primarily contains gypsum

104-7


-------
(CaS04) while the spray dryer FGD by-product contained over half of its total sulfur
as calcium sulfite (CaSC^jl, Calcium sulfite is considered inhibitory to plant growth
and this experiment was designed to evaluate plant growth response to the two FGD
by-products. Optimum growth (15.2 g/ column) after 125 days occurred in the spoil
treated with 10% spray dryer FGD by-products (Table 3). Less growth (8.4 g/column)
was observed when LIMB by-product was applied at the 10% rate Apparently the
pH from the LIMB by-product treatment was too high for growth, even for fescue,
which generally can withstand more extreme conditions than alfalfa. As the
application rates increased, growth decreased few both the LIMB and the spray FGD
diyer by-products. The results reported are only from the first harvest, but if the
growth pattern in the field fallows that observed in the greenhouse, subsequent
harvests, in the columns treated at the 20% FGD by-product application rate, would
be improved over that observed after the first 125 days. Also the rates applied here
exceeded that in the greenhouse. Our best growth responses on the treated Fleming
AML spoil in the greenhouse occurred at the 6 to 12% rate, which is consistent with
what was observed in the field where optimum growth occurred at the 10% rate.
The CaS03 in the spray dryer FGD by-product did not seem to result in any greater
growth inhibitions than that observed for the LIMB by-product (i.e. gypsum
containing by-product). The pH of the spray dryer FGD treated columns was not as
high as where LIMB by-product was applied and CaS03 is much less soluble than
gypsum also minimizing its effect on plant growth.

Conclusions

FGD by-products appear to be highly effective as alkaline amendments for acid soil
and spoil Addition of FGD by-product effectively increased pH to near neutrality or
to slightly alkaline pH and decreased concentrations of soluble A1 and Mn. The
result of these chemical changes is an improved growing medium for plants.
Improved plant growth was, indeed, observed when FGD by-product was mixed
with acid soil or spoil. The addition of 6% sewage sludge with the FGD by-product
created the best conditions for plant growth.

Potential adverse effects on plant growth of high concentrations of soluble salts,
excessively high pH, and trace metal toxicities were not observed, even at rates
slightly in excess required to neutralize the soil or spoil.

Leachate composition indicated that at application rates of 12% or lower,
concentrations of elements of environmental and regulatory concern remained
very low. Most, in fact, were below drinking water standard levels. Boron was an
exception when LIMB by-product was used to amend the spoil. The limiting factor
for application rates of Mg-containing FGD by-products, such as the FFBC by-
product, is more likely to be high soluble salt concentrations, which may inhibit
growth and impact water quality. For the LIMB by-product the factors limiting use
are the initially high pH values that were observed and the potential for enttrmgite
formation. Both would inhibit plant growth. However, it must be stressed that
these limitations occurred only at very high application rates.

There seems to be little potential for adverse effect on water, soil and plant quality
when FGD application rates, based on the amount required to neutralize soil or
spoil acidity, are not exceeded. This last statement is made in the context that the
neutralization potential of the FGD by-products is approximately equivalent to 50%
calcium carbonate FGD by-products with lower neutralization potentials would

104-8


-------
require higher application rates and, thus, their environmental impact would differ
from what is reported here.

Acknowledgments

This research was conducted as part of the "Land Application Uses for Dry FGD By-
products" project which is a cooperative project of the Ohio Agricultural Research
and Development Center, The Ohio State University, The U.S. Geological Survey,
and the Dravo Lime Company. Funding support for this project was obtained from
the Ohio Coal Development Office (Columbus, OH) Grant No. CDO/D-89-35, The
U.S. Department of Energy (Morgantown Energy Technology Center, Morgantown,
WV) Award No. DE-FC21-91MC28060, Dravo lime Company (Pittsburgh, PA) Grant
No. RF768342, Electric Power Research Institute (Palo Alto, CA) Grant No. RP2796-
02, American Electric Power Company (Columbus, OH) Grant No. C-8276, Ohio
Edison Company (Akron, OH), and the Ohio State University (Columbus and
Wooster, OH).

References

1.	Land Application Uses for Dry FGD By-Products, Phase 1 Columbus, OH:

Department of Agronomy, The Ohio State University, April 1993.

2.	C. L Carlson and D. C Adriano. "Environmental Impacts of Coal Combustion

Residues." Journal of Environmental Quality. Vol 22, No. 2, p. 227 (1993).

3.	R. K. Fowler, J. M. Bigham, S. Traina, U. I. Soto, R. G Stehouwer, and E. L.
McCoy. "Properties of Qean Coal Technology By-Products." Agronomy
Abstracts. Madison, WI: American Society of Agronomy, 1993, p. 361.

4.	R. F. Korcak "Fluidized Bed Material as a Lime Substitute for Liming Leached

Mineral Soils." Journal of Environmental Quality. Vol. 9, No. 1, p. 147 (1980).

5.	G. L Term an, V. J. Kilmer, C M. Hunt, and W. Buchanan. "Fluidized Bed Boiler
Waste as a Source of Nutrients and lime." Journal of Environmental Quality.
Vol 7, No. 1, p. 147 (1978).

6.	H. L. Bohn, B. L. McNeal, and B. A O'Connor. Soil Chemistry, 2nd Edition
New York, NY: John Wiley and Sons, 1985.

7.	F. J. Stevenson. Humus Chemistry; Genesis, Composition, Reactions New
York, NY: John Wiley and Sons, 1982,

8.	N. V. Hue, B. R. Craddock, and F. Adams. "Effect of Organic Acids on

Aluminum Toxicity in Subsoils." Soil Science Society of America Journal. Vol
50, No. 1, p. 28 (1986).

9.	E J. Kamprath. "Exchangeable Aluminum as a Criterion for Liming Leached
Mineral Soils." Soil Science Society of America Journal. Vol. 34, No ?, p. 252
(1970).

10.	P. Sutton and W. A Dick. "Reclamation of Acidic Mined Lands in Humid

Areas." Advances in Agronomy. Vol 41, p. 377 (1987).

104-9


-------
11.	P. B. Woodbury. "Trace Elements in Municipal Solid Waste Composts: A

review of Potential Detrimental Effects on Plants, Soil Biota, and Water Quality."
Biomass and Bioenergy. Vol. 3, p. 239 (1992).

12.	R L Chaney, R J. F. Bruins, D. E. Baker, R F Korcak, J. E. Smith, and D. W. Cole.

"Transfer of Sludge-Applied Trace Elements to the Food-Chain," Land
Application of Sludge: Food Chain Implications. Chelsea, Michigan: Lewis
Publishers. 1987, pp. 67-99.

Table 1

Characterization of LIMB, PFBC(cycI), and Sewage Sludge Amendments.

Parameter

LIMB

PFBC(cycl)

Sewage Sludge

Particle size (%)





na*

Sand (0.05-2mm)

0

25.5



Silt (2-50pm)

90

74.1



Clay (<2pm)

10

0.4



Mineralogy (%)





na

Anhydrite (CaS04)

25

22



Calcite (CaCOs)

15

11



Dolomite (CaMg(C03)j)

nd2

23



Lime (CaO)

21

nd



Portlandite (Ca(OH)2)

5

nd



Periclase (MgO)

nd

13



Fly ash

30

32



CaCO, equivalent (%)

59.4

60.3

0

pH (1:1, water)

12,5

10.5

6.5

Total Chemical Analysis







Organic C (%)

na

na

31.2

Ca (%)

35.96

17.53

2.76

Mg (%)

0.60

10.64

0.34

S (%)

5.77

5.21

1.41

A1 (%)

3.52

3.93

3.40

Si (%)

6.58

7.24

na

Fe (%)

5.56

5.17

1.24

Na (%)

0.33

1.03

0.07

N (%)

na

na

3.8

P (%)

0.02

0.02

1.78

K (%)

0.91

0.50

0.15

Ba (%)

0.03

0.02

0.01

As (mg kg"1)

55.1

1.9

<0.03

B (mg kg*1)

233.1

171.2

31.1

Cd (mg kg1)

1.0

1.9

6.3

Cr (mg kg"')

28.0

36.9

315.2

Cxi (mg kg"1)

21.0

52.5

1174

Pb (mg kg1)

16.0

16.0

16.1

y-s

I
1

5.9

6.6

11.2

Ni (mg kg"1)

31.1

52.1

166.4

Se (mg kg"1)

8.1

5.6

<0.3

Zn (mg kg"')

86.0

74.0

1494

"not analyzed

104-10


-------
Table 2

Leachate Concentrations of Elements Regulated With Respect to Land Application of Sewage Sludge,*

Treatment

As

B

Cd

Cr

Cu

Mo

Ni

Se

Zn































lllfti jLj









Soil alone

<0.08

0.13

0.021

<0.004

<0.007

<0.018

0.06

<0.27

0.645

Soil + IX lime requirement (PFBC)

<0.08

0.80

0.019

<0.004

0.071

<0.018

0.04

<0.27

0.012

Soil + 2X lime requirement (PFBC)

<0.08

1.54

0.012

<0.004

0.064

0.024

0.03

<0.27

0.032

Spoil + sewage sludge

<0.08

0.41

<0.003

0.009

0.041

<0.018

0.24

<0.27

3.34

Spoil + sewage sludge + 12% PFBC

<0.08

0.92

<0.003

0.021

0.055

0.089

0.01

<0.27

0.06

Spoil + sewage sludge + 12% LIMB

<0.08

3.73

<0.003

0.008

0.026

<0.018

0.01

<0.27

0.05

' Not listed are Hg and Pb which were below detection limits (0.04 mg/L) in all cases.


-------
Tables

Dry Matter Yields of Alfalfa And Fescue 125 Days After Seeding When Grown on Fleming AML
Spoil Treated With Different Rates of LIMB and Spray Dryer FGD By-Products.

Dry Matter Production (a/column)

Treatment

Alfalfa

Fescue

in j- t

Total

Control

0

0

0

10% UMB

6.8

1.6

8.4

20% LIMB

0.04

0.5

0.54

40% UMB

0

0.1

0.1

10% Spray Dryer

14.2

1.0

15.2

20% Spray Diyer

2.8

0.09

2.89

40% Spray Dryer

0

0.01

0.01

LSD (0.05)

3.9

3.2

3.6

0.0 0.7 1.4	2.8

RATE (weight %)

Figure t, pH, aluminum and manganese leachate concentrations torn Wooster soil
treated with PFBC by-product The first leaching was conducted after _
seeding of the pots to alfalfa or fescue and the last leaching was after nine
months of growth.

104-12


-------
RATE (weight %)

Figure 2. pH, aluminum and manganese leachate concentrations from Fleming

AML spoil treated with PFBC by-product The first leaching was conducted
after seeding of the pots to alfalfa or fescue and the last leaching was after
nine months of growth.

104-13


-------
260

Cft

c

CD

40

°2G CuKgt

Figure 3, Representative X-ray diffraction pattern for Fleming AML spoil treated with 24% PFBC by-product.


-------
0

1

tn

0 3 6	12	24

RATE (weight %)

o

V

PFBC first leaching
LIMB first leaching

PFBC last leaching
LIMB last leaching

12

O

Q.

12

cn

a

LlJ
>-

<
U_

<

12

PFBC(cycl).

PFBC(bed)

o First Harvest
• Third Harvest
v Sixth Harvest

Figure 4 Suite concentrations in leaehates from Wooster soil and Fleming AML Figure 5.
spoil treated with either LIMB or PFBC by-product. The first leaching was
conducted after seeding of the pots to alfalfa or fescue and the last leaching
was after nine months of growth.

0.0 0.7 1.4	2.8

RATE (weight %)

Alfalfa yields (first, third and sixth harvests) from pots containing
Wooster soil treated with LIMB and PFBC (source, bed and cyclone)
by-products.


-------
o

CL

cn

Q

L±J

LlI
3
O
C/5
LlJ

10

6
4
2
10

i	r

LIMB

4 -

0

o First Harvest
• Third Harvest
v Sixth Harvest

+

PFBC(cycI)

12

24

RATE (weight %)

Figure 6. Fescue yields (first, third and sixth harvests) from pots containing Fleming
AML spoil treated with LIMB and PFBC by-products.

104-16


-------
The Stabilization of Orimulsion
Spray Dryer Waste for Landfill Disposal

S. Kuchibotla
Law Engineering, Inc.
Jacksonville, Florida, 32207

EH. Kalajian
Honda Institute of Technology
Department of Civil Engineering
Melbourne, Florida 32901

C-S. Shi eh
Florida Institute of Technology
Department of Environmental Science
Melbourne, Florida 32901

K.R. Olen
Florida Power & Light Co,
Engineering & Advanced Technology
Juno Beach, Florida 33408

Abstract

In this paper the results of a study to develop a fixation process to stabilize
Orimulsion lime spray dryer (LSD) waste are presented. Orimulsion is a candidate
alternative fuel for utility boilers because of its low cost and outstanding handling
and combustion characteristics. An obstacle to commercializing Orimulsion is the
need to install backend environmental control systems to reduce particulate and SOz
emissions. The objective of the study was to determine physical and chemical
properties of the raw Orimulsion LSD waste and its stabilized products and to
determine if the stabilized product is suitable for disposal in landfill. Hie results of
engineering studies indicated that the Orimulsion LSD waste had physical properties
similar to clay soil with low compressibility; however, the Orimulsion LSD waste has
a very low bulk density. The compacted raw Orimulsion LSD waste was non-
pozzolanic and is not recommended for landfill disposal. The stabilized Orimulsion
LSD waste was found to be acceptable for landfill disposal based on its compacted
dry density and compressive strength. Surface hardness of the stabilized Orimulsion
LSD waste product was found to be acceptable for transportation and handling. The
results of chemical studies have shown that stabilization of Orimulsion LSD waste
using Type I Portland cement minimized the leaching of Ca, V, and SO/" to
surrounding aqueous solutions. It is concluded that, using the mix design developed
in the study, Orimulsion LSD waste can be stabilized and disposed of in landfill.

105-1


-------
Intro due tion

Florida Power & Light Co. (FPL) has been, evaluating the possibility of converting one
or more of its utility boilers from residual fuel oil-firing to Orimulsion Orimulsion is
a new fuel form consisting of small amorphous particles of a naturally occurring
bitumen emulsified into water. It exhibits non-Newtonian pseudoplastic flow
properties and physically resembles black latex paint. The bitumen used to formulate
Orimulsion is mined from large reserves in Venezuela's Orinoco River basin.
Currently, Bitumenes Orinoco (BITOR), a subsidiary to the Venezuelan state owned
oil company, Petroleos de Venezuela, S.A., is marketing Orimulsion as an alternative
utility boiler fuel at coal equivalent prices.

Orimulsion resembles a high-sulphur Eastern bituminous coal, in terms of heat and
fuel sulphur contents, 12,700 Btu/lb and 2.90%, respectively. However, unlike the
experience of firing a boiler on coal, Orimulsion produces very little fly ash, e.g.,
about 0.10 gr/DSCF. In addition, Orimulsion fly ash is dominated by submicron
particles of non-pozzolanic magnesium and vanadium oxysulphates. In this regard
Orimulsion resembles a high-vanadium, high-sulphur Venezuelan residual oil, that
had been treated with a magnesium-based fuel additive.

A conversion to Orimulsion would require the installation of environmental control
systems to reduce particulate and S02 emissions. Based on the experience of
reducing S02 flue gas emissions from burning high-sulphur coal, the flue gas
desulphurization (FGD) system of choice would be a wet limestone forced-oxidation
(WLFO) scrubber. Recent information, however, strongly suggested that the life-cycle
costs of a lime spray dryer (LSD) may be competitive against WLFO systems for 90+%
FGD of flue gases with high S02 concentrations (1,2). The major variables to be
compared when evaluating these two competing processes are capital requirements,
the relative costs of limestone and lime, sorbent utilizations, and the costs associated
with solid waste disposal.

With regard to the latter of these variables, WLFO scrubbers produce gypsum, which
can be used for wallboard manufacture, soil conditioning, or safely stacked in an
approved landfill without further processing. The solid waste from a LSD, in
comparison, has no commercial value, as it is primarily calcium sulphite, and must,
therefore, be stabilized and landfilled. Solid waste stabilization is effectively and
inexpensively accomplished in the case of a coal fired plant by mixing the pozzolanic
coal fly ash with the FGD product, and water. The mixture is typically placed in an
approved landfill, where it is compacted and cured thereby forming an
environmentally acceptable waste.

Given the chemical characteristics of Orimulsion fly ash, an economical solid waste
stabilization process had to be developed, before LSD technology could be fully
evaluated. Since Orimulsion and residual oil fly ashes are very similar, it was
believed that the oil ash stabilization process developed at Florida Institute of
Technology (Florida Tech) under FPL sponsorship (3) might provide a basis for a

105-2


-------
process to stabilize Orimulsion LSD solid waste. Accordingly, FPL contracted
investigators at Florida Tech to carry out a study to determine the process
requirements to stabilize Orimulsion LSD solid waste.

The unconfined compressive strengths recommended for landfill disposal of ash
wastes are 100-150 psi (4,5, 6). Thompson et al (7) stated that development of
compressive strengths by spray dryer wastes in excess of 25 psi is considered
adequate to support the weight of the material when used in a fill, as well as the
weight of the trucks and other equipment used for placement and compaction. Spray
dryer wastes with compressive strengths in excess of IS) psi can be expected to
perform acceptably as structural fill material (4).

Preparation of Simulated Lime Spray Dryer Solid Waste

Orimulsion LSD solid waste samples were prepared by Radian Corp., Austin, Texas
using a laboratory-scale spray dryer. To simulate commercial operations a synthetic
flue gas was generated by burning natural gas at high excess air levels such that the
adiabatic flame temperature was essentially the desired spray dryer inlet temperature
of about 325°F. Gaseous SOz was metered into the synthetic flue gas to maintain a
concentration between 2450 to 2500 ppm(v). Feed slurry consisted of slaked lime,
recycled solids from a previous run carried out under similar conditions, and a
quantity of Orimulsion fly ash. The rate of addition of fly ash was adjusted to
approximate a flue gas concentration of about 0.10 gr/DSCF. Inlet gas velocity was
approximately 275 acfm, and the spray dryer was operated at a 22°F approach
temperature.

Despite the unorthodox technique used to introduce the Orimulsion fly ash, it was
believed that the solid wastes produced from the laboratory-scale spray dryer were
reasonably representative of those that would be produced in a full-scale commercial
operation. Accordingly, about 25 kg of waste sample was collected for each of two
desulphurization runs. The aim desulphurization levels were 72.5 and 90%, however,
because the laboratory spray dryer is normally operated at a constant slurry feed rate
to achieve a constant calcium-to-sulphur molar ratio it was difficult to control the test
unit precisely to a predetermined S02 removal level. After adjusting for air in
leakage and moisture, the corrected FGD percentages for the two samples were 64
and 85%.

Calcium utilization for the low and high desulphurization runs were 100 and 94%,
respectively. These utilizations are somewhat higher than those that might be
expected in a full-scale commercial unit because of scale down factors that make the
laboratory-scale unit more efficient The higher efficiency was also demonstrated by
uncharacteristically low calcium-to-sulphur molar ratios of 0.64 for the 64% FGD run
and 0.90 for the 85% FGD case.

105-3


-------
Methodologies

Characterization of Orimulsion LSD Waste. Methods used to characterize the raw
and stabilized Orimulsion LSD wastes are listed in Table 1. The physical properties
determined were moisture content, grain size distribution, specific gravity, Atterberg
limits, bulk density and moisture-density relationships, while the engineering
properties determined were unconfined compressive strength and surface hardness.
Moisture content was determined using procedures outlined by American Society of
Testing Materials (ASTM) C 566-78 (8) 'Total Moisture content of Aggregates by
Drying". Grain size distribution was determined using procedures outlined by ASTM
D 422-63 (8) "Standard Method for Particle Size Analysis of Soils". Specific gravity
was determined using procedures outlined by ASTM D 854-83 (8) "Standard Method
of Test for Specific Gravity of Soils." The Atterberg limits were determined using
procedures outlined by ASTM D 4318-84 (8) 'Standard Test Methods for Liquid limit,
Plastic Limit and Plasticity Index of Soils." The bulk density was determined using
procedures outlined by ASTM D 698-78 (8) and ASTM 1557-78 (8). The moisture-
density relationships was determined using procedures outlined by ASTM D 558-82
(8) "Moisture-Density Relationships of Soil-Cement Mixes." The unconfined
compressive strength tests were conducted in accordance with ASTM D 2166-85 (8)
'Test Method for Unconfined Compressive Strength for Cohesive Soil." The
American Foundrymen's Association method for evaluating the surface hardness of
the baked and cured cores was used to determine the hardness of the samples (9).

The chemistry study was carried out to determine the concentrations of Ca, V, and
S042" in both raw and stabilized Orimulsion LSD wastes. Dried and ground samples
were digested using the hydrofluoric-boric acid (HF-H3B03) technique (10). The
digest was analyzed for Ca and V using the Perkin-Elmer Model 5100 PC Zeeman
atomic absorption spectrophotometer (AAS) equipped with HGA-600 graphite
atomizer and AS-60 autosampler. Sulfate was analyzed using the Shimadzu UV-VIS
Recording Spectrophotometer UV-160A according to the method described in
Standard Methods for the Examination of Water and Wastewater (11).

Stabilization and Fabrication. In the stabilization methodology, Type I cement (10%
or 15%) and sodium metasiHcate commonly known as Metsobeads (0.75%, 1%, 1.25%,
or 1.5%) were used to stabilize both 64% and 85% LSD wastes. It has been
considered that use of Metsobeads will improve the engineering characteristics of
cement products. The moisture contents selected were +3, +5, +10, +20, +35 and +50
percent above Optimum Moisture Content (OMC). The OMC is defined as the
moisture content at which the soil can be compacted to the maximum dry density by
using given compaction (12). To obtain a relationship between the moisture content
and the unconfined compressive strengths of the LSD wastes, samples were
fabricated at various moisture contents relative to the OMC. Based on the results
obtained, the moisture content at the maximum strength could be determined.

The test samples were fabricated using the Harvard Miniature Apparatus (2.81 inches
in height and 1.31 inches inner diameter). A forty pound spring was used in the

105-4


-------
tamper for compaction and twenty-five tamps per layer were applied to each of the
five layers, which approximates Standard Proctor compaction. After compaction, the
samples were wrapped in plastic wrap and placed in plastic boxes. These boxes
(with their top open) were placed in the curing chamber (100% humidity). The air
dry samples were left in the sealed plastic boxes on a laboratory shelf at room
temperature.

Determination of Leaching Characteristics. Leaching characteristics of the selected

stabilized Qrimulsion LSD waste was determined by conducting both column and
tank leaching studies. The column leaching test was conducted according to the
method by Harder et al. (13), which simulated the percolation of natural waters
through a stabilized product The tank leaching test was based on the work by
Duedall et al. (14), van der Sloot et al. (15), an van der Sloot and De Groot (16),
which simulated the immersion of a stabilized product in a aqueous solution.

Results and Discussion

Physical Properties. Both the 64% and 85% Grimulsion LSD wastes had low
moisture contents ranging from 0.1% to 0.5% (Table 2). The mean grain size,
uniformity coefficient and coefficient of gradation as reported in Table 2 showed little
variation between the two Orimulsion LSD wastes. Therefore, the material can be
characterized as uniform and in the silt-clay size range. The specific gravity for the
64% and 85% LSD wastes were found to be 2.50 and 252, respectively (Table 2). The
Atterberg limits for the 64% LSD waste were as follows: liquid limit 35%, plastic limit
10%, and plasticity index 25% (Table 2). For the 85% LSD waste the liquid limit was
36%, plastic limit was 10%, and plasticity index was 26%. The results indicated that
both 64% and 85% LSD wastes had similar plasticity characteristics. The aerated and
settled bulk density for the 64% LSD waste were 27 pcf and 35 pcf, respectively
(Table 2). The aerated and settled bulk density for the 85% LSD waste were 30 pcf
and 35 pcf, respectively. Hie void ratios determined for both the wastes were in the
range 3.5-4.8 and were high compared to the void ratio of typical clay soil which
ranges between 1 to 2. The uniform grain size distribution of the Orimulsion LSD
waste results in poor packing and compaction of the wastes, thus providing low bulk
density and high void ratio.

Based on the Atterberg limits and grain size distribution the 64% and 85%

Orimulsion LSD wastes were classified as CL soil (clay of low compressibility) in the
Unified Soil Classification System according to ASTM D 2487-85 (8) "The Method for
Classification of Soils for Engineering Purposes."

105-5


-------
Table 1

Methods Used to Characterize Orimulsion LSD Waste.

Parameter

Method

Reference

moisture content

ASTM 566-78

8

grain size distribution

ASTM D 422-63

8

specific gravity

ASTM D 854-83

8

Atterberg Ixmits

A^STM D 4=310 — 84

8

bulk density

ASTM D 698-78

8



ASTM 1557-78

8

moisture-density relationships

ASTM D 558-82

8

unconfmed compressivs strength.

ASTM D 2166-85

8

Classification

ASTM D 2487-85

8

surface hardness

The American





y t it 1Y1 ^ ' S





Association Method 9

acid digestion

hf-h3bo3

10

leach.2-Tig* ch3ac t e^u xstxc

c^?lumLk leachxng

13



tank leaching

14, 15, 16

Table 2

Physical Properties of Raw Orimulsion LSD Waste.

Parameter

64% LSD waste

85% LSD

waste





moisture content (%5

0.1-0.5

0.2-0.5

mean particle size (Dso , mm)

0.013

0.011

uniformity coefficient

1.263

1.250

coefficient of gradation

0.88

0.90

specific gravity

2.50

2.52

Atterberg Limits





liquid limit (%)

35

36

plastic limit {%)

10

10

plasticity index (%)

25

26

bulk density {aerated, lb/ft3 }

27

30

g/cm3

0.433

0.480

bulk density (settled, lb/ft3 5

35

35

g/cm3 5

0.561

0.561

void ratio (aerated)

4.8

4.2

void ratio (settled)

3.5

3.5

105-6


-------
Chemical Properties. The concentrations of Ca, V, and S042" in raw Orimulsion LSD
waste are shown in Table 3. The concentrations of Ca in 64% LSD waste and 85%
LSD waste were approximately 25%. The 85% LSD waste showed a slightly lower V
concentration than the 64% LSD waste (0.19% and 0.22%, respectively). Hie sulfate
concentration in 85% LSD waste and 64% LSD waste were similar (17.3% and 16.7%,
respectively).

Enrichment of Ca in Orunulsion LSD waste resulted mainly from the scrubbing
process. In the scrubbing process, a suspension of Ca(OH)2 or CaC03 was used to
remove S02 from the flue gas resulted in the formation of calcium sulfate
hemihydrate (CaS04 -44H20) which was subsequently oxidized to yield calcium
sulfate dihydrate (CaS04 -2H20), i.e., flue gas gypsum (17). Major Ca compounds in
scrubber residues identified by X-ray diffraction are caltite, gypsum, and
hannebachite (CaS04 ~W20) (18,19).

The V concentration measured in the Orimulsion LSD waste was over 10 times
higher than the approximate 71 to 180 jig g"1 concentration found in coal scrubber
by-products (20). The V in the ash may have had an effect on the V concentration in
the scrubber residue because the composition of scrubber residues is, in general,
dependent partly on ash generation (17). Vanadium as V205 is found in flue dust
after fossil fuel combustion (21), which indicates a possible source of V in the
Orimulsion LSD waste.

In the LSD waste, the S042" was mainly derived from the sulfur dioxide removed in
the flue gases. In the scrubbing process, S02 is absorbed to form ultimately
CaS04 -%H20, CaS04 -2HzO or a combination of the two (17). Therefore, the S042"
measured from the acid digest of the Orimulsion LSD waste was correlated to the
amount of SO, removed from the flue gases.

Compaction and Strength of Orimulsion LSD Wastes. Mix designs for stabilizing
Orimulsion LSD wastes, the moisture contents evaluated and the 28-day compressive
strength are shown in Table 4. Factors examined included the content of cement,
water, metsobeads (sodium metasilicate) additives, and curing time. The unconfined
compressive strengths were determined at different compacted moisture contents to
obtain a relationship between the unconfined compressive strength and the moisture
content of the Orimulsion LSD wastes arid to determine the ideal mixing moisture
content. The 28-day compressive strengths for the 10 percent cement stabilized 64%
and 85% Orimulsion LSD waste samples increased between OMC and +3 percent
above OMC. As mentioned above, the OMC is defined as the moisture content at
which the soil can be compacted to the maximum dry density by using given
compaction. This indicates that sufficient water was available for the hydration of
cement to complete between OMC arid +3 percent above OMC. The unconfined
compressive strengths for both the 15 percent cement stabilized 64% and 85%
Orimulsion LSD waste samples exhibited similar properties and gained maximum

105-7


-------
Table 3

Concentration of Ca, V, and S04*~ in Orimulsion LSD Waste.

Elements	64% LSD Waste	85% LSD Waste

Ca (%)	25.1±0.2	25.7±0.2

V	(pg g"1)	2182*7	192Q±6

SO/" (%}	16.7±1.5	17„3±0.6

Standard deviation, ti— 3 .

Table 4

Orimulsion LSD waste Mix Designs Investigated in the Study.

LSD waste cement metsobeads moisture dry	28-d strength.

(%)	{%)	(%}	content density	(psi)b

{%)	(Ibs/cf)a

64	0	0	20	50.7	72

64	10	0	24-74	43.4-55.3	94-172

64	10	0.75-1.5	27	55.4	160-168

64	15	0	26-31	55.0-56.4	200-225

85	0	0	21	50.3	88

85	10	0	25-75	43.1-55.4	71-134

85	10	0.75-1.5	28	54.8-55.0	121-133

85	15	0	29.5-34.5 55.2	155-274

a 62.4 Ibs/cf = 1 g/cm3
b 1 psi = 6.89 kPa

105-8


-------
strengths at 3 percent above their corresponding OMC's. The results indicated that
the moisture content at mixing plays an important role in determination of the
engineering properties. The moisture content at which all the cement stabilized
waste samples obtained their maximum unconfined compressive strength was +3
percent above their corresponding OMC.

The unconfined compressive strengths of the 10 percent cement stabilized 64%
Orimulsion LSD waste samples were higher than the unconfined compressive
strengths obtained for the 10 percent cement stabilized 85 percent Orimulsion LSD
waste samples. As mentioned by Rodriguez et al. (4) the presence of sulfates (in
solid state) in soils decreases the unconfined compressive strengths obtained for the
soil-cement mixes. The presence of higher sulfate content in the 85% Orimulsion LSD
wastes explains the lower compressive strengths for the 10 percent cement stabilized
85% Orimulsion LSD waste samples. Rodriguez et al. (4) also indicated in their
research, the cement content plays an important role in improving the engineering
properties of the stabilized soils. It was found in the study that the cement improved
the unconfined compressive strength of the 15 percent cement stabilized 85% waste
samples, thus providing them higher strengths than the 15 percent cement stabilized
64% waste samples. Increase the content of cement in the mix overrided the effect of
sulfate on stabilization. Metsobeads had been found no effect on improving the
unconfined compressive strengths of the 10 percent cement stabilized Orimulsion
LSD wastes.

The effects of curing condition on the various mix designs were found to have very
little effect on the compressive strength. Samples cured in air had slightly lower
compressive strength (< 10%) than samples cured in a chamber. The compressive
strength of stabilized Orimulsion LSD waste increased with time, however, the
stabilized waste has greater than 60% of its compressive strength developed after 7
days as compared to compressive strengths at 28 days.

The surface hardness of the cement stabilized 64% arid 85% Orimulsion LSD wastes
were found to be in the range of 75 and 85, which classifies them as very hard. The
stabilized mixes can be transported mechanically, with minimum breakage.

Characteristics of Stabilized Orimtils ion LSD Waste. The optimum moisture
content, the maximum dry density and void ratios for the stabilized wastes are listed
in Table 5. As seen in the table there is no significant difference between the 10
percent cement stabilized 64% and 85% Orimulsion LSD wastes and the 15 percent
cement stabilized 64% and 85% Orimulsion LSD wastes. The dry densities for the
stabilized wastes increased approximately 4-6 pcf over the uristabilized wastes and
the OMC for the stabilized wastes increased approximately 4-9 percent over the
uristabilized wastes.

The void ratios for the stabilized 64% and 85% Orimulsion LSD wastes were in the
range of 1.8-1.9. The void ratios for the cement stabilized Orimulsion LSD wastes

were found to have reduced by 85 percent from the void ratios obtained for the raw

105-9


-------
wastes and 10 percent lower than the void ratios obtained lor the unstabilized
Orimulsion LSD waste. This indicates that the void ratio decreased with the addition
of cement, improving the dry density. Dry densities of the Orimulsion LSD waste
increased 66% when compacted at OMC as compared to the settled bulk density
shown in Table 2.

Table 5 also shows the concentrations of Ca, V, and S042' in stabilized Orimulsion
LSD products. Both mixes, e.g., 85% LSD-10% cement (mix 85%-10) and 85% LSD-
15% cement (mix 85%-15), had similar Ca, ¥, and S042" concentrations. The Ca
concentration was approximately 7% higher in the stabilized LSD products than in
the raw LSD waste (85% LSD). The concentration of V in mix 85%-lQ and mix 85%-
15 were 1310 p.g g'1 and 1270 jig g~\ respectively, which were about 600 |ig g'1
decrease from the raw LSD waste. The sulfate concentration, 18.0% in both mixes,
did not change significantly in the stabilized LSD products compared to 17.3% in the
raw LSD waste.

The addition of Portland cement with Ca additives (lime, limestone, and calcium
silicates) contributed to the high concentration of Ca detected in the mixes. The
primary Ca compounds in the stabilized products may be the products of the
hydration of cement, such as calcium hydroxide, calcium sulfate,
and calcium silicates (22), or scrubbing by-products such as CaCO^ CaS04 -2H20, and
CaS04 J/iH,0 (17).

Tank Leaching Study. The results of tank leaching study are expressed as the rate
of metal ions released per unit geometric surface area of the stabilized Orimulsion
LSD product, i.e., metal flux. The flux of leached metal was calculated using the
following equation:

CmxV

J=	

MW x A x t

where

J = Metal flux (mmole/mm2/d)

Cm = Measured concentration (mg/1)

V = Volume of leaching medium (1)

MW = Molecular weight (mg/mmole)

A = Surface area of the leaching sample (mm2)
t = Time lapsed between sample collection (d)

105-10


-------
Tables

Physical Properties arid Concentrations of Ca, V, and S042* in Stabilized Orimulsion
LSD Waste (n=3).

% LSD
Waste

Cement
(%)

OMC

(%)

Max Dry
Density
(lbs/cf)a

Void
Ratio

(%)

Ca

(%)

¥
fag/g)

so42-
(%)

64

0

20

50.7

2.1

NA

NA

NA

64

10

24

55.5

1.9

NA

NA

NA

64

15

26

56.4

1-8

NA

NA

NA

85

0

21

50.3

2.1

NA

NA

NA

85

10

25

55.4

1.9

32.0±2.5

1310±260

18.0±3,0

85

15

29.5

55.2

1.9

32.3±2.6

1270±200

18.0±3.0

a 62.4 lb/cf = 1 g/cm3

Both mixes have nearly identical flux values for Ca (Figure 1). The highest flux
values of Ca occurred within the first 24-hour (5.8 x 10"4 mmole/mm2/d and 7.0 x 10"6
mmole/mm2/d for mix 85%-10 and mix 85%-15, respectively) and decreased
exponentially after 48-hour. The decreasing Ca flux continued throughout the period
of the experiment. At the end of the study, approximately 8% of the total Ca was
released from mix 85%-10 and 10% from mix 85%-15.

For the leaching of V (Figure 2), both mixes exhibited similar flux trends with the
highest flux values occurred within the initial 12-hour (1.3 x 10"4 mmole/mm2/d).
Thereafter, the flux values decreased exponentially and approached -9.8 x 10"*
mmole/mm2/d by the end of the study. Approximately 0.2% of the total V was
released from both mixes by the end of the study.

For the leaching of S042", both mixes also showed very similar flux trends (Figure 3).
The highest flux values were detected initially, which was 1.9 x 10"5 mmole/mm2/d
for mix 85%-15 and 1.6 x 10"5 mmole/mm2/d for mix 85%-10. After initial leaching,
the S042" flux decreased exponentially until the 100-hour period and leveled down
slowly. Approximately 4% of S042" from both mixes was released by the end of the
study.

The flux values detected during the initial 24-hour period for Ca, V, and S042" can be
attributed to the availability of soluble fractions on the surfaces of the stabilized

105-11


-------
products. Shieh et al. (23) reported that leaching of elements from stabilized ash
products was confined to the surface layer less than 1 cm. The leachable Ca
originated from the ca-minerals present in the scrubbing reagents and the added
Portland cement, such as calcium hydroxide (Ksp = 6.5 x 1CT6), gypsum (Ksp = 2.5 x
10"5) and lime (17,24). As the leachable Ca faction on the surface of the stabilized
product becomes depleted, leachable Ca from the inner regions become an important
source contributes to leaching. Leachable Ca within the stabilized product can
diffuse toward the surface and into the leaching medium (25).

Similar leaching characteristic to Ca was also found for V leaching. The rapid
decrease in flux values after the initial 24-hour indicated that a major portion of
leachable V was released initially from the surface of the sample when the stabilized
product was exposed to the test solution.

The soluble form of SO/", such as flue gas gypsum which was available on the
surface and within the stabilized product, was released to the leaching medium.
Similar to Ca and V, the observed rapid decrease in sulfate flux after 24-hour could
be attributed to the depletion of soluble gypsum on the surface of tested samples.
Sulfate release is sustained by the presence of leachable S042" within the stabilized
product The flux continued to decrease, however, demonstrates the effect of
tortuosity and S042" depletion in the outer layers.

Column Leaching Study. The results for column leaching are expressed in terms of
the rate in moles of Ca, V, and S042* per hour. The following equation was used to
calculate the leaching rate (mole hr'1).

Cm x V x MW

R =	

t

where

R = Rate of leaching (mole/hr)

Cm = Measured concentration (g/1)

V = Volume of leachate collected (1)

MW = Molecular weight (g/mole)

t = Time lapsed between sample collection (hr)

The pH in the column leachate was high initially, near 12, and gradually decreased to
pH=ll by the end of the study period (360 hr). The leaching rate of Ca for both
mixes (Figure 4) was highest initially then decreased to a rate of 2 x 1£T5 mole/hr by
the end of the study. Mix 85%-10 showed a higher initial Ca leaching rate (1.1 x W4
mole/hr) than Mix 85%-15. At the end of the study approximately 10% of total Ca
was released from mix 85%-10 and 6% from mix 85%-15.

105-12


-------
High leaching rate of sulfate occurred within the initial 12-hr (24 to 60 x 10"5
mole/hr). The leaching rate was then decreased to less than 5 x 10"* mole/hr near
150-hour and remained level to the end of the study (Figure 5). The trends of
leaching rate for SO/" was very similar to that for Ca. At the end of the study,
approximately 4-5% of total S042" was released from both mixes.

The leaching rate of V for stabilized LSD mixes was somewhat different. Large
fluctuation in leaching rate was observed (Figure 6). At the end of the study,
approximate 1% of .total V was released from mix 85%-10 and 0.5% from mix 85%-15.

For column leaching, as test solution flowed through the samples, more surface area
within the stabilized LSD waste were exposed to the leaching medium.

Consequently, leachable elements originated at various points within the stabilized
product would become mobile when came in contact with test solution. Soluble Ca
within the stabilized product was readily available wherein Ca compounds, such as
calcium hydroxide, gypsum, and unreacted lime, contributed to the dissolved Ca
detected in the leachate.

The leaching characteristic of sulfate was similar to calcium. Soluble forms of S042*,
such as gypsum, was released from the stabilized product. High initial leaching rate
indicated that the majority of soluble sulfate was released immediately into the
leaching medium.

Conclusions

Based on the results of the study, a cement (Type I Portland) content of 10 to 15
percent mixed at a water content of 3 percent above the optimum moisture content
(OMC) is recommended for the stabilization of the Orimulsion LSD wastes to
increase dry density and to meet the unconfined compressive strength and hardness
requirements for landfill disposal. Stabilization of Orimulsion LSD waste minimized
the leaching of Ca, V, and S042". In general, the highest leaching of elements from
stabilized products occurred during the initial exposure periods, thereafter, leaching
decreases exponentially.

105-13


-------
References

1.	R. Rudy, Private Communication.

2.	K. Felsvang, B. Brown, and R. Horn, "Proceeding of title NLA Retrofit FGD
Effective Use of Lime Seminar," Philadelphia, PA, January 9-10,1991.

3.	C. Shieh, I.W. Duedall, E.H. Kalajian, and J.R. WOcox, 'Stabilization of Oil Ash
for Artificial Reefs: An Alternative to the Disposal of Oil Ash Waste." The
Environmental Professional, Vol. 11, pp 64-70 (1989).

4.	A.R. Rodriguez, D.J. Castillo, and F.G1. Sowers. Soil Mechanics in Highway
Engineering. New York: Trantech Publications Company, 1988.

5.	P.H. Wright and R.J. Paquette. Highway Engineering. New York: John Wiley
and Sons, 1987.

6.	O.G. Ingles and B.V. Metcalf. Soil Stabilization: Principles and Practice.
Australia: Butterworths Pty. Limited, 1972.

7.	C.M. Thompson, R.D. Anchord, and G.M. Blythe. "Laboratory Characterization
of Advanced S02 Control By-Products: Spray Dryer Wastes." Prepared for
Electric Power Research Institute, Palo Alto, California, May 1988.

8.	ASTM Annual Book of ASTM Standards. American Society of Testing and
Materials. Philadelphia. Pa., 1987.

9.	E.J. Yoder. Principles of Pavement Design. New York: John Wiley and Sons,
1959.

10.	D. Silberman and G.L. Fisher. "Room-Temperature Dissolution of Coal Fly
Ash for Trace Metal Analysis by Atomic Absorption Spectrometry." Analytica
Chimica Acta 106, pp. 299-307 (1979).

11.	M.A. Franson. Standard Methods for the Examination of Water and
Wastewater. American Public Health Association, Washington DC,
pp. 4-207 - 4-208 (1989).

12.	ASTM Annual Book of ASTM Standards. American Society of Testing and
materials, Baltimore, Maryland, 1982.

13.	P.J. Harder, M.J. Marcinak, N.J. Schlotter, A.L. Labotka, and I.W. Duedall.
"The Fixation of Fly Ash: Physical and Leachate Properties." Consolidated
Edison Company of New York, Inc., New York, pp. 121-169 (1981).

14.	I.W. Duedall, J. S. Buyer, M. G. Heaton, S. A. Oakley, A. Okubo, R. Dayal, M.
Tatro, F. J. Roethel, R. J. Wilke, and J. P. Hershey. "Diffusion of Calcium and
sulfate Ions in Stabilized Coal Ash." In: Waste in the Ocean. Vol. 1. Industrial
and Sewage Wastes in the Ocean. I.W. Duedall, B. H. Ketchum, P. K. Park,
and D. R. Kester, eds., Wiley-Interscience, New York, PP. 375-395 (1983).

15.	H.A. van der Sloot, J. Wijkstra, C. A. Van Stigt, and J.Hoede. "Leaching of
Trace Elements from Coal Ash and Coal-Ash Products." Iru Waste in the
Ocean. Vol 4. Energy Wastes in the Ocean. I. W. Duedall, D. R. Kester, P. K.
Park, and B. H. Ketchum, eds. Wiley-Interscience, New York, pp. 468-497
(1985).

16.	H.A. van der Sloot and G.J. De Groot. "Characterization of Municipal Solid
Waste Incinerator Residues For Utilization: Leaching Properties." Proceedings
of the Ash Utilization and Stabilization Conference (Ash H), Washington (1989).

105-14


-------
17.	I Odler and K.-H. Zysk. "Characterization of Products of Primary Flue Gas
Desulfurization." Ed. Gregory J. McCarthy et al. Materials Research Society
Symposia Proceedings 113, pp. 179-185 (1987).

18.	D. Rai, S.V. Mattigod, C.C. Ainsworth, and J.M. Zachara. "Leaching Behavior
of Fossil Fuel Wastes: Mineralogy and Geochemistry of Calcium". Ed. Gregory
J. McCarthy. Materials Research Society Symposia Proceedings 86, pp. 3-15

(1986).

19.	S.V. Mattigod, D. RaiJ.M. Zachara, and J.E. Amonette. "Mineralogy of
Weathered Flue Gas Desulfurization Sludges." Ed. Gregory J. McCarthy et al.
Materials Research Society Symposia Proceedings 136, pp. 3-8 (1988).

20.	AG. Eklund and D.M. Golden. "Characterization of Solid Wastes from
Advanced SOa Control Technologies for Electric Utilities." Ed. Gregory J.
McCarthy. Materials Research Society Symposia Proceedings 113, pp. 173-177

(1987).

21.	F.A. Cotton and G. Wilkinson. Basic Inorganic Chemistry. New York: John
Wiley and Sons, p. 388 (1976).

22.	A. Lea. The Chemistry of Cement and Concrete, 3rd ed. Arnold, London, pp.
177-249 (1970).

23.	C. Shieh, I.W. Duedall, E.H. Kalajian, and F.J. Roethel. "The Technology of
Energy Waste Stabilization for Utilization in Artificial Reef Construction." Ed.
D.W. Tedder and F.G. Pohland. Emerging Technologies for Hazardous Waste
Treatment American Chemical Society Symposium Series, Washington, DC
(1990).

24.	D.M. Roy, K. Luke, and S. Diamond. ''Characterization of Fly Ash and Its
Reaction in Concrete." Ed. Gregory J. McCarthy et al. Materials Research
Society Symposia Proceedings 43, pp. 3-20 (1985).

25.	T. Edwards and I.W. Duedall. "Dissolution of Calcium From Coal-Waste
Blocks in Freshwater and Seawater." Ed. Iver W. Duedall. Waste In The
Ocean Volume 4: Energy Wastes In The Ocean. New York: Wiley, p. 22 (1985).

Acknowledgements

Hie authors thank Blair A. Kennedy, The New Brunswick Power
Commission, Dalhousie Generating Station, for providing the Orimulsion fly ash that
made this study possible.

105-15


-------
10

~o

o

CD

I

o

X

ll

Mix 85%-10
	 Mix 85%-15

o

0

100 200 300 400
Time (hr)

500 600

Figure 1. Flux of calcium from stabilized Orimulsion LSD waste in the tank leaching
study,

200

I

TJ

CM

0)

o

E 100

<0

I

o

X
LL.

0

0

100

200

300

Time (hr)

400

500

600

Figure 2. Flux of vanadium from stabilized Orimulsion LSD waste in the tank
leaching study.

105-16


-------
CM

0)

o

CD

i

o

x
J=?

LL

100 200 300 400 500

600

Time (hr)

Figure 3. Flux of sulfate from stabilized Orimulsion LSD waste in the tank leaching
study.

120

0)

o

(£>

I

o

0

4->

m
DC

500

Time (hrs)

Figure 4. Leaching rate of calcium from stabilized Orimulsion LSD waste in the
column leaching study.

105-17


-------
30

n

CI)

— 20
O CX}

E

0>

o


-------
WHEN THE REGULATIONS GET TOUGH;
ADVANCED TREATMENT OF FGD BLOWDOWNS

Marek K. Mierzejewski
Infilco Degremont Inc.
2924 Emerywood Parkway
Richmond, Virginia 23294

David C. Olszewski
Keith R. Minnich
Aqua-Chem, Inc.

7800 North 113th Street
Milwaukee, Wisconsin 53201

Abstract

In cases where blowdowns from FGD systems are discharged to the environment,
they must first be treated, principally to remove heavy metals and suspended solids.
The established treatment techniques of metal precipitation and complexation followed
by clarification, while achieving these objectives, do not alter the wastewater's salinity.
If Federal or State regulations limit salt input into the environment, then evaporation of
the wastewater is required. Furthermore, the increasingly strict discharge limits being
set for heavy metals may mandate this zero discharge technology even if no salinity
controls are imposed.

Similarly, if organic additives are used in the absorber, the resulting high Biochemical
Oxygen Demand (BOD) of the blowdown must be removed biologically prior to
discharge.

To date, experience of these more advanced techniques is limited, with only one
evaporator working on an unblended FGD blowdown (in Germany), and with only a
few biological trickling fitters tackling the BOD problem.

This paper reviews key issues and precautions to be considered in designing FGD
wastewater treatment plants to these toughest new standards.

Introduction

The sources and characteristics of FGD wastewaters and techniques of their treatment
have been discussed in an earlier paper(1). Since presentation of that paper in
December 1991, three more contracts for FGD wastewater treatment plants (WWTP)
have been awarded, making a modest total of four such plants in the U.S. dedicated
to the treatment of unblended FGD blowdowns, in response to Phase I of the 1990

106-1


-------
Clean Air Act Amendments. Although only one of these plants has begun operation
(NIPSCO Bailly, in June 1992®), all the others are at varying, but early, stages of
design/construction. Since these plants are normally associated with saleable
gypsum FGD systems, more may be expected during Phase II, where the increasing
cost of waste FGD sludge disposal seems likely to lead to more utilities adopting the
saleable gypsum approach.

Of course, there are several more FGD systems installed, or being installed, than the
number of WWTP, and this reflects the easy first option of blowdown disposal that
utilities consider, namely discharge into ashponds. With tightening regulations, this
option may become less available.

The four U.S. installations have little in common: they are each in a different state, they
are designed to treat blowdowns from different FGD suppliers, and are required to
meet varying standards of treatment (Table 1).

Discharge standards vary from state to state and are being applied increasingly based
on water quality, rather than technology, standards, in effect requiring treatment to
produce an effluent having similar if not lower concentrations of species as the water
body into which it is discharged. Limits have been proposed for metals where an
alternative to the established treatment of these blowdowns has had to be considered.
For example, the New York State Electric and Gas Corporation (NYSEG) had little
option in the case of its Mil liken Station, located on the edge of Cayuga Lake, NY, but
to adopt a zero discharge option.

As regards BOD removal, the FGD WWTP at Pennsylvania Electric Company's
Conemaugh Station will include a biological stage using Sequencing Batch Reactors
(SBRs) to reduce the influent BODs from 290 mg/l (if dibasic acids (DBA) are dosed)
to 25 mg/l, a removal efficiency of over 90%.

Evaporation

Metal limits are one thing: total dissolved solids (TDS) another. Evaporation, yielding a
highly concentrated liquor or a crystalline product, is the technology of choice where
TDS limits are proposed. In Europe, it is debatable whether evaporation would be
proposed for any other purpose. This, however, reflects the generally "easier*
discharge standards that have to be met there as compared to the U.S. On some
projects in the U.S. very low allowable concentrations of heavy metals have been
demanded, considerably lower than in Germany, but with no limit on discharge salinity.
In this case, evaporation has nevertheless had to be considered, since physico-
chemical precipitation and coagulation alone, even using sulfides, may not attain the
set limits.

In Germany, discharges from FGD systems are regulated by the 47 Verwaltungs-
vorschrift (47 VwV) which, although it represents a minimum Federal limit that can be

106-2


-------
locally tightened, seems generous by U.S. standards. By contrast, the U.S. EPA,
through State Departments of Environmental Protection, is imposing standards which,
in many cases, exceed the German limits by one, and sometimes two, orders of
magnitude. The 47 VwV is currently under review, but even the proposed new
discharge limits make an interesting comparison with those for a current U.S. FGD
project (see Table 2).

The implications of these tighter standards are that the existing technology must be
further optimized and fine-tuned, with the additional requirement that the WWTP is
carefully operated. The alternative is to introduce supplementary technology, such as
evaporation, to achieve the limits.

The purpose of an evaporation/crystallization system is principally to minimize the
volume of the FGD blowdown, by producing a concentrated brine: an 80% volume
reduction is typical for an evaporator, and over 90% is possible using salt conversion.

Pretreatomnt Necessary for Evaporation

This pretreatment takes two forms: firstly the treatment that is required to permit
disposal of the salt product (e.g. heavy metal removal, suspended solids removal) and
secondly the pretreatment for the evaporator proper.

If a relatively high purity product is required, then heavy metals need to be removed
by the existing technology described earlier. This necessitates, in effect, a full
physicochemical precipitation/coagulation/clarification plant upstream of the
evaporator itself. Clarification is also required to reduce influent suspended solids to
less than 100 mg/l.

FGD blowdowns are saturated with calcium sulfate and other calcium sails. To
prevent these salts from scaling the evaporator heat transfer surface the evaporator is
"seeded" with CaS04 crystals so that instead of precipitating on the heat transfer
surface the salts precipitate on these crystals. The successful operation of seeded
slurry systems requires that the feed chemistry conforms to several parameters, and
pretreatment by addition of one or more compounds may be required.

A feed/distillate heat exchanger is used to preheat the feed with outgoing distillate to
improve the system energy economy. Anti-scaling agents are metered into the feed to
temporarily increase calcium salt solubility, thereby reducing scaling in the heat
exchanger only and facilitating proper deaerator performance.

Other pretreatments include:

• Acidification to a pH of 5.5 to 6.5 which converts carbonates to dissolved C02
which is removed in the deaerator.

106-3


-------
• Anti-foam dosing, if necessary, to control foaming and reduce solids carryover
into the distillate.

The feed is sprayed into a deaerator to reduce the amount of dissolved gases, it being
particularly important to remove dissolved oxygen to minimize corrosion.

Salt Conversion. Salt conversion is another pretreatment technique that has
advantages in the treatment of FGD blowdowns. This technique replaces the calcium
by sodium ions, allowing the treated FGD blowdown to be evaporated and crystallized
to yield a high purity sodium chloride product. This has the benefit of further reducing
the amount of waste that must be disposed and may yield a salt product that is
saleable. Salt conversion may be required when disposal of the wastes is very
expensive or not an option. Its major drawbacks are the additional processing
equipment required and its capital and operating costs. This option almost doubles
the capital cost when compared to evaporation of the calcium chloride based
wastewater. Although the operating costs of the evaporation systems are slightly
lower, the chemical costs are significantly higher, due to the amount of sodium
carbonate required.

Before salt conversion, tie Mg2+ must be removed by raising the pH, thereby
precipitating Mg(OH)2. Salt conversion itself is performed in a discrete reaction tank
after physicochemical pretreatment. Sodium carbonate is dosed to a mixed reaction
tank and the following reaction takes place:

Ca2+ + NaaCOg - CaC03 + 2Na+

The stoichiometry is such that a considerable amount of soda ash is required. For
example, a 30 gpm (6.8 m3/h) flow, containing 6 g Ca2+/L will consume 2.6 tonnes
Na^COj/d (2.9 US ton/d). The only positive point is that there is no sludge disposal
problem, as the calcium carbonate produced (2.45 tonnes/d in this example) can be
fed into the scrubber with the limestone slurry.

Most Suitable Evaporator Types

Evaporators can be grouped into two broad categories: thin film and forced
circulation. In the first type the brine is distributed over the heat transfer surface in a
thin film, with boiling taking place at the surface of the film. The distribution system
must be designed to completely wet the heat transfer surface. In forced circulation
evaporators, brine at its boiling point is pumped from a flash tank through a heat
exchanger where it is sensibly heated under pressure to prevent boiling in the heat
exchanger. From there, the heated brine is returned to the flash tank where it then
flashes at a lower pressure.

106-4


-------
In both types of evaporator the brine flow required in the heat exchanger exceeds the
feed flow and a recirculation pump provides the necessary flow to the heat exchanger
with the incoming feed being combined with this flow. After boiling takes place a small
portion of the flow is withdrawn as concentrate.

Mechanical vapor compression or multiple effect configurations can be used with
either type of evaporator. In mechanical vapor compression, the evaporated vapor
passes to the suction of a steam compressor: the compressed vapor is then used as
heating steam in the evaporator heat exchanger. In multiple effect evaporation, boiler
steam is used to evaporate vapor in the first effect. This vapor is used as the heating
steam in the second effect, which operates at a lower pressure and temperature than
the first effect. This process can be repeated several more times, until vapor from the
last effect is condensed in a heat exchanger which normally uses cooling water.
Mechanical vapor compression thus recycles the latent heat of vaporization, while
multiple effect evaporation uses the latent heat of vaporization of boiler steam several
times.

Thin film evaporators require much smaller recirculation pumps than forced circulation
evaporators. The smaller pump reduces both the capital and operating cost. This is
particularly significant when high alloy materials are required in hot brine service.

These evaporators are generally limited to non-crystallizing applications. However,
salts that are inversely soluble with temperature, such as CaS04, can be crystallized in
thin film evaporators by using the seeding technique. Forced circulation crystallizers
are used for concentration of brines that are saturated in NaCI, and would therefore be
used in any salt conversion system.

A calcium based FGD purge stream can be concentrated in a thin film evaporator to a
total solids concentration of approximately 33%, which is below the crystallization point
of CaCI2 in a solution at ambient temperature. A filter press system can be used to
remove CaS04 crystals from the 33% CaCI2 solution to minimize the amount of
impurities in the brine. A dry CaCI2 salt can be produced by transferring the 33%
solution to a spray dryer, although the usefulness of obtaining a dry CaCI2 product is
questionable, given this salt's deliquescent nature.A sodium based FGD blowdown
stream can be concentrated in a thin film evaporator to a total solids concentration of
approximately 25%. Further concentration of the sodium based brine can be
accomplished in a forced circulation evaporator. The NaCI crystals, which must be
removed by a dewatering device such as a centrifuge to limit the concentration of the
slurry, can be dried in a drum or fluid bed dryer. As with all aspects of FGD
blowdown handling and treatment, correct selection of the materials of construction is
very important. In evaporation, the higher temperatures and concentrations mandate
materials such as those given in Table 3. Rubber-lined carbon steel can be used as a
substitute material where temperature limitations allow.

106-5


-------
Utility Requirements

Table 4 shows the utility requirements per gpm (0.23 m3/h) of feed, as determined by
product type: 33% total solids calcium chloride solution, calcium based dry salt and
sodium based dry salt (i.e., after salt conversion). Comparative figures are given for
mechanical vapor compression (MVC) and multiple-effect (ME) evaporators. As
mentioned above, the utility costs for the sodium based dry salt option are lowest, but
chemical costs (not shown here) are the highest.

Evaporator Product Disposal

Liquid disposal requirements can be completely eliminated by the addition of
crystallization or spray drying after evaporation. Crystallization is less expensive and
less energy intensive than spray drying but cannot be used with calcium chloride
based wastewater.

Spray drying yields a dry product that is easier to dispose of than that from
crystallization, which yields a wet product (3-10% H20). An additional drying step can
be used to produce a dry product after crystallization. This is normally required if the
product is to be sold such as in the case of salt conversion. Some U.S. utilities are
investigating the possibility of using the liquid calcium chloride solution for road de-
icing. This requires pretreatment to remove heavy metals and other toxic
contaminants.

Experience: Full Scale and Demonstration

Evaporation of FGD blowdowns worldwide has been very limited to date. A power
station in Berlin, Germany, has the only known FGD wastewater evaporators currently
operating. Indeed Germany was expected to be the leader in this new environmental
treatment, but economic problems there have apparently delayed its implementation.
The U.S. has no FGD blowdown evaporators currently operating, although the first
system was purchased earlier this year (by NYSEG) and will be in operation in 1994.
With the current activity in the U.S. FGD market, there are numerous wastewater
evaporators proposed and rt appears the U.S. may become the leader in this field.

Regarding demonstration scale testing, Philipp Muller, the German affiliate of the
Degremont Group, has joint-ventured with Aqua-Chem Inc. to run tests with an
evaporator and crystallizer system at a German power station later this year.

The system has been designed with the flexibility to handle sodium chloride or calcium
chloride based solutions, either unseeded or seeded with gypsum. With the high
grades of materials used (i.e., titanium, Hastelloy, Alloy 20), it can handle a wide range
of chemistries, concentrations and temperatures. It also has many built-in automation

106-6


-------
features to reduce operator attention. The flow diagram for this plant is represented in
Figure 1.

The FGD blowdown, after heavy metals removal and gypsum desaturation, is
pretreated to remove the magnesium and convert the calcium to sodium. The
wastewater is then concentrated in two steps. In the first step, the falling-film
evaporator concentrates the solution to about 25% TDS (just under the solubility limit
for NaCI). Gypsum seed crystals are added if necessary to maintain a seeded slurry
operation. In the second step, the forced circulation evaporator/crystallizer
concentrates the solution further to produce a slurry with suspended salt crystals. A
portion of the slurry is discharged to a filter press to remove the salt crystals and the
other liquid is returned for further concentration. If required, the salt crystals can be
washed in a batch mode for higher purity.

COD and BOD Removal

As well as treating FGD blowdowns to new strict metal limits, the WWTP may need to
be adapted to reduce the concentrations of BOD5 and Chemical Oxygen Demand
(COD). To date, these stages have been required in some WWTP's in Germany and
the Far East, where they are introduced after the clarification stage. The COD of FGD
blowdowns is due principally to unoxidized sulfite ions and trace organics;
environmental authorities in the Far East also require removal of dithionate (S2062")
which can contribute to the COD.

Unreacted sulfite is removed by oxidation; hence, if the influent sulfite level is known to
be low, the first (oxidation) stage can be dispensed with. Trace organics are removed
by adsorption on granular activated carbon in pressure filters. With these two stages
the COD may be reduced from an influent level of 150-200 mg/l to less than 100 mg/l.
However, if dithionate is a regulated species, then ion exchange on a special synthetic
adsorbent is also required.

The BODs of FGD blowdowns is usually very low, and the usual discharge standard of
20-30 mg/l seen in Europe and the U.S. imposes no requirement for special treatment.
Only in the case where organic additives are used in the scrubber is a BOD removal
stage required. Organic additives, used to enhance S02 removal in the scrubber, are
normally mono- or dibasic carboxylic acids (such as formic or adipic acids), which
serve to buffer the scrubber pH to an optimum level. Being highly soluble, these
compounds pass through the scrubber and into the blowdown, from which they must
be removed to meet BODs discharge guarantees. One German FGD supplier's
process is based on a continuous 800 mg/l formic acid feed: in three WWTP treating
the blowdown from Saarberg Holter Umwelttechnik (SHU) scrubbers at Heyden,
Bexbach and Fenne Power Stations, the BOD is removed after physicochemical
treatment by a biological trickling filter.

106-7


-------
The BOD of carboxylic acids, expressed in grammes per gramme of pure substance,
increases as the carbon chain grows. Thus, formic acid has a BOD of 0.24 g/g,
acetic add 0.65 g/g and propionic acid 1.1 g/g. Tests done on DuPont DBA (a
mixture of succinic, glutaric and adipic acids - 4, 5 and 6 carbon chain respectively)
show a BOD similar to propionic acid. Thus, a DBA dose of 300 mg/l may be
expected to give a BODs of about 330 mg/l.

The acids are also readily biodegradable: in deionized water, a 300 mg/l DBA solution
showed an oxygen uptake rate (OUR) of 60-75 mg Og/Lh, evidence of very rapid
biological oxidation. When the DBA was spiked to the same concentration in an FGD
blowdown having a high total dissolved solids content (TDS) the OUR dropped to
20-30 mg 02/l.h, still showing a high biodegradability, similar to that seen with
domestic sewage®.

In common with evaporation of FGD blowdowns, there is very Irttie experience
worldwide with their biological treatment. Except with SHU installations in Germany,
the biological treatment of blowdowns containing organic additives has simply not
been encountered. The simplest carboxylic acid, formic acid, has been the subject of
most investigations, not least because of its extensive use by SHU. In SHU systems,
this acid is dosed at a recommended level of 800 mg/l, but this has been reduced in
several installations to 500 mg/l. Its function is to reduce the pH in the scrubber, to
about pH 4 in the spray zones and pH 5 in the sump, thereby enhancing the reaction
between sulfur dioxide {SOJ and limestone. In this way, a high efficiency removal of
S02 can be maintained at an elevated chloride concentration of say, 40,000 mg/l: in
effect, the organic additive helps to override the decrease in efficiency normally seen
with an increase in chloride concentration.

Pilot tests performed by Bettenworth etal.(4) investigated biological formic acid
oxidation using a two-stage process: a cinder packed trickling filter followed by
filtration through a bed of 1-3 mm grain size Biolite, a high surface area granular
medium.

The pilot was designed so as to maintain both the trickling filter and the biofilter in an
oxic state; with air being diffused into the bottom of the first unit, the effluent entering
the second filter was almost saturated with oxygen, and still contained 3-4 mg 02/l at
the outlet. Over a several week period (to allow for acclimation) the formic add and
chloride concentrations were brought up to 100 mg/l and 9,000 mg/l respectively.
The removal of COD, which is approximately 1.25 times the BOD concentration, never
dropped below 90%, with almost all (>97%) of the removal taking place in the trickling
filter, the second filter acting principally to remove biomass particles.

As with any biological system, the dassic BOD5: N : P nutrient ratio of 100:5:1 was
respected during these tests by appropriate dosing, and should always be followed in
full-scale plants.

106-8


-------
On the basis of these tests, a full-scale WWTP, combing both physicochemical and
biological treatment was installed at Saarbergwerke's Heyden station in Western
Germany in 1985. This was followed by another two WWTP at Bexbach and Fenne
stations.

The trickling filter at Bexbach is a 6 m (20 ft) diameter x 6 m high forced draft unit,
filled to 4.5 m depth with plastic media: physicochemically-treated water is fed over the
media by a rotary distributor. Two pressure filters follow the single trickling filter: the
design filtration velocity is 7 m/h (2.9 U.S. gpm/sq.ft.) through a 1.5 m (5 ft) deep
Biolite bed. (See Figure 2).

A 4-5 times dilution of the FGD wastewater with cooling tower water is practiced at
Bexbach to reduce the chloride and formic acid concentrations entering the biological
treatment and, more importantly, to drop the calcium sulfate concentration below the
equilibrium point. FGD wastewaters are frequently supersaturated in calcium sulfate,
and equilibrium is re-established during physicochemical treatment. However effective
such "desaturatiorf may be, there remains a risk that some calcium sulfate will
precipitate on the packing in the trickling filter. Furthermore, any unreacted lime
remaining after physicochemical treatment may also react with the carbon dioxide
(both atmospheric and respiratory) in the air in the trickling filter. This double risk of a
heavy mineral scale forming on a lightweight, high surface media might be expected to
lead to problems and this indeed has been the case. Firstly, the scale can prevent
good biomass attachment, such that BOD removal efficiency drops. More seriously,
collapse of the media due to the weight of the scale has been encountered in one
plant.

Alternative Biological Treatment

The choice in biological treatment lies between "attached growth", as described above,
or "suspended growth", such as activated sludge, or some combination of the two.

None of these options presents itself as an obvious choice for ail plants. If the
problems of scaling can be overcome (for CaS04 by better desaturation or by dilution
and for CaCOs by add dosing to a pH closer to neutrality) then attached growth is the
preferred method for low BOD wastewaters. In such cases, the trickling filter/biofilter
combination, as at Bexbach can be considered or alternatively, one of the new
generation of oxygenated upflow biofilters available today.

If, however, higher BOD levels, of 200 mg/l and above, are likely to be encountered,
then a suspended growth system is recommended. The argument against using this
option with low BOD wastewaters is that it is difficult to maintain the Mixed Liquor
Volatile Suspended Solids (MLVSS) of 2000-3000 mg/l, i.e., the biomass concentration
necessary to attack the BOD.

106-9


-------
All suspended growth processes depend on a wastewater entering an aeration basin,
where intimate mixing with the biomass and oxidation takes place. The treated
wastewater leaves the basin with the biomass, which must then be recovered, usually
in a clarifier, and returned to the basin to maintain the level of biological activity
necessary there (Figure 3). Solids recovery thus becomes critical: its efficiency
determines the viability of the system. Typically, suspended solids in the overflow from
a secondary clarifier will be in the order of 30 mg/l. An influent BOD5 of 50 mg/1, say,
may yield approximately 30 mg/l of new synthesized biomass: in such a case the
plant will be run on a knife-edge, always hovering between depleting the aeration
basin (where overflow solids > synthesized solids) and just maintaining the MLVSS
(overflow solids = synthesized solids). In actual practice such mathematically
straightforward conditions do not exist, and fluctuations in flow, disruption or
malfunction of operation can all cause excessive solids loss from the system.

The other problem facing suspended growth systems is similar to that which faces its
attached growth counterpart, viz. precipitation of CaS04 and CaC03. Whereas in a
trickling filter this phenomenon is most likely to manifest itself as scale, in a suspended
growth system these compounds may form discrete crystalline precipitates which will
settle readily: the problem then is how to purge them from the system. This requires
that these precipitates be both suspended by the aeration/mixing system and then,
after settling, be removable from the secondary clarifier. The first condition demands
vigorous mixing using static draft tube aerators or jet aeration, with an energy input at
least double the 20-30 W/m3 (75-115 W/1000 gal) energy input used in conventional
activated sludge design. Porous discs, used for fine bubble aeration, are to be strictly
avoided in view of their potential for mineral scaling and blockage in this application.

The second condition requires a scraped clarifier that will remove the solids as soon
as they are settled and not allow them to thicken to an immovable mass. Sequencing
Batch Reactors, where aeration and settlement is alternated sequentially between two
or more tanks (i.e., each tank serving alternately as aeration basin and clarifier), have
been recommended at one U.S. FGD installation, but are as yet without reference.
Their benefit certainly is in that the duration of all sequenced stages can be modified,
after installation, to obtain optimum treatment. The likely disadvantage with SBR's is
that since they are fiat-bottomed, unscraped tanks, mineral sludge will tend to settle
and thicken, gradually reducing the effective volume.

Summary

The tightening restrictions on all wastewater discharges, not just FGD blowdowns,
mandate that existing technologies of wastewater treatment do not stand still, but are
modified and supplemented accordingly. In the case of FGD blowdowns specifically,
the limits on heavy metals may soon be such that evaporation (i.e., the station
becoming zero discharge) will be the rational solution. The problem will, however, be
replaced by another one: hew to dispose of the salt liquor or cake produced (a
question not unlike that of saleable gypsum disposal). The option of discharge into

105-10


-------
ashponds will not be ever present. Likewise, in throwaway systems, the wastewater
treatment problem will not always be "buried" with the sludge; waste FGD sludge
leachate treatment is already under investigation for a Southeastern U.S. utility.

The restrictions put on BODs in the discharge are quite different to those put on heavy
metals and salinity, in that they are avoidable. The facile solution to achieving
compliance may be simply to discontinue the use of the offending organic additive,
since a "typical" FGD blowdown, without added organics, will meet the usual BOD
discharge limits seen in U.S. and Europe of 20-30 mg/l. If, however, enhanced S02
removal is required, and organic additives are used, then treatment to remove the
BOD will be required.

The fact that such treatment is biologically based requires that this stage follow the
physicochemicai removal of heavy metals, without which inhibition of the microbial
processes may occur. Furthermore, the biological removal system requires a constant
feed of substrate (in this case the organic acids) if the bacteria are not to die. This
precludes operation of the FGD system with intermittent or variable application of the
organic additive.

Both these more advanced removal techniques have something in common: there is
very little experience worldwide in either. The first U.S. evaporator installation working
on undiluted FGD blowdown, and one of the first in the world, will be starting up in
1994. Demonstration scale evaporation tests by a U.S. company are also due to start
in Europe this year. As to biological treatment, experience is currently limited to
removal of formic acid at a handful of stations in Germany. (It should be noted that
the zero discharge option of evaporation would also remove concerns over BODs at a
stroke.) According to current forecasts, at least one FGD WWTP in the near future
may be equipped with biological treatment equipment.

The next few years, particularly once Phase II of the Clean Air Act Amendments comes
into operation, should see interesting new developments in this advanced FGD
wastewater treatment technology.

106-11


-------
References

1.	M.K. Mierzejewski, "The Elimination of Pollutants from FGD Wastewaters",
presented at the EPRI/EPA/DOE 1991 S02 Control Symposium,

Washington DC (December 1991).

2.	Michael Sicinski, David LaValle and Marek K. Mierzejewski, "Cleaning the Water
at a Clean Air Plant; Early Operating Data from the NIPSCO Bailly FGD
Wastewater Treatment Plant", Paper lWC-92-50, 53rd International Water
Conference, Pittsburgh, PA (October 1992).

3.	Mark J. Briggs, Radian Corporation, Milwaukee, Wl. [personal communication]

4.	H. Bettenworth, H. Pflug, E. Dieterle and M. Wolkl, "Untersuchungen zum
biologischen Abbau von Ameisensaure im Abwasser von Rauchgasent-
schwefelungsanlagen", VGB Kraftswerktechnik (Germany), Vol. 64, No. 11,
p. 993-999 (1984).

106-12


-------
*

TBS
M +

Table 1

U.S. FGD Wastewater Treatment Plant

Utility

State

Station

FGD System

Oneration

NIPSCO

IN

Bailly

Pure Air

1992

NYSEG

NY

Mi Hi ken

SHU

1994

Atlantic
Electric

NJ

B.L England

GEESI

1994

Penelec

PA

Conemaugh

ABBES

1994

Controlled species are given in this column
Total Suspended Solids
Various metals, heavy and other

Treatment*

pH.TSS
Zero Discharge

pH, TSS, M +

pH, TSS, M + ,
BOD

Table 2.

Comparison of Germany's 47 Verwaltungsvorschrjft (47 VwV)
Discharge Limits with those from a current U.S. Utility FGD Project

Species

47 VwV

u§

Cd

50

100

Hg

50

2

Cr

500

50

Ni

500

30

Cu

500

100

Pb

100

50

Zn

1000

100

V

5000

—

s2-

200

—

Mn

—

500

Se

—

200

As

—

36

All units are parts per billion (jig/l)

106-13


-------
Table 3.

Materials of Construction for Evaporators



cr < 20.0001

cr < 50.000

Cr > 50.000

Tubes

Tl Gr 2

n Gr 2

Ti Gr 2, 12 or 7
Hastelloy C

Upper Tubesheet

Ti-clad

Ti-clad

Ti-clad or
Hastelloy clad

Lower Tubesheet

316 with
2.5 Mo

Ti-clad
6 Mo SST2

Ti-clad

Hastelloy clad

Lower Shell

316 min Mo

6 Mo

Hastelloy clad

Demisters

317

6 Mo

Hastelloy

Piping

317

6 Mo

Hastelloy

Pumps

Cd4MCu3

Cd4MCu

Hastelloy > 180°F
Cd4MCu < 180°F

Upper Channel and
Vapor Separator

316 min Mo

6 Mo or
Hastelloy clad

Hastelloy clad

Distributor

Inconel 625

Inconel 625

Hastelloy

Notes:

1	: All chloride [CP] concentrations in mg/l.

2	; *6 Mo stainless steel = AL6XN, 254SMO, or equivalent

3	; Cd4MCu should contain nitrogen (0.15% min.)

106-14


-------
Steam
kW

(MM BTU/hr)

Electricity
kW

Fuel Gas
kW

(MMBTU/hr)

Cooling Water
m3/h

(gpm)

Table 4.

Utility Requirements By Product Type
per gpm of Feed (7% Total Solids (TS) in Feed)

33% TS Calcium
Based Solution

MCV
0

(0)
9,3

0.11
(0.5)

ME2

0.8

0
(0)

2.95
(13)

Calcium Based
Dry Salt
MVC/ME3 ME

Sodium Based
Dry Salt
MVC ME

44	1^8	cr-f	fi	oq c

(0.150) (0.054) (0.174) (0) (0.135)

6.1

11.7

(0.04)

1.1
(5)

0.8

11.7
(0.04)

3.4

(15)

5.5

1.2

(0.004)

0.07
(0.3)

0.7

1.2

(0.004)

2.52

(11.1)

MVC1 : Mechanical Vapor Compression

ME2 : Multiple-Effect (3-effect for Calcium Based and 4-effect for Sodium
Based)

MVC/ME3: This option requires a combination system.

106-15


-------
o

2

05

Figure 1.

Demonstration Scale Evaporator/Crystallizer for FGD Slowdown Treatment


-------
trickling filter

BLOWER

DILUTION TANK

FROM

PHYSiCOCHEMICAL
TREATMENT

TREATED
WASTEWATER

Figure 2.

Biological Treatment of FGD Blowdown at Bexbach Station

INFLUENT







TREATED

AERATION BASIN

— air

CLARIFIER

WATER

RETURN SLUDGE

WASTE SLUDGE

Figure 3.

Activated Sludge Plant Schematic

106-17


-------
A


-------
DEVELOPMENT OF HASTE WATER CONCENTRATION & SOLIDIFICATION

CVCiPFRf XfAD -utpm fFTTTT? r*lCI flffCTIT T?1TO T 71HTniI "DT liMTn
SIOX&E rUK if IS J. r±jUi!» VsAd IJ£oUJUI: UKi Aill lUu JkrJLj/iNX

S. Tsubouchi
T. Miwada
Chubu Electric Power Co., Inc.
1 Toshin-cho, Higashi-ku
Nagoya, 461-91 Japan

S. Kotake
N. Ukawa
Mitsubishi Heavy Industries Ltd.
2-5-1 Marunouchi
Chiyoda-ku, Tokyo, 100 Japan

Abstract

Wet limestone gypsum FGD system has been widely installed in
coal-fired thermal power stations for environmental protection
by removing the acid-rain causing sulfur oxide compound from
their flue gas. This system inevitably disposes waste water
containing chloride, heavy metals, nitrates and COD
components, which requires processing with a fairly expensive
and complex water treatment system before discharging. In
order to solve this problem, Chubu Electric Power Corp. and
Mitsubishi Heavy Industries have jointly developed a new FGD
waste water treatment system where waste water is collectively
concentrated with the combination of electrodialysis and
evaporation and then solidified with flyash and cement. A
demonstration test with a 20MW equivalent pilot plant
conducted for almost 9 months at the Shin-Nagoya Power Station
adequately proved the efficacy of the system.

Introduction

Recently in coal-burning thermal power stations, large number
of wet type flue gas desulfurization (FGD) plants are being
installed for environmental protection. In desulfurization
plants, sulfurous acid gas in the flue gas is made to be
absorbed by limestone slurry absorbent and recovered as by-
product gypsum while the gypsum filtrate is recycled back into
the system. In this recycled filtrate, a part of the flyash
and halogen etc., contained in the flue gas also get taken in.
Among these, in particular, if cl" ion concentration in the
absorbent fluid becomes high, the materials of the constituent

107-1


-------
equipment and apparatus of the FGD plant get corroded while
its desulfurization efficiency also get affected. As a
countermeasure, a fixed quantity of waste water is constantly
bled out of the absorber to avoid excessive build up of these
harmful components and a waste water treatment facility is
provided to get rid of them.

Presently, in the waste water treatment facility,
corresponding to the characteristics of the waste water
constituent, such high grade treatment systems as; heavy
metals treatment system, fluorine treatment system, biological
nitrogen treatment system, COD treatment system etc., are
provided.

Although these waste water systems have been widely adopted
not only by the electric utilities but by other companies as
well, the fact remains that they are complicated, need large
installation space and are expensive to equip and operate.

To resolve these problems, a desulfurization system without
any waste water discharge has been developed in which the
waste water from the process after being concentrated is
solidified into a stable form by mixing with cement and flyash
to be disposed for land reclamation purposes. The system was
demonstrated with a pilot plant, the summarized results of
which are reported hereunder.

Research, and Development Particulars

The research and development particulars of this facility is
shown in Figure 1. During the elementary research, to carry
out comparative system study in respect of process, material
selection, and economical trial balance, in particular for the
waste water concentration method, laboratory scale tests were
conducted for the three methods; (1) evaporation tower method
using boiler flue gas excess heat (2) method using steam
heating and multi effect evaporator, and (3) combined method
using electrodialyzer and evaporator.

After comparison of the tests made from the technical and
economical aspects, it was confirmed the method (3) to be the
best.

On the basis of these findings, a pilot test plant equivalent
in capacity to a 20 1? electricity generating facility was
installed within the premises of Chubu Electric Company's
Shin-Nagoya thermal power station. The installation and trial
operations of the pilot plant were completed by 1991 year end,
and during the 1992 January-September period it was subjected
to demonstration tests. A general view of the pilot facility
is shown in Figure 2.

As waste water for the test, actual waste water drawn from the
FGD plant of Chubu Electric Power Company's Hekinan Power
Station No.1 coal-burning boiler unit and blended with
chemical additives for adjustment of chemical composition was

107-2


-------
used. Further, the flyash used was also obtained form the
same boiler while the cement used was the market available
common portland cement.

Outline of Waste—Water-Free FGD System Pilot Plant

Pilot plant outline of the waste-water-free FGD system is

shown in Figure 3. The FGD system waste water, typical
composition of which is shown below, is a complex mixture of
flyash, chloride/fluorine ions, heavy metals, nitrogen
compoundsr COD components etc.

Item

pH

SS

Ca2*

Mgz+

Alz+

Fe3+

Properties

5-6
mg/1	500

mg/1 7,000
mg/1 2,500
mg/1	50

mg/1	50

Item	Properties

Na+	mg/1	60

CI"	mg/1	20,000

F"	mg/1	50

T-N	mg/1	440

S042"	mg/1	1,000

COD	mg/1	70

As a pre-treatment, the waste water is filtered to remove the
suspended solids, and then, after being primarily concentrated
with electrodialysis and further secondarily concentrated with
evaporator, is mixed and kneaded with cement and flyash to be
formed into a stable solid substance. Specification of main
equipment for each facility is described hereunder while
outline of each facility is described in the following.

TJnxt

Equipment

Specification

Filtration

Primary
Filter

Model

Filtration area
Treated flow volume

Precoated
Type
1.32 mz
1.4 m3/h

Secondary
Filter

Model

Filtration Area

Treated flow volume

P Or3+*

Type
0.377 mz
1.4 m3/h

Electro-
dialyzer

Electro-
dialyzer

Model	!

Membrane area	:

No.of membrane pairs;

Filter
Press Type
0.368 nr
180 pairs

Secondary
Concentration

Evaporator

Model

ucau llallolcl /ilea

Wiped-Film

Type
0.4 m2

Solidification Mixer	Model

Kneader	Capacity

Vibro-Mixer
188 kg/h

107-3


-------
Filtration Facility

The thickener supernatant, of FGD system waste water used for
this test contains about 500 ppm of suspended solids (SS) . In
order to prevent system performance deterioration due to
clogging and adherence on the electrodialyzer membranes by
these SS compounds, a pre-coated type 2-stage filtration
equipment was provided as a pre-treatment means.

Both the 1st stage and 2nd stage filters used in this
equipment were of ceramic made with their external surface
coated with pearlite as a filtration aid. These filters were
washed at regular intervals.

Electrodialyzer Facility

The filtrate from the filtration equipment is sent to the
electrodialyzer shown in Figure 4. The electrodialyzer is
consisting stacks of narrow compartments through which the
feed solution is pumped. These compartments are separated by
alternating cation-exchange and anion-exchange membranes which
are selectively permeable to positive and negative ions
respectively. The terminal compartments are bounded by
electrodes, for passing direct current through the whole
stack. Plastic-meshes are inserted in the solution
compartments to keep the cation- and anion-exchange membranes
apart and to promote uniform liquid flow.

The "Selemion" membranes of Asahi Glass Co., Ltd, are used for
the pilot test.

When the electrodes are connected to a DC source, ion
migration begins, as shown schematically in Figure 5 for a
group of chambers well within the stack. In each chamber
cations travel from right to left; anions in the opposite
direction. For the chamber D, the anion permeable membrane on
the right does not admit cations from the right, and the
cation-permeable membrane on the left similarly act as a
barrier for the negative ions from the left. As a result, the
filtrate cl" concentration decreases in the chamber D and
increases in the neighboring chambers C.

With this univalent selectivity, S042" ions are kept at low
concentration to enable the gypsum saturated waste liquid
circuit theretofore to be maintained as a unsaturated circuit
in spite of having it subjected to concentration. In
addition, it permits the liquid to be secondarily concentrated
with the evaporator to the final value of 250,000 ppm in terms
of cl" concentration without precipitation of gypsum. The
concentration of the liquid is decided by the density of
current flowing in between the membranes. Increase in current
density increases the concentration degree of the liquid while
a decrease in its value decreases the concentration degree.
As the current density in an electrodialyzer increases, the
migration amount of the ions also increases to bring up the
concentration degree of the liquid and thus the treating
capacity of the equipment. On the other hand, when the

107-4


-------
surplus current density is impressed, electrolysis of water

starts to occur. Further, due to the so-called Donan
equilibrium phenomenon, the concentration of S042" and Ca2+ ions
within the membranes starts to increase to cause gypsum scale
formation. Accordingly, the current density is decided from
the view point of preventing these phenomena.

In the said pilot plant, operation of the electrodialyzer with
high current density has been made possible while to prevent
gypsum scale generation a periodical membrane washing process
has been adopted.

Operation conditions for these were decided beforehand with
laboratory scale test facility and applied to the
demonstration test plant. Summary of these conditions are
stated below.

a.	Membrane current density	s 0.8 A/dm2

b.	Cl concentration of FGD effluent	: 20,000 ppm

c.	Cl concentration of diluted liquid	: 16,500 ppm

d.	Cl concentration of concentrated liquid	: 70,000 ppm

Secondary Concentration Facility

The liquid concentrated in the electrodialyzer is sent to a
secondary concentrating facility for further concentration.
For secondary concentration evaporative concentration is
carried out with a wiped film evaporator using steam as the
heating source. In the evaporator, the supplied liquid drops
from the evaporator top along the heat transfer wall surface
and gradually gets evaporated and concentrated while reaching
the evaporator bottom. During evaporative concentration, the
problem likely to occur is scaling of the heat transfer wall
surface. A means to avoid this is to prevent generation of
gypsum in the concentrated liquid by reducing the afore-stated
solubility product of gypsum in liquid. In combination with
this, low pressure evaporation is carried out employing low
temperature and pressure steam to prevent scaling due to
localized overconcentration of liquid on the heat transfer
wall surface.

Cl concentration for the secondarily concentrated liquid was
set at 250,000 ppm. To control this concentration, the
phenomena of boiling point elevation in conjunction with Cl
concentration in liquid was adopted.

The falling liquid temperature on the evaporator bottom
surface was regulated at a constant to obtain a certain Cl
concentration.

Solidification Facility

The liquid secondarily concentrated with evaporator is mixed
and kneaded with cement and flyash and disposed off as solid.
As mixer-kneader, a vibro mixer was employed.

107-5


-------
Results of Verification Test

Following items were confirmed with the verification test.

Filtration Equipment

The external element surfaces of this filtration equipment are
pearlite coated to facilitate collection and removal of waste
water contaminants. With the collection of contaminants, that
isr with the accumnlated SS loading, there is pressure loss
increase in the filter with time. The filter is therefore
periodically back washed to let the pearlite layer regain its
full capacity. To confirm performance of this filtration
equipment, together with the SS removal capacity it is
important to grasp the correlation of SS loading and pressure
loss.

SS concentration at the respective filter outlets over the 9-
month operation period met with the planned values and were as
shown below.

1st stage filter outlet < 1 ppm
2nd stage filter outlet < 0.2 ppm

During the test period the SS removal percentage has been
always 99.6% which thus met the planned value of 99%.

Electrodialyzer Tests

While carrying out concentration of gypsum saturated FGD
system waste water, following are important:

•	Establishing feasible membrane electric current density
and membrane washing method from the point of scale
prevention on the dialyzer membrane

•	Comprehension of concentration performance change with
time

•	Comprehension of permeability of the ions contained in the
liquid.

In the following explanation xs made about the results
obtained from the pilot test facility and fundamental
research.

Setting of electric current density. A current density of
0. 8A/dm2 was adjudged to be appropriate but as gypsum scaling
was apprehended even with this value periodic washing of the
dialyzer membranes was ascertained to be a necessity. In
Figure 6, CI concentration of the concentrated liquid with
change in electric current density is shown. Both in the
elementary research and pilot plant tests, the CI
concentration of the concentrated liquid with an electric
current density of 0.8A/dmz was approx. 70,000 ppm. Moreover,
within the range of these conditions, the CI concentration of
the concentrated liquid was not affected by diluted liquid

107-6


-------
flow velocity of 4.3 to 6.0 cm/s through, the membrane surface.
With membrane washing, gypsum scale causing Ca and S04 ions
collected within the membranes were discharged outside of the
membranes.

Change of concentration with time. Since the dissolved salts
in this FGD system waste water were mainly calcium chloride,
magnesium chloride and sodium chloride, the electric current
flow efficiency tin based on the permeability of CI ions
through the membranes as standard was used as a measure of the
concentration capacity. Electric current flow efficiency
was defined as follows:

1 CI

_ Qc.Cc.F ^ (io~3)

I.A.MW,

cl

[%I

Where?
Ici 1

Qc
Cc

F
I
A

Non-

standard electric current flow

efficiency of chlorine ions	[%]

j Flow amount of concentrated liquid	[kg/hr]
: Chlorine ion concentration in

concentrated liquid	[ppm]

Faraday constant	[A.hr/Eq]

Electric current density	[A/dm2]

Membrane area	[m2]

Chlorine ion molecules	[g/Eq]

Change of electric current flow efficiency ijCI with time is
shown in Figure 1.

As can be seen from the Figure 7, the electric current flow
efficiency remains practically same as that at the initial
operation stage even after a lapse of 5,400 hours without any
deterioration.

Permeabiflity of each type of ion. As stated before, the main
ions contained in waste water are: Ca2+, Mg2+ and Na* as
cations; and Cl" and S042" as anions. The permeability of these
ions through the dialyzer membranes depends on the particular
characteristics of the ion concerned. This characteristics of
an ion when expressed in a determinate quantity after being
compared with that of the Cl" ions is called the selective
permeability coefficient of the ion and is defined as follows:

rrt 1 _ ^"Cl

Tcl	—

'"Cl

Where;

1%! : Selective permeability coefficient of i component

[	]

Ci : Concentration of i component	[ppm]

Ccl : Concentration of chlorine ions	[ppm]

107-7


-------
Suffixes C and D represent concentrated liquid and diluted
liquid respectively.

Selective permeability coefficient of various components with
change in electric current density is shown in Figure 8. It
can be understood from this figure that the respective
selective permeability coefficient for the cations Ca2%
Mg2+, and Na+ is not affected by the electric current density
and remains almost constant. It also became clear from this
figure that the ion having the largest selective permeability
coefficient, that is, the ion which can permeate most easily
through the membranes among these ions is the Ca2+ ion while as
for the Mgz+ and Na2+ ions, their permeability is almost same as
each other.

With regard to the sulfuric acid SO*"2 ions, the object of
using these membranes in this equipment are:

*	to prevent gypsum scaling of the downstream evaporator

•	to prevent gypsum scaling of the dialyzer membrane itself

Since these membranes are of the univalent anion selective
type, their selective permeability coefficient was small
varying between 0.017 to 0.040.

Secondary Concentration with Evaporator

In this research a wiped film evaporator was used for
secondary concentration of the waste water as the purpose of
this evaporator was to further lower with concentration the
amount of liquid primarily concentrated with the
electrodialyzer in order that the quantity of auxiliary agent
used in the fmal solidification process can be reduced•

In a wiped film evaporator, chlorine concentration in its
concentrated outlet liquid is controlled by regulating the
pressure reduction degree and the concentrated liquid
temperature in side the evaporator but as a change in chlorine
concentration also largely affects the boiling point and
viscosity of the concentrated outlet liquid, it is necessary
to know the correlation of chlorine concentration change and
the change in boiling point and viscosity with this control.
Further, as it is also considered a possibility that with
scale generation there will be deterioration in the heat
transfer capacity of the evaporator with time, a discussion
regarding the change in heat transfer capacity of the
evaporator with continuous operation and physical properties
of the concentrated liquid is made in the following.

Change in heat transfer capacity with time. The change in
overall heat transfer coefficient of the evaporator with time
is shown in Figure 9. Overall heat transfer coefficient Ut was
calculated with the equation below:

Where;

Ut : Overall heat transfer coefficient [W/m2.K]

107-8


-------
Qh ; Overall heat transfer amount	[J/hr]

ATln : Logarthimic mean temperature difference [°C]
A i Heat transfer area	[m2]

Overall heat transfer amount was found from the product of the
condensed water quantity in the condenser used for condensing
the generated steam from the liquid subjected to secondary
concentration and the total of sensible heat and latent heat
found from the evaporation temperature. On the other hand,
the logarthimic mean temperature difference was found from the
temperature of the concentrated liquid and the heating steam.

As can be seen in Figure 9, the overall heat transfer
coefficient was almost constant at 872 [W/mz.K] throughout the
5,000 hours of operation period without showing any large
change with time.

Further, during the inspection of heat transfer surface on
completion of the operation, no scaling on it could be found
which confirmed the long-term stability of the concentration
process of this research carried out with the combination of
electrodialyzer and the wiped film evaporator.

Physical properties of concentrated liquid. During the
continuous operation shown in Figure 9, the liquid

concentration conditions were so adjusted as to let the
chlorine concentration of the concentrated liquid become
between 250,000 to 280,000 ppm as CI. Incidentally, it must
be mentioned that with higher chlorine concentration as the
physical properties of the liquid also change largely proper
attention must be paid in the selection of each type of
pITOCOSS equipment.

In the following mention is made about the changes in the
principal physical properties of the liquid.

(1)	Increase in boiling point

Along with the increase in chlorine concentration, the rise in
boiling point becomes larger; for example, at chlorine
concentration of 250,000 [ppm as CI] it becomes about 16°C (at
water vapor partial pressure Pw=10,133Pa). This means, as the
value of concentrated chlorine concentration becomes higher,
evaporation becomes difficult.

(2)	Specific gravity

Similar to the rise in boiling point, the specific gravity of
the liquid also increases with the increase in chlorine
content of the concentrated liquid. At a concentration of
250,000 ppm as CI, the specific gravity gets to be about 1.4.

(3)	Viscosity

107-9


-------
Viscosity also increases with the increase in concentration of
the concentrated liquid. It is characterized by the fact that
it increases rapidly particularly in the high concentration
region.

Solidification Facility

Purpose of solidification facility is to mix and knead
solidifying agents with the concentrated liquid from the
evaporator and produce a disposable solid. For the production
of an acceptable disposable solid, there are the following
requisites s

•	appropriate selection of solidifying agent mixing
proportions

•	easy handleability of the solidified product

•	appropriate leaching characteristics and strength of the
solidified product to meet environmental regulations

Solidified product obtained from the pilot plant facility is
shown in Figure 10.

Mixing proportions. As solidifying agent , a powdery mixture
of flyash and cement was used. In the selection of mixing
proportion of cement and flyash although good handleability of
the produced solid is an important consideration but in order
to reduce the operation cost it is necessary to decrease the
proportion of cement and replace it with flyash. However, as
the property and composition of flyash change largely
depending on the kind of coal and combustion conditions,
fundamentally it is necessary to select the mixing proportions
on the basis of the kind of coal used. As an example, for the
flyash recovered from the Hekinan Thermal Power Station, the
appropriate mixing proportion for it with concentrated liquid
ana csiDsiit# wss ss xoxxows •

Concentrated liquid s Cement : Flyash = 1.0 : 0.67 : 1.1

Handleability of the produced solids. For satisfactory
handling during the conveying process, the produced solids
should be free of fluidity and should not get bonded with each
other.

Leachability and compressive strength of the toxic components.

In order to be able to handle the produced solxds as
industrial waste, it is necessary to meet the standard values
prescribed in the Prime Minister's Office Ordinance (*1).

The solids produced from this pilot test facility met the
standard values both in respect of strength and leachability.
(*1) Prime Minister's Office Ordinance No. 5

[Ordinance issued by the Prime Minister's Office_to prescribe

107-10


-------
adjudication s tandards for industrial. wastes containing metals

etc. ]

Capital Cost and Operation Cost

In order to evaluate economic aspect of the Wastewater
Concentration and Solidification (WCS) system, investment
costs and operation costs of the system for a coal fired power
generating plant having two boilers of 1,000 MW capacity were
estimated together with the conventional waste water treatment
system under the same process design criteria and economic
criteria.

For the scope of complete wastewater treatment system, the
investment cost and the operation costs (total cost of
utilities, operation and maintenance, finance and management)
of WCS system are found to be 50 while these of the
conventional system are 100.

Conclusion

A waste water solidification system was newly developed for
the desulfurization (FGD) plants to replace the existing waste
water treatment systems. In this system, the waste water from
an FGD system after being concentrated by means of
electrodialysis and evaporation is solidified by mixing with
flyash and cement to produce an environmental pollution-free
disposable product for land reclamation use. In comparison
with the present waste water treatment systems, the newly
developed system is simple, requires less equipment space and
more economical from the point of construction and operation
costs. From the results of the present research it is firmly
believed that this newly developed waste water solidification
system will make large contribution as one of the
technological environmenta1 protection measures for the
thermal power stations hereafter.

107-11


-------
Year

1990

1991

1992

1993

Elementary research









Pilot facility design
and installation









Continuous
demonstration test









Figure 1 Development Milestone

Figure 2 Overall View of the Pilot Facility

NaCi Solution Tank

¦vasal

NaCt

Vaetmn*
Pump I

nl

~d

R«« Liijwi
Rrtidut! Filtrate.

Cake

Evgpsratar

A

- Condenser
4—|

Condensed Water
/Tobe recycled for use *s \
^FGD process cleaning water /

Primary
FiltnrteTank

Diluted liquid Tank Primarily Concentrated
U
-------
Figure 4 Electrodialyzer

:< A K A K A KA

Cathode

Na +













Ca + +













CI- —7



ci-



~ • «



CI-



Jf

SO4-









D

C

D



D

c

S04

7

D

Na +
Ca + ¦

Anode

1 )

I



I



,

,













A :
K :

C :
D :

Anion exchange membrane

K :

C :

D :

Diluted liquid	Concentrated

Liquid

Figure 5 Schematical Construction of Electrodialyzer

107-13


-------
100,000

CI concentration
of C liquid
[ppm]

50,000

Ccc« = 117201 + 65910 tppm]

I: Electric current density
(Data in between the current density of 1 to 3.2 [A/dm2] are
approximated with the least square method)

Measured values in the pilot facility

Pilot facility

Item

Other conditions

Cl concentration of 0 liquid {ppm]

10,800-18,400

Flow velocity of D liquid through
membrane chambers [m/sl

4.3 ~ 6.0

0.5 1.0 1.5	2.0 2.5 3.0

Electric Current Density [A/dm2]

3.5

Figure 6 Cl Concentration of Concentrate with Change
in Electric Current Density

100

Standard electric
current efficiency
of chlorine ions
[%]

~—e	©	®	—32	e			

[Concentration conditions]

Cl concentration of D
liquid [ppm]

13,610-18,400

Temperature of 0 liquid^

40 + 1

Flow velocity of D liquid
through membrane
chambers [cm/s]

6.0

J	1	1	1	L

°0	1000	2000	3000	4000	5000

Operation time Jhr]

Figure 7 Standard Electric Current Efficiency of Cl Ions

107-14


-------
1.0 -

Selective
permeability
Coefficient
Corresponding
to Ci

0.5 •

CD



Key

Component

O

Ca

~

Mg

A

Na

Mg • Na

&
D

O

Symbols with vertical line
indicate values measured
in the pilot facility



CI concentration of 0 liquid
Ippml

10.800- (8.40C

Temperature oi 0 liquid

rci

40 ±1

Flow velocity of 0 liquid
through membrane chambers
[cm/sl

4.3 ~ 5.0

0	'	1	1	'	'	1	1	

0	0.5 1.0	1.5 2.0 2.S	3.0	3.5

Electric current density [A/dm2]

Figure 8 Selective Permeability Coefficient of Components

Overall heat

transfer
coefficient
iW/ma-K]

1000 -

500

Item

Set Value

Pressure drop degree

[mm Hg abs]

100~120

Temperature [®C]

57 ™ 60

Processing amount [kg/hr]

70

CI concentration of concentrated
liquid |ppm as CI]

250,000-280,000

1000

2000

3000

4000

Operation time [hr]

5000

Figure 9 Overall Heat Transfer Coefficient of Evaporator

107-15


-------
7 8 9 023 123* 56789 TOl t 2 3 4 5 6 7 S S j5

Figure 10 Solidified Product

107-16


-------
FGD WASTE COMPOSITIONS AS LANDFILL LINER

C.L. Smith
Conversion Systems, Inc.

200 Welsh Road
Horsham, Pennsylvania 19044

Abstract

Coal burning utilities are and will be generating more flue gas desulphurization (FGD)
wastes at a time when many states are developing more stringent solid waste
regulations. In more and more cases, generators of these wastes are being required
to use some type of low permeable liner to prevent migration of waste into the
environment and degradation of groundwater. These liners, either clay or a
geomembrane, can be inordinately expensive.

Current developments in sulfo-pozzolanic chemistry offers the potential to process the
FGD wastes into a landfill liner, meeting presently promulgated environmental
regulations and providing the same protection from contamination to the environment
as other liners. These compositions contain sufficient compressive strengths and
flexural strengths which adequately compensate for the plasticity exhibited by clay and
geomembrane liners. FGD liners also exhibit a self-healing property, known as
autogenous healing, which seals stress induced fractures similar to clays.

Introduction

The Clean Air Act Amendments (CAAA)of 1990 will generate millions of tons per year
of FGD sludge, primarily calcium sulfate dihydrate and calcium sulfite hemihydrate, as
some utilities will install scrubbers to obtain compliance. Unless a market materializes
to reuse all of the FGD sludge and flyash (collected by the particulate removal systems
on utility stacks), a significant percentage of these waste products will require
disposal -in a solid waste facility. Furthermore, as many states are re-examining their
solid waste regulations to comply with and implement EPA's Subtitle D landfill
requirement, regulations at the state level are becoming more stringent than were
previously in place. In many cases, liners, composed of earthen material,
geomembranes or a combination thereof, are required in the design of a solid waste
disposal facility handling exclusively coal combustion by-products.

108-1


-------
These liner systems serve as a low permeable barrier to protect the ground-water from
contamination caused by the ieachate of the FGD wastes. The capital cost of these
liner systems will cost the utility millions of dollars. Certainly, a liner system that is
more cost effective yet provides equal protection to the environment should be
welcomed by both the utility industry and the regulators.

Regulatory Perspective

The € AAA requires utilities to reduce sulphur oxide emissions, in phases, to eventually
1.25 pounds per million of sulfur oxides (S02) British Thermal Units (BTU's) by the
year 2000, Phase I of the CAAA requires 110 coal burning utilities to obtain an
emission level of 2.5 pounds of S02 per million BTU's by the year 1995 (for the
purposes of this discussion, the extension provisions of the CAAA will not be
discussed since this provision allows the compliance deadline to be extended by two
years, however, it only shifts the starting waste generation date by two years without
affecting the disposal requirements). Utilities that choose stack scrubbing as their
compliance option will be faced with the disposal problem of a sulfate/sulfite filtercake
stream.

The Resource Conservation and Recovery Act (RCRA) has excluded coal combustion
by-products from regulation as a hazardous waste pending the results of studies and
final published rulemaking on the part of the United States Environmental Protection
Agency (EPA). EPA's final ruling is anticipated by August 1993. Unless and until EPA
publishes final regulatory requirements that drastically differ from present practices,
most coal combustion electrical generating facilities will be required to comply with
current sate solid waste disposal regulations.

EPA has also published regulations regarding disposal of municipal solid wastes (40
cfr 258). These regulations, commonly referred to as Subtitle D, require states to
adopt EPA technical requirements by October 9, 1993. (Note: There is a possibility
that EPA will extend the compliance deadline until April 1994).

Many states have extended the basic subtitle D regulations to the industrial sector.

Table I provides a limited selection of some relevant state regulations along with EPA's
Subtitle D requirements.

108-2


-------
TABLE I

SELECTED STATE REQUIREMENTS FOR LININGS16'



California |

Florida |

Illinois

Pennsylvania |

Virginia

Primary

* Landfill underlain

* Composite liner:

• Compacted

• Subbase (6" soil

• Compacted

Design

by natural

60-mil

earth

K<1X105)

clay

Liner

geologic

geomembrane

• 5 feet clay with

• 50 mil

• 2 feet clay with

System

material with

over 3 feet clay,

K <.1X107

geomembrane or

K <.1X107



K <.1X10 6

K <.1X107



1/4" clay mat





• Sufficient

• 24-inch-thick



• Plus 1 ft. clay





thickness to

protective layer



(K<_lX10"e)





prevent leachate

leachate



• Soil cover 18"





movement to

collection



(K> 1X102)





groundwater









Acceptable

• 2 feet of clay

• Double liner,

• Composite

• (Options for

* 60-mil

Alternative

with K <.1X10 6

Lower 60-mil

lining 60-mil

alternative designs

geomembrane

Design



geomembrane

geomembrane,

are included in the

over prepared

Liner



on 6" subbase

3 feet clay with

primary design)

base, or

System



with KXlXlO5

K <.1X107 or



• Other





• Leachate

• Other as



augemented





collection

approved



clay or soils





between liners;





manipulated in





Kj>lX10'e





place; or





* Upper 60-mil





• Double liner,





geomembrane





leachate





• 24" protective





collection





layer, leachate











collection








-------
Waste Quantities Arid Disposal Requirements

To place the potential effect of the impending regulatory changes in perspective, we
should review the volume of waste under consideration. There are approximately 450
coal burning power plants in the U.S. There are approximately 618 operating disposal
sites serving these utilities, while an additional 120 sites will be required by the end of
the decade. To complete the picture, there are approximately 759 sites already
closed.

The breakdown of the 618 operating sites are given in Table II.

TABLE IT"

Surface Impoundment

321

Landfill

273

Minefill

22

Waste Pile

2

Current coal combustion by-product generation is on the order of 90 million tons per
year. As part of Phase I of the CAAA, approximately 17 utilities have committed to
adding scrubbers to control S02 emissions. This will add approximately 4 million tons
per year of sludge to the existing disposal sites. Many utilities have begun planning
for Phase II compliance. While it is unknown as to how many additional scrubbers will
be installed, utilities will require disposal options, such as landfills.

Liners

A liner is a low permeable membrane which minimizes the amount of uncontrolled
leachate available to the environment. Ultimately, the liner, regardless of its
construction, will allow minimal amounts of leachate to enter the sub-soil and
groundwater. It is the job of the landfill designer to incorporate risk based factors into
the landfill design, and to minimize the exposure a receptor will receive at some
designated compliance point hydrologicaily downgradient from the landfill. Some of
the factors the landfill designer must incorporate into the design include the following:

1)	Compatibility of the liner with the materials being disposed of in the liner;

2)	The hydraulic conductivity of the liner;

3)	The ability of the liner not to crack (span sub-surface voids, settling, etc);

4)	The ability of the liner not to contribute to the quality or quantity of the leachate
generated;

108-4


-------
5) Risk based factors, such as distance to receptors, that will ensure that

groundwater will not exceed Maximum Contaminant Levels or Drinking Water
Standards (whichever is required at a particular site) at a compliance point
designated by the appropriate regulatory agency. Contaminate transport time
to the receptor may be a factor to consider in this subcategory.

Liners are composed primarily of either earthen material (such as clays) or
geomembranes (such as high density polyethylene (HDPE). Typical clay liners are
constructed by importing a sufficient quantity of material to properly compact and
construct a liner at least 5 feet thick with a hydraulic conductivity of
1 x 10'7 cm/sec or less over the entire area to be used for landfilling. If the utility is
fortunate to have in-situ clay, the cost to construct the liner is minimized. More
common is the practice of obtaining clay from a nearby source which requires
obtaining mineral rights to a property, mining the vein, transporting the clay to the site
and constructing the liner. The cost for this activity varies depending on the local
geological conditions, property values and landfill size, and reasonably has a range
between 5 and 20 million dollars for a typical disposal site.

A geomembrane only liner is usually between 30 to 60 mil thick and may be either a
single or double layer depending upon state requirements. These liners are effective
in preventing the passage of leachate, however, they are susceptible to puncture and
seam leakage and require expert installation to ensure proper construction. These
liner systems may be more cost effective given local availability of sources of clay and
the price to mine and transport the clay.

Composite liner systems are composed of a 30 mil layer of a geomembrane and
between 2 to 5 feet of low permeable earthen materials. In addition to the
construction concerns for the individual material, it is imperative that good contact be
made between the geomembrane and the earthen portions of the liner. Poor contact
between the two can allow for increased lateral flow from any hole in the
geomembrane increasing the leakage rate from the landfill as a whole.

FGD Liners

Liners can be manufactured by properly blending FGD scrubber sludge, flyash and
other additive(s). The blend initially results in a damp soil consistency near optimum
density eventually curing into a cementitious mass. Placement of the material is
comparable with clays. After placement, compaction is performed to allow the
pozzolanic and sulfo-pozzolanic reactions to take place bringing about a reduction in
the hydraulic conductivity of the blend. Formulations in the laboratory have
demonstrated permeability coefficients in the ranges of 10'7 to 10'8 cm/sec.
Permeability values of the fully cured stabilized FGD blends can reach into the 10"9
cm/sec range. The cost of these liners tends to be a function of the desired

108-5


-------
permeability. These liners will cost less than their clay or geomembrane equivalent
since approximately 95% of the liner will be constructed from waste materials being
generated by the utility.

As the liner is constructed, the nature of the pozzolanic bonding allows for increasing
the thickness without compromising the liner's integrity due to cold joints. Pozzolanic
reactions continue to take place over a long time. The placement of new liner grade
material on top of already placed liner grade material will result in vertical as well as
horizontal chemical bonding. The final liner effectively behaves as a monolith.

Furthermore, the landfill designer can vary the thickness of the liner providing for
equivalent to greater protection (as compared with common lining materials) of the
groundwater. The greater the thickness, the greater the vertical barrier preventing
percolation into the groundwater, the greater the amount of time it will take a
contaminant to reach a potential receptor.

Another beneficial property of stabilized FGD materials is the eventual development of
the liner into a cementitious mass with an unconfined compressive strength in the 500
to 1100 psi range. The corresponding tensile strength values are approximately 16%
of the compressive strengths or 80-175 psi. These strengths serve to bridge sub-
surface voids and support the weight of landfilled material (including heavy
equipment).

Cracking Of FGD Liners

Since stabilized FGD wastes will form a cementitious product, a brief discussion of the
potential for the stabilized FGD liner to crack thus providing a pathway for leachate to
enter the groundwater is appropriate. A close examination of the mechanism of
concrete cracking and comparison to the properties of the stabilized wastes should
satisfy the concerns of the designer that the liner will not crack under normal
operating conditions.

Concrete cracks for a variety of reasons, including shrinkage from moisture loss and
temperature changes of the mass. Both of these factors cause significant stresses
within the quick curing (28 days) concrete mass. The inability of the concrete mass to
handle the internal stresses causes the external surface to crack to relieve the
stress/strain buildup.

Stabilized- FGD materials are placed as part of the liner at a high percent solids
(approximately 75% or more) unlike concrete which is typically wet placed (poured).
While some of the residual moisture is continually consumed in the slow curing
pozzolanic reaction, very little water is lost to evaporation. Furthermore, since the
stabilized FGD wastes results in a mildly expansive reaction, similar to shrinkage
compensated concrete, cracking caused by shrinkage does not occur. Stresses

108-6


-------
normally caused by temperature differentials do not occur because the slow curing of
the stabilized FGD mass allows the liner to absorb the stresses.

The above is confirmed with field data from Conversion Systems, Inc. (CSI) operation
of the landfill at the Seminole Electric Cooperative, Inc. in Palatka, Florida. (Figure 2)
CSI, as part of the operating contract, regularly cored the landfill and prepared reports
on the condition of the landfill. Analysis of the cores showed little decline in the
moisture content of the cores over time. The cores were also visually examined for
cracks and none were recorded.

Performance of Stabilized FGD Waste Compositions

From 1985 to 1989, four water storage ponds at the Palatka facility that were
previously lined with plastic were found to be leaking; all four were replaced with a five
foot thick lining of the FGD waste composition. The four ponds, each of which
averages about 1/4 acre area, have performed well to date. This is a more rigorous
test as compared to a landfill lining, in that hydraulic gradient builds up slowly via
rainfall in a typical landfill, while the hydraulic gradient was applied to the ponds within
a very few weeks. Figures 3, 4, and 5 show two of these ponds.

After 6 months, cores were removed from segments of the pond lining, in a manner
not compromising the integrity of the linings. The average permeability coefficient
values (after 6 months of cure) was 1.8xi0'8 cm/sec. Unconfined compressive
strength values averaged 679 psi. Hence, the performance and durability of FGD
composition liners have been demonstrated.

Fissures

Small fissures which may develop are corrected by the autogenous healing
characteristic of sulfo-pozzolanic materials. Due to differential settling or uneven
loading in the initial months, some fissures may develop. The autogenous healing
characteristics of the sulfo-pozzolanic materials are capable of bonding the segments
and restoring the hydraulic conductivity initially developed while increasing the
strength of the liner mass.

CSI empirically evaluated the autogenous healing characteristic by preparing a wide
range of stabilized FGD compositions at optimum moisture and density. These
Proctor cylinders were cured at 100% relative humidity for 100°F for seven days.

These cylinders were fractured and then recombined and returned for curing. The
data in Figure 1 shows that the sulfo-pozzolanic chemistry effectively regenerates
strength significantly beyond the original cylinders tested. Table III provides the same
data as in Figure 1, but includes specific strength values and provides the number of
data points used in obtaining the averages shown in Figure 1.

108-7


-------
AUTOGENOUS HEALING

o
<2°

oo

2 3 4 5 6 ?

10 11 12

EQUIVALENT FIELD CURE (Months)

FIGURE 1: Strength Regeneration in FGD Compositions


-------
TABLEffl

FIELD
EQUIVALENT
CURE
(Months)

CURE AFTER
"BREAK"
(Months)

UNCONFINED
COMPRESSIVE
STRENGTH
(AV PSD

DATA POINTS
IN

AVERAGE

1

0

453

43

1

1

949

42

3

0

1042

60

3

3

1195

60

6

0

1151

39

6

6

1271

39

Environmental Impact Of FGD Liners

EPA, as well as many sates, evaluate the potential hazards of a landfill through the
use of fate and transport models. These models will calculate potential exposure to
hazardous chemicals released from a landfill. The use of these predictive techniques
is appropriate in evaluating the environmental impact of an FGD liner system for utility
combustion wastes. Using EPRI's Fossil Fuel Combustion Waste Leaching (FOWL)
model in conjunction with EPA's Hydroiogic Evaluation of Landfill Performance (HELP)
and Composite Model for Landfills (CML), a thorough evaluation of the site design can
be performed demonstrating to the regulators that the final site design will have a non-
adverse environmental impact on a potential receptor.

Conclusions

Stabilized FGD materials can be successfully utilized as a liner for a landfill of coal
combustion by-products. Key design parameters, such as hydraulic conductivity and
thickness, can be varied to provide the utility a cost effective design without an
adverse environmental impact. Design adjustments can be driven by modelling results
thus assuring a potential receptor is not exposed to hazardous materials above the
appropriate state standard. The sulfo-pozzoianic chemical reactions taking place
within the chemical matrix will continue to lower the permeability over time while
providing sufficient compressive and flexural strengths to avoid breakage of the liner
from spanning sub-surface voids and differential sub-surface settling.

108-9


-------
Reference

1.	"Utilities May have To Clean Up Old Coal Waste Sites", Power Engineering,

May 1993, p. 10

2.	R.G. Knight and E.H Rothfuss. FGD By-Product Disposal Manual. Palo Alto,
CA: EPRI. 1993, pp 7-56-7-61.

3.	Hairy M. Freeman (editor). Standard Handbook of Hazardous Waste Treatment
and Disposal. New York, New York: McGraw - Hill 1991, pp. 10.68 - 10.70.

4.	Robert A. Corbitt. Standard Handbook of Environmental Engineering. New
York, New York: McGraw-Hill 1990, pp. 8.123-8.125

5.	R.H. Cook and D.C. Kocunik, "Examination of State Regulations for Dry
Disposal Facilities for Coal-Fired Plant Wastes," Paper No. 54-1, presented at
the American Power Conference, Chicago, IL(April,1993)

6.	C.L. Smith, "Physical Aspects of FGD By-products" International Journal of
Environmental Issues in Minerals and Energy Industry Vol 1 No. 1, p. 37(1992).

7.	Code of Federal Regulations, Title 40, Part 260

8.	Code of Federal Regulations, Title 40, Part 258

9.	J, P. Giroud and R. Bonapack, "Leakage Through Liners Constructed with
Geomembranes," Part I: Geomembrane Liners, Geotextiles and
Geomembranes, 8 (1) (1989) 27-67.

10.	Foundation Engineering Handbook; edited by Hans Winkerton and Hsai-
Yang Fong; Van Nostrand Reinhold Company; 1975.

11.	M.Anderson; G. Sharpe; D.Allen; H. Southgate; and R. Dean; "Laboratory
Evaluations of Stabilized Flue Gas Desulfurization Sludge and Aggregate
Mixtures;" presented at the Seventh International Ash Utilization and Exposition;
Orlando, FL; 1985.

12.	R. Smock; "UtilitiesMay Have to Clean Up Old Coal Waste Sites," Power
Engineering, May 1993, Page 10.

108-10


-------
Figure 2: Florida Utility With 8+ Million Ton Monolithic FGD
Waste - Fly Ash-Lime Landfill

Figure 3; Water Storage Ponds Lined With FGD Waste-Fly Ash-Lime
Composition

108-11


-------
Figure 4: Closer View of Pond Lined With FGD Waste-Fly Ash-Lime
Composition

Figure 5; Comparable View of Another Fond

108-12


-------
TECHNICAL REPORT DATA

(Please read Instructions on the reverse before completi II11 III II 1|||| 111 III 1 1 1 111

1. REPORT NO. 2.

EPA-600/R-95-015d

3. f III HII || Hill 11mi ii 1111

PB95-179255

4. TITLE AND SUBTITLE

Proceedings: 1993 SOg Control Symposium, Volume
4. Sessions 7, 8A, and 8B

S. REPORT DATE

February 1995

6. PERFORMING ORGANIZATION CODE

7. AUTHOR(S)

Miscellaneous

8. PERFORMING ORGANIZATION REPORT NO.

TR-103289-VI, -V2, and -V3

(EPRI)

9, PERFORMING ORGANIZATION NAME AND ADDRESS

See Block 12

10, PROGRAM ELEMENT NO.

11. CONTRACT/GRANT NO.

NA (Inhouse)

12. SPONSORING AGENCY NAME AND ADDRESS

EPA, Office of Research and Development

Air and Energy Engineering Research Laboratory
Research Triangle Park, NC 27711

13. TYPE OF REPORT AND PERIOD COVERED

Final; 8/93-8/94

14. SPONSORING AGENCY CODE

EPA/600/13

is. supplementary notes j^eERL project officer is Brian K. Gullett, Mail Drop 4, 919/541-
1534.

i6. abstract repOI»fc documents more than 100 presentations at the 1993 SG2 Control
Symposium, in Boston, MA, August 24-27, 1993. The presentations covered a wide
range of topics: industry's strategies for dealing with the Clean Air Act Amendments
of 1990, including Phase I strategies, the emission allowance trading system, and
retrofit construction; additiv|.es, materials, and operating issues for wet flue gas
desulfurization (FGD); clean 'coal demonstration programs; the effect of FGD sys-
tems on air toxics; dry FGD technologies of spray drying and furnace sorb en t injec-
tion; applied sulfur dioxide (S02) control research results, and emerging acid rain
control technologies; and waste disposal issues. The presentations covered results
obtained from full-scale demonstration/operation to pilot and bench scale work.

\o

17. KEY WORDS AND DOCUMENT ANALYSIS

a. DESCRIPTORS

b.IDENTIFIERS/OPEN ENDED TERMS

c. COSATI Field/Group

Pollution Toxicity
Sulfur Dioxide Spray Drying
Additives Furnaces
Flue Gases Sorb en ts
Desulfurization Waste Disposal
Coal

Pollution Control

Stationary Sources
Acid Rain

13 B 06T
07b 13 H

11G 13 A
2 IB

07A.07D 15E
2 ID

18. distribution STATEMENT

Release to Public

19. SECURITY CLASS (ThisReport}

Unclassified

21. NO. OF PAGES

526

20. SECURITY CLASS (Thispage)

Unclassified

22. PRICE

EPA Form 2220-1 (9-73]


-------