EPA/600/9-85/020a
July 1985
PROCEEDINGS: FIRST JOINT SYMPOSIUM ON
DRY S02 AND SIMULTANEOUS S02/N0X CONTROL TECHNOLOGIES
Volume 1. Fundamental Research and Process Development
Symposium Cochairpersons:
M. W. McElroy (EPRI) and R. D. Stern (EPA)
Acurex Corporation
555 Clyde Avenue
Mountain View, CA 94039
1
EPA Contract 68-02-3933
EPRI Contract RP2533-3
EPA Project Officer:
P. Jeff Chappell
Air and Energy Engineering Research Laboratory
Research Triangle Park. NC 27711
AIR AND ENERGY ENGINEERING RESEARCH LABORATORY
OFFICE OF RESEARCH AND DEVELOPMENT
U.S. ENVIRONMENTAL PROTECTION AGENCY
RESEARCH TRIANGLE PARK, NC 27711
-------
Forty six papers describing recent advances 1n dry sorbent injection technologies
for SO2 control were presented at the 1st Joint Symposium on Dry SO2 and
Simultaneous SOg/NOx Control Technologies. These papers covered the following
topics: fundamental research; pilot-scale development of furnace injection; burners
for simultaneous S02/N0X control; post-furnace SO2 removal; process integration and
economics; sorbent availability and costs; and field applications and full-scale
testing.
-------
PREFACE
The 1st Joint Symposium on Dry SO2 and Simultaneous S02/N0X Control Technologies was
held November 13 through 16, 1984 in San Diego, California. This symposium, jointly
sponsored by EPRI and EPA, was the first meeting of its kind devoted solely to the
discussion of emissions control processes based on dry injection of calcium or
sodium sorbents to meet SO2 and N0X regulations for coal-fired power plants.
Specific processes that were discussed included: direct furnace injection of
calcium-based sorbents, sorbent injection combined with low-NOx burners for
simultaneous S02/N0X control, and post-furnace Injection of calcium and sodium
sorbents. The objective of the symposium was to provide a timely forum for the
exchange of data and information on the current status and plans for these emerging
technologies.
Forty six papers were presented beginning with a keynote address on acid rain
strategies and control technology Implications, followed by overviews of the EPRI,
EPA, and Canadian programs, and the utility perspective for dry control
technologies. Other papers focused on the latest advances 1n fundamental research
and process design, power plant integration and economics, field applications, and
fuTl-scale testing. A panel of representatives from architect-engineering firms,
boiler manufacturers, and utility companies discussed the impact of dry SO2 control
processes on new and existing power plants.
The speakers included EPRI and EPA staff menbers as well as representa ti ves from
utility companies, manufacturers of utility boilers and process equipment, sorbent —
suppliers, and research and development qroups conducting Investigations sponsored
by EPRI, EPA, and Others. Participants from West Germany, France, The Netherlands,
Austria, Canada, and Japan provided a worldwide update on technological developments
and an international perspective on SO2 and S02/N0X control issues.
The Cochairmen of the symposium were Michael W, McElroy, Subprogram Manager of
EPRI's Air Quality Control Program in the Coal Combustion Systems Division and
Richard D. Stern, Chief of EPA's LIMB Applications Branch of the Industrial
Environmental Research Laboratory.* The welcoming address was given by John Hamrick,
Vice President of Customer Service for San Diego Gas & Electric and the keynote
address was given by Donald J. Ehreth, Deputy Assistant Administrator, Office of
Research and Development, EPA.
The symposium proceedings has been published in two volumes:
• Volume 1: Fundamental Research and Process Development
— Session I: Introduction
— Session II: Fundamental Research
— Session III: Pilot-Scale Development of Furnace Injection
— Session IV: Burners for Simultaneous S02/N0x Control
— Session V: Post-Furnace SO2 Removal
(*) ^ow, the Air and Energy Engineering Research Laboratory.
iii
-------
Volume 2: Power Plant Integration, Economics, and Full-Scale Experience
— Session VI: Process Integration and Economics
-- Session VII: Sorbent Availability and Costs
— Session VIII: Field Applications and Full-Scale Testing
iv
-------
CONTENTS
VOLUME 1
FUNDAMENTAL RESEARCH AND PROCESS DEVELOPMENT
Paper Page
SESSION I: INTRODUCTION
Chairman, Richard Stern, EPA, IERL/RTP and
Michael McElroy EPRI
1 "Acid Deposition Strategies and Implications for Control
Technology Requirements," D. J. Ehreth* 1-1
2 "The EPRI Program — Background and Motivation," J. S. Maulbetsch . . 2-1
3 "EPA's LIMB RAD Program - Evolution, Status, and Plans," G. B. Martin
and J. H. Abbott* 3-1
4 "Overview of Canadian Research, Development and Demonstration
Program for Low N0X/S02 Control Technologies," W. A. Warfe and
G. K. Lee 4-1
5 "The Utility Perspective on Dry SO2 Control Technologies,
G. P. Green 5-1
SESSION II (PART 1): FUNDAMENTAL RESEARCH
Chairman, Kerry Bowers, Southern Company
Services
6 "EPA Experimental Studies of the Mechanisms of Sulfur Capture by
Limestone," R. H. Borgwardt, K. R. Bruce, and J. Blake* 6-1
7 "Flow Reactor Study of Calcination and Sulfation," V. P. Roman,
L. J. Muzlo, M. W. McElroy, K. W. Bowers, and D. T. Gallaspy 7-1
8 "Calcium-Based Sorbents for Dry Injection," J. L. Thompson 8-1
SESSION II (PART 2): FUNDAMENTAL RESEARCH
Chairman, Dennis Drehmel, EPA, IERL/RTP
9 "Laboratory-Scale Production and Characterization of High Surface
Area Sorbents," D. A. Klrchgessner* 9-1
10 "Reactivity of Calcium-Based Sorbents for SO2 Control," J. A. Cole,
J. C. Kramlich, G. S. Samuelsen, W. R. Seeker, and G. D. Silcox* . . . 10-1
*See EPA disclaimer on page ix
v
-------
CONTENTS
Paper Page
11 "Bench Scale Evaluation of Sulfur-Sorbent Reactions,"
D. M. Slaughter, G. D. Silcox, P. M. Lemieux, G. H. Newton,
and D. W. Pershing* 11-1
12 "Evaluation of SO2 Removal by Furnace Limestone Injection with
Tangentially Fired Low-NOx Burner," K. Tokuda, M, Sakai, T. Sengoku,
N. Murakami, M. W. McElroy, and K. Mouri 12-1
13 "Performance of Sorbents With and Without Additives, Injected into
a Small Innovative Furnace," S. L. Rakes, G. T. Joseph, and
J. M. Lorrain* 13-1
SESSION III: PILOT-SCALE DEVELOPMENT OF FURNACE INJECTION
Chairman, Michael McElroy, EPRI
14 "Pilot-Scale Characterization of a Dry Calcium-Based Sorbent SO2
Control Technique Combined with a Low-N0x Tangentially Fired System,"
J. T. Kelly, S. Ohmine, R. Martin, and D. C. Drehmel* 14-1
15 "Boiler Simulator Studies on Sorbent Utilization for SO2 Control,"
B. J. Overmoe, S. L. Chen, L. Ho, W. R. Seeker, M. P. Heap, and
D. W. Pershing* 15-1
16 "Studies of Sorbent Calcination and S02~Sorbent Reactions in a
Pilot-Scale Furnace," R. Beittel, J. P. Gooch, E. B. Dismukes, and
L. J. Kuzio 16-1
17 "Recent IFRF Fundamental and Pilot Scale Studies on the Direct
Sorbent Injection Process," S. Bortz and P. Flament 17-1
18 "Demonstration of Boiler Limestone Injection in an Industrial
Boiler," C. E. Fink, N. S. Harding, B. J. Koch, D. C. McCoy,
R. M. Statnick, and T. J. Hassell 18-1
19 "Pilot-Scale Studies of In-Furnace Hydrated Lime Injection for
Flue Gas SO2 Emission Control," G. F. Weber, M. H. Bobman, and
G. L. Schelkoph 19-1
20 "Bench Scale Process Evaluation of In-Furnace N0X and S0X Reduction
by Reburning and Sorbent Injection," S. B. Greene, S. L. Chen,
D. W. Pershing, M. P. Heap, and W. R. Seeker* 20-1
SESSION IV: BURNERS FOR SIMULTANEOUS S02/N0X CONTROL
Chairman, G. Blair Martin, EPA, IERL/RTP
21 "Evaluation of Low-N0x Burners for SO2 Control," R. Payne and
A. R. Abele* 21-1
22 "Limestone Injection With an Internally Staged Low-N0x Burner,"
J. Vatsky and E. S. Schindler (Paper not submitted) 21—16 •
*See EPA disclaimer on page ix
VI
-------
CONTENTS
Note: There are no pages for section 22
Paper Page
23 "Development of Internally Staged Burners for LIMB," G. C. England,
R. Payne, and J. Clough* 23-1
SESSION V: POST-FURNACE S02 REMOVAL
Chairman, Dan Giovanni, Electric Power Technologies, Inc.
24 "Characterization of Alternate Sodium Sorbents for Fabric Fiber
SO2 Capture," R. Hooper 24-1
25 "Dry Injection Scrubbing of Flue Gases With the SHU Process,"
M. Schutz 25-1
26 "Flue Gas Desulfurization by Combined Furnace Limestone Injection
and Dry Scrubbing," L. E. Sawyers, P. V. Smith, and T. B. Hurst . . . 26-1
27 "Pilot Evaluation of Combined SO2 and Particulate Removal on a
Fabric Filter," F. G. Pohl, M. W. McElroy, and R. Rhudy 27-1
VOLUME 2
POWER PLANT INTEGRATION, ECONOMICS, AND FULL-SCALE EXPERIENCE
SESSION VI: PROCESS INTEGRATION AND ECONOMICS
Chairman, David Lachapelle, EPA, IERL/RTP
28 "Fireside Consequences of Furnace Limestone Injection for SO2
Capture," G. J. Goetz, M. D. Mlrolli, and D. Eskinazi 28-1
29 "Effects of Furnace Sorbent Injection on Fly Ash Characteristics
and Electrostatic Precipitator Performance," R. S. Dahlin,
J. P. Gooch, and J. D. Kilgroe* 29-1
30 "Evaluation of Temperature Histories 1n the Radiant and Convectlve
Zones of a Pulverized Coal-Fired Steam Generator," B. M. Cetegen,
J. L. Reese, K. Kurucz, W. Richter, and D. G. Lachapelle* 30-1
31 "Impact of Sorbent Injection on Power Plant Heat Rates,"
D. Y. Giovanni 31-1
32 "Boiler Design Criteria for Dry Sorbent S0X Control with Low-NOx
Burners," A. Kokkinos, D. C. Borio, R. W. Koucky, J. P. Clark,
C. Y. Sun, and D. G. Lachapelle* 32-1
33 "Wall-Fired Boiler Design Criteria for Dry Sorbent SO2 Control
With Low-NOx Burners," R. K. Mongeon, J. P. Mustonen, and
D. G. Lachapelle* 33-1
34 "Dry Sorbent Emission Control Prototype Conceptual Design and
Cost Study," D. T. Gallaspy 34-1
35 "EPA's LIMB Cost Model: Development and Comparative Case Studies,"
D. G. Lachapelle, N. Kaplan, and J. Chappell* 35-1
*See EPA disclaimer on page ix
Vll
-------
CONTENTS
Paper Page
SESSION VII: SORBENT AVAILABILITY AND COSTS
Chairman, Richard Hooper, EPRI
36 "An Update on the Application of L1me Products for SO2 Removal,"
D. D. Hoffman and D. H. Stowe, Jr 36-1
37 "Dry Injection for FGD Sodium-Based Sorbents: Availability and
Economic Evaluation," R. M. Wright 37-1
38 "Sodium Bicarbonate for Sulfur Dioxide Emission Control,"
R. Shaffery 38-1
SESSION VIII: FIELD APPLICATIONS AND FULL-SCALE TESTING
Chairman, Richard Stern, EPA, IERL/RTP
39 "Reduction of S02~Em1ssion in Brown Coal Combustion: Results From
Research and Large Scale Demonstration," K. R. G. He1n and
G. Kirchen 39-1
40 "Reduction of SO2 Emissions From a Coal Fired Power Station by
Direct Injection of Calcium Sorbents in Furnace," H. Brice,
G. Chelu, G. Flament, R. Manhaval, and M. Vandycke 40-1
41 "Direct Desulfurlzation at the 700-MW Weiher III Unit,"
M. Y. Chugtai (Paper not submitted) 41-1
42 "Laboratory Tests, Field Trials, and Application of Furnace
Limestone Injection 1n Austria," G. Staudlnger and
H. Schrofelbauer 42-1
43 "Experience With Furnace Injection of Pressure Hydrated Lime at
the 50-MW Hoot Lake Station," H. Ness, T. P. Dorchak, J. R. Reese,
and V. Menze 43-1
44 "EPA Wall-F1red LIMB Demonstration," R. V. Hendrlks* 44-1
45 "The Homer City Experience in Developing a LIMB Process for
Use with Coal Preparation," D. W. Carey, D. I. Cessna, and
J. H. T1ce 45-1
46 "N0X/S02 Control Experience at Saskatchewan Corporation's
Boundary Dam G.S. — Unit #6," R. D. Winship and J. A. Haynes .... 46-1
UNPRESENTED PAPERS
47 ; "SuctionPyrometry Tests on Innovative Furnace," S. L. Rakes
and G. T. Joseph* 47-1
48 "Surface Characterization and Microanalysis of Sorbents
and Ash/Sorbent Mixtures," R. S. Dahlin and D. A. Klrchgessner* . . . 48-1
APPENDIX A — LIST OF ATTENDEES A-l
*See EPA disclaimer on page ix
viii
-------
EPA DISCLAIMER
Papers Identified by an asterisk (*) 1n the Table of Contents were funded by the
U.S. Environmental Protection Agency (EPA) and have been reviewed 1n accordance with
EPA peer and administrative review policies and approved by EPA for presentation and
publication. The contents of other papers do not necessarily reflect the views of
the EPA and no official EPA endorsement should be inferred.
ix
-------
SESSION I: INTRODUCTION
Chairman, Richard Stern, EPA, IERL/RTP and Michael McElroy, EPRI
-------
ACID DEPOSITION STRATEGIES AND
IMPLICATIONS FOR CONTROL TECHNOLOGY REQUIREMENTS
Donald J. Ehreth
Deputy Assistant Administrator
Office of Research and Development
U.S. Environmental Protection Agency
Washington, DC 20460
It is really a pleasure for me to be here today. My pleasure steins
principally from the importance of the topic we have assembled here to discuss,
acid deposition.
I would like to begin by describing my personal introduction to the topic,
so you can get a better feel for the factors that have guided EPA's interest in
acid rain control. Then, I plan to bring you up to date on the history of the
work going on to help solve the problem. I'll close my talk with just enough
details of some of the presentations that are before us to whet your appetite
and prepare you to consider several approaches to solutions being considered by
both EPA and the remainder of the engineering research community, both here and
abroad.
Although EPA and the other involved segments of the scientific community
had already expressed considerable concern about the effects of acid deposition,
my involvement peaked about July 27, 1983, when, as Director of the Office of
Environmental Engineering and Technology (OEET), I was asked to convene a panel
of experts representing industry, boi1er manufacturers, engineering firms,
academia, research consulting firms, public policy groups, the power industry,
burner manufacturers, manufacturing and chemical industry representatives, and
(of course) EPRI.
Prompted by the receipt of $5 million from Congress in the Limestone
Injection Multistage Burner (LIMB) area, we convened the panel of 19 people,
many of whom are here with us today. I asked the panel three questions:
(1) Is the relative emphasis between the LIMB research and development
program, and the wall-fired boiler LIMB demonstration project as
envisioned by EPA appropriate?
(2) Is the LIMB demonstration feasible within the timeframe proposed by
EPA? (EPA proposed that a design manual be available by 1988 so that
"somebody" can put this thing on the street after its being demonstrated
at a reasonable boiler size.)
(3) What is the probability of a successful LIMB demonstration?
At that time, public interest regarding acid rain had been steadily growing.
Congress was interested. Congress was active. Congress was excited. Many
bills (no less than 15 of which specified some degree of emission reduction, to
1-1
-------
indicate an opening for control technology solutions) were floating through
Congress. EPA's LIMB program at that time was focused on research and develop-
ment, leaving the demonstration and commercialization to the private sector.
Basically, we were going to turn over EPA's prototype and pilot plant data to
the private sector, which would then assume the responsibility for commercial-
ization. That is, in fact, the philosophy of the Administration: to encourage
the private sector to commercialize solutions to pollution problems, as well
as other problems facing the nation.
EPA laid out very ambitious goals for both degree of removal and cost for
LIMB. We had established a goal of 50 to 60 percent reductions of SO2 and N0X
for retrofit systems. We were estimating capital cost savings of 70% and
overall cost savings of up to 50% for LIMB vs state-of-the-art flue gas desul-
furization technology. Someone had asked if the claims were valid. Were they
figments of our imagination, or were they real? That's why we convened the
panel. We asked them to help us. They were unbiased. They had nothing to
gain or lose from the exercise. I was delighted to work with such a fine
group. They were candid. It was a one-day session. No one left until the
final "gong." From my perspective, it was an incredibly successful venture.
Following the panel meeting, I wrote a summarizing report. Let me summarize
the panel's comments responding to the three questions I asked initially.
Question 1: Appropriateness of the relative emphasis between LIMB research
and development and the demonstration project envisioned by EPA: Of the panel,
37 percent supported the proposed approach; the balance split almost equally
between more research and development emphasis (21 percent) and more demonstration
(26 percent). Three panelists (15 percent) did not express any clear-cut
opinion.
Question 2: Feasibility of the demonstration within the proposed timeframe:
All panelists supported the need for at least one demonstration to prove the
technology. Many questioned the adequacy of a single demonstration to cover
all ooiler/size variables. That the schedule should be at least as fast as
that proposed by EPA was supported by 69 percent of the panel. Some of those
attending saw 1990 as looming on the horizon, and 1988 as a bit too tight.
Most (53 percent) of the panelists agreed that EPA's schedule was fairly reason-
able. Some favored delaying the schedule for a year or two. In fact, between
16 and 26 percent of the group favored delays in place of any acceleration at
all. In brief, there was a broad range of opinions on this question; however,
for the most part, the panel concurred with EPA's proposed schedule.
Question 3: The probability of success, defined as 50 percent SO2 removal
jsing high sulfur coal within 3 years: Nearly three-quarters of the panelists
indicated a 50/50 probability of success, with estimates ranging as high as
70/30. Although 21 percent of the panel did not express a quantitative opinion,
no strongly negative opinions were expressed. In fact, several panelists were
quite positive concerning the potential of the technology.
The panel's conclusions followed an elaborate in-depth presentation of
EPA's latest data. In fact, some of the data had just been generated the day
before, and presenters were arranging graphics and drawing conclusions on the
plane from the West Coast the night before.
Something interesting happened at the meeting, something that we hadn't
anticipated, but something that led to bigger and better things. In addition
to answering the three questions posed by EPA, the panelists evidenced interest
1-2
-------
in a presentation by the Southern Company/EPRI program. Southern's research
favored an intermediate step of using a 20- to 80-MW boiler in the research and
development before the demonstration. Nearly 40 percent of the panelists
recommended this approach be considered, although individual views varied
significantly. In other words, many people thought that EPA's proposed demon-
stration timeframe was acceptable, but suggested adding a smaller unit before
the final demonstration.
A number of the panelists expressed the need for close communication
and coordination between EPA and EPRI. In the past, they said, these
organizations have net informally and infrequently on LIMB. The need for
a more formal mechanism was apparent: the Office of Management and 3udget
(CMB) seemed to know what each of us was planning to do. We'd meet with
CMB; they'd tell us what EPRI was doing. EPRI would meet with 0MB; they'd
tell them what E°A was doing. But although there was communication at the
working level we never formally got to the same table at the same time.
Furthermore, although neither of us had claimed all the technical expertise
available to solve the SO2/N0X control technology program, many of us
thought that (collectively) we could go a long way to develop sensible,
technically feasible, and cost effective solutions to the mounting public
i ssue.
George Green, whom you will hear from in just a little while, told
lie recently what had to be done. George told me unhesitatingly that
there just wasn't enough money for EPRI and EPA to work independently.
He said that there had to be a way for EPA and E?RI to pool resources and
work together. The bottom line was that EPA and EPRI did get together
and, I believe, both learned to value compromise and cooperation.
We are testing a smaller boiler than EPA had originally intended to test.
This option was strongly supported by EPRI and others in the July 1933 meeting,
-rankly, we at EPA did not support it at that time. Furthermore, out of that
meeting grew a joint steering committee for field projects and collection of
information on foreign technology and demonstrations. We are developing a
cooperative research agenda. We have technical coordination and information
exchange meetings every 6 months (two of which have already been held). I believe
that this symposium is a manifestation of EPRI's and EPA's joint commitment to
respond to the acid rain issue with sensible, technically sound, and cost-effec-
tive research strategies.
EPRI and EPA both support these goals; and today's attendance shows that
we are not alone. The problems are many and there is enough room here for all
of us, but let's continue to share needs, goals, research status, discussion of
technical issues and, most importantly, technology transfer. It is no secret
to us, the air pollution control technology professionals, that the
nation's attention has been focused on acid rain policy discussions within
Congress and the Executive Branch for the past several years. For this symposium
to focus on that debate would be a major loss of time and talent. Let's continue
to focus on technical issues. If and when Congress passes legislation that is
signed by the President, the clock starts ticking and the hour of crisis begins.
That's been the experience in the past; and the future should be the same.
The theme of today's conference is to continue what has been started.
At that meeting in July 1983, the attendees (not just EPA) got a glimpse
of EPRI's and industry's plan; however, only a few participants at that
meeting presented real data or plans relating to their own LIMB-like
control technology. Things have changed since then. We are building on a
1-3
-------
spirit of cooperation, born in 1980 and 1981 when we (EPRI and EPA) began to
jointly sponsor FGD and N0X control technology symposia, and to jointly fund
projects to improve electrostatic precipitators (ESPs).
EPA recently implemented the State Acid Rain (STAR) Program. It is an attempt
at contingency planning, an attempt at front-end planning and analysis of
issues which could flow from an acid rain control program if one is enacted.
STAR will focus exclusively on management and administrative problems and
solutions. Research needs might flow from it, but definition of research needs
is the problem for us and the Interagency National Acid Rain Program. The
National Acid Rain Program is aimed at establishing the scientific underpinning
for the cause and effect (or source-receptor) debate. The control technology
research effort might also be considered contingency planning, and that's why
the government has research in LIMB, coal cleaning, fluidized bed furnaces, and
other areas.
The various acid deposition bills that were introduced in the 98th Congress
included approximately 15 Pleasures which required emission reductions. At this
time, I am not sure of what legislation will be introduced by the new Congress,
what it will contain, nor do I want to speculate. As I said before, that is
not the purpose of this conference. It is sufficient to say that the issue
will appear again. Let's assume that the fundamental feature, emission reduction,
will remain. Therefore, research in this area continues to be appropriate.
Congress obviously feels strongly: they appropriated 55 million in 1984; in
1985 they gave us an additional 5-5.5 million; and with that money we plan to
demonstrate wall-fired boilers and set the stage for demonstrating tangentially
fired boilers. LIMB technology, if fully developed to our expectations, would
be uniquely suited to take advantage of a flexible control implementation plan.
Assuming that simultaneous SO2 and N0X controls would be allowed, LIMB, at a 2 to
1 N0x/S0? tradeoff, would afford a very attractive cost benefit. We would project
that LIMB would be considered equivalent to 70% percent efficient FGD if it
achieved 60 percent reductions of both SO2 and N0X, or equivalent to a 60 percent
efficient FGD if it obtained only 50 percent control of each pollutant.
During the next few days, you will hear over 45 papers on all aspects of
dry SC2 and simultaneous SO2 and N0X controls. The topics include, fundamental
research stressing process chemistry and sorbent, gas mixing, pilot-scale develop-
ment, the burner configuration for N0X and SO2 removal, post-furnace SO2 removal,
and sorbent availability and cost which, by the way, can account for 20 to 50
percent of projected annual revenue requirements. You will hear about field
application and testing. You'll hear about key research findings.
For example, we feel that sufficient data are already available for proof
of concept. There is a very narrow window establishing efficiency at somewhere
between 1600 and 2300°F (871 and 1260°C) and sorbents should not be exposed to
higher temperatures. The residence time in that temperature window is extremely
short, perhaps less than 1 second. Therefore, mixing is critical. We found
out recently that we probably would require high surface area sorbents to meet
and possibly exceed our performance goals. Because of the ever-present potential
for slagging or fouling or ESP loading or overloading, high surface area sorbents
offer the possibility to reduce the need to add significant quantities of solids.
However, there is always some good news accompanying such warning: you'll be
hearing that the impact of sorbents is manageable. Full-scale demonstrations
will also be discussed, all completed recently or on-going both in the domestic
and foreign scene: DOE's project, Conoco's project, and projects in Germany,
France, Canada, and Austria. And I know that we have visitors here from Sweden,
Japan, and other countries.
1-4
-------
These few days of exchange can bring all of us to the same level of under-
standing and knowledge- Then we can move on toward achieving a consensus on
the remaining problems and, finally, agree on the R&D that is needed. This is
essential; as George Green has said, "there just isn't enough money to go
around." Supportable performance and cost data are needed if this technology
is to be factored into planning. Industry and government have a unique opportU'
nity to work together as we've already started to do. I especially applaud the
organizations of this symposium who assembled this gro'up and who did the leg
work, the hard work, and the late night work to put it together. And last, but
not least, I draw special attention to those participants from Germany, France,
Canada, and Austria. I look forward to the results of this outstanding display
o* cooperation and am proud to have played even a small role in it.
1-5
-------
THE EPRI PROGRAM — BACKGROUND AND MOTIVATION
John S. Maulbetsch
Electric Power Research Institute
Coal Combustion Systems Division
3412 Hi 11 view Avenue
Palo Alto, California 94303
Good morning and welcome on behalf of EPRI to this 1st Joint Symposium on Dry SO2
Control and Simultaneous SOj/NO* Control Technologies. It is a pleasure for us to
participate in this conference with EPA and to continue the tradition of giving
joint attention to important technologies for air quality control. I think this
conference may be rememberd as a landmark session in that it is the first of its
type devoted to just these particular technologies. As Mike McElroy noted while
introducing me, I am new to this research area. So to the degree that these are
emerging technologies, I guess I might be described as an emerging technologist.
My charge this morning is to give EPRI's perspective on these dry control options.
To begin, let me say that there is clearly a fundamental benefit to having a variety
of approaches to any particular problem. One of the obvious character!sties of the
electric utility industry is its diversity. Across the industry, there is a
diversity of financial situations; a diversity of regulations that have to be
contended with; a diversity of fuels, sizes of plants, ages of plants, availability
of water resources, availability of space for the management of solid byproducts;
and a diversity in the degree to which each utility is prepared to accept the risks
of technological innovation. Therefore, when you pose the question, for example, of
how to deal with acid rain related pollution control requirements, you get, as you
would expect, a great diversity of opinion. No one solution need be universally
preferred at all times and in all places in order to be useful. But each potential
solution does have to have a niche, and one of the things that we have to be
attentive to as we do our research is identification and characterization of these
niches. SO2 control can be achieved in many ways — from plant retirement to fuel
switching, fuel blending, fuel cleaning, or flue gas treatment — and the
technolgoies we are considering here today, particularly furnace sorbent injection,
are fundamentally different ways. Personally, I think furnace sorbent injection is
an especially attractive approach in that it represents intervention in the process
of power production to control pollutants at the point of their liberation or
formation. This represents another strategy in the arsenal of technologies
available to the industry.
If there is a theme to my remarks today, it is that dry SO2 control and simultaneous
N0x/S02 control have been and still are grounded in perceptions. As our research
continues, this is changing of course. But, whether this change is coming as
rapidly as the forces which may lead to decisions to implement the technologies is
not entirely clear. Decisions to implement are coming fast upon us, and they may
have to be made without all the facts in hand. If so, they will necessarily be made
on the basis of current perceptions.
2-1
-------
One important perception is that these technologies are simple and inexpensive. Now
that's bound to be a plus. They represent, I think, a part of a trend away from
end-of-the-pipe technologies that are perceived as complicated and expensive. I
have a colleague at EPRI in the health effects group who in a previous incarnation
was involved in a massive survey of health care services in the United States and
around the world. Emphasis in the survey was on the relationship between levels of
health care and economic welfare. After several years, several million surveys, and
the expenditure of a great deal of money, the survey team held a press conference to
reveal its finding. The finding can be summed up in one sentence: "It is better to
be well and rich than to be sick and poor." I guess the technological equivalent of
that is that it is better to be simple and inexpensive than to be complicated and
costly. Technological simplicity, however, implies more than just conceptual or
mechanical simplicity. It implies in sone sense an absence of risk, or at least an
understanding of the risks involved, and a sense that you are not buying into
problems that are not obvious from the beginning.
As to the perception of low cost, I think there is general agreement that these
technologies represent a substantially reduced first cost. However, the degree to
wnich they represent a reduced total cost depends strongly on the performance
expectations you hold for them and what you think you are going to need to do to
reduce whatever uncertainties and risks that are associated with them. So there are
some countervailing concerns, and some of these concerns are based on historical
experience with technologies which were less than completely satisfactory. In this
vein, Mr. Stern spoke earlier about being a born-again dry injection technologist,
and harped back to the experience in the 1960's at TVA's Shawnee plant where dry
injection did not perform as well as might have been anticipated, and where the
effect on the boiler was less favorable than anticipated.
A moment ago 1 talked about risk, and how simplicity might be interpreted as
implying an absence of risk, or certainly of reduced risk. This is not to say there
are no strong perceptions of risk with these technologies, however. For example,
there is the risk of whether the results we are seeing at the bench scale and the
pilot scale can be achieved at full scale. If the achievable performance at larger
scale is less than expected, then the economics of the process get called into
question. There is also the risk of exacerbating particulate control problems by
adding large quantities of additional material which must be taken out in
particulate control devices. This concern is especially critical in retrofit
si tuations.
Another risk is that the generation of solid byproducts of a different physical and
chenical composition may make waste management both in and outside the plant more
difficult. Finally, and specifically with furnace sorbent injection, there is the
risk of fouling the boiler by intervening directly in the combustion and heat
transfer process.
Given these considerations, these perceptions, what then has been and is EPRI1s
attitude toward these technologies? In a word, they clearly look like they are
worth a shot. If the risk can be dealt with — and as we proceed with our work I
think we are coming close to the conclusion that they can be dealt with — then the
hoped-for results look well worth the effort of an intensive research, development,
and demonstration program.
To take furnace injection as an example, let me give you a bit of the history and
scope of our EPRI research effort. We started off 2 or 3 years ago with some proof
of concept work in conjunction with MHI on a tangentially fired pilot-scale boiler
in Japan. This work was done at a modest scale, but it confirmed the expectation
that performance was good. We moved from there to more general bench-scale work in
cooperation with Southern Company Services and Southern Research Institute. Here we
tried to learn more about the fundamentals of the process — where one ought to do
2-2
-------
the injecting, what one ought to do with regard to the choice and preparation of the
sorbents, and what temperature was appropriate. At the same time this was going on,
similar rekindling of interest was going on at EPA and elsewhere. In mid-1983, we
tried to bring the programs together in a formal way, and in the course of doing
that we all learned some things. We learned more about performance and what governs
it, we learned more about where to inject the reagent, about how you go about
choosing the reagent, and about how you go about preparing the reagent. We have
learned something, I think, about the relationship between the injection process and
the combustion process, and about the mixing in the postcombustion region that gives
us yet a better chance for success. Probably the most significant question
remaining is the risk associated with the boiler. In this regard, we and others
will be continuing to work at the bench scale and at the pilot scale to learn more
about the fundamental mechanisms involved,°the hoped-for result being, of course, to
increase our confidence and our ability to scale results up to large sizes.
We are also about to initiate a substantial program to look at waste management
issues associated with these technologies as well as with atmospheric fluidized bed
coal cleaning and other nonstandard coal combustion processes. Given that these
technologies involve new, fundamentally different wastes, our goal is to see
whether we are buyng into anything fundamentally different and more complex in the
in-plant management of waste, its transport, containment, and the degree to which
injection may affect the potential for waste utilization. We have seen waste
utilization start to displace disposal for standard flyash, and it would be
regrettable, I think, to somehow interrupt that trend by not being prepared for it.
Some of these issues can only be dealt with at the demonstration scale, so it is
important to move as quickly as we can out of the laboratory and out of the pilot
scale into field demonstration. It is with this in mind that we look forward to
formal cooperation with EPA in its recently announced demonstration with Babcock &
Wilcox and Ohio Edison. We are also looking vigorously right now for other
demonstration opportunities. We think 1t 1s important to conduct demonstrations in
a variety of situations using a variety of fuels and a variety of boiler types under
a variety of utility operating conditions. So, we will be working hard over the
next few months to identify some opportunities for this type of demonstration. We
hope to have companion demonstrations to go along with the EPA work by perhaps
1986.
Many of my remarks here today, and those of previous speakers, have emphasized
furnace injection. Part of my rationale for this is that I will be followed here in
a few minutes by George Green who will talk in detail about backend, or baghouse
injection of both sodium- and calcium-based sorbents. This technology is a little
further along, really at the stage of commercialization, and we at EPRI have a sort
of fatherly interest in it, having been associated with bringing it all the way from
the laboratory to the commercial arena. Development of flue gas dry sorbent
injection technology is almost a textbook example, I think, of what a well thought
out research program and the cooperation of the right kind of parties can accomplish
in a relatively short period of time.
Let me conclude simply by saying that there are a nurrber of interesting technologies
on the table here, all of which I think have a real potential to make a difference
in the way the industry we serve does business. I am optimistic about their
potential; I hope we are successful in achieving it; and I hope we have a good time
in the process. I look forward to working with all of you in the years to come as
we continue in this direction.
2-3
-------
EPA's LIMB R&D PROGRAM - EVOLUTION, STATUS, AND PLANS
G. Blair Martin and James H. Abbott
Industrial Environmental Research Laboratory
U.S. Environmental Protection Agency
Research Triangle Park, NC 27711
ABSTRACT
The LIMB R&D program has provided a detailed understand^ng of tne key processes
governing sulfur capture with sorbents. While it appears that 1irrestcne alone will
not achieve program goals, several other promising soroents have been identified.
Basec on the R&D results and cost estinates use cf these sorbents, LIMB shows
substantial promise as a S0X and N0X ccrtrol technology for retrofit applications.
The ongoing R&D program should resolve the remaining tecnnical questions and provide
a basis for widespread private sector commercialization.
This paper provides a brief history of sorbent injection technology, synops'zes the
status of LIMB R&D, and discusses future program plans.
INTRODUCTION
The EPA is developing Limestone Injection with Multistage Burners (LIMB) as a
potential low ccst control technology for S0X and N0X. which are believed to be two
of the major precursors of acid precipitation. The LIMB program is structured to
provide an understanding of the controlling factors in the process and to establish
a oasis fcr private sector decisions on commercialization. Tne purpose of this paper
is: 1) to provide a brief history of previous and other current sorbent furrace
injection effcts; 2] to summarize the status of the LIMB program; and 3} to outline
planned research ard development. Most of the technical subjects discussed in this
paper will be discussed in greater detail by other symposium presentations.
BACKGROUND
Sulfur oxides {S0X) and r itrogen oxides (N0X) are two rrajor polljtants resulting "^om
the conbustion of fuels. Coal fired utility boilers account for about 70 percent of
the S0X and 2C-25 percent o~ the N0X emissions in the United States. For the 180,000
MW of coal f:red boiler capacity east of the Mississippi River, this amounts to
approximately 16 million tons* of SO2 and 4-5 million tons 0^ NCX per year. Only
about 10 percent of these ooHers are subject to NSPS contro's for S0X and N0X.
Therefore to accomplish any significant reduction in S0X and N0X requires a retrofit
of existing boilers which may have a remaining useful lifeof from 5 to 30 years. The
vast majority 0* these are wal 1-f i red and tangentiaTy fired boilers.
x 1 Ton = 907 kg
3-1
-------
The seriousness of the acicl rain problem and the importance of various acid orecursor
sources are still under debate; however, control of SO2 from utility boilers is a
major elenent in all proposed strategies. In addition N0X is increasingly being
linked to forest damage mechanisms. Proposed reductions of SO2 range from 5 to 12
nillion tons per year, and the issue of an N0X offset is being debated.
Control Technology Options
The choice of emission control strategies will have a significant effect not only on
the abi1ity to achieve any mandated reduction but also on its cost to the nation, 'he
final decision cn the technology mix will be based on the availability, specific
performance, cost, and overall economic impact on the nation, including socio-
economic factors such as displacement in the work force. Among the choices
commercially available are coal switching, coal cleaning, and various types of'flue
gas desulfurization (FGD) systems. In addition, some early analyses indicated that
a low capital cost technology would be attractive even at moderate.SO? removal (e.g.,
5C percent control). Although there are several approaches being developed, none of
the potential 1 ow cost alternatives have been demonstrated. One such rapidly
emerging technology is LIMB which is based on injection of sorbents into the boiler
fo*- direct capture of SO2 from the combust-on gases.
History of Sorbent Injection
There is a considerable body of bac
-------
times stoi ch -'onetr ic calc' um-to-sul fur ratio was of the order of 20 percent and
that, for captures approaching 50 percent, as much as seven times stoichiometric was
required. There are a number cf reasons quoted for this relative lack of success.
The two mentioned most often are: (1) dead burning of the limestone, and (2)
insufficient mixing. Dead burning is the phenomenon of heating the limestone to a
temperature above which the fresh calcine, lime (CaO), recrystal1izes, causing the
reactivity of this lime (as influenced by the surface area) to decrease drama-
tically. Under these high temperature conditions the potential for capture is thus
reduced, and there would be a decomposition of any calcium-sulfur complexes that may
have formed before the dead burning occurred. In terms of mixing, the theory was
that, since the limestone was injected with relatively cold jets -nto streams of hot
gases rising from the combustion zone, insufficient mixing and contacting of the S0X
with the sorbent occurred. Whatever is the correct explanation, it should be noted
that the success in this particular effort on a wall-fired boiler was very limited.
In addition, boiler operability problems, including convective pass and preheating
plugging, were observed.
The other efforts, which may relate more directly to the conditions of concern to the
_;vE program, were the experiments carried out in both tangentia'ly and arch-fired
boilers. In these boilers the mixing of air and fuel is delayed and, as a result,
it is theorized that the peak temperatures are lower. In addition to this, the
boilers are slow mixing devices where the fuel exists in a fuel-rich zone for a
significant period cf time before mixing with the air. For this reason both of these
systems in an uncontrolled mode give relatively low leve's of N0X compared to the
wall boilers cf the same vintage. For both the tan gen t i a 11y ard arch-fired boilers,
sul*'jr capture levels approaching 50 percent for stoichiometric ratios of calcium-
to-suifjr less than two have been reported in the literature.
Germar Experience. Within the past few years two activities related to sorbent
i njection for control of 50x have been initiated in the Federal Republic of Germany.
Information related to these boilers was also considered in structuring the LIMB
program. While the German work is germane to the current LIMB program, differences
in fjel characteristics and boiler designs Co not allow direct extrapolation to U.S.
boilers.
Rheiniscn-Westfaliscnes Electrizitatswerk (RWE), the major brown coal burning
utility ;n the Federal Republic of Germany, has performed the most advanced German
effort. The properties of brown coal are between those of peat and lignite, and the
b»-own coal is mined by a stripping technique involving a large geographical area.
The characteristics of brown coal are high moisture (up to 50 percent), ash with a
highly fouling characteristic because of high alkali metal content, and relatively
"ow sulfur. The RWE work was initiated because of concern about the possibi1ity of
sulfur emissions standards for German utilities. Under normal operation the alkali
in the brown coal captures a significant percent of the sulfur; however, the
composition of fuel changes in such a way that, as the fuel sulfur increases,
generally the alkali decreases. Under the scenario of maximum sulfur and minimum
alkali, the inherent capture would probaoly be insufficient to meet the anticipated
German regulations. For this reason RWE looked into the possibility of incremental
alkali metal addition for additional capture to meet the standard. The initial work
was carried out on a small down-fired research facility where sorbent effects, fuel
effects, and operating variables were examined. Based on this work, the conclusion
was that tne sorbents in the order of decreasing effect:veness were calcium
hydroxide, calcium carbonate (limestone), and calcium oxide (precalcined lime).
While the activities of the calcium carbonate and calcium hydroxide were close
together, the act;vity of calcium oxide was considerably less. Based on the results
3-3
-------
of this testing tne system was installed and tested on a 60 MW (electrical) boiler
in the RWE system.
Based on tne success at the 50 MW scale, RWE has installed this technique on a 300
MW brown-coal-fired power plant. As opposed to the relatively simple and cuic'<
installation used at the Fortuna Station (60 MW), the 300 MW unit required a
signi^'cartly nore extensive equipment investment. Inc^ded in the effort were the
provision of a rail siding for supply of the limestone or calcium soroent to be used
in the system, bunkering capacity, and the possibility of additional pulverization
capability. The estimated cost of th;s effort is about $12 million or ^oughly
$40/kW. It should also be noted that, because of the precipitator design and the
reported changes in the ash characteristics, the anticipation is that no changes in
the precipitators w" 11 be required to comply with the recessary particulate removal
efficiencies. 'herefore, no incremental costs for upgrade cf the particulate
collection system are included in this number.
L&C Steinmjller GmbH, one of tne major German boiler manufacturers, has explored the
applicability of the distributed mixing burner concept to their boilers. Following
some "imited pilot-scale testing at IFRF, Steinmuller exercised a target of
opportunity at a 700 MW boiler located in the Saar region cf Germany. During an
outage scheduled for other mairtenance the burner zone of the boiler was modified
to incorporate four tertiary ports in the wall arojnd each of the existing burners.
The N0X performance of the boiler was reduced from a baseline number of abound 600
ppm at 6 percent excess oxygen to around 25C ppm at 6 percent excess oxygen.
L&C Steinmuller has also done extensive pilot-scale work cn sorbent injection with
its staged mixing burner. 3ased on experimental results at 10 to 100 % 10^ Btu/hr*,
LCS has installed a sorbent injection system at the We:her III boiler. Performance
evaluation is scheduled for late 198^ to early 1985.
EPA Distributed Mixing Burner Testing. In 1979 a limited series of pilot-scale
tests were carried out at the 10 x 10° Btu/hr scale using sorbent injection through
the distributed mixing burner. This testing was carried out based on theoretical
analyses suggesting that the presence of fuel-rich conditions and delayed heat
release leaoing to lower peak temperatures rright give a beneficial effect for
sorbent removal of S0X in second-generation low- N0X burners. For this particular
testirg, the S0X removal was evaluated based on gas-phase measurements of SO? in the
flue gas, and no attempt was made to close the sulfur balance.
The testing was carried in the small watertube simulator (SWS) located at the EERC,
El Toro, facility. The burner was a dual-throat distributed mixing burner that had
been used for developing basic design criteria for second generation staged low-NOx
burners. For this test series, pre-pulverized sorbents were used. The sorbents
were mixed with the coal, and tne coal/sorbent mixture was passed through the
pulverizer. This affected the pulverization cf the coal and the intimate mixing cf
the sorbent with the coal. The sorbent used included limestone, sodium carbonate,
trona (a mixture of sodium bicarbonate and sodium carbonate), calcium oxide (pre-
calcined limestone), and several naturally occurring calcium- and sodium-containing
minerals. Test results indicated that the effective additives were limestone,
* 1 Btu = 1.06
-------
sodium bicarbonate, and trona. For limestone the observed removals ranged from
approximately 50 percent at a calciurr-to-sulfur stoichiometry of 1 to 80 percent at
a calcium-to-sulfur stoichiometry of 3. For the sodium-based additives the removal
ranged from 40 percent at a sodium-to-sulfur stoichiometry of 2 up to 70 percent for
a sodium-tc-sulfur stoichionetry of 4. While the absolute levels of emission
reduction should be viewed with caution, the results gave sufficient incentive to
Dursue the technology further. As a result of these tests, a number of hypotheses
were formulated as to the critical parameters in the reaction of limestone with
sulfur in distributed mixing burners. These include:
(1) Intimate "mixing of the sorbent with tne coal leads to a high degree of
contacting of the active sorbent with the sulfur early in the flame.
(2) Contacting of the calcium and si^fur under fuel-rich conditions leads to the
formaticr of calcium sulfide, which is stable to higher temperatures.
(3) Peak temperatures in the burner are be 1 ieved to be reduced.. In addition, the
SWS has a relatively cold firebox. Both of these lead to lower peak temperatures
and therefore a higher likelihood of retention of the sulfur by the sorbent once
captured.
(4) The in-flame calcination of calcium carbonate leads to a potentially higher
activity of the sorbent and an enhancement of surface area during the
calcination. When this process occurs in intimate contact with the sulfur
specie, the capture may be enhanced. The relatively poor results with pre-
calcined 1imestcre tended to substantiate this hypothesis.
(5) There is evidence that the limestone exiting the combustion zone has a high
level of residual activity which could lead to subsequent capture of sulfur in
the convective passes of the SWS.
Structure Of The Program
This background information was used in structuring the LIMB program, which was
initiated by the EPA in 1981. LIMB combines sorbent injection for S0X control with
lcw-N0x burners for N0X control. Low-N0x burners of various designs have been
developed by both EPA and private industry and are capable of retrofit applications.
The S0X control by sorbent injection is an emerging technology wh'ch has been
developed by the EPA. The reaction of S0X with sorbents (i.e., limestone and other
alkaline solids) is well krown under proper conditions (e.g., wet rGD). LIMB is
based on injection of a sorbent directly into the furnace and its subsequent
reaction with gas-phase SO2 to form a dry calcium sulfate. The amount of SO2 that
can be captured is dependent on the type and amount of sorbent, its mixing with
combustion gases and fly ash in the furnace, and its thermal history. The relative
simp"! icity of the technology lends itself to a relatively low cost retrofit on a wide
var i ety of systems. The program has beer structured to give the best probability
0* achieving the stated goals of moderate S0X and N0X control (50-60 percent) at low
cost with applicability to the major portion of the existing ooiler population. A
secondary objective is to improve the S0X removal efficiency to 70-90 pencent for
new source's, with possible retrofit in selected cases. To achieve this, work has
been concentrated in four major areas, as discussed below.
Generic R&D. The program is centered around generic R&D to provide a complete
understanding of the important factors in sulfur capture by sorbents. T h i s work is
performed in experimental systems which simulate conditions in a boiler without
Qeing subject to hardware constraints. The results to date have shown that the
3-5
-------
sj1 fur capture is strongly dependent on conditions within the boiler which determine
not only the activation of the sorbent but a"1,so the resultant reaction with S02-
Tnis work has shown the necessity for considering sorbents other than limestone
(e.g. dolomitic and calcitic hydrates). The generic R&D is currently concentrated
in improved sorbent activity for retrofit systems, although it also has application
to new systems.
Prototype Test ing. The results of the small-scale generic R&D must be scaled up to
practical systems. As an intermediate step prior to a demonstration prototype,
testing is carried out in large experimental systems. In addit'on small boilers may
be jsed to provide R&D data on scale-up, operability, and reliability.
Wall-fired Boilers Demonstration. Wall-fired boilers, one of the two major types of
boilers, are sold by three manufacturers: Babcock and Wilcox, Foster Wheeler, and
Riley Stoker. The wall-fired boiler, whicn is a major source of SO? and N0X, has
been the subject of the initial LIYB develooment becajse both the R&D background and
large-scale experimental faci1 ities were available. The demonstration program for
a represent ative wall-f ired ut i1 i ty boiler was initiated in FY 84 and is d i scussed in
more detail below.
Generalization of the Technology. To be widely accepted, a well defined set of
criteria *or application of LIMB to a wide range of boiler designs, coals, and
sorbents is necessary. A limited numoer of demonstrations alone may not be
sufficient for widespread private sector corrmercialization. A key element of the
overall program is to produce the required information and methodologies for
applyi ng LIMB R&D to s i te spec if i c design dec i s ions for any g iven boiler. This will
be accomplished by a combination of modelling techniques and supporting measure-
ments on operating boilers.
RESEARCH AND DEVELOPMENT PROGRAM STATUS AND PLANS
This section discusses the current status and R&D plans for each of the four major
program areas discussed in the program description above.
Generic R&D
The LIM3 program has been based on the fact that a comolete understanding of the
process is necessary to give the maximum probability of successful commerciali-
zation by the private sector. Generic R&D is relatively independent of the
hardware-specific constraints of practical boilers and provides information es-
sential for application of LIMB to all boiler designs. This section prov ides a brief
description of the current status, ^ollowed by a detailed discussion of plans for
FY85 and following years.
Current Status. The R&D has provided an excellent insight on the effects of critical
process parameters on. SO2 capture. It has shown how these oarameters affect sorbent
activation and subsequent sulfur capture as a function of combustion system
conditions. It has also provided an understanding of fly ash/sorbent mixture
characteristics as related to slagging, fouling, and particulate capture. As a
result of these findings, it has been concluded that limestone alone will not
achieve the i-IYB sulfur capture goals for many units in the U.S. boiler population.
However, it has also provided at least two alternate sorbent aopncaches capable of
meeting or substantially exceeding the goal of 50-60 percent capture. These
3-6
-------
approaches, which are high surface area sorbents and promoters to enhance sorbent
activ'ty, are discussed below:
Major Generic R&D Areas. The continuing small-scale R&D needs are divided into
three major areas:
1. Inhojse research has identified high surface area sorbents as a key factor
in obtaining high sulfur capture. It has been shown that high sjr'ace area can
be generated external to the combustion orocess and/or in situ. These
materials, which are processed limestone (e.g., hydrates), have shown the
potential for sulfur capture in excess of 70 percent. The planned R&D addresses
methods for obtaining highly reactive sorbents, for optimizing reaction
conditions to achieve maximun capture, and for minimizing sorbent costs.
2. Another key factor is the interaction of sorbents with mineral matter which
can either enhance or degrade the sorbent reactivity. The most promising
results indicate that it may be possible to add small amounts of relatively
inexpensive, innocuous promoters (mineral compounds) which will enhance the
sorbent activity. Sulfjr captures approaching those o* high surface area
sorbents have been achieved with promoted limestone in lirrited bench scale
exper;ments. It also appears tnat promoters can significantly improve the
oerformance of high surface area sorbents (e.g., hydrates). A significant
effort is necessary to understand the enhancement mechanisms and to provide the
basis *or jse in practical systems. It should be noted that a similar
understanding is necessary for other sorbent/mineral matter interactions which
can inhibit sulfur capture and whkh affect slagging, fouling, and collection
cnaracteristics of the particu'ate.
3. Process analysis has indicated substantial benefits may be derived from
recycle of unreacted sorbent and promoters. In addition, utilization of the
spent sorbent and fly ash has significant potential economic benefits. Pilot
scale R&D is necessary to evaluate the engineering feasibility of these process
opt ions.
Prototype Testing
Extensive prototype testing of wall-fired boiler lcw-N0x burners has been conducted
to evaluate sulfjr capture potential with injection of conventional sorbents. A
substantial data base exists for both N0X and S0X control potential of a number of
exDer;mental and commercial burners. Any additioral work will be in support of site
specific decisions for the wall-fired demonstration.
The emphasis will be shifted to tangentially fired prototype systems and to use of
improved sorbents, which are identified in generic R&D, for all systems. Tangential
prototype work will be initiated in a large-scale experimental facility with a
filing system prcducing a vortex flow field typical of that boiler class. In
acdition, a cooperative testing R&D program will be conducted on a srra"• 1 boiler (20-
4C i'"U') to eval uate su 1 fur captjre potent ial, operab i 1 i ty, and rel iab i 1 i ty over short
periods with fuel and sorbent flexibility. The prototype work will concentrate on
evaluation of ootimum scrbents and injection methods to maximize SO2 capture.
Wall-fired Demonstration
The contract for the wall-fired demonstration was awarded in September 1984 to
Babcock & Wilcox Company. LIMB will De installed on a 105 MW single-wal1 -fired unit
3-7
-------
at the Edgewater station of Ohio Edison Company. The final site specific design for
the installation will be completed in February 1986. Long term testing over a 1 year
period will begin in July 1937, ard a report document;ng the performance evaluation
will be completed late in 1988. The funding to complete the effort is provided in
the FY85 budget, and no outyear contingency funds are identified.
Techrclogy Generalization
For ultimate widespread use of the LIMB technology, the R&D results must be
integrated with tie ful'-scale boiler demonstration results to provide guidance for
commercialization by the private sector. The program includes: 1) process analysis
to evaluate applicability and economics for specific systens; and 2) process
modelling to provide a methodology useful for s;te specific designs. The orocess
analysis emphasizes LIMB system options for application to different boiler classes
in the population and for minimizing the cost per unit SO2 removal. The process
modelling will provide component models for thermal history, sorbent activation and
reaction, injection, and mixing.
SUMMARY
The LIMB R&D program has provided a detailed understanding of the key processes
governing sulfur capture with sorbents. While it appears that limestone alone will
not achieve program goals, several other promising sorbents have been identified.
Based on the R&D results and cost estimates of the use of these sorbents, LIMB shows
substantial promise as a S0X and N0X control technology for retrofit app¦icat ions.
The ongoing R&D program should resolve the rerna'ning technical questions and provide
a basis for wide-spread private sector commercialization.
This paper provides a brief history of sorbent injection technology, synopsizes the
status of LItvB R&D, and discusses future program plans.
3-8
-------
OVERVIEW OF CANADIAN RESEARCH, DEVELOPMENT
AND DEMONSTRATION PROGRAM FOR LOW N0X/S02
CONTROL TECHNOLOGIES
W.A. Warfe, and G.K. Lee
Energy, Mines & Resources Canada
580 Booth Street
Ottawa, Ontario K1A 0E4
ABSTRACT
One of the major concerns, associated with the expanded use of coal for
heat and electricity, is the emission to atnosphere of the acid rain
precursors, N0X and SO2 -
This paper outlines the technologies, the status of research,
development and demonstration activities and future plans for Low
N0x/S02 control technologies in Canada. It includes federal
government activities as well as those of the Canadian Electrical
Associ ation.
INTRODUCTION
In Canada we receive some criticism that because we have no flue gas
scrubbers on our utility industry we are doing little or nothing about
acid rain as one of the major concerns associated with the expanded use
of coal for heat and electricity. I therefore welcome this opportunity
to present to you our Canadian strategy and program for control of the
acid rain precursors, N0X and SO2 emissions to the atmosphere. In
1982, North American utility boilers emitted over 6.8 million tonnes of
N0X and 15.8 million tonnes of SO2 with Canadian utility sources
accounting for about 3.5% of the N0X and 5.3% of the SO2 emissions.
Of the total SO2 emissions in Canada 60% is smelter originated and 15%
from utility sources.
STRATEGY
Our strategy for utilities is based on the concept of cost effective
control of acid gas emissions from fossi1-fuel-fired thermal generation
wi th the realization that direct involvement by the utility industry is
essential for the success of our program.
Energy, Mines and Resources Canada embarked on two paths: one to develop
control technologies for existing sources and the other for new sources.
A-l
-------
Control technologies for existing sources includes coal beneficiation,
coal/v/ater mixtures, absorbent injection in low N0X designed burners
and advanced slagging conbustors. New source technologies centre on
fluidized bed combustion with a preference for circulating fluid bed in
the utility application.
The federal source of funding for development and demonstration of these
new technologies is mainly from the Coal Utilization Program,
administered by the Coal Division of Energy, Mines & Resources (EMR).
It was created in October 1980 to provide $150 million until March 31,
1986 and directed mainly to getting Maritime utilities off oil. Other
sources of funds come from base annual reserves such as Canada Centre
for Mineral and Energy Technology (CANMET) and the Office of Energy
Research and Development (OERD). CANMET through its Energy Research
Laboratories such as the Combustion and Carbonization Research
Laboratory (CCRL) provide the Coal Division vvith technical support in
their various projects.
Approximately one-third of the Canadian Electrical Association (CEA)
research funds is provided by EMR who have representation on the various
research committees.
FUNDING
PROGRAM
The federal program to reduce N0X and SO2 is as follows:
Project
Qbjecti ve
Federal
Fundi ng
Maritime Coal
Benefi ci ati on
To assess and develop the
benefi ci ati on potential of
both New Brunswick and
Nova Scotia coals.
$2.1 million
over 3 years.
Coal Water
Mixtures
To develop a liquid coal
fuel that wi 11 replace
oil as a fuel in utility
boilers.
Currently $7.2
mi 11 i on wi th
another $8.0
mi 11ion
budgeted for
the next 2
years.
Low N0x/S02
Burner Demonstration
at CFB Gagetown
To demonstrate low NOx/
SO2 burner technology
at an industrial scale.
$1 .5 mi 11i on
International Energy Canada, Denmark and Sweden
Agency - Agreement agreed to validate various
in N0X and SO2 non-US coals for N0X an(j
$0.3 mi 11 i on
SO2 emissions utilizing
the advanced low N0X
burner concept.
4-2
-------
TransAlta Utilities
Low NOX/S02
Pi 1 ot Seal e
Development
Atmospheric Bubbling
F1nidi zed Bed Tech-
nology Demonstration
Canadian Forces Base
Surxierside and
materials testing
at Point Tupper
Pilot plant.
Coal/Oil Shale
Ci rculati ng
FI in di zed Bed
Technology
Denonstrati on,
Chatham,
New Brunswick.
EMR/CANMET
F1uidi zed-Bed
Research and
Development.
To evaluate the Rockwell
process and its applica-
bility to Western and
Eastern Canadian coals.
To demonstrate atmospheric
bubbling f1ui di zed bed
technology at the
industrial scale.
To demonstrate the
circulating fl ui di zed
bed technology at 22 MWe
utilizing high sulphur
coal and Oil Shale as
sorbent.
To evaluate N0X and SO2
suppression in bubbling
and circulating fluidized-
bed with Canadian coals and
sorbents.
$0.3 mi 11 i on
Sumnersi de -
$18.0 mi 1 lion
Point Tupper -
$8.0 mi 11i on
$36.0 mi 11ion
$2 .0 mi 11 ion
Total Federal Funding:
$83.4 mi 111 on
STATUS OF PROJECTS
Coal beneficiation, coal/water mixtures and the Point Tupper materials
testing project being peripheral to the theme of this symposium will not
be reported in this paper.
Canadian Forces Base Gagetown Demonstration
As part of an on-going effort to expand the use of high-sulphur Maritime
coal for heat and electricity with minimal environmental impact, CANMET
has initiated a project to demonstrate limestone injection, multi-stage
burner technology (LIMB) for substantially reducing acid gas enissions
from pulverized-coal-fired boilers. The project, which is being carried
out with the active participation of the Department of National Defence
and the analytical support of Environment Canada, involves the
retrofitting of an existing 20 MW hot water generator at CFB Gagetown
with two "staged-mi xi ng" burners. These are designed to inhibit the
formation of nitrogen oxides by flame modification and to suppress the
emissions of sulphur oxides by limestone injection with no reduction in
boi 1 er ef fi ci ency.
Each burner is rated at 10 MW-th anc' 1S designed to achieve a 50%
reduction in N0X and SO2 emissions while burning a 3% sulphur
eastern Canadian coal. The installation of equipment on Unit No.2 at
CFB Gagetown was completed in April 1984 and burner shakedown trials
under the supervision of the contractor (Volcano Inc.) are in progress.
4-3
-------
Final acceptance tests are scheduled for December 1984 when weather
conditions i/ill permit full load operation of the hot water generator.
The Gagetown project, which is the first full-scale denonstration of
this new burner concept to North America will provide potential users as
well as research and regulatory agencies with advance information on the
operational benefits of this new technology.
TransAlta Utilities Low N0X/S02 Pilot-Scale Development Program
The program was initiated on December 16, 1982 when Rockwell and the
first of several utilities signed Participation Agreements. Current
participants are Houston Lighting and Power Company, Niagara Mohawk
Power Corporation, Southern California Edison Company, TransAlta
Utilities and Wisconsin Public Service Corporation. The Canadian
Electrical Association participates through TransAlta.
The program, developed by Rockwell International, utilizes advanced
combustion concepts for both retrofit and new installations in an
attempt to reduce SO2 emissions by up to 70% and to suppress N0X
emissions to less than 100 ppm.
Pilot plant testing at both the 4.7 MW and 7.3 MW scale have confirmed
the design concept for the Rockwell burner and further work to verify
specific technology features is in progress. Preliminary design work
for the commercial scale 30 MWth demonstration burner has been initiated.
The demonstration program will be a three year, two-phased program t/hich
is planned for Unit No. 1 (66 MWe) at TransAlta Utilities
Corporation's wabamun Generating Plant.
In the first phase, the 30 MWth burner complete vri th slag separator will
be designed, fabricated, installed and tested over a 6 month period.
Tv/o additional burners will be subsequently installed, without slag
separators, and intensive testing of the composite Rockwell Low
N0x/S02 combustion system in 11 follow for a period of about one year.
A phased approach is used to permit the demonstration testing and
shakedown so that any required modifications can be incorporated into
the second and third units. Total cost of the project is approximately
$10 mi 111 on Canadi an.
International Energy Agency (IEA) Project for Control
of Nitrogen Oxides Emission During Coal Combustion
CANMET, on behalf of Canada, participates in a major project sponsored
by the International Energy Agency. The project, co-funded by Canada,
Denmark and Sweden with US EPA guidance and support, involves the
validation and optimization of advanced burner concepts under a
three-stage agreement. The work is being performed by the Energy and
Environmental Research Corporation at Santa Ana, California.
Stage I, completed in March 1982, consisted of bench-scale furnace
trials on 45 coals including 9 from Canada, to elucidate the mechanisms
of N0X formation from fuel nitrogen under pre-mixed and staged
combustion conditions. It was determined that much of the fuel nitrogen
4-4
-------
in the volatile natter can be transformed to N2 instead of NO. The
conversion of fuel nitrogen to NO was 30% to 40" in conventional flames
but only 7% to 14% in staged flames. Nitrogen retained in the char
showed about 20% conversion to NO and appeared to be relatively
independent of local oxygen concentration. Reductions in sulphur
emissions of 50% by sorbent injection into tertiary combustion air were
achievable wi th Ca/S ratios of 2/1.
Stage II, now in progress, consists of combustion trials in three
progressively larger furnaces (50, 90 and 100 x 10^ Btu's) to generate
data for extrapolation to full-scale burner designs. Four coals, two
from Canada, are being evaluated for N0X production and two of the
four coals, including one from Canada, have been evaluated for
simultaneous reductions of SO2 and N0X using limestone sorbents.
You will be hearing about this work later on from Energy and
Environmental Research Corporation, so I will not go into further
detai1s.
Chatham Coal/Oil Shale Demonstration
The New Brunswick Electric Power Commission has dedicated its Chatham
generating station consisting of one 12 MWe B£W boiler and one
22 MWp CE boiler to research in advanced burner concepts and
circulating fluidized bed technologies until March 1988 under an
agreement with the Coal Division of Energy, Mines and Resources.
A pilot-scale combustion rig with a staged combustion burner
(1x10^ Btu/hr.) modelled on the IEA work done through EPA at Energy
and Environmental Research, California is being erected at the Chatham
station to study the potential of achieving acceptable acid gas
emissions by co-firing coal with local oil shale. A comparison will be
done with firing coal and limestone. A successful demonstration with
011 shale as a sorbent could lead to further demonstration on the
12 MWe boiler at Chatham or the 20 MW^ boiler at CFB Gagetown.
A 22 MUe circulating fluidized boiler to be manufactured by CE
Canada-Lurgi will be installed at Chatham to tie into the existing
turbine. It is expected to be commissioned in the fall of 1986.
Initially it will fire high sulphur (8%) New Brunswick coal and
limestone to establish a baseline for comparison with the same coal and
oil shale. Tne oil shale has approximately 2100 BTU's/lb. and an ash
analysis of CaO 14.9% and MgO 5.5% present in the original mineral
matter as a carbonate.
Canadian Forces Base Sunnerside AFB Demonstration
EMR in collaboration with Defense Canada is demonstrating atmospheric
bubbling fluidized bed combustion at CFB Summerside, PEI, heating
plant. Construction of the equipment, which consists of two
Foster-Wheeler boilers rated at 18 tph of steam each, together with all
anci 11 iaries, was completed in 1982. In the following 1-1/2 heating
season commissioning trials have been underway, some modifications have
been made, and the boilers have each accumulated well over 1,000 hours
of operating time.
4-5
-------
The boilers have been tested with three different Cape Breton coals
ranging from 4 to 6% sulphur, and have net guarantees of capacity,
efficiency and emission control vri th all of then.
Further denonstration tests during the 1984/85 and 1985/86 heating
seasons will include co-firing of wood chips and coal, and evaluation of
load response characteristics. It is also hoped to evaluate the effects
of limestone size consist on sulphur capture, and to conduct combustion
tests vri th a high ash fuel, possibly coal washery rejects. Under the
contract the wastes are being characterized and specific uses in PEI are
oei ng i denti fi ed.
CANADIAN ELECTRICAL ASSOCIATION (CEA) SUPPORTED ACTIVITIES
Ontario Hydro, a major utility in North America, has for several years
been examining methods to reduce N0X emission levels from its
Nanticoke Generating Station. This generating station comprises 8x500
MWe natural circulating boilers of Babcock and Wilcox manufacture,
each with 40 opposed wall firing conventional circular coal burners.
In 1980, with the financial assistance of the Canadian Electrical
Association Ontario Hydro employed the services of B&W (Canada) to
undertake detailed design, manufacture, and installation of
modifications to eight burners (one row). Subsequent testing proved
that these modified burners performed satisfactorily and reduced H0X
emissions by 20% below the unmodified burner emissions.
Based on this moderate success, Ontario Hydro in 1983 converted all of
number 5 and 6 units at Nanticoke. This resulted in a high carbon
carryover in the ash. As a result of the high carbon, Ontario Hydro had
B&W do some modelling at their Alliance laboratories looking at possibly
40% N0X reduction with less than 6" carbon carryover. Modified burner
changes based on the modelling is planned for October 1984.
Ontario Hydro is also considering converting some burners on their
corner-fired units at Lakeview to the C-E Low N0X Concentric Firing
System and experimenting with finely powdered limestone injection.
TransAlta Utilities, Calgary, Alberta as a member of the consortium
interested in developing the Rockwell burner received the support of the
Canadian Electrical Association Generation R&D Committee. To enable
them to have a more prominent position in the development process, the
Alberta/Canada Energy Resources Research Fund is supporting, pending
successful pilot scale trials at Rockwell's California facility, the
demonstration of a 30 MWth burner at TransAlta's Uabamun generating
stati on.
The CEA Generation R&D Committee also supported the installation of the
Combustion Engineering Low Nox Concentric Firing System at
Saskatchewan Power Corporation's Boundary Dam Generating Station 300 MW
lignite fired Unit #6. Dave Winship, CE Canada, will be giving a report
on the results later on in the symposium.
4-6
-------
CONCLUSIONS
So you see, Canada is not only concerned about acid gas emissions, but
is doing something about it. Canadian utility's contribution of SO2
to acid rain is only 5.3% of the total for North American utilities, but
our efforts to control this environr.iental problem is a major priority.
Our RD&D activities address both short-term regional concerns as well as
the longer term issues associated wi th the expanded use of coal.
4-7
-------
THE UTILITY PERSPECTIVE ON DRV S02 CONTROL TECHNOLOGIES
George P. Green
Public Service Co. of Colorado
550 15th Street
Denver, Colorado 80202
ABSTRACT
Tnis paper summarizes current understanding and provides comparative descriptions of
two promising techniques for dry SO2 control in utility applications: flue gas
sodium sorbent injection, and in-furnace calcium sorbent injection. In both cases,
dry sorbents are injected in powdered form and react chemically with SO2 t0 form a
dry particulate waste which potentially can be collected in eitner a baghouse or
electrostatic precipitator (ESP). Two electric utilities, Public Service Co. of
Colorado and tie Colorado Springs Dept. of Public Utilities, have announced firm
plans to employ flue gas sodium sorbent technology. In-furnace calcium soroent
injection is still under active development and no long term utility commitment to
this tecnnology has yet been made in the United States. Both control techniques
snow potential as reliable, efficient, and economic SO2 control options, giving
utilities greater flexibility in meeting their environmental control
responsloilities.
INTRODUCTION
Tnis paper summarizes current understanding and provides comparative descriptions of
two promising techniques for dry S02 control in utility applications: flue gas
sodium sorbent injection, and in-furnace calcium sorbent injection. In both cases,
dry sorbents are injected in powdered form and react chemically with SO2 to form a
dry particulate waste. Wit.n sodium soroent injection, the waste sodium material is
collected along with particulate fly ash in a fabric filter dust collector
(baghouse). dith calcium sorbent injection, the calcium waste material is collected
in either a baghouse or electrostatic precipitator (ESP). Figure 1 is a schematic
of a coal-fired power plant snowing, for both types of sorbents, their injection,
SO2 capture, and solid waste removal sites. This paper discusses the advantages and
disadvantages of these two control technologies and tne implications and prospects
for utility application.
Sulfur-dioxide emission control is presently required for all new coal-fired power
plants. The EPA New Source Performance Standards stipulate 10% SO2 removal for
plants burning low-sulfur coal, and 90% removal for those burning high-sulfur coal.
In Doth cases, emissions inay not exceed 1.2 lb per million 3tu of heat input to the
boiler.
Conventional wet scrubbing is an effective option in meeting these standards.
However, wet scrubbers are costly -- accounting for as much as 25% of the capital
s l
-------
ana operating costs of a new 1Q0Q MW plant -- and are often problematic to operate,
requiring complex hardware and the use of substantial quantities of water. Costly
liquid waste treatment systems are required and result in additional complexities
and expense.
Spray drying, the second generation in SO2 control technology, may be a lower,
capital cost alternative to wet scrubbing. While this technology has eliminated the
need for liquid waste treatment systems, it is still necessary to prepare a wet
slurry for injection.
Dry sorbent injection control techniques simplify the sulfur dioxide removal process
by eliminating the wet slurry, liquid waste treatment system, and extensive
additional hardware, thus resulting in lower comparative capital costs. Also, tnese
systems use equipment already familiar to power plant operators. Operating costs
can be significant, depending largely on reagent cost, but overall system cost in
virtually any imaginable scenario will be considerably less than that for wet
scrubbers. Tnis is tne case because with dry control techniques sorbent cost and
utilization efficiency have the greatest impact on total system costs -- not system
nardware, as with wet scrubbers and spray dryers. Further, dry sorbent reagent
costs are expected to decrease as suppliers expand their efforts to make new, more
efficient products available at lower prices. Other advantages of dry sorbent
injection systems include:
• System simplicity contributes to improved reliability.
• Power costs are lower.
• Scaling and corrosion are minimal.
• Flue gas reheating is not necessary.
• •• Systems can be. retrofitted to boilers with baghouses and ESPs.
• There is combined particulate and 30? collection.
Public Service Co. of Colorado (PSCC) has recently committed to all-dry flue gas
sodium sorbent injection on a new 500 MW coal-fired boiler, Pawnee Unit 2. Although
no firm date has been announced for construction of Pawnee 2, for engineering and
planning purposes it is scheduled to begin service in 1990 burning western, low-
sulfur subbituminous coal. PSCC's confidence in flue gas sodium sorbent injection
stems from extensive testing, both bench-scale and in the field, most specifically
in two years of testing at PSCC's 22 MW Cameo Station Unit 1. PSCC and tne Colorado
Springs Dept. of Public Utilities are currently the only electric utilities to
announce Firm plans to employ flue gas sodium sorbent injection.
In-furnace calcium sorbent injection is still under active development and no long
tern utility commitment to this technology has yet occurred in the United States.
Laboratory test results and recent large-scale exploratory tests indicate the high
potential of this process as a low-cost option for S02 control. Continued
development and large-scale demonstrations planned over the next few years are
expected to confirm the SO2 removal potential and resolve remaining cost and plant
impact issues.
Figures 2 and 3 illustrate the variety of options available to utilities in the
selection of particulate and SO-? control technologies, and reflect the complexity of
the decision-making process in this area.
5-2
-------
FLUE GAS SODIJM SORSENT INJECTION
The three sodium sorbents which have received most utility arid supplier attention to
date are nahcolite (naturally occurring sodium bicarbonate, NaHC03), trona
(naturally occurring NaHCOj • Na?C03 • 2 H^O), and sodium sesquicarbonate (a trona
analogue, NaHC03 " Na2C03 ' 2 ^O).
In the dry-injection process, the sodium sorbent is pulverized and fed into the flue
gas stream (nominally at a temperature of 300°F) ahead of a baghouse and downstream
of the air heater. In the ductwork, the sodium bicarbonate in the sorbent particles
decomposes to sodium carbonate CNa2'C03) in a "popcorn" fashion, forming an open,
porous nicrostructure exposing more particle surface area. As shown in Figure 4,
tne sorbent reacts with the SO? in tne 9as> forming sodium sulfate, and
subsequently is collected along with the fly ash as part of the dustcake in the
baghouse. Contacting of sorbent and SO? on the dustcake results in further SO2
removal. Typically, 20-30% of the 502 collection occurs in the ductwork, and the
remaining 70-83% in the baghouse.
Historical Perspective and Process Development
It nas been known for some time that sodium reagents chemically react with SO2 to
form sodium sulfate and sodium sulfite. Until recently, however, this reaction was
not seriously considered for SO2 control in coal-fired power plants because it was
thought there was no practical way to bring the two chemicals into contact long
enough to react, and then to collect the by-products. However, the emergence of the
baghouse in the utility industry resolved this issue by providing both efficient
particulate collection and allowing for an extended period of S02~s°rbent contact.
Identification of an appropriate sodium-based reagent has also been an issue in
development of this technology. Initial reagent testing was conducted with
nahcolite, the preferred alternative because of its high sodium bicarbonate content
ana higher SO? collection efficiencies — in excess of 70% removal. Over 30 billion
tons of nahcolite 'nave been identified in tne western United States. However, the
resource is locked in oil snale formations, and tne recent slackening of interest in
oil shale has made its future availability uncertain. While long-range prospects
for mining nahcolite are unclear, efforts are under way to produce nahcolite through
solution mining techniques.
In contrast, trona is commercially available in large quantities, and recent tests
have shown that it too can achieve 70% SO2 removal. In excess of 85 billion tons of
trona are estimated to exist in the western United States, and tne material is now
being mined as a source of soda ash (.N^2C03J f°r glass-making by a number of
companies. Trona availability for SO2 collection has been enhanced Dy the recent
downturn in its demand Dy glass manufacturers due to the decline in new building
construction and the use of less expensive glassware substitutes. In addition,
trona suppliers are now undertaking research and development programs of their own
to improve the mineral's SO2 collection characteristics for utility application.
These investigations are focusing on both sodium sesquicarDonate and other
proprietary compounds.
The first utility testing of flue gas sodium sorbent injection used nahcolite as the
reagent and was conducted in the mid-1960s by Southern California Edison Co. at its
Alamitos station on an oil-fired boiler equipped with a baghouse. Although SO2
removal was not quantified, results were said to be encouraging( _L>?J . Subsequent
tests in the late lybOs and early 1970s oy Wheelabrator-Frye in conjunction with the
Public Service Co. of Indiana and, separately, with Colorado-Ute Electric
Association, by the Air Preheater Co. with Public Service Electric and Gas Co. of
5-3
-------
New Jersey (under contract with the National Air Pollution Control Administration),
and by Superior Oil Co., with PSCC at its Cherokee Station, showed that a variety of
sodium based reagents — including nahcolite, sodium caroonate, and commercial and
predecQ;nposed sodium Dicarbonate -- could acnieve SOo removal rates in the range of
7Q-90^1,l>->—J. It is important to note, however, that sodium caroonate, or soda
ash, has proven to be largely ineffective as an SO? adsorbent.
In tne late 1970s, the Electric Power Research Institute (EPRI) sponsored a bench-
scale experimental investigation of sodium sorbent injection specifically to
quantify SO2 removal with low-sulfur coals^i). This testing, conducted by
-------
PSCC has performed an economic analysis for Pawnee Unit 2, comparing costs for spray
dryer and dry injection systems. Tne study assumed startup of Pawnee in 1990, a 30-
year plant life, and the consumption of 70,000 tons/year of sorbent. Costs included
the particulate control system for both spray dryer and dry-injection systems, as
well as costs for the balance-of-plant systems required with either SOo removal
technology. and were found to be unquestionably size-, site-, and schedule-
specific^/. Costs we^e calculated assuming the use of trona as the dry sorbent;
costs for nahcolite would be expected to be less due to a lower reagent requirement.
The total capital investment for the dry-sorbent system was $57 million,
substantially less than the $120 million estimated capital investment for the spray
dryer system at this site. Tne dry-sorbent system is expected to nave higher annual
O&M costs, however, mainly due to the larger quantities of reagent needed to
accomplish the same SO2 removal. In spite of this projected higher cost for
reagent, the dry injection system has an $8 million lower annual evaluated cost,
which includes the annualized capital cost plus tne annualized O&M cost.
Issues In Utility Application
Although flue gas injection of trona has proven to be the economic choice for SO2
control at Pawnee, a number of issues are appropriate for more detailed discussion
and analysis by other utilities considering the technology.
Waste Pisposal. Sodium salts produced by the reaction of SO? with sodium-based dry
sorbents are very soluble in water. Althojgh the weight fraction of sodium in the
waste product -- which also includes fly ash -- is less than 10%, precautions must
be taken to avoid leaching of sodium into ground or surface waters. Clay-lined
holding ponds preclude leaching and can be utilized to address this concern.
Alternatively, a technique to fix sodium ions in the residue, rendering it
insoluble, is currently under development. Sjch a process could potentially permit
disposal of fly ash, spent sorbent, and bottom ash together in a conventional
landfill. In a plant originally constructed with an ESP, a baghouse and dry
injection system could be added for S02 control. Under this approach, the ESP
collects the major portion of the fly ash while the bagnouse collects all of the
spent sorbent and the remainder of the fly ash. The spent sodium/fly ash mixture is
then disposed of separately from the fly ash, thus allowing fly ash to be utilized
as a concrete additive and return of the spent sodium to its origin.
'Jse With High-Sulfur Coal. To date, dry sorbent injection has been demonstrated
only on coals with sulfur contents below 0.8%. However, recent work by sorbent
suppliers indicates that the upper limit may reach as high as 2-3%. Even if
sufficiently nigh SO2 collection efficiencies could be obtained on nigh-sulfur coals
-- standards as high as 90% may be required -- this would necessitate the use of
very large amounts of sodium reagent and potentially increase reagent costs to the
point where the process would no longer be economic.
Use With ESPs. Most research to date has been conducted using baghouses for
particulate collection. Preheating sorbents prior to injection in system ductwork
is a concept now being explored to facilitate the use of this technique with ESPs.
The objective is to increase reactivity of the sorbent such that SO2 collection
efficiencies in the ductwork alone are acceptable. In-duct SO? collection of 20-30%
has already been demonstrated, and 50% appears possible.
-------
Policy Issues. The question of whether flue qas sodium sorbent injection can and
wi 1 1 qua 11fy as best available control technology (8ACT) as defined by the EPA is
anotner potentially important issue in considering the technology. Preliminary
discussion with regulatory agencies, combined with the obvious simplicity and low
cost of the process, nave resulted in favorable reaction.
IN-FJRNACE CALCIUM S0R3ENT INJECTION
witn in-furnace sorbent injection, a pulverized calcium-based material such as
limestone (CaC03) or calcium hydroxide (nydrated lime Ca(.OH)2) injected directly
into the furnace cavity of a coal-fired boiler. As shown in Figure 6, tne sorbent
rapidly decomposes (a limestone sorbent, for example, releasing CO?; hydrated lime
releasing H?0), to form a porous inicrostructure with greater exposed surface area.
Tne resulting lime particles are highly reactive and chemically combine in
suspension with SO? to form solid calcium sulfate (CaSO*). This calcium sulfate,
along with any unrsacted lime, is tnen collected with fly ash in a baghouse or ESP.
In essence, this process attempts to apply to pulverized coal-fired ooilers an SO?
control technique similar in overall chemistry to that used in fluidized bed boilers
-- albeit under distinctly different temperature, residence time, and combustion
conditions.
Historical Perspective and Process Development
The concept of furnace sorbent injection originated in the early lybOs. It was felt
at the time that direct injection of limestone followed, for example, oy wet
particulate scrubbing was the least complicated and most economic procedure for
meeting anticipated 50? and particulate removal requi raiments, However, trial tests
in laboratory furnaces and full-scale utility boilers in the United States, Europe,
and Japan generally failed to demonstrate sufficient SO? removal at practical
sorbent-to-sulfur ratios. SO? removal during tests at utility boilers typically
ranged from 15-40%, well below the target values of 80-90%. Removal efficiency was
also found to be highly dependent on boiler design and operation, as well as on the
type of sorbent and injection system used. Potential adverse effects on boiler
performance, principally increased slagging and fouling of boiler heat transfer
surfaces and degraded ESP performance, were also noted. Given these concerns and
the lower-than-anticipated SO? removals, testing was essentially abandoned by the
early 1970sCJL9).
Interest in the technology was rekindled in the mid-1970s with research performed
in West Germany. This work arose out of a need to develop practical methods for
incremental SO? removal in *lest German plants burning lignite coal. The experiments
showed better results than any previous tests and triggered a multi-year development
program in that country, including testing of bituminous coals. In the United
States, the EPA announcement of the LIMB concept in the late 1970s was a major
factor in accelerating interest in this country.
Another factor now contributing to the resurgence of interest in in-furnace dry
sorbent injection is the growing incentive for developing low-cost incremental SO?
controls applicable to both existing and new power plants. The potential also
exists for combining in-furnace sorbent injection with other SO? control
technologies. For instance, use in conjunction with coal cleaning or coal blending
may provide flexibility in achieving SO? emission compliance or allow the purchase
of less expensive, higher-sulfur coals for existing units. Also, integration of the
process with other flue gas treatment systems may provide an overall SO^j control
capaoility adequate to meet higher requirements for new plants.
5-6
-------
Recent experimental work in the United States and Europe indicates the possibility
of achieving high SO^ removal efficiencies at practical sorbent injection rates if
sorbent characteristics and injection conditions are properly controlled.
Figure 7 shows percentage SO? removals for pressure hydrated lime, calcium
hydroxide, and limestone at various Ca/S molar ratios based on pilot laboratory
tests jointly sponsored by EPRI and Southern Company Services. These tests and
supporting bench-scale experiments, engineering studies, and information exchanges
will set the groundwork for utility tests of in-furnace sorbent injection in the 40
to 100 HW range. Demonstrations of this scale are planned as a necessary
intermediate step in the development process prior to application at full-size
uti1i ty boilers .
Parallel R&D programs are under way at EPA, DOE, and several other utility,
industrial, and manufacturing organizations, as well as in other countries,
including Canada, West Germany, France, The Netherlands, Austria, and Japan. These
various programs span a wide range of laboratory and field test conditions and
involve a variety of sorbents, coals, and injection system designs. Sorbent
suppliers are also involved. As a result, within the next two years a tremendous
amount of new information on process cnemistry, S0£ removal performance, and system
design will be made available.
Economi cs
Since this process uses existing furnace cavities and particulate control devices
for S0£ reactions and removal, capital costs are expected to be much lower than
those associated with wet scrubbers. However, the developmental status of the
technology and uncertainties about final process design and plant impact issues
makes it difficult to accurately estimate costs. Preliminary estimates range
between approximately S30 and $80/kW for retrofits. Costs for new unit applications
may be at the lower end of this range.
Operating costs for these systems depend strongly on sorbent utilization and coal
sulfur content. Calcium-based materials are readily available at low cost.
However, current utilization efficiency, which to date has been 20-30% for limestone
and up to 40% for lime, is less than for sodium sorbents, thereby increasing costs
for sorbent supply and solid waste handling and disposal. Current efforts to
improve calcium utilization, if effective, could reduce in-furnace injection
operating costs appreciaoly.
Issues in Uti1ity Application
A number of issues remain to be resolved before in-furnace calcium sorbent injection
can be considered a practical control option for full-scale utility application.
Most of these issues are of a technical nature and, althougn none appear
insurmountable, the costs involved in solving them are as yet unknown.
Impact on ESPs. In-furnace calcium sorbent injection can double or triple solids
loading to the boiler, depending on coal sulfur content, sorbent type, and Ca/S
ratio. Greater solids loading, in turn, increases ESP particle inlet loading, while
the calcium compounds increase resistivity in the ESP. These factors together can
degrade ESP performance and can result in nigner particulate emissions, especially
for marginally performing units. Efforts are now underway by vendors and others to
quantify the problem and verify retrofit upgrading techniques, sucn as chemical and
tnermal flue gas conditioning. There is no evidence at this time that in-furnace
sorbent injection adversely affects fabric filter bag'nouse operation.
5-7
-------
Ash Handling and Disposal. In-furnace calcium sorbent injection produces solid
wastes greatly different from conventional ash. Unreacted sorbent, fly ash, and
reaction by-products have different disposal characteristics and requirements which
must be better understood. Specifically, the possible need for special handling,
transport, pretreatment, and disposal of the by-products must be clarified before
the technology can be widely employed.
Boiler Slagging and Fouling. Increased solids loading to the boiler can change the
chemical and physical properties of the ash, thereby raising concerns about
increased slagging and fouling on furnace heat transfer surfaces. Research to date
indicates that these deposits likely can be managed with conventional sootblowing
equipment. However, additional sootblowers and/or more frequent use may be
required^!!). Field test activities planned and currently in progress will provide
further insight into these issues.
System Design. While in-furnace calcium sorbent injection appears to be a feasible
means to control S0£ emissions, much work remains tobe done to optimize the
process. This includes determination of optimum injection system design, the most
efficient type of sorbent(s), and of a means for enhancing sorbent utilization.
Recent studies have identified the upper furnace cavity as the best site for calcium
sorbent injection^U'lz). Injection in this post-flame area effectively decouples
tne process from the combustion system, thereby facilitating retrofit and
significantly expanding the number of boilers in which the technology could be
appli ed.
With regard to sorbent-SC^ captjre efficiency and enhancement, most R&D to date has
focjsed on evaluating commercial dolomitic and high calcium limestone and lime
sorbent materials. The efficiency or reactivity of tnese compounds has been
evaluated under a variety of test conditions and furnace designs. Results suggest
that the size and surface area of the calcined sorbent play a major role in SO2
capture efficiency. There are also indications that chemical constituents and
physical micro-structure of the sorbent are important. Some of these factors are
strongly influenced by the thermal history of the sorbent in the furnace. Future
work will focus on understanding the •"ole of these various factors, how tney are
influenced by the thermal or chemical environments that the sorbent is exposed to
and, ultimately, how tnese environments can be created or manipulated to enhance
sorbent characteristics for optimum SO2 capture.
Policy Issues. The question of whether in-furnace calcium sorbent injection can and
will qualify for BACT status is another potentially important issue in considering
the technology.
CONCLUSION
Evolution of SO^ control technologies for utilities from the wet scrubber to the
spray dryer and dry injection processes shows a progressive development towards '
simpler, more economic systems. Planned installation of a flue gas sodium injection
system at PSCC's Pawnee Station will represent a major step in development of that
technology. Additionally, demonstrated increased interest on tne part of suppliers
in providing new, improved sodium sorbents to the utility market is encouraging. As
5-8
-------
regards i'i-fjrnace calcium injection, laboratory results to date are optimistic for
development of a commercially viable system, and field demonstrations scheduled to
be underway in tne next few years will represent a major milestone towards
conmerci ali zati on.
Emergence of these technologies as reliable, efficient, and economic SO? control
options gives utilities greater flexibility in meeting their environmental control
responsibilities, and presents a real opportunity for reducing trie cost of
electricity to their customers.
acknowledgments
r°o W f t0 thdnk the Coal Corruption Systems Division of the
t.1 e^tnc Power Researcn Institute for its assistance in the preparation of
A!S0' ^duard0 E- Gon"les, professional Engineer, ana Robert D.
Gat^s, Supervisor, ,*1ecnamcal engineering, of the Puolic Service Co of
Lo.oraoo provided important background and data on ?SCC's evaluation of flue
gas sodium sorbent injection for Pawnee Unit 2
5-9
-------
REFtRENC ES
1. F.A. Bagwell, L.F. Cox, E.A. Pirsh, "Design and Operating Experience: A
Filterhouse Installed on an 0i1-Fired Boiler," JAPCA 19:149 (1969).
2. "Evaluation of Dry Alkalis for Removing Sulfur Dioxide from Boiler Flue
GasesFP-207 , EPRI, Palo Alto, CA, October, 1975.
3. "Edwardsport Test Report," Wneelabrator-Frye Inc., 1969.
4. H. Liu, R. Chafee, "Evaluation of Fabric Filter as Chemical Contactor for
Control of Sulfur Dioxide from Flue Gas," NAPCA Contract PH22-68-51, Air
Preheater Company, Inc., Wellsville, NY, December 1969.
5. '"r'nase I Testwor*, High-Temperature Injection of Nahcolite for SO2 Control,"
Superior Oil Company, Denver, CO, November 1974.
6. L.J. Muzio, J.K. Arand, "bench-Scale Study of t'ne Dry Removal of SO? with
Nahcolite and Trona," CS-1744, EPRI, Palo Alto, CA, March 1981.
7. L.J. Muzio, T.W. Sonnichsen, "Demonstration of SO? Removal on a 22 MW Coal-Fired
Utility Soiler by Dry Injection of Nahcolite," R?Ib82-2, EPRI, Palo Alto, CA,
April 1982.
3. R.'aI. Scheck, D.J. Naulty, A.E.E. Gallagher, R.P. Grimm, D.A. McDowell, R.J.
Keeth, J.E.Miranda, "Economic Evaluation of Dry Injection FGD Technology,"
RP1682-1, EPRI, Palo Alto, CA, 1984.
9. G.P. Green, R.C. Carr, R.G. Hooper, "Technical and Economical Evaluation of Dry-
Sorbent Injection for SO? Control Using Sodium Compounds," Public Service Co. of
Colorado, Denver, CO, 1984.
ID. A. Kokkinos, et al, "Feasibility of Furnace Injection of Limestone for 50^
Control," in Proceedings of the 1982 Joint Symposium on Stationary Combustion
N0X Control, Dallas, TX, November 1-4, 1983. EPRI CS-3182, Vol. 1, July 1983.
11. EPRI Research Project RP1836-1 in progress.
12. Y. Takahashi, et al, "Evaluation of Tangential Fired Low-N0x Burner" and A.
Kokkinos, et al, "Feasibility Study of a Low-NO^ Retrofittable Firing System
witn U.S. Coals," in Proceedings of t'ne 1982 Joint Symposium on Stationary
Combustion N0X Control, Dallas, TX, November 1-4, 1983. EPRI CS-3182, Vol. 1,
July 1933.
5-10
-------
BOILER
DP* CAIC'UM
SOR8ENT
INJECTION
coal and
*E>TED A R
c#so.
SODIUM
REACT (Oft
SORBENT
INJECTION
PARTICULATE
COLLECTION
HEA"ER
NtrSO, and
CtSO
COLLECTION
Nt,SO
REACTION
combustion
ZONE
CLEANED Flue gas
STACK
Figure 1. Schematic of a coal-fired power plant,
sorbent injection and removal sites.
showing sodium and calcium
^Exiallng
I
' Ca
Add
Wei
Irvfurnaca
FGD
Upgrade
•SCA '
FGC
Upgrade
Existing
Na Inieci
Ca Na
In furnace
ln-furnac«
Ca'Na
.
. Na ' :
Ca
to
Cairyover
FF
toFF
0 PARTICULATE COLLECTION
~ DRY SO, COLLECTION
~ WET SO, COLLECTION
Figure 2. Comercial and developmental retrofit options for particulate and
SO2 control technologies.
5-11
-------
WET FGD
WITH
ESP or FF
SPRAY
DRYER
WITH
FF
NaINJECT
TO
FF
Ca/Na
TO
FF
O DRY SO, COLLECTION
~ WET SOj COLLECTION
Figure 3. Commercial and developmental particulate and S02 control options
for utilities constructing new units.
Duiluk* Ftbrtc I
Captured SO* on sotteni
Reactitt sortMnt
ut GM
Cfeangu
PinicuLii# msttw
5<*b«fit Injection
Figure 4. Flue gas sodium sorbent injection -- S02 reaction and capture
mechanisms.
5-12
-------
7k
nahcolite
NAHCOLITE (1960*
O NAHCOLITE (1901-82)
GREENO NAHCOLITE
WYOMING TRONA
A OWENS LAKE TRONA
O SESQUICARBONATE
<3> SODA ASH
uj GO
TRONA
SODA ASH
0.5 10 15
NORMALIZED STOICHIOMETRIC RATIO
Figure 5.
S02 removal as a function of normalized stoichiometric ratio
(PiCC's 22 MW Cameo Station).
Heating
0 CaCO,
~ CaSO,
~ CaO
Calcination
Sclntertng
Sulfation
SO,
» \f
r<
so,
Pore Plugging (?)
Figure 6. In-furnace calcium sorbent injection SO2 capture mechanism.
5-13
-------
CALCIUM
HYDROXIDE
PRESSURE
HYDRATED
LIME /
DOLOMITE
LIMESTONE
1.0 * 10" BTU'HR.
3.5% S BITUMINOUS COAL
2050'F INJECTION
CALCIUM/SULFUR MOLAR RATIO
Figure 7. SO2 removal as a function of calcium/sulfur molar ratio.
5-14
-------
SESSION II (PART 1): FUNDAMENTAL RESEARCH
Chairman, Kerry Bowers, Southern Company Services
6-i
-------
EPA EXPERIMENTAL STUDIES OF THE MECHANISMS
OF SULFUR CAPTURE BY LIMESTONE
R. H. Borgwardt
U.S. Environmental Protection Agency
Office of Research and Development
Industrial Environmental Research Laboratory
Research Triangle Park, NC 27711
K. R. Bruce and J. Blake
Northrop Services, Inc.
Research Triangle Park, NC 27709
ABSTRACT
Reaction kinetics of limestone particles were measured under conditions that
eliminate pore-diffusion and interparticle-diffusion effects. Included in these
laboratory studies were: the reaction of H2S and sulfur with CaC03, the calcina-
tion of CaC03 to CaO, the reaction of CaO with H2S and COS, and the reaction of
CaO with S02- The results show that nascent lime formed immediately after CaC03
decomposition has a specific surface area of about 80 m^/g. in all cases, the
reactivity of the CaO increased with the square of the B.E.T. surface area. The
reactivity is markedly affected by the presence of foreign metal oxides or salts
on the CaO surface; carbonates and sulfates of the alkali metals are effective
additives for promoting CaO reactivity under laboratory conditions.
INTRODUCTION
The objectives and rationale of the IERL-RTP in-house experimental studies were
previously outlined at the 1982 Dallas Symposium.^' Because sulfur capture is
theoretically feasible in the reducing zone of the limestone-injection multi-
stage burner (LIMB) and CaS formation is free of limitations on maximum conver-
sion, our efforts began with reactions involving H2S, using uncalcined limestone
particles. Those results have been reported in detail elsewhere.^) The second
phase of work concentrated on the calcination kinetics of small limestone parti-
cles (1 to 90 urn) and measurements of B.E.T. surface area produced by "flash cal-
cination" in the absence of CO2. The results of that study are currently in pub-
lication. f3' The third area of investigation was reactions of CaO to form CaS
under reducing conditions.*^ That work, like our subsequent studies involving
CaO, was focused on the effect of specific surface area as the main experimental
parameter. The reactor design developed for these studies 5) js unique by the
fact that particles as small as 1 pm can be examined, thus eliminating pore dif-
fusion resistance--only under such circumstances is the surface area effect fully
revealed. Elimination of diffusion resistances has also made possible, perhaps
for the first time, unobscured measurements of the effect of gas phase concentra-
6-1
-------
tion on reaction rate. These new experimental methods were also applied to the
study of the reaction of CaO with SO2 under oxidizing conditions which has been
recently completed.(6)
Measurements of CaO sintering rate and the effect of reaction promoters on SO2
capture are now in progress.
SUMMARY OF SIGNIFICANT RESULTS
Calci nation
Earlier studies of the calcination kinetics of limestone particles were limited
to 90-um particle size.'?) Our objective was to examine smaller particles more
appropriate to LIMB and extend the range down to 1 um. This was accomplished
with two different reactor types having very dissimilar heat- and mass-transfer
characteristics. The data from both systems were correlated over a temperature
range of 516 to 1000°C by a kinetic model based on the B.E.T. surface area of
the undecomposed CaC03. The kinetic parameters were the same for the two di-
verse types of limestones evaluated: 49 kcal/mol activation energy and a rate
constant of 2.5 x 10"® mol/cm^-sec at 670°C. The rate data were correlatable on
the basis of reaction kinetics alone over a range of nearly 5 orders of magni-
tude.
CaO Specific Surface Area
Ishihara 9) reported surface areas averaging 22 m2/g for "instantaneous" cal-
cination of limestone ranginq from 3.4- to 49-ym particle size in an unspecified
atmosphere. Coutant et al. *7', in a contemporaneous study carried out for the
EPA, found CaO surface areas as high as 53 m^/g CaO when 50-im particles were cal-
cined in 0.15 sec at 1230°C, dropping to about 21 m2/g after 1.3 sec. Coutant's
calcinations were made under non-isothermal conditions in combustion flue gas con-
taining 10 percent CO2; although the conditions were carefully controlled, the re-
liability of the B.E.T. measurements was uncertain due to the impossibility of
completely avoiding recarbonation of CaO during sample collection. Consequently,
one objective of our work was to determine CaO surface areas generated in the ab-
sence of COg using nitrogen entrainment. The results showed that the specific
surface area of CaO is 50-60 m^/g when 10-pm limestone particles were calcined
isothermally at temperatures up to 1075°c in 0.6 sec, and can reach 90 m^/g when
calcined at 600°C. Our results also verified that sintering occurs rapidly at
higher temperatures, reducing the surface area to 25 m^/g in an isothermal resi-
dence time of 0.5 sec at 1150°C. The conclusion reached is that the nascent CaO
formed immediately after CaC03 decomposition has a grain radius, rg, of about
150 A and that these grains coalesce and grow at a rate that increases rapidly
with temperature. The grain growth caused by this sintering process reduces the
specific area, Sg, because
where pCaO is the absolute density of calcium oxide.
6-2
-------
Surface Area and Reactivity
The high specific surface area produced by rapid calcination is important to
the kinetics of sulfur capture. Much of our effort has been directed toward
quantitative evaluation of its importance. Early studies of low-surface area
CaO (with relatively large particles) had indicated a possible linear relation
between surface area and reaction rate noj> suggesting a chemically controlled
process. The maximum surface area that could be produced by methods then avail-
able was less than 10 m^/g. Using the new experimental techniques for produc-
ing nascent CaO, we have extended the kinetic measurements to a surface area of
80 m^/g, using particles small enough to eliminate pore diffusion resistance
and thus improve our ability to evaluate the surface area effect.
When pore diffusion resistance is absent, a CaO grain within a lime particle
will react with a gas at a rate that is most probably controlled either by the
Intrinsic kinetics of the reaction at the surface of the unreacted CaO core or
by diffusion through the product layer surrounding that core. The conversion
(X) vs. time (t) responses expected for these respective situations are given
for spherical grains by:
1 - (1 - X) 1/3 = kct (2)
1 - 3 (1 - X) 2/3 + 2 (1 - X) = kdt (3)
where kc and k
-------
Sintering
Our investigation of CaO sintering confirms the strong effect of temperature on
the rate of grain growth and the resulting loss of surface area. The sintering
rate of 1-pm limestone particles, after calcination under differential condi-
tions to yield maximum initial surface area, is shown in Figure 4. The effect
of temperature on the initial Sg value was weak, if at all, over the temperature
range that our measurements are considered reliable: 600° to 950°C. In this
range, the initial surface area was 79 _+ 6 m^/g, indicating an initial CaO grain
radius of about 110 A. Once formed, the grains grow at a rate that increases
rapidly with temperature as indicated by the surface area measurements of Figure
4. A continuous flow of N2 was passed through the reactor during these tests.
The rate of sintering is strongly affected by CO2. This is illustrated by the
data of Figure 5 which shows a series of tests in which CO2 was added in varying
concentrations to the N? that passed through the CaO during the sintering period.
(The calcination was initially completed with a pure-N2 sweep.) In all tests the
temperature and CO2 partial pressure were maintained well outside the region
where recarbonatlon could occur. We conclude that CO2 catalyzes the CaO sinter-
ing process. This effect can qualitatively account for the lower CaO surface
areas found by Coutant, et al. compared to the 50-60 m^/g that we obtained by
nitrogen entrainment in the EPA flow reactor. Since Coutant's calcinations were
made in flue gas containing 10 percent CO2, the lower surface areas obtained
probably resulted primarily from its effect on sintering rather than recarbona-
tion during sample collection.
SO? Model
Our measurements of the Ca0/S02 reaction kinetics were correlated ^ by the fol-
lowing relationship:
kd = 1.993 Sg2 (pSo2)°-62 exp (-36600/RT) (5)
over a range of Sg from 2 to 63 m2/g, S0£ partial pressures (Pjo-) from 0.00012
to 0.01 atm, and temperatures (T) ranging from 1033 to 1148 K. 2A comparison of
Eq. (5) with pilot furnace injection data of Ishihara yields agreement with-
in the expected limits of accuracy of available Sg data for high-temperature cal-
cination in flue gas.
IMPLICATIONS OF THE KINETIC ANALYSES: RATE LIMITING MECHANISM
The above sumiary discusses two independent results of our studies which indicate
that product layer diffusion determines the maximum rate of sulfur capture by CaO.
They are: (1) the shape of the X vs. t response curves, and (2) the strong de-
pendence on specific surface area. It should be possible to deduce the nature
of the diffusion process occurring within the product layer from measurements of
the effects of temperature and of gas phase concentration on the reaction rate.
The obvious mode of transport would be molecular diffusion of H2S, COS, or SO2
through pores or cracks in the product layer. If this were so, the overall rate
would be expected to Increase in direct proportion to the partial pressure of the
reacting gas, and the effect of temperature would be relatively slight, reflect-
ing the 1.5-power temperature dependence of the gas diffusion coefficient. An
Arrhenius-type plot of rate data for such a case would yield an apparent activa-
tion energy of only 5 kcal/mol.
6-4
-------
The results of our investigations of the temperature effect consistently show
good agreement with the Arrhenius relationship over the full range of tempera-
tures studied: 760 to 1175°C for SO2; and 600 to 900°C for H2S and COS. The
plots yield apparent activation energies of 37 and 31 kcal/mol, respectively,
implying that the diffusing species are not gaseous. The observed results are
most consistent with a mechanism of ionic diffusion in the solid state. In this
case, the diffusivity of the product layer increases exponentially with tempera-
ture in the same Arrhenius relationship that defines chemical reaction rate:
Ionic diffusion: De = D0 exp (-E/RT) (6)
Chemical reaction: kc = k0 exp (-E/RT)
and showing similar, high values of the activation energy, E, that normally
characterize chemically controlled reactions. Without information on the inde-
pendent effect of surface area on the rate of reaction, it would not be possible
to distinguish between the two analogous processes.
The effect of gas concentration 1s also revealing. Under reaction conditions
that ensure differential operation and eliminate pore diffusion, the reaction
rate of CaO was found to be less than first order in all cases. For SO2 and H2S,
the rate increased with only the 0.6 power of partial pressure. For COS, the
reaction rate was independent of partial pressure as indicated in Figure 6, pro-
viding the clearest evidence against gas phase diffusion.
From the four lines of evidence discussed above, it seems most reasonable to con-
clude that the diffusion process controlling the rate of transport through the
product layer involves the movement of ions, not gas. This conclusion is con-
sistent with other recent investigations of CaO reactions. ('!» 13) Figure 7
illustrates how the solid-state process may occur: oxygen ions diffuse from the
CaO core through the product layer surrounding the grains and react with the ad-
sorbed gas at the product layer surface. The product ion counterdiffuses toward
the CaO core, maintaining charge balance.
REACTION PROMOTERS
The rate of ionic diffusion is determined by the concentration of defects in the
atomic lattice (14). jn this case, the lattice structure of the product layer.
Defects are induced either thermally—which is responsible for the exponential
temperature dependence of Eq. (6) — or by the substitution of ions of different
valence into the lattice. When an ion having a valence different from calcium
is incorporated in the growing product layer, an oppositely charged defect is
created in the lattice to preserve electroneutrality. It is characteristic of
such "substitutional defects" to enhance diffusion most strongly in the lower
temperature regions where thermal defects are less dominant.( '5") it is not un-
reasonable to expect, therefore, that certain foreign metal oxides or salts
might enhance the reactivity of CaO when present on its surface. If incorpor-
ated into the growing product layer, the defects induced by the foreign ions
could increase ionic mobility and reduce the activation energy of diffusion
through the product layer. Faster sulfur capture would result.
CI2O3
Our study of promoters was begun shortly after the discovery, by the Energy and
Environmental Research Corp., that ^03 significantly enhances SO2 capture in
6-5
-------
pilot furnace tests. When that result was confirmed by the Southern Research
Institute, we undertook a laboratory study to determine: (1) if the effect of
O2O3 is due to catalysis of SO2 oxidation, and (2) whether its effect could be
duplicated by other additives.
One could hypothesize that O2O3 increases the rate of SO? capture by any of
three likely mechanisms: (1) by reducing the rate of CaO sintering, (2) by in-
creasing the rate of SO2 oxidation, or (3) by increasing the rate of ion diffu-
sion. The first possibility was quickly eliminated by measurements of the sin-
tering rate of CaO prepared from limestone containing 5 percent O2O3 additive.
At 700°C, it reduced the specific surface area 60 percent relative to untreated
limestone that was calcined and sintered at the same conditions. When the lime
stone contained only 1 percent O2O3 additive, it yielded CaO having 22 percent
lower surface area.
Since 0^03 accelerates sintering, the procedure adopted to minimize this inter
ference with the SO2 reactivity measurements was to pre-calcine a batch of 3.0
mz/g CaO without additive (90 min. at 980°C) then mix 9 wt percent CrjOq (equiv-
alent to a 5 percent mixture with limestone) by grinding in a mortar. Samples
of the mixture were then exposed to 3000 ppm SO2 + 5 percent O2 (balance N2) in
a differential reactor at 1125°C, following a 2 m1n. heatup period. The conver-
sion vs. time response was compared with the response obtained with similar
tests of 3.0 m^/g CaO that was ground in the mortar, but no promoter added. A
quantitative measure of the promoter's effect was determined as the ratio of k
-------
Included in the list are V2O5 and Fe203, both known catalysts of SO2 oxidation.
Their low effectiveness as CaO reactivity promoters, compared to non-catalysts
such as Na2S04, is further evidence that catalysis is not involved in the mech-
ani sm.
Table 1 also shows that Cl" ion inhibits the reaction; both NaCl and FeC^ re-
duced reactivity while compounds containing the same cations, but no Cl", in-
creased reactivity. Earlier experiments have shown that CaC^ is a strong
inhibitor of the H;>S/CaC03 reaction under anoxic conditions, suggesting a mech-
anism similar to the one operating here.
Figure 8 shows the effect of varying the amount of promoter added for Na?C03.
The doses were equivalent to 1, 2, and 5 wt. percent addition to limestone.
The degree of enhancement increased with the amount added, but not in direct
proportion. It is important to note that the contribution of the Na2C03 to SO2
capture is relatively minor: at the 5-percent level of additive, it accounts for
about 7.5 percent of the total sulfur capture while the CaO conversion doubled.
800°C. If promoters function by the lattice-defect mechanism, a reduction of the
activation energy should be apparent. Another series of tests was therefore con-
ducted at 800°C to permit comparisons with the reactivities measured at 1125°C.
These tests were focused on elements of Group IA of the periodic table which were
identified as potentially most effective by the initial screening at 1125°C. Fig-
ure 9 shows the conversion vs. time responses measured for several of these addi-
tives. Table 2 lists the k
-------
Diffusion through the product layer appears to occur by a solid-state process
involving ionic transport, not by gas diffusion.
The reactivity of CaO with SO2 can be markedly affected by the presence of for-
eign metal oxides or salts on Its surface.
The observed effects of promoters are not consistent with a mechanism involv-
ing catalysis of SO2 oxidation* The effects are qualitatively consistent with
a mechanism which assumes that defects are induced in the growing product layer,
increasing the 1on1c diffusivity.
Carbonates and sulfates of the alkali metals were the most effective practical
promoters found to date, according to the screening procedures used in our
1aboratory.
The reactivity of CaO with COS is zero order with respect to COS concentration.
CaO sintering is accelerated by CO2 and by most promoters of the Ca0/S02 reaction.
REFERENCES
1. R. H. Borgwardt, G. R. GilHs and N. F. Roache. "IERL-RTP Research on Sul-
fur Capture 1n LIMB," presented at EPA/EPRI Joint Symposium on Stationary
Combustion N0X Control, Dallas, TX, November 1982.
2. R. H. Borgwardt and N. F. Roache. "Reaction of H2S and Sulfur with Lime-
stone Particles," Ind. Eng. Chem., Process Pes. Dev., 23, 724-8 (1984).
3. R. H. Borgwardt. "Calcination Kinetics and Surface Area of Dispersed Lime-
stone Particles," AIChE J., in press.
4. R. H. Borgwardt, N. F. Roache and K. R. Bruce. "Surface Area of Calcium
Oxide and Kinetics of Calcium Sulfide Formation," Environ. Progress, 3,
129-35 (1984). : ~
5. R. H. Borgwardt, N. F. Roache and K. R. Bruce. "Method for Variation of
Grain Size in Studies of Gas-Solid Reactions Involving CaO," IERL-RTP paper,
Research Triangle Park, NC (1984).
6. R. H. Borgwardt and K. R. Bruce. "Effect of Specific Surface Area on the
Reactivity of CaO with SO2," IERL-RTP paper, Research Triangle Park, NC
(1984).
7. Coutant, et al. "Investigation of the Reactivity of Limestone and Dolomite
for Capturing SO2 from Flue Gas," EPA report APTD 0802 (NTIS PB 204-385),
U.S. EPA, Industrial Environmental Research Laboratory, Research Triangle
Park, NC, October 1971.
8. Y. Ishihara. "Kinetics of the Reaction of Calcined Limestone with Sulfur
Dioxide in Combustion Gases," presented at EPA Dry Limestone Injection
Process Symposium, June 1970.
9. Y. Ishihara and H. Hukusawa. "Studies on Sulfur Oxides Removal from Flue
Gas by Dry Limestone Injection Process (III)," Nenryo Kyokaishi (J. Fuel
Soc. Japan), 54, 321 (1975).
6-8
-------
10. R. H. Borgwardt and R. D. Harvey. "Properties of Carbonate Rocks Related
to SO2 Reactivity," Environ. Sci. & Techno!., J5, 350 ( 1972).
11. M. Hartman and 0. Trnka. "Influence of Temperature on the Reactivity of
Limestone Particles with Sulfur Dioxide," Chem. Eng. Sci., 35, 1 189 ( 1980)
12. S. K. Bhatia and D. D. Perlmutter. "The Effect of Pore Structure on Fluid
Solid Reactions: Application to the S0?-Lime Reaction," AIChE J., 27,
226 (1981).
13. S. K. Bhatia and D. D. Perlmutter. "Effect of the Product Layer on the
Kinetics of the C02-Lime Reaction," AIChE J. 29, 79 (1983).
14. W. Jost. "Diffusion in Solids, Gases, Liquids," Academic Press, New York,
1960.
15. F. Beniere and C. R. A. Catlow. "Mass Transport in Solids," Plenum Press,
New York, 1983.
6-9
-------
Tine (t), seconds
0 50 100 IjO 2CC 25C 300
100
C-C02 j see
+ 2(1
0.C00213 sec
40
eap.
7 CC
7 00
960
0
400
200
6GC
300
1000
TIzie (t), seconds
Reproduced from
best available copy. Mfop
Figure 1. Response characteristics of the reaction of 1-um CaO
particles with sulfur gases.
6-10
-------
2
2.E.T. Surface Area of Calcine, :u
100
3000
GCC
C03
1000
1C00 o
"lope
r j
lope
« 100
100
30
An
10CC
100
10
n
E.T. Surface Area of Calcine
•p
U1
Figure 2. Test for product-layer diffusion control.
6-11
-------
Symbol
-xposure
tl-ae, sec
T enp.,
°C
Trouuct
O
900
soo
CaGO.
4
A
600
7 00
Ca GO, + CaGO .
J 4
¦
30
700
Ca 3
_] 1 I 1 1 1
10 20 30 40 50 60 70
2
Surface Area of Calcine, m /g
Reproduced from
best available copy.
Figure 3. CaO conversion as a function of its initial specific
surface area. Closed symbols = 5000 ppm H2S + 45% H2. Open
symbols = 3000 ppm SO2 + 5% 03.
6-12
-------
600 C
80
700 C
tN
•H
U
i—H
a
CJ
a
u
ei
m
u
3
C.1
C,
20
1000 c
0
10
20
30
40
6C
50
3inter Time, ainutes
Figure 4. Sintering rate of CaO in the differential reactor.
Pure nitrogen sweep during sinter period.
6-13
-------
•o
I
3
u
c
15. C
l-l
3
cl
0
40
10
20
30
50
60
Sinter Time, minutes
Figure 5. Effect of CO2 on the sintering rate of CaO at 900°C.
6-14
-------
1 I I I I I I I I I I I I I I II I
o
Cr
XT
-O-
O
O
J—i—i i i i 11 i i i i » i i i
90 1OC0 10000 30000
cor Concentration, pjia.
Figure 6. Reaction rate of CaO with COS as a function of
gas-phase concentration. Temperature = 700°C.
6-15
-------
Reproduced from
best available copy.
U)
f1 \
\o)
Figure 7. Possible mechanism of ionic diffusion in the
product layer: (a) H2S reaction, (b) SO2 reaction.
6-16
-------
ore Volune
0.0090 sec
a CO, AJded
10 20 30 40 50 60
Tine, seconds
Figure 8. Effect of sodium carbonate on the reactivity of
3.0-nf/g CaO at 1125°C. 3000 ppm S02 + 5% 02.
6-17
-------
Reproduced from ^pl
best available copy.
o Additive
100
30C
~.e, seconds
Figure 9. Comparison of the effect of alkali-metal salts on
the reactivity of 3.0 n^/g CaO at 800°C. Additive = 9 wt. %,
CaO basis (5% limestone basis). 3000 ppm SO2 + 02*
6-18
-------
Table 1
PROMOTER SCREENING TESTS AT 1125°C
(9 wt. percent in 3.0 nr/g CaO, 3000 ppm SO2 + S% 0^)
Ratio With Additive
Additive kn Without Additive
Rb£S04
4.6
Cr203
4.5
Na2C03
4.3
Rb2C03
4.2
K2SO4
4.1
NaHC03
4.0
NaHS04
4.0
k2co3
4.0
Na5p3°10
3.9
Na2S04
3.7
Trona
3 .5
KHCO3
3.0
Na2Si03
2.5
M0O3
2.3
L i9SO4"H2O
Ti&2
2.2
1 .9
H3BO3
1 .8
N i CO3
1 .8
N32B407
1 .6
P2O5
1 .6
Fe203
1 .4
AT 2°3
1.4
Silicic Acid
1 .3
MgO
1 .2
Li2C03
1 .1
V2O5
1 .1
ZnO
1 .0
None
1 .0
Sb20s
0.8
CuO
0.8
NaF
0.7
NaCl
0.4
FeCl3
0.3
NaBr
0.2
6-19
-------
Table 2
PROMOTER TESTS AT 800°C
(9 wt. percent in 3.0 nr/g CaO, 3000 ppm SO2 + 5* O2)
Ratio kn With Additive
Additive Without Additive
L i 2CO3
200
Li2S0<.'H20
Na5p30lO
59
21
Na2C03
20
Na2S04
13
KHCO3
13
Rb2C03
11
k2co3
9
K2SO4
8
Cr203
7
NaHC03
6
Table 3
SINTERING TESTS OF S02-REACTI0N PROMOTERS
wt. percent in limestone, calcined and sintered at 800°C)
CaO Specific Surface Area
Addi ti ve (m^/g)
None 37
Al2 O3 30
M0O3 23
Na2C03 20
Na2S04 18
Na2C03 (repeat) 17
NaHC03 15.0
NaHC03 (repeat) 15.2
Rb2C03 13
K2CO3 12
Cr203 11
L i 2 CO3 1.4
L12CO3 (repeat) 4.2
6-20
-------
FLOW REACTOR STUDY OF CALCINATION AND SULFATION*
V. P. Roman and L. J. Muzio**
KVB, Inc.
Irvine, California 92714
M. W. McElroy
EPRI
Palo Alto, California 94303
K. M. Bowers and D. T. Gallaspy
Southern Company Services
Birmingham, Alabama 35202
ABSTRACT
The renewed interest in direct furnace injection of dry sorbents for SOo control
from coal fired boilers has prompted bench scale studies of the calcination and
sulfation of calcium compounds. The bench scale study was conducted in a one-
dimensional flow reactor. The objective of the study was to determine the inter-
relationships of the calcination and sulfation processes and how these processes
are influenced by the type of sorbent material, temperature, and residence time.
Four materials were studied: two limestones, a calcium hydroxide, and a pressure
hydrated dolomitic lime. Calcination and sulfation of these sorbents were inves-
tigated over the temperature range of 1400-2000°F and residence times of 0.20 to
U.75 seconds. The reaction environment consisted of combustion products from a
natural gas or hydrogen fired burner doped with SO2 and CO2.
* The work presented
Research Institute
**Currently with the
in this paper has been sponsored by the Electric Power
and Southern Company Services.
Fossil Energy Research Corporation
7-1
-------
FLOW REACTOR STUDY OF CALCINATION AND SULFATION
INTRODUCTION
Dry sorbent emission control is currently undergoing renewed interest as an S0£
control approach for coal fired boilers. While this technology was investigated
during the 196U's and early 1970's the results were unsuccessful in achieving
NSPS goals. Success of this current development activity hinges on gaining a
better understanding of the process fundamentals. EPRI and Southern Company
Services are currently supporting the development of dry sorbent emission control
(USEC).
At the onset of the program, little fundamental information existed on the funda-
mentals of the calcination and sulfation processes for conditions representative
of utility boilers. Previous studies (Ref. 1-5) had studied calcination and
sulfation using rather large particles and in experimental systems with either
characteristically long residence times; or in pilot scale systems where the
conditions were not uniform or well specified. To support the current develop-
ment program, bench scale experiments were conducted to provide a fundamental
basis to interpret the pilot scale work being conducted as part of the DSEC
development program.
The objective of the study was to develop an understanding of the interrelation-
ship between the calcination and sulfation processes and the effects of tempera-
ture and residence time on the processes for a variety of calcium based sorbents.
This was considered important in gaining an understanding of the mechanisms that
either limit high sorbent utilization or can be taken advantage of in yielding
high utilization in a utility boiler environment. In particular the bench tests
were structured to determine: (1) the rate of calcination of both limestones and
calcium hydroxides in an idealized combustion gas environment; (2) the evolution
of surface area of the calcine during the calcination process; and (3) the rate
of sulfation as a function of temperature, residence time, and sorbent type.
EXPERIMENTAL APPARATUS
A one dimensional entrained flow reactor was used for the study. A schematic
diagram of the experimental apparatus is shown in Figure 1. The system is
comprised of an isothermal flow reactor, a sorbent injection system, and a
sampling system. The flow reactor (Figure 2) consists of a 1.94 inch (I.D.)
stainless steel tube, three feet long, which is heated electrically. The reactor
was operated with the inside diameter defined by the stainless steel tube or
lined with a high alumina refractory. The electrical heaters were controlled
using thermocouples located in the reactor gas stream. A gas burner, using
either natural gas or hydrogen, was used to generate gases for the reactor. In
addition a variety of dopant gases, typically COo and SO2, were added to the gas
stream at the burner. The temperature at the entrance to the reactor test sec-
tion was controlled by varying the location of the burner in the horizontal
section and by the electrical heaters on the horizontal section (Figure 2).
7-2
-------
Te.nperatures within the test section were controlled by the electrical heaters.
In the current configuration the flow reactor could be operated over the temper-
ature range of nominally 1200°F-2100°F. All flows to the reactor were measured
with calibrated rotameters.
The flow reactor is operated at a flowrate of approximately 1.4 scfm, which at a
temperature of 1800°F results in a velocity of 6.5 ft/sec and a residence time
over the maximum reactor length of 0.75 seconds. Residence time was varied by
varying the reactor length from 1.5 to 3 feet or varying the gas flow through the
reactor.
A fluidized bed feeder was used to feed the sorbents to the reactor. Two fluid-
ized bed feeder designs were used; a mechanically fluidized system similar to
that employed by Borgwardt (Ref. 6) or a sprouted fluidized bed that was mounted
on a load cell. This latter arrangement could be used to more accurately deter-
mine the sorbent feed rate. For the results reported in this paper sorbent feed
rates were on the order of 10 grams/hr. The sorbent was injected into the
reactor through a water cooled injector to insure that all reaction occurs within
the test section.
The sorbent sampling system is comprised of a sample probe, particle collector,
vane pump and a gas sample system. The probe is water cooled with the capability
of injecting inert gas into the sample stream to quench the reactions. All of the
reactor gas flows through the sample probe. The solid sample is collected with
either a cyclone or an alundum thimble. A gas sample is extracted upstream of
the solids collector and is continuously analyzed for Oj, CO^, CO and S02.
Basically two types of experiments were conducted during the study:
Calcination experiments where the flow reactor was operated
without SO2 and the rate of calcination and the evolution of
calcine surface area determined.
Sulfation experiments where SOg was added to the flow reactor gas
stream and the extent of sulfation determined. During these
tests the flow reactor was operated in a dilute mode where the
Ca/S ratio was much less than unity. In this way the sorbent
particle was exposed to a constant SO2 concentration over the
entire residence time in the reactor test section.
All data analysis was based on characteristics of the collected solids. The
solids were analyzed for carbon and hydrogen with a Perkin-Elmer elemental
analyzer and the sulfate content determined by ion chromatography*. The extent
of calcination of the calcium carbonate materials was determined from the carbon
content of the sample and calcination of the hydrated sorbents determined from
the hydrogen concentration of the sample. The extent of sulfation (or sorbent
utilization) was determined from the sulfate content.
The BET surface area of the samples was measured using a Micromeritics Flow Sorb
analyzer. This instrument performs a single point determination of the BET area.
* Chemical analysis of the samples were perfomed by the Southern Research
Institute.
-------
SORBENT CHARACTERISTICS
A variety of sorbents were used during the study including two natural lime-
stones, a calcium hydroxide, and a pressure hydrated dolomitic lime. These same
sorbents are being used in a pilot scale study of dry sorbent emission control
(Ref. 7). The chemical and physical properties of the sorbents are given in
Table 1. The two limesones were selected to represent a range of morphological
properties. The St. Genevieve limestone is a fairly crystalline material with a
surface area of 1.7 m /gm while the Marianna limestone has a substantially higher
initial surface area, 6.3 nr/gm. The.CaJOHJg and pressure hydrated dolomitic
lime were selected as being typical, commercially available, hydroxide compounds.
RESULTS
The test program involved separate tests of the calcination and sulfation of
the sorbents. These results will be discussed separately below. It should be
pointed out that the test program is currently ongoing and the results presented
in this paper should be viewed as preliminary.
C a 1c inati on
The calcination results of the tests performed with the St. Genevieve limestone
in a flue gas environment containing 12 percent CO2 are shown in Figure 3a. As
expected the calcination rates increase as a function of temperature. At 2000°F
80 percent of the calcination occurred in less than 0.5 seconds. The Marianna
calcined more rapidly (80 percent in 0.3 seconds at 2000°F) and shows similar
temperature dependence with increasing calcination rates with increasing
temperature. Both, the Marianna and the St. Genevieve limestones exhibit no
calcination at 1400°F in an environment of 12 percent CO^. This is expected
since the equi1iDrium partial pressure of C0£ at 1400°F is 0.12 atm. Removing
the CO2 from the gas stream (by firing hydrogen) results in calcination of the
limestone at 1400°F as shown in Figure 4. At a temperature of 1800°F changing
the bulk gas CO2 concentration from 0 to 12 percent has no apparent effect on
calcination. Again this result is expected since the equilibrium partial
pressure of CO2 for the calcination reaction is 0.8 atm at 1800°F.
Borgwardt (Ref. 6) suggested that in a kinetically limited regime calcination
rates are proportional to initial BET surface area with an activation energy of
48 Kcal/moles. For the conditions studied in this program the higher surface
area stone calcined more rapidly even though the particle size of the Marianna
limestone was 3.5 times as large as the St. Genevieve stone. However, both the
Marianna and St. Genevieve stones exhibited slower calcination rates than
reported in Ref. 6. The porosity of these stones may have resulted in a CO2
diffusion limitations within the particle which lowered the overall rate of
calcination relative to kinetic limitations.
Figure 5 shows the results of the calcination tests with calcium hydroxide. From
these results it is not possiole to identify any temperature dependence of the
calcination rate. This is thought to be an artifact of the analysis since the
degree of calcination is determined by the mass fraction of hydrogen which
accounts for only three percent of the total mass of the uncalcined material.
Any contamination by moisture or modest error in the analysis results in rela-
tively large discrepency in the calculated degree of calcination. It is evident,
however, that the calcination rate of the calcium hydroxide is faster than that
of the limestones with at least 80 percent calcination occuring in 0.2 seconds.
The pressure hydrated dolomitic lime exhibited calcination rates comparable to
the calcium hydroxide.
7-4
-------
The evolution of surface area for the Marianna limestone (Figure 6) shows that
the surface area of the material increases with reactor residence time to a peak
value and then decreases. Highest surface areas are generated at the lower tem-
perature conditions. During calcination the surface area increased by over a
factor of about two (from 6.3 m^/gm to 15 m^/gin).
The surface area of the pressure hydrated dolomitic lime (Figure 7a) is seen to
be an increasing function of reactor residence time during calcination. The
evolution of surface area is also seen to be a function of reactor temperature
with surface areas of 37 nr/gm being produced at 1800°F and residence times of
0.6 seconds. This effect is not observed during the calcination of calcium
hydroxide. Figure 7b shows that the BET surface area for the calcium hydroxide
does not increase as a function of residence time but actually decreases some-
what. Since calcination results with the calcium hydroxide were not obtained for
residence time less than 0.2 seconds, higher surface areas could have been pro-
duced early in the calcination process and that grain growth and sintering of the
porous CaO structure occurred at residence times earlier than 0.2 seconds.
Calcination of the pressure hydrated dolomitic lime in the flow reactor resulted
in substantial increases in BET surface area (Figure 7a). To investigate the
effect of more moderate heating on surface area, small quantities of the material
were heated at temperatures of 300-400°C for 16 hours in a nitrogen sweep gas.
The resulting weight loss and BET surface areas are shown in Figure 8a and 8b
respectively. As seen in Figure 8a, the majority of the weight loss can be
associated with the dehydration of one water molecule from the dolomitic
hydroxide (a somewhat greater weight loss occurred with heating at the 400°C
condition). Thermodynamic considerations suggest that the dehydrated water is
associated with the magnesium. Surface areas in excess of 110 m2/gm resulted
from the heating process. Further, data at 350°C and 400°C suggest that the
surface area continues to increase up to the point where the one ^0 molecule
is dehydrated. With further heating and weight loss (heating times greater than
nine hours at 400°C) the surface area begins to decrease. Sulfation tests of
these partially precalcined materials is currently in progress.
Sulfat i on
Sulfation of the sorbents was investigated over the same conditions as the cal-
cination tests. During these experiments, the sorbent feed rate was kept at a
value such that Ca/S < 1 in order that the sorbent particle experienced a con-
stant SO2 level throughout its residence time in the reactor. In general, the
sulfation data exhibited a higher degree of variability than the calcination
data, and the source of this variability is still under investigation. The
results of the sulfation of the Marianna limestone and calcium hydroxide are
shown in Figure 9a and 9b respectively. For the sulfation of Marianna limestone
two points are noteworthy. First, the general level of calcium utilization
(e.g., fraction of the calcium converted to sulfate) is on the order of 15 to
20 percent. Secondly, there is very little effect of temperature over the
range of 1600 to 2000°F. The effect of residence time is difficult to establish
altnough it appears that beyond 0.25 seconds, the utilization only increases by
about 15-17 percentage points per second. More data is needed to more defi-
nitively establish the residence time effects.
The sulfation of the calcium hydroxide, shown in Figure 9b exhibits a somewhat
higner utilization than the limestone. Again, the variability observed in the
data to date preclude drawing definite conclusions on the effect of temperature
or residence time. Temperature, as with the limestone appears to have a minimal
7-5
-------
effect over the range studied and the majority of the sulfation appears to have
occurred within 0.25 - 0.4 seconds. This is consistent with the surface areas
generated during calcination of Ca(0H)2 (Figure 7b); all the surface areas were
in the range of 10 to 15 rrr/gm.
More interesting results were obtained with the pressure hydrated dolomitic lime
as shown in Figure 10. This material exhibited utilizations on the order of 20
to 25 percent when the reactor was operated at 1600-1800°F with sulfation occur-
ring within a residence time of 0.3 seconds. When the temperature was increased
to 20(JL)°F , the utilization increased to the range of 50-55 percent for residence
times between 0.25 and 0.35 seconds. Further tests are in progress to extend
this range of residence times at this higher temperature condition. These trends
with residence times and temperature are consistent with pilot scale work
reported in another paper in this symposium (Ref. 7).
A question still remains as to what limits high utilization of these calcium
based sorbents; kinetics and reactive surface area, pore plugging, or solid
diffusion. To provide some insight into the limiting mechanism the surface area
of the sulfated pressure hydrated dolomitic lime was determined and is plotted as
a function of utilization in Figure 11. During the sulfation process the surface
area decreases from a calcine surface area of 25-37 nr/gm to under 10 [Tr/gm as
the utilization exceeds 50 percent. If all of the CaO surface area consisted of
grains of a uniform diameter and each grain sulfated uniformly, the change in
surface area should follow the shaded band in Figure 11. The charge in surface
area is significantly greater, suggesting either: (1) the concept of uniform
sulfation of a grain is not correct and that visualizing the process as sulfation
of a pore where a larger change in surface area per unit of sulfation occurs is
more correct, or (2) pore plugging limits access to active surface, or (3) the
sulfated layers from adjacent grains merge to decrease access to reactive
surface.
SUMMARY
The following summary points can be made from the bench sacle results to date:
Limestone calcination rates at temperatures over 1800°F are
relatively fast with 80 percent calcination occurring within a
residence time of 0.3 seconds for the Marianna limestone, and 0.5
seconds for the St. Genevieve limestone.
Calcination rates are a function of the initial surface area of
the limestone.
Limestone calcination rates measured in this study were slower
than rates determined under conditions where kinetics dominates
(Ref. 6) suggesting a CO^ diffusion limitation within the
particle.
Surface area of the limestone calcine increases during calcina-
tion; for the Marianna limestone the surface area increased by a
factor of three. Calcine surface area decreases as calcination
becomes complete.
Calcination rates of the hydrated materials are more rapid than
the carbonates, with the majority of the calcination occurring in
1 ess than 0.2 seconds.
7-6
-------
During calcination of the calcium hydroxide the calcine surface
area was less than the initial surface area at residence times
greater than 0.2 seconds; while the calcine surface area of the
pressure hydrated dolomitic lime increased with residence time
and temperature.
Sulfation of the limestones and Ca(OH)2 was relatively insen-
sitive to temperature and residence times with utilizations in
the range of 12-20 percent and 15-27 percent respectively. The
majority of the sulfation occured within a residence time of
0.25 seconds.
Utilization of the pressure hydrated dolomitic lime increased as
the temperature increased from 1800°F to 2000°F. Utilization at
2000°F was on the order of 50-55 percent.
ACKNOWLEDGMENTS
Tnis work was sponsored by the Electric Power Research Institute and Southern
Company Services. The authors are grateful to F. Garcia and R. Himes for helping
to conduct the experiments. The efforts of the Southern Research Institute for
performing the chemical analysis of the samples is also greatly appreciated. We
also appreciate the ongoing discussions and interaction with R. Rush of Southern
Company Services, and D. Giovanni. These discussions have been instrumental in
structuring the work for the present program.
7-7
-------
REFERENCES
1. A. E. Potter, "Sulfur Uxide Capacity of Limestones," American Chemical
Society, Vol. 48, No. 9, 1969.
Z. K. H. Borgwardt and R. D. Harvey, "Properties of Carbonate Rocks Related to
SUo Reactivity," Environmental Science and Technology, Vol. 7, No. 4, April
1972.
3. R. C. Attig and P. Sedor, "Additive Injection for Sulfur Dioxide Control, A
Pilot Plant Study," National Air Pollution Control Association Report No.
5460, March 1970.
4. R. H. Borwardt, "Kinetics of the Reactional SO2 with Calcined Limestone,"
American Chemical Society, Vol. 4, No. 1, January 1970.
5. R. W. Coutant, et al., "Investigation of the Reactivity of Limestone and
Dolomite for Capturing S02 from Flue Gas," Final Report, NAPCA Contract, No.
PH-86-67-115, October 1971.
6. R. H. Borgwardt, "Calcination Kinetics and Surface Area of Dispersed
Limestone Particles," submitted for publication 1984.
7. R. Beittel, et al., "Development of Calcium Based Dry Sorbent Emission
Control," 1st Joint Symposium on Dry SO2 and Simultaneous S02/N0X Control
Technologies, November 1984 (this symposium)
7-8
-------
I
jE
SORBENT INJECTOR
BURNER
SAMPLE PROBE
PARTICLE
COLLECTOR
GAS ANALYSIS
9
PUMP
CD-
CO COg SO2
& t? %
AIR NAT. HU
GAS d
DILUTION
AIR
FLUIDIZED
FEEDER
SWEEP AR
0-
0-
%
AIR SUPPLY
Figure 1. Flow Reactor System
-------
Water-Cooled
Sorbent Injection
Electrical Heaters
Thermocouple
Thermo-
couple
Quench
Gas Burner
falPlj Insulation
Ceramic Liner
310 Stainless Steel
Water-Cooled, Gas Quench Probe
Figure 2. Flow Reactor Details
7-10
-------
EXTENT OF CALCINATION, %
to ^ Oi CD
EXTENT OF CALCINATION. %
o
o
In
G">
n>
3
rp
<
(D
<
O)
(D
O
fD
-------
100
# 80
o
5
Z 60
o
-J
<
o
LL
O 40
LD
h
X
IU 20
I I
OPEN SYMBOLS - 12% C02
CLOSED SYMBOLS - 0% C02
/
/
/
/
0.2
~
0.4
~
1400°F
J.
0.6
1800 F
~
1 400°F
0.8
1.0
RESIDENCE TIME, SEC
Figure 4. Effect of Gas Composition on Calcination of Marianna Limestone
7-12
-------
100
O 1600°F
~ 1800^
A2000°F
0.2 0.4 0.6 0.8
RESIDENCE TIME, SEC
1.0
Figure 5. Calcination of Calcium Hydroxide
7-13
-------
20
(50)
(48)
(53)
E
O)
(65)
(60)
177 )
E
2000°F
35 1°-
GC
<
H M
LU O
CD 5r
(81)
( ) EXTENT OF CALCINATION
0.8
1.0
0
0.6
0.2
0.4
RESIDENCE TIME, SEC
Figure 6. Evolution of Surface Area during Calcination of Man'anna Limestone
7_14
-------
LU
ffi
10
iaoo°F
1600°F
1400°F
J.
-L.
0.2 04 0.6 0.8
RESIDENCE TIME. SEC
1 0
(a) Pressure Hydrated Dolomitic Lime
40
E
a
—.
CM
LU
C
<
LU
10
30 -
20 -
10 -
o 1600°F
~ ieoo°F
A 2000°F
£]
Do
-L
_1_
0.2 0.4 0.6 0.8
RESIDENCE TIME. SEC
10
(b) Calcium Hydroxide
Figure 7. Evolution of Surface Area during Calcination
7-15
-------
¦ °aiSr
0.95
~—
X
O
- 0.90
5
Ui
> 0.85
UJ
(E 0.80
A.
-A
*^2^— - _
¦LOSS OF
ONE H20
. 300°C
350°C
O o
-A
400°C
u—o
0.75
4 8 12
TIME OF HEATING (HRS)
16
(a) Weight Loss
120
350°C
100
E
ai
400°C
80
£
co
-------
ao
1
\
1 1
40
O 1600°F
~ 1800°F
A 2000°F
¦
*
z"
o
30
-
f-
<
J
»-
20
10
0
q
1
0
1
>
DCP
- - A 1
0 0 2 0 4 0.6 0.8 1.0
RESIDENCE TIME, SEC
(a) Marianna Limestone
0 2 0.4 0.6 0 8
RESIDENCE TIME. SEC
Figure 9.
(b) Calcium Hydroxide
Sulfation of Limestone and Calcium Hydroxide
7-17
-------
100
80
z- 60
o
£
N
-I
i= 40
20
T
o 1600°F
~ 1800°F
A 2000°F
/
/
i'
)L I L_ L
0.2 0.4 0.6 0.8 1.0
RESIDENCE TIME, SEC
Figure 10. Sulfation of Pressure Hydrated Dolomitic Lime
7-18
-------
RANGE OF CALCINE SURFACE AREA
E
o>
CM
E
m
<
LU
DC
<
H
LU
ffl
<
s
O u-
N
Q
111 20
s
CO
10
UNIFORM SULFATION OF GRAINS
X
20 40 60
UTILIZATION, %
80
100
Figure il. Change in surface area during sulfation of pressure
hydrated dolomitic lime.
7-19
-------
Table 1
SOKBENT CHARACTERISTICS
St. Genevieve
Li mestone
CaCO-j
Marianna
Li mestone
CaCO,
Calcium
Hydroxi de
Ca(0H)2
Pressure Hydrated
Dolomitic Lime
Ca(0H)2 Mg(0H)2
CaU^, wt%
Mg 0, wtX
Particle size,
MMU, ) m
Density, gm/cm^
Surface Area BET
m2/gm
53.1
0.9
8.4
2.71
1 .7
50.1
1.0
30
2.70
6.3
72.2
2.0
2.3
2.24
14.3
43.1
30.3
1.3
2.30
19.6
7-20
-------
CALCIUM-BASED SORBENTS FOR DRY INJECTION
Jeffery L. Thompson
Dravo Lime Company
3600 Neville Road
Pittsburgh, Pennsylvania 15225
ABSTRACT
Dry injection technologies for SO2 capture require that the mass mean diameter of
the sorbent particle be relatively small, less than 20^, to obtain removal rates
high enough to make the process economically competitive with wet scrubbers. The
least expensive method of producing fine sorbent particles is by hydrating lime.
Typical commercial hydrates have a mass mean diameter less than S/n . New studies
of hydrate particles used in SO2 capture show that the reactant layer is less
than 1G00 A thick, the subject of this paper. These studies define the limiting
mechanism for sorbent utilization, i.e. the conversion ratio of CaO to CaSO,;.
Moreover, this understanding of the SO2 adsorption process points to how hydration
production methods might be modified to enhance SO2 capture.
CHARACTERIZING PARTICLES
"Calcium-based sorbents" for all practical considerations means limestone or its
derivative products: lime, essentially calcium oxide, CaO; and/or hydrate,
Ca(0H)o. There are also industrially precipitated calcium carbonates, CaCOj, and
natural ocean bed deposits of calcareous sands. For dry injection into boiler
zones where the temperature exceeds 1000°C both the carbonate and hydrate formu-
lations will calcine to the oxide, CaO. The SO2 control reaction thus occurs
with calcia, or lime, i.e., CaO, irrespective of the initial calcium composition.
Much of the physical and chemical description of CaO is dependent upon its
processing pathway, and some of those properties are critical to the adsorption
of SO2: specific surface area; mass mean diameter of the particles; and, chemical
composition and physical chemistry of the particle surface. In order to ascertain
what are desirable attributes of sorbent particles, it is necessary to understand
how SO2 is adsorbed onto CaO.
In a broad review of the existing data for SO2 adsorption into lime, there are two
quasi related trends which are obvious: the fractional conversion of CaO to CaS04
is inversely related to particle size; and as the particle size decreases the
specific surface area plays an increasing important role in SO2 capture. Large
particles, bigger than 30 or 40 microns, perform poorly in terms of S02 capture
and are resistant to improvement by any means, e.g. time-temperature profile in
an SO2 laden gas, SO2 concentration, and increasing the specific surface area of
the sorbent.
8-1
-------
Diffusion limitations of an unreacted core model suggest, and empirical data
supports the idea, that SOj capture is improved by using small particles with a
high specific surface area. Thus the limit of Ca conversion is at least a combined
function of particle morphology and geometry.
Powders that are produced by grinding and/or pulverizing may be reasonably well
approximated in mathematical treatments by spheres or regular geometric solids
with an aspect ratio (length:width:height) close to 1; powdered limestone and the
derivative lime that is calcined from it are examples of such materials. Instru-
ments that measure particle size for example assume that the particles have an
aspect ratio close to 1 in the calculations used to determine size distribution.
Such an assumption for hydrate particles has little correspondence to reality: the
particles are often serpentine, highly irregular in shape and are commonly
clustered into built-up forms; they may have aspect ratios as high as 50:1:0.1.
If the calculated surface area of a material (the surface area of spheres having
diameters defined by the particle size distribution) is substantially less than the
measured specific surface area then the conventional view is often that the
particles must be porous. In the case of hydrate particles the measurement of
particle size distribution by means of the usual commercially available instruments
nay be somewhat misleading in that the instruments do not account for irregular
shapes with significant length to diameter ratios; i.e., the very nature of
hydrate. An accurate characterization of hydrate requires SEM micrographs, and
often TEN micrographs and surface analysis with SIMS in addition to particle size
and BET specific surface area.
In the pursuit of small particles with high specific surface areas, it is necessary
to constantly keep in mind that the entire dry sorbent injection technology is
driven by economics: wet scrubbers using Mg enhanced lime perform extremely well:
99+% SO2 removal at a stoichiometry of essentially 1, they are, however, expensive
to install. The cost advantage that a dry injection system might have deteriorates
rapidly as the unit cost of sorbent used to capture SO^ increases. The quest thus
is one of low cost, small particles with a high specific surface area, with
emphasis on "low cost". Obtaining small particles, 100% less than 20^, by
grinding down rocks is not an attractive proposition, particularly in view of the
fact that the typical commercial hydrate has an mmd of less than 10^, usually
around 2
-------
price around $160-180, vs. S40-5O for commercial lime. Experiments that have
produced calcines with 100 m2/g or more specific surface area have used conditions
so exotic (combinations of very small particles, approaching a 1^ mmd, and/or very
high sweep gas rates) that the CaO cost would be in the range of 20
-------
binary CaO/CaSO^ reactant layer to at least a quaternary system of CaO • CaS04 -
NaoSO^j - Na20. In reality the coal combustion flue gases contain chlorine as well,
ana the sorbent surface contains some NaCl as an impurity so that the overall
system phase diagrams have a considerable liquidous and/or two phase liquid/solid
regions. The surface system is also more complex than quaternary: CaSC>4f CaCl2»
Na2C03, KC1, CaC03.
The mechanisms for mixing additives and sorbents are three:
(A) The additive and sorbent are mixed and injected into the boiler where
the additive vaporizes, then condenses out of the vapor state onto
the sorbent particle, but also on the fly ash, boiler walls, tubes,
etc. Transfer of the additive to the sorbent surface is not very
efficient; NaHCOj is in this category.
(B) .The additive and sorbent are mixed and injected into the boiler where
the additive melts and/or will form a solid-state compound with CaO.
To be effective this requires that the sorbent and additive particles
are small enough and mixed intimately enough that they cling together
and/or collide in the boiler.
(C) The additive is applied to the surface of the sorbent particle prior
to injection. This can be done by washing limestone with a solution
of the additive or in the case of hydrate the additive can be included
in the water used for hydration. If the additive is more soluble
than Ca(0H)2 then with a carefully designed hydrator the additive
will be largely contained on the surface of the hydrate particle.
The size, shape, surface morphology and chemistry of hydrate particles are deter-
mined by a combination of:
(A) Reaction rate of the water and lime.
(B) The time lapse to evaporate any excess water.
(C) Additives in the water and impurities in the original lime.
Hydration kinetics, i.e. crystallization of Ca(0H)2 from CaO + H2O, defines the
basic particle morphology. There seems to be almost a basic unit structure of
hydrate when lime and water are mixed and there is any appreciable delay (more
than 1 or 2 seconds) before the heat of reaction vaporizes the excess moisture: an
obloid shape, 200 to 1000 A long with an aspect ratio of about 2, i.e. its diameter
is roughly half its length. Liquid-mixed hydrate particles up to centimeters in
size when viewed in SEM micrographs all appear to be built-up collections of these
same basic pieces.
The morphology of hydrate particles resulting from steam and lime, or a water
spray (fog) striking an excess of CaO yields particles with vastly different
surface structures than the liquid mixed particles have. There is yet another
variable, additives, which alters the particle surface even more.
The demonstrated efficacy of additives used with hydrates and their limited pene-
tration into the reacted sorbent particles emphasizes the fact that the surface
characteristics and morphology of the sorbent particle are the critical deter-
minants in capturing SO2. While it is essential that the sorbent particles are
small, it is probably critical that the particles be small (]^, or less) in only one
dimension, not all 3.
8-4
-------
There is also the prospect that for boiler injection some combined oxide forms may
have a higher affinity for sulfur dioxide than CaO with just a modified surface.
Work in the steel industry on ladle metallurgy for desulfurizing specialty steels
would suggest certain combinations of lime, alumina and silicates might be a
promising avenue of investigation. The "designed formula lime" can be made by
briquetting combinations of limestone, hydrate and/or hydrate with additives,
bauxite, etc. which is then fired in a commercial lime kiln; the product is
hydrated as the most economical method of obtaining very fine particles, and then
injected into the boiler where the hydrate reverts to the oxide. To date only a
few formulations have been produced by this method and are yet to be tested in a
combustor.
The possible combinations of compounds, and their concentrations in a designed
lime product, in addition to a variety of hydration methods yields a myriad of
potential sorbents to evaluate. In view of the fact that neither the unreacted
core model (for roughly spherical particles), nor a porous solid model are
realistic representations of hydrates and their derivative oxide forms it may seem
that there is little alternative but to matrix through the list of possible
sorbents that could be manufactured. However, it seems much more efficient to
attempt to model the system: computer graphics, particularly fractals, along with
descriptions of behavior from physics, chemistry and thermodynamics have the
potential of creating a technique for evaluating the sequence of "designed 1ime-
hydration-dehydration/sulfation."
8-5
-------
& so,
m
1
100
80
60
40
%
20
&
hydrated do 1oline
(5.2)
:e w/additives
hydrate {3.1)
1 He st an e
w/addi:ives
1imestone
(6.25)
100 1C
specific surface
area
Sg in nv9
0.1 1 10
rrd sarticle size
10L
10J
Figure 1. Relationship of measured particle size, BET specific
surface and sulfur dioxide removal in a conbustor. The sorbent
to captured sulfur dioxide ratios are on a weight basis.
8-6
-------
600
500 HWe
31 S coal
500
-s • fraction of sulfur dioxide captured
Q
or
D
400
H
a
<
u
UJ
O
X
O
300
o
( 100)
5
K
200
SORBEN" COST
(J/TON)^-'
I 00
50
20D
100
300
capital cost
*/KW OF GENERATING CAPACJTT
Figure 2. Effect of reagent cost added to capital cost
(15-year depreciation) and its effect on cost of captured SO?.
To offer significant advantages over wet scrubbers, a dry
injection technology must have an installation cost half that
of wet scrubbers and use a reagent that costs less than $150
per ton.
8-7
-------
Reagent cost
S/ton SO2
captured
press, h/drated
dol ol me
V
wet scrubbers
cr.si te
zd 1 c mation
super scrbert
gcdl
lmestone
/addU ives
23
40
60
33
d502
f:b
Ons11e pruct.'i^ng
Sa rben t
Reagent Cost
Sorbent
Trar.spcrut ion
Co It
CO 5! dt
5/"on S3?
J/Ton
5/~on S3rcent
$/Tor Sor-cent
11 i ty
C
-------
Sorbent Cost
S/Ton (FOS ?lant)
1CGG
secondary
calcia
100
11 me
hycrate
10 -
.1
1.0
10
'.0
2
Sg in m /g mmd l'r^
Figure 4. Calculated costs of obtaining small particles with
high specific surface areas. Hydrated lime is a notable anomoly
in the system, yielding small particles, reasonably large specific
area at a significantly lower per ton cost than any other pathway
8-9
-------
e
Ar
>-
-i r
Sputter
4
3
2
reactant layer
thickness
resctart
1 ayer
tni ckness
1000
800
400
600
Death in X
Figure 5. Depth profile of reacted sorbent particles showing
sulfate reactant layer concentration and thickness. Data was
taken with SIMS-ISS system (Advanced R&D, St. Paul, Minn.) and
treated mathematically after viewing SEM micrographs cf sorbent
particles (before and after ion milling) to calculate an
apparent reactant layer depth.
8-10
-------
Figure 6. Common view of hydrate particle from a
commercially produced hydrate product with an mmd of
2.5 microns and a specific surface area that averages
19 m2/g.
H23
CO-
surface
IZ -LJ
& l/l
-------
additives
crushing/
screening
additives
grind/
pulveri ze
calciner
hydra tor
hydra tor
additives
CaO
additives
additives
calcine
CaO
hydratcr
designed hydrate
designad sarhent
{hydrated)
Figure 8
8-12
-------
SESSION II (PART 2): FUNDAMENTAL RESEARCH
Chairman, Dennis Drehmel, EPA, IERL/RTP
9-i
-------
LABORATORY-SCALE PRODUCTION AND CHARACTERIZATION
OF HIGH SURFACE AREA SORBENTS
D. A. Kirchgessner
U.S. Environmental Protection Agency
Office of Research and Development
Industrial Environmental Research Laboratory
Research Triangle Park, NC 27711
ABSTRACT
The objectives of the research described are to produce high surface area sorbents
in an apparatus amenable to scale-up, and to determine those characteristics in
the raw carbonates which correlate to the development of high surface areas in the
calcined product. Thus far, limestones crushed to minus 600 wm have been calcined
to produce maximum surface areas of from 40 to 58 m^/g in a laboratory-scale flu-
idized bed reactor with a pulsed air flow. A dolomite similarly treated has pro-
duced a surface area of 74 m^/g.
Particle size and elemental analyses of the raw stones have been performed, and
a significant negative correlation between maximum surface area development and
iron content has been noted. No correlation between surface area development
and petrographic properties has yet been demonstrated.
INTRODUCTION
The U.S. Environmental Protection Agency is developing the limestone-injection
multistage burner (LIMB) process as a cost effective means of achieving the simul-
taneous reduction of SO2 and N0X emissions from coal-fired boilers. As originally
envisioned, the process was to control N0X by staged combustion and SO2 by injec-
tion of limestone with the coal. It was expected that the limestone would calcine
to CaO in the higher temperature zone of the boiler and combine with SO2 at lower
temperatures downstream. Subsequent experimental evidence suggests, however, that
optimal sulfur capture may not occur due to imperfect calcination of the limestone
and/or to an interaction between the sorbent and mineral matter in the coal. Ad-
ditionally, experimental evidence on sulfation kinetics shows that SO2 capture in-
creases with the square of the sorbent surface area.^' One effective alternative,
therefore, appears to be calcination of limestone under more favorable conditions
outside the boiler and Injection of the resulting high surface area sorbent into
the boiler at the upper portion of the sulfation temperature zone. In support of
this approach to the LIMB process, the ongoing research described here has the fol-
lowing objectives: 1) to develop a laboratory-scale method, amenable to scale-up,
of calcining limestones to high surface areas (35-45 m^/g); 2) to determine the
physical properties of the sorbents produced; and 3) to identify properties in the
9-1
-------
raw stone which correlate to the development of high surface areas in the calcined
product.
CALCINATION
The investigation of methods to produce high surface area calcined sorbents pro-
gressed in two phases. The first phase consisted of a series of simple screening
experiments to delineate the parameters thought to be important 1n governing max-
imum surface area. In this case the parameters were: 1) sorbent characteristics;
2) time; 3) temperature; 4) particle dispersion; 5) rate of flowing medium; and
6) composition of flowing medium. The second phase consisted of constructing and
operating a fused quartz fluidized bed calcination unit to optimize surface area
development.
Screening Studies
Samples for these experiments were drawn from a collection of 19 pulverized lime-
stones and dolomites that were already on hand. Most of these stones consisted
only of minus 44 ym material. Those that had a wider size range were screened
at 44 ym to eliminate larger particle sizes. This was the only sample prepara-
tion performed on these materials.
To determine the degree to which sorbent characteristics govern surface area de-
velopment, 5 g of each stone from the above collection was calcined at 850°C for
30 minutes 1n a quartz crucible. The atmosphere was still air. The resulting
surface areas are recorded in Figure 1. The fact that dolomites as a group
achieved higher surface areas than the limestones was expected, but it was
somewhat surprising to note the inconsistent performance of chalks and marls.
Research by Harvey, et al.'2) has suggested that these types of materials rou-
tinely produce high pore volumes and surface areas. The important result for
this part of the screening effort is that, even among the limestones, the surface
areas vary by a factor of nearly three. It appears that, aside from the distinc-
tion mentioned above between limestones and dolomites, sorbent characteristics
play a more significant role in surface area development than was initially ex-
pected.
The next series of experiments focused on calcination conditions which might
affect the maximum surface area produced. For this series two stones were
selected from the above group: El Dorado limestone and dolomite 99, hereafter
referred to as Kaiser dolomite. In each case the stones were calcined over a
series of temperatures, comparing one other variable at a time.
In the first case only the means of dispersing the sample was varied. Of each
stone, 2 g was calcined in a still air environment for 30 minutes at tempera-
tures ranging from 750 to 900°C. The samples were placed 1n a quartz cruci-
ble for one run and dispersed on quartz wool for the other. As shown in Figure
2, both the limestone and dolomite achieved higher maximum surface areas, and
achieved these maxima at lower temperatures when the sample was dispersed on
wool. Dispersion therefore was assumed to play a positive role in surface area
development.
In the second case 2 g of each stone was again calcined at temperatures from
700 to 850°C for 30 minutes. All samples were now suspended on quartz wool
9-2
-------
with a still air atmosphere used in one run and flowing air used in the other.
Figure 3 shows that limestone achieves a higher maximum under the flowing regime
and that dolomite is tending to achieve its maximum at a lower temperature.
A flowing air regime, therefore, appeared to be beneficial in achieving higher
surface area.
The results of the last screening experiment are shown in Figure 4. In this case
2 g of material dispersed on quartz wool was calcined for 30 minutes at tempera-
tures from 700 to 850°C. In one run a flowing air regime was used, and in the
second run flowing nitrogen was used. Both gases had been dried. The nitrogen
atmosphere appears to have been beneficial in producing surface area in both
stones; the dolomite achieved a higher maximum, and the limestone reached its
maximum at a lower temperature.
The screening studies, despite their simplicity, held important implications for
later work. It seemed clear that characteristics of the raw limestones could sig-
nificantly affect surface area development and should be identified if possible.
The importance of particle dispersion and a flowing regime suggested that fluid-
ized bed calcination, already available commercially, might be adapted to the
production of high surface area sorbents. The remainder of the research Investi-
gated these preliminary findings.
Fluidized Bed Studies
A new suite of samples was required for the fluidized bed experiments since the
minus 44 pm material was too fine for fluidization and more costly for commer-
cial production than a coarser size range. Discussions with 11 me industry repre-
sentatives revealed that a minus 600 pm size fraction could be produced with
state-of-the-art crushing and screening equipment. To obtain this size fraction
for laboratory use, quantities of 2-5 cm stone were obtained from commercial lime
or limestone suppliers, crushed in a Bico jaw crusher, ground in a Bico disc mill,
and screened at 28 mesh (600 ym). Materials were then dried at 200°C for 8 hours
and riffled and/or coned and quartered to 15 g sample size for calcination.
Figure 5 shows an exploded view and stock components of the fused quartz fluid-
ized bed calciner designed to be operated in a Lindberg muffle furnace to tem-
peratures of about 950°C. A constant flow of either air or nitrogen was supplied
at 5 L/min, supplemented with a flow of 4 L/min, pulsed for 1 sec durations at 1
sec intervals to maintain fluidization.
Figure 6 shows the results of a suite of 15 g, minus 600 ym samples fluidized
in nitrogen for 30 minutes at temperatures ranging from 700 to 850°C. This
is assumed to be a base case in which conditions were optimized. All stones
achieved their maxima in the 750 to 800°C range, with the single dolomite reach-
ing a surface area of 69 m2/g. The limestone maxima ranged from 40 to 50 m^/g
with an average of 45 m2/g.
Figure 7 shows the results of a similar experiment, but with air instead of nitro-
gen as the fluidizing medium and with two additional stones from the screening
experiments included for comparison. The results are both more erratic than the
base case and somewhat unexpected. The screening experiments suggested that
higher surface areas would be produced in a nitrogen atmosphere, but during flu-
idized calcination in air some surface areas decreased and others increased rela-
tive to the results achieved with nitrogen. The average of 44 m^/g for the
limestones is not significantly lower than the base case.
9-3
-------
Figure 8 shows the results of reducing the flu1d1zat1on time to 15 minutes in a
nitrogen atmosphere. The data set must be regarded as incomplete since only the
Clifton and Maysville stones were TOO percent calcined at 850°C. The remaining
stones can be expected to gain additional surface area before reaching their
maxima at 100 percent calcination, but even at this point the limestones had
achieved an average surface area of 40 rn^/g.
CHARACTERIZATION
Calcined Sorbents
B.E.T. and TGA. The principal analytical techniques employed in this investiga-
tion for determining the sorbent properties of interest were the B.E.T. (Brunauer,
EnHnett, and Teller) method of surface area analysis and thermogravimetric analysis
(TGA) for calculating the degree of calcincation. A Micromeritics Flow Sorb 2300
B.E.T. system was routinely used for single-point surface area analysis. The
reader is referred to in-depth discussions on the nitrogen adsorption method of
surface area analysis in Gregg and Sing (3) for details of the technique. TGA
analyses for carbon dioxide (degree of calcination) were performed on a DuPont
Series 99 thermogravimetric analyzer and used primarily as a check on the rela-
tive accuracy of the surface area measurements. The expected behavior is that
surface area will increase with increasing calcination until 100 percent calcina-
tion is reached. From that point on, the sintering process will act to decrease
surface area by collapsing the internal pore structure. Raw TGA and B.E.T. data
for the fluidization experiments shown in Figures 6-8 are displayed in Table 1.
Another property of interest, and perhaps greater importance, is the distribu-
tion of surface area among the various pore sizes in the calcined sorbent. It
has been assumed that much of the surface area in the high surface area materials
is concentrated in the range of smaller pore diameters. If these pores are too
small, however, they may quickly become inaccessible to the SO2 molecules due to
blockage by the sulfation production. It seems reasonable to speculate, there-
fore, that the "ideal" sorbent will eventually be defined by some optimal combi-
nation of surface area and pore size.
Surface area as a function of pore diameter can be determined either by mercury
porosimetry or by multipoint analysis of desorption curves from the B.E.T. in-
strument. Again, the reader is referred to Gregg and Sing for details of this
analysis. To determine the utility of these methods, a sample of Longview lime-
stone calcined to about 50 mvg was analyzed on a Micromeri tics Model 2100E and
by a commercial laboratory using the mercury porosimetry technique. The results
of these analyses are shown in Figures 9 and 10, respectively. Extreme care
must be exercised in reading these plots since they cannot be interpreted as
normal histograms. The bars are of unequal width because they are determined
by the availability of data points on the horizontal axis, and extrapolation be-
tween these points is not appropriate. In Figure 9, for example, the wide 2
percent bar extending from 250 to 480 A* indicates that 2 percent of the total
surface area is found within this total range of pore diameters; not that 2
percent is found within each category of the range as in a normal histogram.
The B.E.T. plot in Figure 9 suggests that the surface area is concentrated in the
range of pore diameters from 80 to 140 A, while the porosimetry data in Figure 10
* 10,000 A = 1 m m
9-4
1
-------
shows the range of highest concentration to be 180 to 260 A. The utility of this
type of analysis appears obvious, but the required accuracy for the purposes of
this research and the available accuracy from the two techniques must be deter-
mined before it is extensively applied or firm conclusions drawn from results.
SEM and TEM. In an effort to adequately visualize the calcined products and par-
ticularly to demonstrate the pore structure that was being created, a research
grade Camscan Series 4 scanning electron microscope (SEM) with a maximum resolu-
tion of about 30 A was employed. Reagent grade calcite samples were calcined to
surface areas of from 12 to 28 rn^/g and viewed with the instrument. Figure 11
shows the results. Two Important observations can be made from the photographs:
1) the external expression of the pore structure appears as an irregular series
of fissures or cracks; and 2) there appears to be a directional nature to the
pore development, perhaps governed by the orientation of twin planes within the
crystal structure. If these observations prove to be applicable to other calcined
limestones, they have interesting implications. The implication of the fissure-
like nature of the pore structure may be that the assumption of cylindrical pores
used in current sulfation models and 1n calculating derived values from B.E.T and
porosimetry analyses is incorrect. The second observation leads to the interest-
ing speculation that, if calcination (pore development) has a preferred direction,
then so may sulfation. If this is true then stone crushed to a particle size of
less than the crystal size may calcine to particles having a uniform directional
orientation of both porosity and sulfation. This type of product may prove to be
more effectively sulfated than one consisting of particles containing numerous
randomly oriented crystals.
Figure 12 may lend support to the idea of a directional orientation to the por-
osity although its interpretation is not yet clear. It is a transmission elec-
tron micrograph (TEM) of one of the Maysville calcines. The crystals on the left
are hydrated due to exposure, while those on the right appear to be lime. The
dark areas may represent internal porosity preferentially oriented along the long
axis of the crystals.
Raw Stone
Petrography. At present two samples of each of the raw stones have been thin-
sectioned, stained to distinguish calcite from dolomite, and described petro-
graphically. Large amounts of time have not been expended in point counting or
statistically validating grain size. This may be warranted when the properties
of particular interest are determined. No correlation has been observed thus
far between petrographic characteristics and surface area development in the
calcined product. Nomenclature used in classification is after Folk.
Maysville stone (Figure 13) is a biolithite consisting of Tetradium coral frag-
ments replaced primarily by untwinned sparry calcite ranging in crystal size up
to 0.8 irm. Echinoid plates and spines, some partially replaced by chalcedony,
are also abundant. Pelleted intraclasts add to the framework while micrite and
pelleted micrite comprise the matrix. Dolomite rhombs are ubiquitous.
Longview limestone (Figure 14) is classified as an intrasparrudite or a pelspar-
ite. It consists of alternating thin (1.5 to 3.0 mm) layers of pelleted micrite,
pellets plus pelleted intraclasts, and intraclasts only, depending on the degree
9-5
-------
of sorting. Spaces between clasts are filled with twinned sparry calcite up to
1.0 mm in crystal size. No fossils are present.
Clifton stone (Figure 15) is a biomicrudite consisting largely of Rudistid
mollusc shell fragments. The matrix consists of micrite and some shell hash.
Generally untwinned sparry calcite up to 0.7 mm in crystal size is replacing
some of the shell material as well as the matrix. The stone is extremely por-
ous with as much as 15-20 percent open voids.
Round Rock limestone (Figure 16) shows two distinct lithologies. One is a bio-
pelmicrite with a mollusc shell fragment framework. Interstices are filled with
nearly equal portions of pelleted micrite and coarse (up to 1.00 mm) untwinned
spar. The other is approximately half fossiliferous intrasparite and half fossil-
iferous intramicrite. Foraminifera are abundant with some echinoid fragments.
Grains are generally rounded.
Genstar dolomite (Figure 17) also consists of two markedly different lithologies.
The first one consists only of intergrown crystals up to 0.1 mm with all sedimen-
tary features apparently eradicated by dolomitization. The second lithology is
a dolomitized pelmicrite with alternating layers of micrite and large pellets
(0.4 mm) in a micrite matrix. Some sparry material is found in pelleted layers
and as crack filling. Ostracod and bryozoan fragments are present.
Particle Size. Analyses of particle size distributions in the minus 600 um ma-
terial used for the fluidized bed experiments were performed using an Allen-
Bradley Sonic Siever and are presented in Table 2. Figure 18 shows a scatter-
plot of the median particle diameters versus the maximum surface areas achieved
in both air and in nitrogen. No significant correlation is found using this
limited amount of data although there is an indication that coarser sized
material tends to develop lower surface areas. It would be expected that, for a
given set of calcination conditions, an optimal particle size exists. Correla-
tions will be sought when additional data are available.
Elemental Analysis. Elemental contents of the minus 600 um materials were de-
termined using x-ray fluorescence and are presented in Table 3. Figure 19 shows
a scatterplot of the maximum surface areas achieved in air and in nitrogen ver-
sus iron content. A significant negative correlation is found with iron, and this
is the only element that produced a significant correlation. The interpretation
is not entirely clear in this case since all of the stones, to varying degrees,
must have been contaminated with iron during the grinding process.
CONCLUSIONS
The results of this research to date demonstrate that it is technically feasible
to produce high surface sorbents using a fluidized bed calcination technique at
the laboratory scale. The implication is that the commercially available fluid-
ized bed calcination units may be capable, perhaps with modification, of produc-
ing sorbents having high surface areas. Additional work will be done to deter-
mine the effect on surface area of reducing air flow to the minimum needed for
fluidization, and of using a simulated combustion gas for the fluidizing medium.
Characterization of calcined reagent grade calcite by SEM and TEM suggests a fis-
sure-like pore structure and a directional nature to the calcination controlled
by crystallographic properties. If these observations are borne out on other cal-
cined materials, they imply that the current assumption of cylindrical pores used
9-6
-------
in sulfation models and various analytical techniques is incorrect. They imply
further that, with the proper combination of crystal size in the limestone and
particle size in the material to be calcined, perhaps some advantage can be made
of the apparent directional nature of calcination.
Particle size and elemental analyses of the raw crushed stones show an apparent
significant negative correlation only between surface area and iron content. As
additional data is generated, other correlations will be sought. Petrographic in-
vestigations of^ the raw stones will be focused on the relative amounts of sparry
calcite present, as well as its crystal size and crystallographic properties such
as twinning.
9-7
-------
ACKNOWLEDGMENTS
Connie Turlington of Northrop Services, Inc., assisted by Linda Harry, was re-
sponsible for generating the laboratory data. Their efforts and technical input
have proven invaluable. Thomas Nuhfer of Carnegie-Mellon University provided the
SEM photos. Howard Wagenblast and Margaret Nasta of Mellon Institute directed
the TEM analysis. Limestone samples were donated through the cooperation of the
following companies: Commercial Minerals Company, Inc., Newark, CA; Chemical
Lime, Inc., Fort Worth, TX; Round Rock Lime Co., Blum, TX; Longview L1me Co.,
Saginaw, AL; Dravo Lime, Maysville, KY; and Genstar Cement and Lime Co., Hender-
son, NV. Daniel A. Textoris, Department of Geology, University of North Carolina,
Chapel Hill, NC, lent guidance in the petrographic interpretations.
REFERENCES
1. Private communication, R. H. Borgwardt (EPA/IERL-RTP) to D. A. Kirchgessner,
October 1984.
2. R. D. Harvey, R. R. Frost and J. Thomas. "Petrographic Characteristics and
Physical Properties of Marls, Chalks, Shells and Their Calcines Related to
Desulfurization of Flue Gases." EPA-650/2-73-044 (NTIS PB226321), Septem-
ber 1973.
3. S. J. Gregg and K. S. W. Sing. Adsorption, Surface Area and Porosity. New
York, NY: Academic Press, 1967.
4. R. L. Folk. Petrology of Sedimentary Rocks. Austin, TX: Hemphill's, 1968.
9-8
-------
SURFACE AREA (m2/g)
CONDITIONS: 850 °C
30 MINUTES IN AIR
NO FLOW
5g (-44/im) _
Figure 1. B.E.T. surface areas of -44 urn materials.
60
U>
N
£ 40
<
LU
cc
<
LIJ
o
<
LL
CC
V)
20
—
QUARTZ CRUCIBLE - STILL AIR - 2 g (-44/im)
QUARTZ WOOL - STILL AIR - 2 g (-44/im)
30 min
KAISER DOL
^JWOOL)
—
KAISER DOL
¦ — —1
(CRUCIBLE)
^^-""'ELDORADO LS ~
(WOOL)
ELDORADO LS
(CRUCIBLE)
1 1
750 BOO 850
TEMPERATURE (°C)
Figure 2. Results of calcination in crucibles
vs. calcination on quartz wool.
900
9-9
-------
60
OJ
QUARTZ WOOL - STILL AIR - 2 g (-44^m)
QUARTZ WOOL - FLOWING AIR - 2 g (-44/7m)
30 min
N
.I. 40
<
IXJ
DC
<
id
O
<
u.
fiC
D
(/)
KAISER DOL
IFLOWING)
20
ELDORADO LS
FLOWING
KAISER DOL
(STILL)
ELDORADO LS
(STILL)
700
750 800
TEMPERATURE (°C)
850
Figure 3. Results of calcination in still air
vs. flowing air.
60
CM
s
<
iu
AC
<
IftJ
o
<
u.
fiC
3
CO
40
QUARTZ WOOL - FLOWING AIR - 2 g (-44^/m)
QUARTZ WOOL - FLOWING N2 - 2 g (-44^m)
KAISER DOL
(Jl2>
20
ELDORADO LS
(N2)
! 30 min
KAISER DOL
(AIR)
ELDORADO LS
(AIR)
700
760 800
TEMPERATURE (°C)
850
Figure 4. Results of calcination in flowing air
vs. flowing nitrogen.
9-10
-------
Type "S" Thermocouple
12/5 Ball Joint
40m
40/35 Standard
Taper
40 60f,m
Coarse Frit
5 mm i.d Quartz
10/30 Standard
Taper
Figure 5. Fused quartz fluidized bed calcination unit.
D>
v.
<
LU
80
60
40
QUANTITY: 15 g (-600/vm)
LU
O
<
Li. 20
CC
if)
0
650
GENSTAR DOL
LONGVIEW LS
CLIFTON LS
MAYSVILLE LS
ROUND ROCK LS
AVQ. MAX S.A. FOR LS = 45 m2/g
700 750 800
TEMPERATURE (°C)
850
Figure 6. Results of calcination by fluidization
in nitrogen for 30 minutes.
9-11
-------
80
60
CM
E
<
UJ
< 40
UJ
O
20
(/)
QUANTITY: 15 g (-600/ym)
KAISER DOL
EL DORADO LS
GENSTAR DOL
LONGVIEW LS
CLIFTON LS
ROUND ROCK LS
MAYSVILLE LS
AVG. MAX S.A. FOR LS = 44 m2/g
0
650
700 750 800
TEMPERATURE (°C)
850
Figure 7. Results of calcination by fluidization
in air for 30 minutes.
80
QUANTITY: 15 g (-600^m)
60
40
20
0
650
AVG. MAX SJL FOR LS s 39m*/g
GENSTAP DOL
LONGVIEW LS
ROUND ROCK LS
MAYSVILLE LS
CLIFTON LS
700 750 800
TEMPERATURE (°C)
850
Figure 8. Results of calcination by fluidization
in nitrogen for IS minutes.
9-12
-------
a
LU
h-
<
o
o
CO
(/)
<
LU
O
<
z
<
O
LU
QC
O
CL
l/> £
-J
<
I-
o
LONGVIEW 750/30/AIR
BET
0 40 80 120 160 200 240 280 320
PORE DIAM. (A°)
b0 400 440 480 520
Figure 9. Percent of total surface area associated
with pore size increments by B.E.T.
LONGVIEW 750/30/AIR
MERCURY POROSIMETRY
mmm:
IXI LU
150 200 250
PORE DIAM. (A°)
450
Figure 10. Percent of total surface area associated
with pore size increments by porosimetry.
9-13
-------
Reproduced from
bcsi available copy
3 0 00R
Figure 11. SEM photographs of calcined reagent grade calcite
9-14
-------
Figure 12. TEM photograph of calcined Maysville limestone.
Figure 13. Photomicrographs of thin-sectioned
Maysville limestone (picture width = 2 mm).
9-15
-------
Figure 14. Photomicrographs of thin-sectioned
Longview limestone (picture width = 2 mm).
Figure 15. Photomicrographs of thin-sectioned
Clifton limestone (picture width = 2 mm).
9-16
-------
Figure 16. Photomicrographs of thin-sectioned
Round Rock limestone (picture width = 2 mm).
Figure 17. Photomicrographs of thin-sectioned
Genstar dolomite (picture width = 2 mm).
9-17
-------
* DOI.V FLUID tz ED 80 min
• ib. r m km and m n2
_ 80
CM
E
<
HI
c
<
UJ
o
<
u.
cc
3
(0
¦
X
<
70
60
50
40
30
230
250 270 290 310
MED. DIAM. OF RAW STONE (//m)
330
Figure 18. Scatterplot of maximum surface areas in air and
in nitrogen vs. median particle diameter of raw stone.
50
— 60
cr>
CM
<
LU
0c
<
LU
o
<
Li-
en
D
a>
x
<
2
40
30
i-WBBW
400
5"
¦i
¦
bwB
h3I
lldi«iiil
.. i'i:a....,^.J -j - t
600 800 1000
IRON BY XRF (ppm)
m
i
1200
Figure 19. Scatterplot of maximum surface areas 1n air
and in nitrogen vs. iron content of stone.
9-18
-------
Table 1
B.E.T.
AND TGA ANALYSES OF
FLUIDIZED
BED SAMPLES
Time
Temp.
May
'sville
CIi fton
Round
Rock
Longview
Genstar
El Dorado
Kaiser
Atm. (min.)
(°C)
TGA*
r BET**
TGA
BET
TGA
BET
TGA
BET
TGA
BET
TGA
BET
TGA BET
700
25
12.9
49
13.5
26
12.4
39
21.5
67
46.5
750
97
40.0
95
44.8
94
38.2
92
47.0
100
69.2
N? 30
800
100
39.8
100
49.9
100
39.3
100
49.8
100
64.2
850
100
33.6
100
44.8
100
31.9
100
45.7
100
57.6
700
32
17.0
58
17.6
32
11.0
42
20.1
85
33.1
13
24.8
6 7.6
750
95
39.8
67
35.1
67
31.7
94
43.1
100
64.8
100
58.1
59 37.1
Ai r 30
800
100
37.4
100
39.8
100
43.1
100
53.0
100
57.8
100
53.5
100 74.0
850
100
34.3
100
40.3
100
38.6
100
48.0
100
54.2
100
56.6
100 63.6
700
20
27.0
750
15
5.4
28
4.9
28
9.7
19
18.0
31
24.0
N2 15
800
94
24.0
58
16.0
77
20.0
94
22.0
42
23.0
850
100
39.0
100
28.0
95
45.0
96
47.0
98
59.0
* Percent calcination
** m^/g
-------
Tabl e
2
SIZE DISTRIBUTION OF
FLUIDIZED BED
SAMPLES*
Screen Size (ym)
Genstar
Maysville
Longview
CI ifton
Round Rock
Kai ser
El Dora*
595
5.4
7.6
4.2
3.8
3.9
1.6
3.9
354
40.1
38.0
36.1
30.5
36.6
29.4
33.1
250
12.3
14.7
12.7
16.2
15.8
21.2
15.0
149
16.5
17.1
19.6
20.4
17.9
21.2
17.4
88
8.6
9.7
11.3
11.9
10.9
11.7
11.7
44
9.7
9.0
9.2
10.4
9.3
11.0
12.5
<44
6.7
3.9
6.0
6.6
5.2
2.8
5.2
Median Diam.
311
325
278
255
300
250
267
* Expressed as percentage of material equal to or greater than screen size
but less than next larger screen size.
-------
Table 3
X-RAY FLUORESCENCE ANALYSES OF FLUIDIZED BED SAMPLES*
Elements
Genstar
Maysville
Longview
Cli fton
Round Rock
Kaiser
El Dorado
Magnesium (%)
11.8
1.49
0.44
0.35
0.23
9.01
0.36
Aluminum
720
1800
1000
460
970
64
480
Silicon
2200
7500
2200
1400
2300
1300
1500
Phosphorous
0.0
0.0
7
0.0
0.0
140
490
Sulfur
190
510
300
280
250
130
190
Chlorine
740
280
5
4
0.0
75
0.0
Potassium
300
600
240
29
16
1
29
Calcium (%)
23.4
39.8
39.0
41.1
40.7
20.6
37.0
Titanium
110
9400
57
19
31
28
55
Iron
2500
1100
510
810
1200
1600
450
Nickel
21
11
8
13
74
12
17
Copper
46
38
35
39
52
21
42
Zinc
14
0.0
0.0
0.0
0.0
2
4
Rubidium
54
73
72
79
75
5
70
Stronti um
82
31
180
160
370
93
230
Barium
70
120
0.0
18
0.0
0.0
0.0
* ppm except when noted otherwise.
-------
REACTIVITY OF CALCIUM-BASED
SORBENTS FOR SO2 CONTROL
J. A. Cole, J. C. Kramlich,
G. S. Samuel sen, W. R. Seeker, G. D. Silcox
Energy ana Environmental Research Corporation
18 Mason
Irvine, California 92718
ABSTRACT
Laboratory-seale control 1ed-temperature experiments were used to study aspects
of SO2 capture Dy limestone sorbents in a flame-gas environment. Experimental
parameters were sorbent type, temperature, residence time, and the effects of
mineral additives, or promoters, on sorDent reactivity. The data revealed that
isothermal capture is greatest at 1000°C, and that above 1000°C sintering of the
limestone can occur which reduces the sorbent utilization. High surface area
precalcined sorbents achieved moderately higher ultimate utilizations than their
parent carbonates, but their real advantage was more rapid sulfation at lower
temperatures where raw stones were limited by calcination. At 900 ana 1000°C
the time for calcination of carbonate sorbents was significant. Pressure
hyarated (type S) dolomite limes consistently achieved the highest utilizations.
The results suggest that--at ideal sulfation conditions (1000°C, isothermal
residence times greater than 1 second, no deactivation of the sorbent by coal
ash mineral s)--the best calcium utilizations achievable would be about 25-302.
with the raw limestone tested (Vicron 45-3), about 30-35% with the raw dolomite
tested, and about 40% with precalcined dolomite (precalcined to a surface area
of 60 m^/g) and with pressure-slaked aolomitic lime. The addition of 0^03,
alkali metal salts, and certain other promoters increased the utilization of
limestone. 0^03 effected a factor of 3.5 increase 1n utilization after
calcination at 1600-1700°C.
INTRODUCTION
Dry-limestone injection provides an economically attractive means for
controlling SO2 emissions in pulverizea-coaI-fired (p.c.) utility boilers. The
process is especially attractive as a retrofit for older boilers because of the
potentially low capital costs relative to other SO2 control technologies (1).
In addition, the raw material 1s readily available and relatively Inexpensive
(2). Early demonstration tests l'n the late 1960's and early 1970's, however,
met with little success (3) in that acceptable levels of SO2 removal could not
be achieved, even with a large stoichiometric excess of limestone. The present
effort was undertaken in order to investigate the conditions which will optimize
10-1
-------
S02 removal from flue gases by calcium-based dry sorbent injection. The
conditions which were explored included time, temperature, sorbent type and
preparation, and the effects of mineral additives on sorDent performance.
Previous efforts have examined the overall process of SOg capture within the
confines of the conditions present in full-scale systems (4,5). Both
calcination (6) and sulfation kinetics (7,8) have also been studied in
fundamental experiments and have Deen addressed theoretically (9-11). The
calcination of various limestones has been examined with regard to the
properties -- surface area, porosity, pore size, etc. -- which are most likely
to affect the reaction with SO2 (12-16). Lastly, the interactions of limestones
with other minerals (17, 18) which are found in coals and their effects on SO2
capture, Doth positive (19, 20) and adverse (14, 21), have been investigated in
recent years.
The present paper examines tne SO2 capture process in light of past efforts. It
aoes so under experimental conditions which simulated the thermal and chemical
environment of a p.c. utility boiler. The experimental objectives were to
provide conditions for sorbent injection which were representative of those in
large-scale systems, but which were well characterized, uniform, and
reproducible. Tne needs addressed by this work were:
$ To determine the so-called "reactivity" of a wide variety of sorbents.
Reactivity refers to the ability of a sorbent to uptake sulfur under
dispersed-phase high-temperature conditions which would exist in
utility furnaces. Tne reactivities of a number of sorbents including
high surface area precalcines were determined in a single experimental
system.
• To examine the reactions of calcination and sulfation in the dispersed
phase at the high temperatures and short residence times which are
representative of those encountered in p.c. utility boilers.
$ To study the effects of mineral matter interactions on the ability of
limestone to remove SC^-
• To observe physical changes which could be linked to either reactivity
or the ability of the sorbent to uptake sulfur.
The approach applied to these needs was to inject sorbents into the high
temperature environment of a laboratory gas flame. Both isothermal and non-
isothermal environments were provided from which the sorbents were sampled and
analysed to determine their physical and chemical properties. The sorbents were
sulfated both in-situ during calcination and external to the reactor. In the
external sulfation experiment, sulfation occurred in the dispersed phase, in an
environment which was independent of the reactor conditions, thereby decoupling
sulfation from heatup and calcination. An attempt was then made to link the
physical properties of the sorbents to their abilities to absorb sulfur under a
variety of conditions.
10-2
-------
EXPERIMENTAL
Apparatus ana Conditions
Flame Thermal Decomposition. Two experimental reactors were employed in this
p rog ram. TWe first is a Flame reactor which was used to examine flame
processing of sorbents at high temperatures (>1200°C) and short times (<250 ms).
This reactor (Figure 1) consists of a porous-Dronze-plug flat-flame Durner
downfired into a 10 cm x 10 cm stainless steel chimney with optical,
thermocouple, ana solids-sampling access. Limestones were injected into the
reactor through a 0.64 cm i.a. copper tuDe mounted axially through the burner.
Solids were entrained in a fuel/air premixture prior to injection in order to
ensure rapid heating to peak reactor temperature. Heat loss through the reactor
walls then resulted in a steep temperature dropoff. Injection of the solids was
accomplished with a Smith-type fluidized Ded feeder and off-take tuDe assemDly
(22, 23).
Reactivity toward SO2 was determined using a dispersed-phase quartz SO2
reactivity proDe. Samples of dispersed sorDent--now calcined in the flame
reactor, out not yet sulfated—were collected at the top of the proDe and
quenched slowly (10^ K/s) to 650°C. At this point, the calcined soroent sample
stream entered a reaction zone which was heated to 1100°C. In the reaction
zone, 6 percent SO2 was added and the sample stream allowed to react for 0.6 s.
The sample then was quenched and collected on a glass fiber filter at 130°C.
Isothermal Reactor. The second reactor is a dispersed-phase isothermal reactor
(ITR). rhe ITR provides a relatively long (up to 3.0 s) isothermal zone in
which soroent chemistry can be studied as a function of time, temperature, and
environment. This reactor is unique because it provides a large volume for
dispersal of sorDents at reasonable feed rates. This is necessary to permit
solids sampling for chemical analysis within practical time frames. The ITR
(Figure 2) is an electrically heated, drop-tube furnace into which flame gases
are downfired. It has a heated length of 90 cm and accommodates a 10 cm
diameter alumina reaction tuDe. The ITR nas a maximum wall temperature of
1500°C.
The test gas for the ITR is produced Dy a burner identical to the one used with
tne flame thermal decomposition reactor. Limestone was injected into the ITR
along the axis of the reactor through the burner. The limestones were injected
from a 1.1 nni i.d. tube, which produced a turbulent jet, effectively and rapidly
dispersing the materials over a wide cross-section of the reactor. Residence
times and heating rates of the particle streams were calculated based on
confined jet mixing theory (24) and convective and radiative heat transfer
calculations. Heating rates were on the order of 10^ K/s and total (end of
reactor) residence times of 1.2 - 1.6 s were available for the experiments
described here. Solids sampling from the ITR was accomplished with an
isokinetic water-cooled stainless steel probe. Sorbents were quenched rapidly
in the probe and collected on a glass fiber filter located at the base of the
probe. Sampling times and temperatures were selected so that sulfur uptake by
the sorbent on the filter holder was insignificant. The probe is 1.2 m long and
enables sample collection within 40 cm of the soroent injection location, thus
permitting sampling of the dispersed soroent after short (— 250 ms) residence
times.
10-3
-------
Temperature profiles in the ITR are shown in Figure 3. Temperatures were
measured using a 0.025 run diameter type S thermocouple. Radiation corrections
to the thermocouple readings were applied only for the flame thermal
decomposition reactor and for non-isothermal conditions in the ITR. Otherwise
the corrections were smaller than +5 K. The profiles in Figure 3 are all for
nyarogen/air flames. Methane was use"d for flame temperatures aDove 1350°C.
SorDent Preca1cination Apparatus. In order to generate high-surface-area
precalcinea materials tor testing in Doth the TDR and ITR, the transpirated Ded
calciner shown in Figure 4 was developed. The apparatus in Figure 4 is a 20-cm
diameter stainless-steel can placed inside a large box furnace. The can is
fitted with a heavy lid which has a single hole for thermocouple access and to
allow sweep gas and CO2 to escape. Limestone is spread in a thin bed on a 400
mesh stainless steel screen. Beneath the screen 1s a spiral made from 6.4 mm
stainless steel tubing. The spiral has small holes drilled on the underside and
acts as a flow distributor for the transpi ration gas. The transpi ration gas,
N2, is preheated by passing it through a long coil of stainless steel tuDing
located inside the box furnace.
Operation of the transpirated bed calciner consists of passing nitrogen through
the Dea of powdered limestone at high temperature. The mass of limestone and
flow of nitrogen are balanced to provide the shortest practical calcination time
at the lowest possible temperature. Typical operating conditions are 700°C, 60
g limestone, 0.55 1/s N2- Carbon dioxide evolved resulting from calcination 1s
swept away from the limestone by the nitrogen, thereby lowering the local CO2
concentration and accelerating the calcination rate. For both Vicron 45-3 and
D3002 dolomite, the optimal calcination time was near 75 min.
Soroents
Limestone samples were characterlzeo both before and after injection using
several analytical techniques which are listed in Table 1. Most of the raw
materials were analysed for chemical composition, particle size distribution,
and specific surface area. Samples collected from the reactors were analysed
for carDon (carDonate), hydrogen (hydroxide), total sulfur (sulfate), and total
calcium. From these measurements the extent of calcination and calcium
utilization (percent calcium as sulfate) were determined for most samples.
The soroents which are discussed here are listed 1n Table 2 along with some
physical and chemical properties. Vicron 45-3 and D3002 are, respectively, the
baseline calcite and dolomite in this study. They are comparable in mean size
(see Table 3 for size data) and specific surface area and both are high-purity
minerals. It was from these limestones that the V40 and D60 precalcines were
produced. Surface areas indicated for the precalcines are typical. However,
the materials were produced 1n small batches, and surface areas varied between
batches. The Type S material Is a pressure-slaked dolomitic lime manufactured
by Warner.
-------
RESULTS AND DISCUSSION
Reactivity of Flame Treated SorDents
Figure 5 displays histograms of the reactivity for eight sorDents flame-treated
in the thermal decomposition reactor at peak temperatures of a) 1200°C and b)
I50Q°C and then sulfated in the dispersed-phase reactivity probe at the standard
0.6 s/6% SO2 condition. For all sorDents tested, the reactivity decreased
precipitously when the flame temperature was increased. Previous measurements
have indicated that this can De attriDuted to decreased specific surface area
due to more rapid sintering at increased temperature. However, the relative
reactivity of the flame-treated sorDents is insensitive to flame temperature;
only tne D3002 dolomite and the precalcine changed positions at the higher
temperature. The limestones were generally the less reactive after flame
treatment, followed Dy hydroxides; and the dolomites were the most reactive. A
single precalcine (produced from Vicron limestone with a specific surface area
of 34 m'/gm) was tested for reactivity after flame treatment. This precalcine
was found to De more reactive than the raw stone from which it was produced.
Thus precalcining does have the potential for providing an increased reactivity
that will not De completely lost when flame-treated for short times (<200 ms).
Isothermal Reactor
In order to investigate simultaneous calcination and sulfation of sorDents for
longer times, as might occur for sorDents Injected in the upper furnace, tne
isothermal reactor (ITR) was employed. Calcium utilization was measured as a
function of residence time in the ITR for five sorbents at temperatures of 900,
1000, 1100, and 1200°C. In each case the initial SO2 concentration in the
Durned gases was 3600 ppm, and the sorDent feed rate was adjusted to ensure a
calcium-to-sulfur ratio (Ca/S) less than 1.0 so that the measured calcium
utilization would not be affected Dy SO2 depletion in the reactor.
Effects of Residence Time and SorDent Types. At 900°C (Figure 6) the capture
levels ot the precalcines and Type S are all greater than those of the raw
sorDents, D3002 and Vicron 45-3. In part, this is because the raw sorDents
calcined slowly at this relatively low temperature, which reduced the availaDle
calcium. The Type S sorDent is suspected as having a lower calcination
temperature as well as a less endothermic calcination reaction than the raw
sorbents since it is a hydroxide. It may in fact calcine so quickly at 900°C
that the calcination reaction presents no impediment to sulfation. The
preca lines (D60 and V40) are not expected to sinter (lose surface area) rapidly
at this temperature. In surrmary, the precalcines begin to sulfate more rapidly
than the raw sorDents (within 0.5 s) because they do not experience a lag time
for the calcination reaction; they reach a higher ultimate utilization (at 1.0-
1.5 s) Decause their initial high surface areas do not sinter so rapidly at this
temperature, thus remaining available for sulfation.
At 1000°C (Figure 7) the relative order of reactivity has changed to: Vicron
45-3 < V40 < D3002 < D60 < Type S. This reflects a large increase in the
relative reactivity of D3002. At 1.5 s the calcium utilization of D3002 is
approaching that of the D60 precalcine; and, from 0.75 s on, the V40 maintains
aDout 10 percent greater utilization than Vicron 45-3. Both of the raw sorbents
exniDit some delay in SO2 uptake, again due to calcination, but it is not as
severe as that experienced at 900°C. In summary, the precalcines reach somewhat
10-5
-------
hi g he r ultimate utilization levels than the raw sorbents Dut the precalcines
achieve those levels in a substantially shorter time. All five sorDents display
a dramatic increase in reactivity Detween 900 ana 1000°C; evidently, the
improvement in sulfation kinetics with this increase in temperature more than
compensates for any increase in the rate of sintering (deactivation).
Temperature Effects. Figure 8 summarizes the sulfur capture of the five
sorDents as a function of temperature. The data shown in Figure 8 were taken
from utilization profiles (analogous to Figures 6 and 7) at the residence time
of 1.0 s. There is very little uncertainty associated with the ranking in
Figure 8 Decause the slopes of the calcium utilization profiles all were shallow
at 1.0 s. What has not Deen taken into account is the delay of the onset of
sulfation for Vicron 45-3 and D3002 due to slow calcination at 900 and 1000°C.
Accounting for the delay would alter the shapes of the temperature/utilization
curves In Figure 8 somewhat; however, it would not De reflective of the ultimate
result of low-temperature injection into a p.c. utility Doiler where calcination
times may De a factor.
The most significant aspect of Figure 8 is the appearance of a maximum in the
utilization achieved as a function of temperature. The location of the true
maximum appears to be very near 1000°C Dut may De different for each sorDent.
The maximum 1s thought to De a result of the tradeoff Detween sintering and
reaction kinetics. It is interesting that the optimum temperature is the same
for five different sorDents.
Sintering Rates. A grain model was used to estimate sintering rates for D60 and
V4U based on the ITR sulfation data. The grain model is a mathematical model
aescriDing the sulfation of sorDent particles, assuming that the individual CaO
grains within the particle are reacting according to shrinking core theory
(25,26). According to the model, the rate of sulfation at any selected level of
sorDent utilization is dependent upon the initial surface area (at zero
utilization). By using the model to calculate the effective initial area for
V40 ana D60 at the various utilizations (at the various residence times) shown
in Figure 7 for I000°C (and in the analogous curves for 1100°C), it was possible
to estimate how the initial surface areas of these precalcines were disappearing
due to sintering as a function of residence time in the ITR. The open symbols
in Figure 9 show the results of these sintering rate calculations. As shown in
the figure, the effective initial surface area decreases rapidly due to
sintering for Doth precalcines, especially at the higher temperature. For
example, at 1IQ0°C, the precalcines have lost half their effective initial
surface area in less than 0.25 s. The apparent sintering at 1000°C is not as
severe. For example, almost 1.0 s is required at the lower temperature in order
to lose half of the initial precalcine surface area; significant sulfation could
occur within that time before the area is lost.
In addition to the physical loss of BET surface area due to thermal sintering,
there would be a loss of reactive surface area due to sulfation (as reactive CaO
sites were converted to CaS04). The closed symDols in Figure 9 show the
calculated loss of "reactive surface area" due to sulfation as a function of
residence time, based upon the sulfation levels measured in the ITR experiments
(e.g., Figure 7). As indicated by the comparison of the closed and open symDols
in Figure 9, the loss of area due to thermal sintering 1s much more severe than
the loss of reactive area due to sulfation of CaO sites. (The sulfation
10-6
-------
reaction might also reduce the BET surface area by plugging the interstices
between grains, as the CaO structure expands to become CaS04- The grain model
does not account for such sulfate plugging. In fact, the BET area loss
indicated by the open symbols in Figure 9 is due not only to thermal sintering,
but also to sulfate plugging of the interstices or pores.
Sorbent Reactivity Promoters
Figure 10 shows calcium utilization for Vicron 45-3 injected alone, and mixed
with 6 percent-by-weight (>203, into the ITR. For these tests, the ITR was
operated at a constant furnace temperature of 11Q0°C, out the flame temperatures
were varied as shown, resulting in a ramped temperature profile from flame
temperature to 1100°C. All tests employed 3600 ppm SO2. As expected, the
utilization of Vicron 45-3 decreased with increasing flame temperature. This
reflects both a decrease in surface area upon calcination, and in some cases, a
snorter residence time in the sulfation window where sulfation will occur with
reasonable kinetics (approximately 1250 to 1000°C). With the Vicron /C r 2O3
mixture, however, the utilization initially increased as the flame temperature
was increased. Subsequently, the utilization decreased until, at a 1950°C flame
temperature, the utilization was nearly equal to that of Vicron 45-3 alone. By
comparing the utilization of Vicron 45-3 with that of the Vicron/Cr203 mixture,
enhancement factors were obtained as shown in Figure 11. The enhancement factor
is a relative increase in utilization due to the addition of (>203. The
enhancement is greatest, a factor of 3.5, at 1600-1700°C flame temperature.
Both the temperature and the magnitude agree with data obtained in a bench-scale
boiler simulator furnace (27).
Additional Minerals. Fourteen additional minerals were examined in a series of
screening tests 1n order to determine their potentials as sorbent reactivity
promoters. The materials, in 5 percent-by-weight mixtures with Vicron 45-3,
were exposed to 3600 ppm SO2 under the 1100°C isothermal condition as well as
with a 1360°C flame fired into the ITR at a furnace temperature of 1!00°C.
Figure 12 shows the results of isothermal tests at 1100°C in bar-graph form.
Tne open section of each bar is the utilization achieved 1n 0.92 s; the shaded
portion represents the additional utilization up to 1.4 s. The horizontal lines
are the averages of four replicates of the utilizations measured for unpromoted
Vicron 45-3 at the two residence times. Down-pointing vertical arrows adjacent
to the data for hJa2SO4, K2SO4, and M0S2 show what their calcium utilizations
would be if the sulfur initially present in the additives remained with the
additive, and was not released by the additive and captured Dy the calcium. For
the U2CO3 mixture, the utilization measured at 1.4 s was lower than that at
0.92 s.
Every additive tested, except M0S2, caused a net increase 1n utilization
(compared with unpromoted Vicron) after 1.4 s. The magnitudes of the increases
are not as great as were observed with O2O3; however, all of the results are
above the 95 percent confidence limit based on the four replicate samples of
Vicron 45-3 collected at 1.4 s.
The same promoters were tested under nonisothermal conditions using a 1360°C
flame with the ITR still at 11Q0°C. Results of these tests are shown in Figure
10-7
-------
13. The calcium utilization Dy Vicron 45-3 was considerably lower at this
conaition than at 1I00°C. This was due in part to thermal deactivation Dut also
stems from the substantially shorter time that the sorDent had in the sulfation
window- The total residence time of the sorDent in the flame was 0.6 s;
however, much of this time the temperature was above 1250°C. Calcium
utilization Dy many of the mixtures was quite high compared with Vicron 45-3.
Only three additives, T102> M0S2, ana ^2^5* fiad no beneficial effect on calcium
utilization within 95 percent confidence. After correction for its initial
sulfur content, utilization Dy the M0S2 sample actually lay below the 95 percent
confidence interval for the five Vicron 45-3 samples collected. Platinum and
V2O5, Doth oxidation catalysts, had little effect. The alkali metal salts as a
group showed the most promise as promoters; and lithium, the lightest alkali,
produced the greatest enhancement.
CONCLUSIONS
The reactivity of f 1 ame-treated sorbents was found to De strongly affected Dy
Doth temperature and sorDent type. The results of this work, showing a loss of
reactivity with increasing calcination temperature, were consistent with results
seen in a prior study (28). The general order of sorbent reactivity, in terms
of calcium utilization, was: dolomites > hydroxides > calcites.
For simultaneous calcination and sulfation under lower temperature isothermal
conditions, precalcines ana a pressure-hydrated dolomite had the greatest
initial reactivity. At longer times and higher temperatures, however, the
advantages of precalcines diminished. In general, dolomitic materials were more
reactive than calcitic stones; and the most reactive material tested was a
pressure-sl aked dolomitic lime. The advantage of the dolomitic lime may have
Deen due, in part, to a small mean particle size (aDout 1 m mass mean, compared
to aDout 10 m for the calcite and dolomite) suggesting a need for Detter
characterization of sorbent size and size distribution. For all of the sorDents
tested, the optimum temperature for isothermal SO2 capture was 1000°C. The
precalcines sintered rapidly at higher temperature, offsetting any increased
reaction rate and diminishing their reactivity.
SorDent reactivity was enhanced Dy the addition of several mineral additives.
The greatest improvement was seen with ^03. An injection temperature of 1600-
1700°C induced the maximum increase in capture with 0^03, relative to
unpromoted limestone at that temperature. Alkali metal salts also promoted SO2
sorption Dy lime. With the alkalis, however, a suDstantial improvement in
utilization was oDserved only by using an injection temperature of 1360°C.
However, the promoted utilizations at 1360°C were still generally less than
unpromoted utilization at 1100°C. Enhancement of SO2 sorption by alkalis
increased with decreasing formula weight of the additive.
ACKNOWLEDGEMENTS
This work was supported Dy the U. S. EPA under Contract 68-02-3633. D. B.
Henschel was the EPA Project Officer. The efforts and contriDutions of EER
staff, R. K. LaFond, T. C. Grogan, and R. D. Blethen, to this study are
gratefully acknowledged.
10-8
-------
REFERENCES
1. M. E. Kelly ana S. A. Shareef. Second Survey of Dry SO2 Control Systems,
Durham, NC: Radian Corporation, EPA-600/7-81-018 (NT IS PB81-157919),
February 1981.
2. R. S. Boynton. Chemistry and Technology of Limestone. 2nd ed., New York:
J. Wiley ana Sons, 1980.
3. R. C. Attig ana P. Seaor. Additive Injection for Sulfur Dioxide Control,
A Pilot Plant Study, Alliance, OH: BaDcock and Wilcox, APTD-1176 (NT IS
PB226761), March 1970.
4. R. W. Coutant, et al. Investigation of the Reactivity of Limestone and
Dolomite for Capturing SO2 from Flue Gas, Columbus, OH: Battelle Columbus
Laboratories, APTD-0802 (NT IS PB204385), OctoDer 1971.
5. Case, et al. Testing and Evaluation of Experimental Wall-Fired Furnaces
to Determine Optimum Means to Reduce Emissions of Nitrogen ana Sulfur
Oxides, Irvine, CA: Energy ana Environmental Research Corporation, Final
Report, EPA Contract 68-02-3921.
6. R. H. Borgwarat. "Calcination Kinetics and Surface Area of Dispersed
Limestone Particles," AIChE Journal, Vol. 31, No. 1, January 1985, pp.
KJ3-111.
7. R. H. Borgwarat. "Kinetics of the Reaction of SO2 with Calcined
Limestone," Environmental Science and Technology, Vol. 4, No. 1, January
1970, pp. 59-6X
8. R. H. Borgwarat and R. D. Harvey. "Properties of Carbonate Rocks Related
tc SO2 Reactivity," Environmental Science ana Technology, Vol. 6, No. 4,
April 1972, pp. 350-60"
9. S. K. Bhatia and D. D. Perlmutter. "A Ranaom Pore Moael for Fluia-Solia
Reactions: I. Isothermal, Kinetic Control," AIChE Journal, Vol. 26, No. 3,
May 1980, pp. 379-86.
10 S. K. Bhatia ana D. D. Perlmutter. "The Effect of Pore Structure on
Fluid-Solid Reactions: Application to the S02-Lime Reaction," AIChE
Journal, Vol. 27, No. 2, March 1981, pp. 226-34.
11. S. K. Bhatia and D. D. Perlmutter, "A Random Pore Model for Fluia-Solia
Reactions: II. Diffusion ana Transport Effects," AIChE Journal, Vol. 27,
No. 2, March 1981, pp. 247-54.
12. G. H. McClellan. Physical Characterization of Calcined and Sulfatea
Limestones. Presented at The Fourth Dry Limestone Symposium.
Gilbertsville, KY, June 1970.
13. R. D. Harvey. Petrographic and Mi neral ogical Characteristics of
Carbonate Rocks Related to Sulfur Dioxide Sorption in Flue Gases, Urbana,
IL: Illinois State Geological Survey, APTD-0920 (NTIS PB206487), July
1971.
10-9
-------
14. D. R. Glasson. "Reactivity of Lime ana Related Oxides," Journal of
Applied Chemistry, Vol. 17, April 1967, pp. 91-6.
15. D. R. Glasson ana P. O'Neill. "Reactivity of Lime ana Related Materials
with Sulphur Dioxide," International Conference on Thermal Analysis, Vol.
1, 1980, pp. 517-22.
16. D. Beruto, et al. "Characterization of the Porous CaO Particles Formea
by Decomposition of CaC03 ana Ca(0H)2 in Vacuum," Journal of the American
Ceramic Society, Vol. 63, Nos. 7-8, July-August 1980, pp. 439-43.
17. F. E. Huggins, et al. "Correlation Between Ash-Fusion Temperatures and
Ternary Equilibrium Phase Diagrams," Fuel, Vol. 60, No. 7, July 1981, pp.
577-84.
18- L. J. WiDberley and T. F. Wall. "Alkali-Ash Reactions and Deposit
Formation in Pulverizea-Coal-Firea Boilers: Experimental Aspects of
Soaium Silicate Formation ana the Formation of Deposits," Fuel, Vol. 61,
No. 1, January 1982, pp. 93-9.
19. R. T. Yang and M. T. Shen. AIChE Journal, Vol. 25, No. 5, May 1979, pp.
811-9.
20. G. C. Frazier and E. J. Baain. Chemically Improved Limestones for
Fluidizea Bea Coal ComDustion. Presented at The Meeting of the Central
States Section of the Combustion Institute. Lexington, KY, March 1983.
21. D. C. Baker ana A. Attar. "Sulfur Pollution from Coal Combustion.
Effect of the Mineral Matter Components of Coal on the Thermal
Stabilities of Sulfated Ash ana Calcium Sulfate." Environmental Science
and Technology, Vol. 15, No. 3, March 1981, pp. 288-93";
22. R. J. Hamor and I. W. Smith. "Fluidizing Feeders for Fine Particles at
Low Stable Flows." Fuel, Vol. 50, 1971, pp. 394-404.
23. R. A. A'ltenkirch, et al . "Fluidized Bed Feeding of Pulverized Coal,"
Powder Technology, Vol. 20, 1978, pp. 189-96.
24. M. A. Fiela, et al. Combustion of Pulverized Coal, Leatherheaa, England:
British Coal Utilization Research Association, 1967.
25. M. Hartman and R. W. Coughlin. "Reaction of Sulfur Dioxide with
Limestones ana the Grain Moael," AIChE Journal, Vol. 22, No. 3, May 1976,
pp. 490-498.
26. M. Hartman ana 0. Trnka. "Influences of Temperature on the Reactivity of
Limestone Particles with Sulfur Dioxide," Chemical Engineering Science,
Vol. 35, 1980, pp. 1189-1194.
27. B. J. Overmoe, et al. Boiler Simulator Stuaies on Sorbent Utilization
for SO2 Control. Proceedings: 1st Joint Symposium on Dry SO2 and
Simultaneous S02/N0X Control Technologies, San Diego, CA, November 1984.
28. J. A. Cole, et al . Fundamental Stuaies of Soroent Calcination and
Sulfation for SO2 Control from Coal-Firea Boilers, Irvine, CA: Energy and
Environment Research Corporation, Final Report, EPA Contract
68-02-3633, October 1983.
10-10
-------
Table 1
ANALYTICAL PROCEDURES
PROCEDURES DETERMINATION
Brunauer, Emmett, Teller Specific Surface Area
^2 absorption isotherm
Perkin-Elmer 240B Carbon, hydrogen determination
Extent of calcination,
hydrati on
Leco SC32 Instrumental
Sulfur Analyser
ASTM D2795
Cnelometric titration
Sedigraph (X-ray
sedimentation)
Mercury instrusion
pori slmetry
distribution
Tot_dJ Sul_fur
(S , SO3 , SO4 , S)
Total Calcium
Particle size distri-
bution, mean size
Pore size distribution,
true porosity, porosity
10-11
-------
Table 2
PHYSICAL AND CHEMICAL PROPERTIES OF LIMESTONE SORBENTS
Material
Vicron 45-3
Dolomite (Pfizer)
D3002 Dolomite
D60 Precalcine
V40 Precalcine
Type S (Warner)
M and M
Limestone
German
Hyaroxi ae
RWE
Hyaroxi ae
Fi sher
Hyaroxi ae
Composi tion
CaC03
CaC03*MgC03
CaC03*MgC03
CaO«MgO
CaO
Ca(0H)2*Mg(0H)2
CaC03
Ca(OH)2
Ca(OH) 2
Cat OH )2
Mean
Size
34
12
1.0
6
12.5
Surface
Area
m2/g
0.6
0.9
0.54
60.67
41.45
18.20
1.1
14
26
13
Chemical Analysis,
wt%
Ca
Mg
39.0 0.49
21.0 12.1
24.8 11.3
28.0 15.9
54.0 0.3
10-12
-------
TaDle 3
PARTICLE SIZE DISTRIBUTIONS OF RAW SORBENTS
Equivalent Spnerical Mass Percent Finer
Diameter, Vicron 45-3 D3002 Dolomite
80 100 100
50 100 99
30 95 97
20 80 80
10 46 42.5
8 37 35
5 24 21
3 14.0 13.5
2 8.5 10
1 3.0 5.5
0.8 2.5 4.5
0.5 1.0 3.0
0.3 — 1.0
10-13
-------
a fU« lorbon;s
Sort»*i**. 1n;«;tor tgce
Hart thtrtM;
SKwcGSltlon reactor
su'fur
iMc cs injection
HeiCtmiy jnj&e
ft !
/!«t flaw
_ -"Im* t*e*te4 ior&#i*s
Cu«!"t{ prcoe
Heee renvi
LS
I
I
,.1'cae^g 'urnace
¦^3 T®^S* M«"VCy»lt» filter
Figure 1. Schematic of the dispersed-phase SO2
reactivity probe as installed in the
flame reactor.
Sorbent 'njectior.
Reaction tube
Water-csoled probe
Filter housing
Figure 2. Schematic of the isothermal reactor
with the burner and water-cooled probe installed.
10-14
-------
o
e
0)
J-
3
4->
fO
S_
Ol
CL
E
-------
a) 1200QC
Injection
Figure 5. Reactivity ranking of eight sorbents injected into the
thermal decomposition reactor at 1200 and 1500°C. Sulfation zone
(reactivity probe) conditions were 1100°C, 0.6 s, 6% SO^-
10-16
-------
50
4->
| 40
5-
QJ
O.
c- 30
c
4->
(O
M
r 20
+¦>
l 10
U
13
<_>
a
i 1 r
O Type S Warner
A D60
_ o D3002
A V40
~ Vicrori 45-3
- Air Flame
•900°C
3600 ppm S0o
O -
0.25 0.50 0.75 TTO 1725f75 T775
Residence Time, s
Figure 6. Calcium utilization profiles for five sorbents
at 900°C.
50
£ 40
30
20
10
0
1000°C
- Air Flame
- 3500 ppm SO2
_L
I
OType
AD60
O 03002
4 V40
~ Vicron 45
Warner
0.25 0.50 0.75 1.0 1.25
Residence Time, s
1.5
Figure 7. Calcium utilization profiles for five
sorbents at 1000°C, 3600 ppm S0_
10-17
-------
3
T-
u
to
o
10
& D60
• D3002
a V40
¦ Vicron 45-3
1.0s
900 1000 1100
Temperature, °C
1200
Figure 8. Relative levels of calcium utilization
at 1.0s by five sorbents as a function of isothermal
reaction/calcination temperature.
i-
CD
i-
O)
Q_
rc
OJ
i-
-------
-------
20,
ID
a. 15
A
C
o
ro
¦S 10
.
LJ>
n4
II i1 ¦ i. 11
R.i inn ¦ i ¦
Figure 12. Calcium utilization by Vicron 45-3 in the presence of promoters
under isothermal conditions in the ITR. 1100'C, 3600 ppm SO2• The open
portion of each bar presents utilization achieved in 0.92 s. The shaded
portion shows the additional utilization at 1.4 s. For Li^CO^ the data
at 0.92 s were hiqher than at 1.4 s. Arrows adjacent to Nij-SO. and MoS,,
designate the correction for the sulfur content of the additives.
I
10-20
-------
20
C
0>
u
i-
a;
a.
15
ra
rsi
£ 10
E
3
O
to
cU
1
i
If)
O
O, CN1
U i
<->|z
ro
o
o
ra
CO
o
o
ro
o
o
X
O
o
CNJ
o
CO
CNJ
ro
^r
o
CO
CNJ
CO
o
o
uo
CNJ , UO
O • CO o
S- I O • csj
us: >
LD
o
a.
5
o
CNJ
a.
o
ra 1
5
t
LD
«3"
C
21
-------
BENCH SCALE EVALUATION OF SULFUR-SORBENT REACTIONS
D. M. Slaughter, G. D. Silcox,
P. M. Lemieux, Gerry H. Newton, and D. W. Pershing
Department of Chemical Engineering
University of Utah
Salt Lake City, Utah 84112
ABSTRACT
High temperature, isothermal data on calcination and SO2 capture were obtained
as a function of temperature, residence time, and Ca/S molar ratio for a wide
variety of sorbents including limestones, dolomite, and slaked limes. The
calcination results include extent of both calcination and surface area and
define the relationship between thermal environment and sorbent characteristics.
The experimental sulfation data indicate that sulfur capture is strongly depen-
dent on the general class of sorbent. Reaction zone temperature was also found
to critically influence the overall effectiveness of sulfur capture by sorbent
injection; as the local temperature increases, the rate of heterogeneous chemi-
cal reaction and diffusion increase but they are ultimately compensated for by a
decrease in initial sorbent surface area due to desurfacing during flash calci-
nation. The results from the experimental studies are compared with theoretical
predictions using a combined diffusion/heterogeneous chemical reaction model
which was developed based on a grain formulation.
BACKGROUND
Introduction
Considerable attention is being focused on the control of SO2 emissions from the
combustion of sulfur-containing coals. The desire to reduce SO2 emissions stems
partly from their suspected role in acid rain formation. The addition of
pulverized, dry, calcium-containing sorbents that react with SO2 in the boiler
or furnace to produce a solid product is potentially an economically attractive
control method. The overall reaction associated with this process under oxi-
dizing conditions is:
CaO + S02 + j 02 = CaS04 (1)
Extensive early investigations of this control technique(l-4) defined the
overall characteristics of the process, but yielded some conflicting and
inconclusive results as to its effectiveness. These studies made clear the need
for additional high temperature calcination and sulfation data with well-
characterized sorbent particles.
The elementary reactions involved in the process represented by Equation (1) are
11-1
-------
unknown. Previous experimental studies(2) have indicated a weak dependence on
oxygen concentration. The effect of 502 concentration is complicated by the
presence of coupled diffusion and heterogeneous reaction steps; apparent orders
of one(5), one-half(2), and zero(6) have been reported.
Purpose and Scope
The overall objective of the present study was to obtain high temperature sulfur
capture data under conditions where the thermal environment and the sorbent pro-
perties were well-characterized. A 65 kW refractory-walled reactor was used to
obtain isothermal capture data as a function of temperature for calcitic and
dolomitic sorbents in both carbonate and hydroxide forms. The results of these
experiments were compared to a grain model which included pore and product layer
diffusion as well as heterogeneous chemical reaction. Sulfation kinetics were
assumed to be zero order with respect to 502-
EXPERIMENTAL SYSTEMS
The isothermal reactor which was used to obtain the calcination and sulfation
data is illustrated in Figure 1. The refractory-walled chamber is 3.0 m high
with an inside diameter of 15 cm. The main natural-gas-fired burner is attached
to the horizontal extension at the bottom of the furnace. This 65 kW burner
supplies the hot combustion products which flow vertically upward within the
reactor. An independent wall and gas temperature control is provided by a pair
of 30 kW natural-gas auxiliary burners which are attached to the upper portion
of the furnace and fired downward in annular channels around the main combustion
chamber. All natural-gas and air streams supplied to the furnace are precisely
metered with high accuracy rotameters. SO2 is metered into the inlet gas stream
to allow independent control of sulfur partial pressure.
The sorbent was metered with a K-tron twin-screw auger feeder, and was
transported through an eductor to a water-cooled sorbent injector located at the
bottom of the vertical reaction zone. A "showerhead" injector design was used
to ensure rapid mixing of the sorbent with the combustion gas stream; both cold
flow visualization and hot tracer studies were used to verify the extent of
mixing.
The furnace exhaust was continuously monitored for CO, CO2, O2, and NO. SO2
concentrations were continuously determined by withdrawing a gas sample through
a controlled temperature, stainless steel, phase discrimination probe and
pumping it to a DuPont UV 400 Photometric analyzer. Sample conditioning prior
to the instrumentation consisted of two glass-wool filters, an electrically
heated line to maintain the sample at 130" C, two Perma-pure dryers to selec-
tively remove water from the flue gas, and a 60 ym, stainless steel filter.
Solid sorbent samples for subsequent surface area analysis were collected with a
stainless steel, controlled temperature probe followed immediately by a high
temperature filter. During the sampling, all parts of the system were main-
tained above 230* C to minimize hydration. At the conclusion of the sampling
sequence, N2 was forced through the system until the filter temperature was
reduced to ambient and all samples were stored under nitrogen. Both the phase
discrimination SO2 probe and the solids collection probe were located at the top
of the furnace just above the end of the controlled temperature zone and
parallel to the flow.
11-2
-------
MATHEMATICAL MODELING OF THE S02-LIME REACTION
The 502-liiTie reaction is a fluid-solid system which can be analyzed by assuming
that the gaseous reactant, SO2, must undergo the following:
diffusion from the bulk gas to the particle's surface;
pore diffusion to the particle's interior;
solid state diffusion through the product layer (CaSO^); and
reaction at the product/lime interface.
A grain model formulation was adopted for this study based on high resolution
STEM photograohs of calcined dolomite particles which showed a grain-like struc-
ture with 100A cracks. The modeling of the sulfation reaction by Hartman and
Coughlin(7) for fluidized beds was used as a starting point. This model is
based on the assumption that porous lime particles can be represented as conglo-
merates of nonporous spherical grains of pure CaO. Note that the major consti-
tuents of dolomite are CaCOj and MgCOj. Dolomite calcines to form CaO and MgO.
In this study, the MgCOj was also assumed to form spherical grains of pure MgO.
The reaction of the CaO grains with SO2 was assumed to occur by the shrinking
core model. The MgO does not react under the conditions of this study. Because
of the differences in the molar volumes of pure CaO and CaSO^ (17 and 46 cm^
gmol_i, respectively], the sulfation reaction necessarily involves the growth
of the grains and the concomitant decrease in the porosity of the particle.
The following assuirptions were made in the development of the grain models
the lime particles are fully calcined;
the grains of CaO and MgO are of the same average size;
the diffusion of reactant to the particle surface is not
rate limiting;
Equation (1J is irreversible and zero order with respect
to SO2;
the sole reaction product is CaSO^;
the CaO grains react by the shrinking core model, and the
MgO grains do not react; and
excess oxygen is present.
The last assumption allows any reaction rate dependence on oxygen to be included
as a constant in the rate term. The details of the overall mathematical for-
mulation and numerical solution procedure have been described elsewhere(8).
DISCUSSION OF EXPERIMENTAL RESULTS
Sorbent Activation
Prior to the sulfation experiments, studies were conducted to establish the
influence of reaction zone residence time and temperature on the calcination and
subsequent activity of the sorbent materials. Each dry sorbent was injected at
the beginning of the reaction zone (furnace bottom), and a solid sample was
collected at the exit of the controlled temperature zone for subsequent surface
area characterization. No SO2 was present in these experiments; the surface
areas reported are those associated with unreacted calcium oxide. Six sorbents
were used in these and subsequent sulfation experiments: two high-purity large-
grain limestones, two slaked limes, a 56% calcium carbonate dolomite, and a
high-pressure slaked dolomite. Table 1 summarizes the chemical and physical
11-3
-------
properties of these materials prior to their introduction into the reactor.
Figure 2 summarizes the data obtained on surface area as a function of residence
time within the reactor for the six sorbent materials. In these experiments the
reactor temperature was maintained constant at 980* C ^20" C (1800* F). (For
the purposes of data reduction, the MgO and the CaO in the dolomitic materials
were assumed to contribute surface areas based on their molar volumes and weight
percentages.) As the calcination proceeded at 980* C with the limestones, the
surface area developed rapidly; the initial increase occurred in less than 250
msec. The sorbents were more than 60316 calcined at the first sampling point.
The Vicron limestone exhibited a definite peak in surface area at approximately
300 msec and then the surface area decayed slightly. No obvious peak was
observed with the Rollovit limestone.
Figure 2-B shows similar results for two slaked lime materials. These calcitic
hydroxides began with much higher initial surface areas and were approximately
80% dehydrated by the first sampling point. A local maximum in surface area
could exist within the first 200 msec; however, it was not possible to observe
this behavior due to the limitations of the experimental apparatus. In general
the slaked limes exhibited relatively little change in surface area, even at
residence times up to 750 msec at 980* C. In fact, the only major difference
between the calcitic carbonates and hydroxides was in initial surface area;
after approximately 500 msec both the extent of calcination and surface area
were very similar for all four of these materials (Figures 2-A and B).
Figure 2-C shows similar results with the dolomitic materials (both the natural
dolomite, CaCOj.MgCOj and the pressure-slaked dolomite, Ca(0H)2.Mg(0H)2). As
in the case of the calcitic materials, the slaked dolomite began with a much
higher initial surface area and exhibited a definite peak in surface area at
approximately 300 msec. However, beyond approximately 500 msec the properties
of both dolomitic materials appear to be very similar.
Figure 3 summarizes the data which were obtained for all six sorbents on surface
area as a function of reactor temperature for constant reactor residence time of
460 msec. Figure 3-A shows the results for the two limestone sorbents and indi-
cates that, while there are subtle differences between the sorbents, both stones
behaved quite similarly. After 460 msec at 770* C calcination was less than 20%
complete and the surface area had only begun to develop. By 870* C the calcina-
tion was approximately 60% complete and the surface area of both sorbents had
reached nearly 30 m^/g. Further increases in the reactor temperature resulted
in ultimate decreases in the surface area at the 460 msec sampling point. At
temperatures typical of large utility boiler injection schemes (2200* F) the
surface area for these limestones would be approximately 10 to 20 m^/g once the
materials were fully calcined.
Figure 3-B shows similar data for the calcium hydroxides and indicates that, as
in the case of the residence time studies, surface area is only weakly dependent
on temperature. This may be because dehydration occurs extremely rapidly and
all of the data shown in Figure 3-B represent essentially fully dehydrated par-
ticles; even at 770" C more than 70% of the material was dehydrated at 460 msec.
Figure 3-3 also shows that both hydroxides behave very similarly with respect to
the influence of temperature.
Data for the dolomitic sorbents are shown in Figure 3-C. The surface areas for
the pressurs-slaked dolomite were large at all conditions; calcination at 870* C
produced a peak surface area of nearly 50 m^/g, a reasonably high value for
dolomite which had been flash calcined in combustion gases. Further increases
in reactor temperature resulted in a significant decrease in the available sur-
11-4
-------
face area. Beyond approximately 1000* C the pressure-slaked dolomite became
essentially indistinguishable from the naturally occurring carbonate by the 460
msec sampling point (in terms of surface area). Physical inspection under the
STEM did, however, indicate that the materials had retained their large initial
difference in particle size; the pressure-slaked sorbent appeared to have a mean
particle size less than 1 um, while the natural occurring dolomite had a mean
particle size of 20 um.
In surrmary, the sorbent activation studies suggest that for each sorbent there
exists an optimum reactor residence time and peak temperature for maximum sur-
face area during flash calcination. There appear to be subtle differences
between sorbent within the same class (ie. calcitic carbonates); however, the
major differences occur between classes and in general dolomitic materials which
produce higher surface areas than calcitic materials. In general hydrated sor-
bents produce higher surface areas under conditions where the calcination times
are very limited, but beyond approximately 300 msec there does not appear to be
a great difference between limestones and slaked limes.
Sulfation
Figure 4 shows the SO2 capture as a function of reactor temperature over the
same temperature range as the calcination studies for a calcium-to-sulfur ratio
of 2.0 and a gas-phase SO2 concentration (dry) of 2700 ppm. These data indicate
that captures slightly in excess of 20% were achieved with the limestone sorbent
(Vicron) in 460 msec. As reaction zone temperature was increased, the overall
capture decreased dramatically. The solid line in Figures 4 shows the predic-
tions made with the grain model using the measured thermal conditions and the
measured surface areas. A 5 um mean particle size was assumed for the lime-
stone based on measurements of flash-calcined Vicron by Seeker and Cole(9).
No adjustments were made to improve the agreement between the experimental
results and the model predictions; all of the diffusion coefficients and kinetic
rates were based on the prior fundamental work of Borgwardt(5). In general the
grain model appears to be capable of approximately predicting the actual
measured performance based on elementary kinetic measurements.
The overall trend shown in Figure 4, decreasing 502 capture with increasing
reaction zone temperature, is almost certainly the result of a competition bet-
ween sorbent desurfacing (sintering) which reduces the sites available for reac-
tion with SO2 and the diffusion surface reaction mechanism of the sulfation
process. Figure 5-A shows a summary of the available data regarding the effect
of reaction zone temperature on the surface area of flash-calcined Vicron in
bench-scale combustion environments. This figure demonstrates that the results
obtained in this work are consistent with those obtained by previous investiga-
tors and that the surface area available for reaction with SO2 decreases drama-
tically as the peak calcination temperature is increased. Figure 5-B shows
model predictions at a constant surface area of 15 m2/g and indicates that, were
surface area not a strong function of reactor temperature, sulfation would
greatly increase with increasing temperature due to both increased pore and pro-
duct layer diffusion and increased chemical reaction rates. It is the com-
petition between these processes that produces the overall thermal effect shown
in Figure 4, and the model does approximate these effects correctly.
It should also be noted that in the practical system a more favorable thermal
environment (lower temperature) can be achieved by injecting the sorbent farther
downstream from the burner zone; however, in actual field hardware there is an
additional penalty associated with this because of a decreased downstream resi-
dence time. Unlike the idealized experiments illustrated in Figure 4, calcina-
11-5
-------
tion temperature and sulfation zone residence time are not independent in a
practical system.
Figure 6 summarizes the data obtained on the influence of reactor temperature
for all six of the sorbents. These data were all obtained with a sulfation zone
residence time of 460 msec and a gas-phase SO2 concentration of 2700 ppm. The
results are reported in terms of percent calcium utilization; in the case of the
dolomitic material, the magnesium was assumed to remain unreacted based on early
fundamental studies and recent work by Overmoe et al.(10). In general all of
the calcitic materials (both the carbonates and the hydroxides] exhibited the
temperature dependence previously discussed for the Vicron limestone; as the
reactor temperature increased, the overall SO2 capture decreased. For this
relatively short sulfation zone residence time the peak calcium utilizations
were between 10 and 20%.
In general the dolomitic materials exhibited much higher calcium utilizations.
At both temperatures tested, the pressure-slaked dolomitic lime from Genstar
produced calcium utilizations of approximately 35% (which would correspond to
70% SO2 capture at a calcium-to-sulfur ratio of 2.0). These materials also
appeared to be less temperature sensitive than their calcitic analogues. This
effect is almost certainly due to the presence of the magnesium in the crystal
structure and may relate to the prevention of pore closure.
Figure 7 shows a composite plot of calcium utilization as a function of sur-
face area for the case with 460 msec residence time at 1090* C. This figure
indicates that in general the higher calcium utilizations (higher SO2 captures)
are associated with an increase in sorbent surface area. These data also indi-
cate that other factors (e.g. in situ particle size distribution, pore size
distribution) are of some importance in terms of SO2 capture by dry sorbent
injection. Additional work is needed to clarify the role of these secondary
effects.
CONCLUSIONS
The results of this study suggest that the fundamental sulfation data obtained
by Borgwardt can be extrapolated to combustion conditions typical of pulverized
coal boilers and used in conjunction with an appropriate diffusion/heterogeneous
chemical kinetic model to predict SO2 capture by dry sorbent materials.
Temperature in the reaction zone was found to have a critical influence on SO2
capture because it controls the reactivity of the sorbent (as characterized pri-
marily by the surface area) and the overall rates of diffusion and heterogeneous
reaction. Increasing termperature increases both the kinetic and diffusion
rates; however, it significantly decreases the surface area of the sorbent at
the end of the flash calcination process and, thereby, reduces the sites
available for reaction with S02^
As expected, dehydration of slaked limes occurs significantly more rapidly than
calcination of carbonates, but beyond approximately 500 msec at 900" C the sur-
face areas of all the calcitic materials used in this investigation were simi-
lar. In general both the dolomitic carbonate and the dolomitic hydroxide
produced significantly higher surface areas at all conditions tested, and this
resulted in overall higher SO2 captures for these materials. Calcium utiliza-
tions in excess of 35% can be achieved in less than 500 msec with pressure-
slaked dolomitic sorbents.
11-6
-------
ACKNOWLEDGMENTS
The authors gratefully acknowledge the assistance of Robert H. Borgwardt, G.
Blair Martin, Dennis C. Drehmel, and David A. Kirchgessner of the U.S.
Environmental Protection Agency and P. L. Case, W. R. Seeker, and M. P.
Heap of the Energy and Environmental Research Corporation. This work was
completed under EPA Cooperative Agreement CR-811001.
REFERENCES
1. TVA Shawnee Steam Plant, KY, Field data obtained by EPA (1972).
2. Coutant, R. W., R. E. Barrett, R. Simon, B. E. Campbell and E. H.
Lougher, "investigation of the Reactivity of Limestone and Dolomite
for Capturing SO2 from Flue Gas (Summary Report)," EPA Report APTD 0621
(NTIS PB 196749), November 1970.
3. Attig, R. C. and P. Sedor, "Additive Injection for Sulfur Dioxide
Control: A Pilot Plant Study," EPA Report APTD 1176 (NTIS PB 226761),
March 1970.
4. Whitten, C. M. and R. G. Hagstrom, "Pilot Plant Moving Grate Furnace
Study of Limestone-Oolomite for Control of Sulfur Oxide in Combustion
Flue Gas," EPA Report APTD 1264 (NTIS PB 184944), May 1969.
5. Borgwardt, R. H., Environ. Sci. Technol., 4, 59 (1970).
6. Ishihara, Yoshimi, "Kinetics of the Reaction of Calcined Limestone
with Sulfur Dioxide in Combustion Gases," presented at the Dry
Limestone Injection Process Symposium, Gilbertsville, KY, June 22-26,
1970.
7. Hartman, M. and R. W. Coughlin, A.I.Ch.E.J., 22, 490 (1976).
8. Slaughter, D. M., G. D. Silcox, and D. W. Pershing, "influence of
Sorbent Composition on SO2 Capture Efficiency in LIMB Applications,"
paper 100-1 presented at the 1984 Annual AIChE Meeting, San Francisco,
CA, November 1984.
9. Seeker, W. R. and J. Cole, EER, Private Communication (1983).
10. Overmoe, B. J. et al., "Boiler Simulator Studies on Sorbent Utilization
for SO2 Control," presented at the 1st Joint Symposium on Dry SO2 and
Simultaneous S02/^0X Control Technologies, San Diego, CA, November 1984.
11-7
-------
Auxi1iary
Burner
Gas and Solid Phase
Sampling Probes
Exhaust
Auxi1iary Burner
J
. v • •. ¦¦ •"
Ma l n
Gas Burner
^ Sorbent Injector
Figure 1. Schematic of Isothermal Reactor
11-8
-------
1 1 1
1 1 1
1 1 1
A. Limestones
B. Hydroxides
C. Dolomites
0— Vicron
Steinmuller
Rollovit
Genstar-N
k
1
. A k -
¦ ¦
¦
•
f ~
f
| ~ * * <
~ ~~ %
-
|^— Genstar-S
Dolomite
1 1 1 1
1 1 1 1
i ,
0 0.50 0 0.50 0 0.50 1
Reactor Residence Time (sec)
Figure 2. Development of Surface Area at 980°C (1800°F).
-------
50
30
10
1 1
• i
1 1
A. Limestones
B. Hydroxides
C. Dolomites
% — Vicron
Steinmuller
I
^ _ Rollovi t
^ Genstar-N
-
~
•
1 * 4
* k
k A
¦ ~
~ •
A •
~ ~ A
~
~ ~ -
m Genstar-S
• 4
^ _ Dolomi te
~
i i
• ¦
' '
1400 1800 2200 1800 2200 1800 2200 °F
I I I I I I I I I
900 1100 1300 900 1100 1300 900 1100 1300 °C
Reactor Temperature
Figure 3. Influence of Calcination Temperature on Surface Area Development (460 ms residence time).
-------
1 1
1
1
50
#— Experimental Data
40
-
_ Model Prediction
-
30
-
-
20
•
10
-
•
i i
I
i
1400
1800
1
I
2000
900 1100
Reactor Temperature
Figure 4. Sulfur Capture: Comparison of Theory and Experiment
(Ca/S = 2,0, SO^ = 2700 ppm, Residence Time = 460 ms)
-------
40 -
20
i r
0 — This study
A- Ref. 10
X
X
X
X
Model Prediction
(Const, surface area)
RT= 0.46
Ca/S= 2.0 sec
50^= 2700 ppm
SA= 15 m^/g
X
40
30
20
10
1400
X
1800
2200
2600
1400
1800
2200
2600
X
X
X
900
1100
1300 1500 900
Reactor Temperature
Figure j. Individual Effects of Thermal history (Vicron)
1100
1300
1500
-------
1 r
C. Dolomites
50
A. Limestones
Vicron
A Rollovit
30
10
_L
B. Hydroxides
Steinmuller
a Gens tar-N
*
-L
^ — Genstar-S
am Dolomite
1400
1800
2200 1400
1800
2200 1400
1800
2200
J
900
1100
1300
900
1100
I .
1300
900
1100
1300
Reactor Temperature
Figure 6. Influence of Thermal Environment on Calcium Utilization (SO2 - 2700 ppm,
Residence Time = 460 ms)
-------
Model Prediction
Surface Area (m /g)
Figure 7. Correlation of Calcium Utilization With Surface Area
(SO^ = 2700 ppm, Temperature = 1090 C = 2000 F, Residence Time = 460 ms)
-------
TABLE 1. SORBENT CHARACTERISTICS
Sorbent
Dolomite
Genstar-S
Genstar-N
SM Ca(0H)2
Vicron 45-3
Rollovit
Type
Dolomite
Dolomitic
Pressurized
Hydrated
Lime
Hydrated
Lime
Hydrated
Lime
Calcite
Chalk
Composition
CaC03
MgC03
Ca(0H)2
Mg(0H)2
Ca(0H)2
Ca(0H)2
CaC03
CaC03
Top Size, pm
<100
30
30
45,40
<40
Mean Size, pm
34.0,40
1.4
3.0
3.5
11,12
6
Bottom Size, um
1.0
0.3
0.3,0.8
Density, g/cm5
2.855
2.28
2.30
2.35
2.757
2.37
Initial Surface
Area, m2/g
0.64,0.60
30
20
13.5,12.1
0.9,0.6
1.73
Surface Area at
1000* C, m2/g
47.1,44.2
25.3
44.0,41.0
43.8
Ca, %
21.0
29
51.0
39.2
37
Mg, %
12.1
16
0.485
0.50
-------
EVALUATION OF SOz REMOVAL BY FURNACE LIMESTONE INJECTION
WITH TANGENTIALLY FIRED LOW-NOx BURNER
K. Tokuda, M. Sakai, T. Sengoku and N. Murakami
Mitsubishi Heavy Industries, Ltd.
M. W. McElroy
Electric Power Research Institute
K. Mouri
Electric Power Development Co. (Japan)
ABSTRACT
The Mitsubishi Heavy Industries (MHI) contracted with the Electric Power Research
Institute (EPRI) and Electric Power Development Company (EPDC, Japan) to evaluate
the S02-removal effectiveness of furnace limestone injection when applied in com-
bination with a low-NOx burner.
The evaluation included a series of furnace limestone injection tests conducted at
MHI1s 4 ton/hour pulverized coal-fired test furnace equipped with the low-NOx PM
burner developed by MHI. The results showed 30 to 40 percent reduction of SO2 at
a Ca/S molar ratio of 2:1 while maintaining low-NOx combustion conditions compati-
ble with full-scale utility boilers and operating practices.
Basic (bench-scale) tests were carried out prior to the combustion tests, using an
electrically heated flow reactor. The resutls from the basic tests were applied
in interpreting the results from the combustion tests.
12-1
-------
EVALUATION OF SO2 REMOVAL BY FURNACE LIMESTONE INJECTION
WITH TANGENTIALLY FIRED LOW-NOx BURNER
INTRODUCTION
The growing problem of atmospheric pollution by nitrogen oxides (NOx) and sulfur
oxides (SOx) in stack gases of pulverized coal-fired power plants has generated a
large amount of research and development activity aimed at demonstrating suitable
emission reduction technologies.
Of the several processes that could decrease the NOx and SOx emissions, the injec-
tion of limestone or other calcuim compounds into furnaces equipped with low-.NOx
burners is simplest and requires the least investment.
In this process, SO2 reacts with limestone within the furnace to form solid sulfate
particles which are then removed from the flue gas in the conventional particulate
control device.
This paper shows the results of the experimental evaluation of SO2 removal by
furnace limestone injection when applied in combination with the low-NOx PM burner
technology (_1), {2), (3_). The evaluation centered around tests at the MHI's 4 ton/
hour combustion test furnace (nominally 12.5 MW) which is designed to maintain com-
bustion conditions comparable to full-scale utility boilers and operating practice.
SO2 REMOVAL CHARACTERISTICS BY LIMESTONE WITH REACTOR TUBE TEST APPARATUS
Basic limestone injection research on additive type, composition, surface area,
point of injection and Ca/S molar ratio was conducted in 1970's (4), (5), (6), (2),
(8). Approximately 30 percent SO2 reduction was achieved at the average Ca/S molar
ratio of 1.5:1 (9). This work was followed by full-scale testing by the Tennessee
Valley Authority [TVA] at their Shawnee Steam Plant (10.). These studies indicated
approximately 20 percent SO2 capture at the Ca/S molar ratio of 2:1.
The general reaction describing sulfur capture under oxidizing condition shown in
the following formula has been studied extensively by several researchers (11), (12).
CaO + SO2 + j O2 -> CaSOi* (1-1)
However', none of these studies duplicated the time/temperature conditions that
prevail in pulverized-coal flames. Borgwardt (_13) has suggested that the reactions
formulated in (1-2) and (1-3) could be significant under fuel-rich conditions.
CaCOa + H2S CaS + HzO + CO2 (1-2)
CaO + H2S - CaS + H2O ' (1-3)
Low NOx coal burners produce extensive fuel-rich regions in addition to the usual
fuel-lean burnout regions of the flame. Consequently, there are two possible modes
by which calcium-based sorbents may capture sulfur species in a pulverized coal-
12-2
-------
fired boiler operating under low NOx conditions.
In the fuel-lean region the reaction (1-1) may proceed because the lower peak tem-
perature will minimize a loss of surface area (deadburning). In the fuel-rich
region the formation of calcium sulfide may be significant.
However, Pershing (_14) showed that sulfur capture should be enhanced under fuel-
rich conditions but sulfur retention in the coal char and its regeneration during
burnout could negate this benefit.
As reviewed above, there are some discrepancies in the opinions of different re-
searchers. These may be due to the differences in the test equipment or test
conditions employed.
To clarify sulfur capture mechanisms, basic data on process fundamentals was felt
to be necessary. For this purpose the basic tests were performed.
Test Apparatus and Test Method
Fig. 1 shows the test apparatus used. The body of the test furnace is composed of
two cylindrical electric furnaces vertically connected, and inside them are located
two ceramic reactor tubes each 24 millimeters in diameter and 1000 millimeters in
length connected end to end with a silica joint. The primary reactor (upper elec-
tric furnace) is for the reaction of limestone with S02, and the secondary reactor
(lower electric furnace) is for the complete conversion of CaC03 to COz at higher
gas temperatures (>1400°C).
The quantity of limestone injected is calculated from the measured CO2 concentra-
tion in the gas. Pulverized limestone with a prepared particle diameter of under
56 micro-meters (250 meshes) is stored in a vessel at the top of the apparatus and
dropped into the top of the primary reactor with two vibrators.
Test Results and Discussions
For the presentation of the results the removal rate of SO2 (nS02) for a specific
Ca/S molar ratio is defined as percent of inlet SO2 removed.
Fig. 2-1, Fig. 2-2 and Fig. 2-3 show the effects of the Ca/S molar ratio, gas tem-
peratures and residence time on the SO2 removal rate, respectively.
These figures indicate the following:
(1) SO2 removal efficiency increases with Ca/S molar ratios from 0.0 to
4.0, but begins to level off at Ca/S molar ratios larger than about
4.0.
(2) Optimum SOz removal conditions in the tunnel flow type furnace used
is different from that found in a fluidized bed furnace. The optimum
gas temperature in Fig. 2-2 is from 950 degrees to 1100 degrees
centigrade. In case of SO2 removal in fluidized bed combustion, the
optimum gas temperature is from 850 degrees to 950 degrees centi-
grade. It could be considered that the difference in the optimum
gas temperature between pulverized coal combustion and fluidized bed
combustion is based on the great difference in limestone reaction
time with SO2.
12-3
-------
(3) Maximum SO2 removal efficiencies are 30 percent at the residence
time of 0.54 seconds and 80 percent at that of 1.54 seconds respec-
tively.
These test results show a qualitative agreement with the results from other pilot-
scale and bench-scale tests (5J , (6), (14).
Fig. 2-4 shows the effect of t'ne limestone fineness on the SO2 removal rate. The
SOz removal rate at Ca/S molar ratio of 2.0 increased to 47 percent from 40 percent
with increased fineness. From Fig. 2-4, the SO2 removal rate is proportional to
about one-third power of limestone fineness. Ishihara (_7) showed that the reac-
tivity of limestone was proportional to one-third to one-fourth power of limestone
fineness. Ishihara's test result shows a quantitatively excellent agreement with
that shown in Fig. 2-4.
Fig. 2-5 shows the effect of SO2 inlet concentration on the SO2 removal by lime-
stone. The SO2 removal rate is increased with the increment of sulfur content in
combustion gases and is proportional to one-fifth to one-sixth power of sulfur
content at 1.0 to 2.0 of Ca/S molar ratio. But, the value of power slightly
changes with the Ca/S molar ratio, the residence time and so on. Meanwhile,
Ishihara (_7) showed that the reaction rate of limestone was proportional to 0.3 to
0.5 power of sulfur content. It would be possible to understand that the differ-
ence in the value of power is based on a different Ca/S molar ratio, residence time
and reaction temperature.
As mentioned before, it was suggested that the SO2 removal rate should be increased
due to the reducing atmosphere in addition to lower combustion temperatures in low-
NOx combustion condition (13). The reactivity of limestone with H2S in the zero
percent oxygen atmosphere was investigated, simulating the actual pulverized coal
combustion with over fire air. Here, it was assumed that sulfur species in the
reduced atmosphere was only H2S with COS left out of consideration.
Fig. 2-6 shows the reactivity of H2S with limestone in the zero percent oxygen
atmosphere as compared with that of S02 in the oxidized atmosphere as shown in
Fig. 2-1. Although the oxygen content in the reaction zone of H2S with limestone
was zero percent, the oxygen content at the inlet of the second reactor tube, where
limestone was calcined perfectly, was kept constant at three percent by adding
oxygen as over fire air for measuring the concentration of H2S as SO2. From Fig.
2-6, the reactivity of HzS with limestone is around half of that of SO2 in the
oxidized atmosphere.
Fig. 2-7 shows the effect of the inlet gas composition on the SO2 removal rate as
compared with the result of the base condition shown in Fig. 2-1. It is clarified
that the S02 removal rate of the inlet gas containing Hz, CO and zero percent oxy-
gen is lower than that of the base condition.
Moreover, the SO2 removal rate of the inlet gas containing one percent H2 is about
40 percent with and/or without limestone. And, this reduction of SO2 could be
understood to suggest that SO2 decomposes after its reaction with H2 while solid
sulfur is formed and sticks onto the inside surface of the reactor tube. Moreover,
it means that the S02 removal by limestone is almost zero in the reduced atmosphere
with H2.
Fig. 2-8 shows the effect of methane contained in the inlet gas on the SO2 removal
rate. By addition of 1000 ppm of methane to the inlet gas, the SO2 removal rate
decreased slightly and there was no difference in the optimum gas temperature for
the SO2 removal.
12-4
-------
Moreover, it was reported (15) that the SO2 removal rate also decreased with the
decrement of air ratio in the primary combustion zone in the fluidized bed boiler.
Since the SO2 removal rate is considerably lowered in the reduced and/or zero per-
cent oxygen atmosphere, as mentioned above, it is clarified that the reaction of
limestone with sulfur species such as H2S in the reduced atmosphere is not desir-
able to keep a high SO2 removal rate.
ANALYSES OF COALS AND LIMESTONE TESTED
The two bituminous coals selected for the limestone injection test provide a wide
range of sulfur contents. The results of analyses of the two coals are shown in
Table 1.
The limestone of Japanese Ikura was selected for the limestone injection test as a
high quality of CaC03. The result of analysis of the limestone is shown in Table 2.
Three values of limestone fineness were selected for the limestone injection tests.
The finenesses are 80, 97 and 100 percent through 200 mesh.
FACILITIES FOR COMBUSTION TEST
Tangential Fired Low-NOx PM Burner
The low-NOx PM burner, which had been previously tested and verified (_1), (2J, {3),
was selected as a suitable low-NOx burner for the limestone injection tests. The
basic concept of this burner is as follows:
(1) Pulverized coal in the primary combustion zone (that is the combustion
zone for volatile matter in coal) should be rapidly burned under
stable and high temperature condition with an adequate residence time
to give a low NOx level and good stability.
(2) The air/coal mixture to the burner should be divided into two
streams, a coal-rich stream and a coal-lean stream, and be burned
separately through different nozzles contained in the same burner.
(3) The coal-rich flame and coal-lean flame combined produce low NOx.
The former has a good ignition stability and the latter essentially
reduces unburned carbon in fly ash due to a high oxygen content.
(4) With the above, by combining coal-rich flame and coal-lean flame,
pulverized coal can be burned stably and efficiently (with low un-
burned carbon) with an extremely low NOx emission level.
Fig. 3-1 shows the structure of a coal-fired low-NOx PM burner.
12-5
-------
Combustion Test Equipment
Fig. 3-2 shows a flow sheet of the 4 ton/hour (nominally 12.5 MW) test furnace.
The body of the furnace is a sea-water-cooled double-walled cylinder of 4.4 meters
in inside diameter and 20 meters in length. A part of its inner surface is lined
with refractory to keep the furnace temperature at around the level of the actual
boiler furnace so that the flame ignition characteristics, the degree of NOx forma
tion and the percentage of unburned carbon in fly ash in the actual boiler can be
simulated.
Measuring Instruments
Measured items, measuring instruments and measuring method used are shown in
Table 3.
Flue gas was sampled for analysis of SO2 at the following locations: (1) along the
furnace flame axis, (2) furnace outlet, (3) nulticyclone outlet, (4) bag filter
inlet and (5) bag filter outlet.
Fig. 3-2 also shows the location of SO2 measuring point in the 4 ton/hour test
furnace. Overall sulfur material balance has been calculated, based on the analyses
of the solid and the gaseous sulfur samples at each section mentioned above.
TEST METHOD
The limestone injection conditions that were changed included the limestone-inject-
ing method, the Ca/S molar ratio, the limestone fineness, the combustion rate and
the oxygen content in the flue gas.
Moreover, the amount of over fire air (OFA, which was introduced downstream of the
flame) and the amount of the recirculated gas flow through the "Shield" gas re-
circulation nozzle (SGR) were changed.
The burner primary air temperature and pulverized coal fineness remained constant
at 80 degrees centigrade and 70 percent through 200 mesh, respectively.
Fig. 3-3 shows the various arrangements of injecting systems used for the pulverized
coal and/or limestone. The type (a) shown in Fig. 3-3 is the coal/Iimestone mixing
system. The primary air is led to below t'ne pulverized coal bin and the limestone
bin, and passes to the burner, carrying pulverized-coal and limestone fed through
the coal feeder and the limestone feeder, respectively. Here, the pulverized coal
and t'ne limestone were well mixed by the exhaust fan.
The type (b) shown in Fig. 3-3 is the system through the auxiliary air nozzle. The
pulverized-coal and limestone are separately transported to the burner from each
bin. And, the limestone is injected into the furnace from the auxiliary air nozzle
provided above the coal-rich coal nozzle.
The type (c) shown in Fig. 3-3 is the system through the 0FA nozzle. In this
system, the limestone is transported from the limestone bin to the OFA nozzle by
the limestone primary air. And, the limestone is injected into the furnace from
the three OFA nozzles provided around the furnace body after being mixed with the
over fire air.
The type (d) shown in Fig. 3-3 is the system through the gas recirculation (GR)
nozzle. In this system, the limestone is transported by the recirculated flue gas
from the limestone bin to the GR nozzles located around the furnace body downstream
12-6
-------
of the OFA nozzle. The limestone is injected into the furnace from the three GR
nozzles after being mixed with the recirculated flue gases.
TEST RESULTS
Optimum Limestone Injecting Method
Four types of the limestone injecting methods, as mentioned before, were tested with
high sulfur coal. Here, the SO2 removal rate is defined as percent reduction of the
SOs gas concentration in flue gas compared to the value of that without limestone
i njection.
Fig. 4-1 shows the comparison of the SO2 removal rates with each type of injecting
method. Although the SO2 removal rate increases with the Ca/S molar ratio in all
types of injecting methods, there are great differences in the SO2 renoval rates
with the same Ca/S molar ratio in the four types of injecting methods. It is
clarified from Fig. 4-1 that the type (c), which is through the OFA nozzle system,
is the most suitable system to remove S02 gases in the furnace.
Since the oxygen concentration of the type (c) in the reaction zone of SO2 and CaO
is highest due to the mixing of limestone with the total amount of over fire air
and the S02 removal rate increases with the oxygen concentration in the reaction
zone as mentioned before, it may be understood that the SO2 removal rate of the
type (c) shows the highest value in those of all types of injection methods.
SO2 Removal Characteristics
The relation of SO2 removal rates to limestone injection conditions and/or burner
operating conditions are shown in Fig. 5-1 to Fig. 5-3.
Effect of Ca/S Molar Ratio on SO2 Removal Rate. Fig. 5-1 shows the relation of the
SO2 removal rate to the Ca/S molar ratio with the high sulfur (HS) and the low
sulfur (LS) content coals. These tests were made by changing the limestone flow
rate only.
Although the SO2 removal rate increases with the increment of the Ca/S molar ratio,
this upturn rate gradually flattens with the further increment of the Ca/S molar
ratio.
The SO2 removal rate with HS coal is higher than that with LS coal. The S02 remov-
al rate at a Ca/S molar ratio of 2.0 is 44 and 35 percent, respectively.
From these values of the SO2 removal rates, the SO2 removal rate is proportional to
one-fifth power of the sulfur content in coal. And, the results of the basic
information tests shown in Fig. 2-5 show a similar dependence on the sulfur content.
Earlier studies have shown that the sulfur capture is proportional to from one-
third power to the square root of the sulfur content (, (6) , (_7). The differ-
ences in proportional coefficients may be explained due to differences in other
test conditions.
Effect of Firing Rate on SO2 Removal Rate. Fig. 5-2 shows the relation of the SO2
removal rate to the firing rate in the furnace with the HS coal and the LS coal.
The SO2 removal rate increases as the firing rate decreases when the firing rate is
in the range from 120 to 100 percent and decreases as the firing rate decreases
when it is in the range from 100 to 80 percent. And, the SO2 removal rate in-
creases about three percent with the increment of over fire air from 20 to 25
12-7
-------
percent, as shown in Fig. 5-2.
It is considered that the increment of the SOz removal rate with the decrement of
the firing rate in the range from 120 to 100 percent is due to the increment of the
residence time, which means the reaction time with lime and SO2 gas, in the reac-
tion zone. Moreover, it is considered that the decrement of the SO2 removal rate
with the decrement of the firing rate in the range from 100 to 80 percent is due
to the decrement of the gas temperature in the limestone injection zone.
Effect of Over Fire Air on SO2 Removal Rate. Fig. 5-3 shows the relation of the
SO2 removal rate to the amount of over fire air flow with the HS coal and the LS
coal. The SO2 removal rate increases with the increment of the over fire air flow.
In this test, the limestone was injected into the furnace after being mixed with
the over fire air. It is considered that the increment of the SO2 removal rate to
a higher percent of 0FA is based on a higher concentration of oxygen in the reac-
tion zone of lime and SO2. Changes in sorbent dispersion and mixing may also have
an effect.
SO2 Distribution and Sulfur Material Balance
Fig. 6-1 to Fig. 6-2 show the SO2 concentration distribution along the flame axis
with the two types of injecting methods. Here, the measuring points are shown in
Fig. 3-2 as mentioned before.
Fig. 6-1 shows the result with the coal/limestone mixing system of the type (a).
In this case, the limestone is injected into the furnace from the burner front.
The SO2 concentration gradually decreases from the burner front to the point 17 to
18 meters away from the burner front along the flame axis.
Fig. 6-2 shows the result with limestone injection through the 0FA nozzle of type
(c). In this case, the limestone is injected into the furnace from the 0FA nozzles
which are provided at the furnace side wall and located about 13 meters away from
the burner front along the flame axis. As shown in Fig. 6-2, the SOz concentration
abruptly decreases near the limestone injection port along the flame axis.
Fig. 6-3 shows a typical result from the sulfur mass ballances which were performed
during the test program as a check on S02 measurements. In this figure, the inlet
SO2 means the SO2 concentration calculated from the sulfur content in coal as shown
in Table 1. The sulfur content in the coal showed slightly scattered values from
day to day. It is felt that the error in sulfur material balance can be attributed
to this and to unavoidable errors in sampling of ash. Nevertheless, a reasonably
good sulfur balance was achieved.
Since the SO2 removal rate is based on gaseous sulfur, which means SO2 gas content
in flue gas with and without limestone injection, it is considered that the SO2
removal rates mentioned before have a considerably high degree of precision.
Flame Temperature and Burner Flame Condition
The flane temperature distribution measured by optical pyrometer in the direction
of the flame axis comparing the firing rates with the HS coal and the LS coal are
shown in Fig. 7-1.
From this figure, the flame temperature at the 0FA port is from 1000 to 1100 degrees
centigrade.
12-8
-------
Burner flame conditions with the type (c) of injecting method is shown in Fig. 7-2.
This figure shows the ignition condition and the flame pattern of the HS coal with
the limestone injection using the pulverized-coal-fired low-NOx PM burner.
From this photograph it is obvious that the pulverized-coal-fired low-NOx PM burner
is able to keep not only a low NOx level but also stable ignition and good combus-
tion in the volatile matter combustion zone.
NOx Emission Characteristics
The relation of NOx emission to Ex.02 (oxygen concentration in flue gas) comparing
the cases with and without limestone injection for HS coal and the LS coal are
shown in Fig. 8-1 to Fig. 8-2 respectively.
NOx values with the injection of 20% OFA at four percent of oxygen concentration
are around 110 ppm as shown in Fig. 8-1 and Fig. 8-2. Moreover, NOx value decreases
about four ppm by the injection of the limestone. And, as shown in Fig. 8-1 to
Fig. 8-2, the unburned carbon in fly ash is very low and slightly increases with the
decrement of Ex.02.
INTERPRETATION OF TEST RESULTS AND DISCUSSION
When predicting the SO2 removal rate in an actual boiler with the limestone injec-
tion system, it is required to understand the following items which were clarified
in the series of the combustion tests.
• The optimum limestone injecting method is the injection through the
over fire air nozzle system after mixing of the limestone with over
fire air.
• The SO2 removal rate increases with the increment of the over fire
air flow mixed with the limestone and the NOx value decreases at the
same time.
• There are three important parameters affecting SO2 removal rate:
the gas temperature, the residence time and the oxygen concentration
in the reaction zone.
T'ne SO2 removal rate increases with the increment of oxygen content
in flue gas, and with the increment of the residence time. The lat-
ter may be interpreted as meaning a reduction of combustion rate in
the furnace.
• The SO2 removal rate increases with the increment of the sulfur
content in coal.
As mentioned before, the SO2 removal rate is greatly affected by the gas tempera-
tures and the residence time in the reaction zone of SO2 with limestone.
For the reasons mentioned above, it is most important to evaluate the relation
between the gas temperature (or the flame temperature) and the residence time in
the test furnace and/or an actual boiler for predicting the SO2 reduction by the
furnace limestone injection.
Fig. 9-1 shows the relation between the gas temperatures and the residence time at
a full load of the actual pulverized-coal-fired boiler which was designed for
bituminous coal. Here, the gas temperatures show the design values, calculated
from the heat balance of the boiler.
12-9
-------
And also, Fig. 9-2 shows the relation between the gas temperatures and the resi-
dence time at the 120 percent load with the PM burner used in the 4 ton/hour
pulverized-coal fired test furnace. Here, the gas temperatures were measured by
optical pyrometer for the combustion zone and by thermo-couple for the rear path
zone.
Since Fig. 9-1 and Fig. 9-2 show good coincidences of the relations of gas tempera-
tures and the residence time in the test furnace and the actual boiler, it would be
considered that the SO2 removal rate at full load of the actual pulverized coal
fired boiler designed for bituminous coal coincides with the SO2 removal rate at
120 percent load in the test furnace which is shown in Fig. 5-2.
Moreover, if the combustion rate and/or the heat release rate is lower as in the
lignite fired boiler, the SO2 removal rate is higher than that with the typical
steaming-coal fired boiler.
Although no consideration is given to the
gases, this is a very important factor in
actual pulverized-coal fired boiler.
mixing performance of limestone with flue
predicting the S02 removal rate in an
A soecial consideration should be given to the mixing of limestone with the flue
gas in an actual boiler furnace.
CONCLUSIONS
In a series of furnace limestone injection tests conducted at the MHI's 4 ton/hour
test furnace with the pulverized-coal fired low-NOx PM burner, it has been shown
that 30 to 40 percent reduction of S02 at the Ca/S rr.olar ratio of 2:1 is possible
while maintaining low-NOx combustion conditions. The optimum limestone injection
location was through the overfire air ports located away from the main combustion
zone. These results suggest that the limestone injection process can be decoupled
from burner design and operation. Nevertheless, it appears that furnace limestone
injection and low-NOx combustion are compatible with each other for providing
simultaneous NOx and SO2 removal.
These results can be used along with considerations of the design and operating
conditions of the actual boiler in predicting the performance of limestone injec-
tion at ful1-scale.
The writers would feel greatly rewarded for their efforts if this paper should
present a clue to the pollution control of pulverized-coal fired boilers.
12-10
-------
REFERENCES
1. Tokuda, K. et al., EPA/EPRI Joint Symp. on Stationary Combustion NOx Control,
November 1982.
2. Tokuda, K. et al., Mitsubishi Technical Review, October 1981.
3. Kawamura, T., et al ., EPA/EPR I Joint Symp. on Stationary Combustion NOx
Control, 1980.
4. Goldschmidt, K., Third Limestone Symposium, Florida, December 1967.
5. Tanaka, K., Third Limestone Symposium, Florida, December 1967.
6. Ishihara, V., Third Limestone Symposium, Florida, December 1967.
7. Ishihara, V., EPA Dry Limestone Injection Process Symposium, Kentucky, June
1970.
8. Coutant, R. W., et al., Final Report for EPA Contract No. CPA 70-111, October
1971.
9. Attig, R. C., HEW Order 4078-01, March 1970.
10. Full-Scale Desulfurization of Stack by Dry Linestone Injection, EPA Report
650/2-73-019-a.
11. Coutant, R. W., et al., Batelle Memorial Institute Final Report, EPA Contract
PH-86-67-115, 1970.
12. Sorgwardt, R. H., NAPCA 4, 59, 1970.
13. Borgwardt, R. H., Presentation at the EPA SPO Contractors Meeting, North
Carolina, October 1981.
14. Pershing, D. W., et al., Spring Technical Meeting, Central States Section, The
Combustion Institute, Ohio, March 1982.
15. Williams, P. T., Fourth International Conference on F1uidization, Japan, May
1983.
12-11
-------
Vibrator
Vibrator '.2
Pulverized limestone bin
Resde^ee t —>e — I 54s*e
contraler
Pulverized limestone
'eed nozzle
V Drator 1!
Pulverized limestone
Ras-aence tirre=l 08s«
ShccW Bbsorbe
^ Ceramic reactor lube !i
2MX lOOCmm
Res dftnte tm« = G 54sec —
lilei SO? : : OOOC3T1
InleOi 3 ?{
Rtsidftnce C
t«r.e (sec): 0.54' 1 .06: ; 54
~a meter
IOOOC
SO? meter
Ca/S rrolar rat o
-t«J} Quartz jon'.
H?5 i COS
Fig. 2-1 Relation between SO2 remova; and Ca/S rroiar ratio
(Gas temperature in primary reactor = 1OQO'C)
emic reactor lut>e i2)
! 24*x IOOOt-b
Os mete
Inlet SO?
IQOCac
Inlet Oj 3%
Ca/S **&'¦&' ratio 2
Rt&idante 5* sec
COj meter
Filter
R«s»d«nc« Mat
Glass ball filter-^
Vacuurr pjrr 3
F g. 1 Reactor tube lest facility fo- SO? removal
by pu;verized limestone
700 800 9CC '000
Gas te~iperature -r. p-inary reactor (C)
Fig 2-2 Relation between SO2 removal and gas temperature
(Ca/S rro ar ratio = 2)
0 5
I 0
Res'dence
Fig. 2-3 Relation between SOa removal and residence time
(Ca/S molar ratio = 2)
12-12
-------
iOOr
Irlp! SO; iGjOppf
Inlet 0? 3°C
Rendttce '.line 1 OSset
t Tg ' MOD'C
63
Less than 25Cfncsb«i
f'or". 2C0 tc ?50"^eshes
3 4
Ca/S ^olor ratio
Fig 2-4 Effect on fineness of limestone with SO? removal
!niet SOj
, 0^ 2C00c5-r
; £ :_500p3t« ;
¦rile L Or 3 ;
Residence I ~e ! 1 06**i
O
o !
SO»-2CC'33rp'
SOj — SOoDpra
V
Ce/S rr.olar rat;o
Fig 2-5 Rela tion between SO2 removal and Ca/S molar rat o
with different SO* concentration
ICOr
BO-
O
i/>
1 Inlet
Gas Temp
C i BDjopt
: iooo'c
6C0PP*"
; 1 IOC C
O ?D0dpt
: i2QC'c
Inl^t Od
c %
Res»deic« t
im* C 5
-------
2EE
m
iS:
Weak
Fig 3-1 Conf.guraiio-i of pjIwBrt/ed cca'-<'f#fl lew-NOi PM bti'iie'
H f
%
V
O^a
¦tt
id!
Type fa) Can 'itfresto*-* mung avste-n
# Meas-jrems ectn?
In
No r gas
cooi»r
D9& -
1
r
1BL
fM&>£
>* K? •
)•: i *CJ
• ?• s» <•
Test fwrracs
No I (as
cccier
—7*>
^ h •'>:
04 CyCcr«
SROB'at a* ft/>
FDf
lM,
-fils
Fig 3-2 Fie* she«'. sf MHi ( t '.jn.'htw: pulvtf.-'icC caa'"fi'*c tast fgrnace
and local.an of SO* p-.easi.nrg pcrr»
Typ» U) ' Sy*t*r. th'oygh ?.*»» 0^4 ncj«:a
B«ci'cwi*'.«4 'kj» g*»
Type (b) System the euila^y ai- nozzle
Typ« (d) Sv»l«'Tl th« GR ^Oll *
r g 3"3 fnj«el>rg system of the cca; anc tn« hmejl:>^e
Type xdj
Type (d)
O 20
?CC nesh
liPiSfctori*'
Coal
HS
Firing rata
?9
100
OF A
%
2D
SGR
%
M n
Ei Oa
%
: 4 0
%
c?
2 3
Ca'S moiar ratio
Fig. 4-1 Comparison of SO* removal rates with types of
injecting methods
12-14
-------
c
e
* A
>
a
€
Coal
Ir.jectir.g Ty5* ""ype
me'.iod 'c; (c]
Flnn8 no - 130
<5
in
OF A C,
• J00T,»5h
of ; Fir^ngss
Cd/S fnc)ar rat
Thru 200^es^
of l>™«»tonc)
ED ;oo
Fnrg rale {%\
SyTibe
i O
•
A <
, Injecting
1 ire thed
iTtf'
~r*°*
^ yp«
(c) :
Or A {%)
i 20
25
2\
:SGi {%}
; Min
Mir
Mir
E*0* {%)
: *0
* 0
4.0
C«/S {-)
: 2 0
2 0
2 0
; so
97
89
Fig. 5-2 Effect of *'r>ng rate on SOa removal rate
v
a
HS
c
: Injecting ; Type ; Type Type
: me'.hafl , (cj j (c) • (c)
£
Firing
SGR
V r Wi
* (% Thru JCC mesh
d< hmastene)
97-80 '80-' 00
Fig. 5-3 Effect of OFA on SO? removal rate
12-15
-------
300Cr
Symbol
1 0
C
I
(Suffi* : Oi%]
L2000 -
J S
__
I .J
4 6
[Injecting method: Type (a)
«.3
~ 2
-n-
! Fnng rate {%) j
! Ca/5 Mola- 'alio:
-<±r
7 2
6 5
_Z_
7 5
IC 3
—D_
100
2.24
•0 4
10 4
1000 -
. 01
, m
o
(V
03
10 OPA 15 20 25
D stance fron burner (rr)
Furnace I
Fig. 6-1 SO* distribution along the flame axis [Type (a))
3c:c
200C
Symbol
Type (c)
Injecting method
Firing rate K%)
ra to
Ca/5 Mela
[Suffix
1000 -
OFA
15 2,0 25
Distance fron burner
F-rnace
Fig 6-2 SO2 distribution along the flane axis [Type (c)]
12-16
-------
-
Type (c)
+
(2564)
(2625)
(2351)
:
SOz in
gas
(1440)
SOj in
SOa in
coal
gas
(2564)
(2192)
SOa >n
ash
(1185)
SO2 in esh (159)
y.
Without
With
m Inlet SOj
limestone
limestone
Fig. 6-3 Mass balance of sulfur with HS coal
at furnace outlet
Synbcl
Type ic)
HS Coal (Muke)
Injecting rrelhoc
Coal
F ri-ig rate (°g
100
120
' 500
{%)
OFA
SGR
E* Oj
Min
1400
Cs/S Mola* rat o
® 1200
800
OFA
600,
Dist ance *ron* Burner (m)
Fig 7-1 Flame temperature distribution along the flame axis
with different firing rates
12-17
-------
Front view of the flame around the burner front
Fig, 7-2 Combustion flame with injecting nrethod in the syste-n througn
the OFA nozzle (Type (c))
Symbol
Coal
Fr^ng rate
I O •_ I ~ ¦ J A A 1
MS CoaS (Miikp) j
V-. 120 " :?o" 1 ioj I
SGR
%! Mir. Mm. Min
E*.0?
Cs;S Moia- ratio
J
i
i
ro 1
, 1
° I
1 i
-7 100--
Unbjrnec C " ny asn
Fig. 8-1 NOx and unburned C in fiy ash
vs. Ex.Os with HS coal
(Limestone injecting rrethod I Type (c) 5
o
z
Synbol
o • ~ ¦
A ~
Ccai
LS Coal (Plateau)
F-
ing rate | °c
IPO 100
120
OFA \%
20 20
20
sgr : %
M>n Min
N'ir
Ex.Ch °/c
—
—
Ca S
Molar rat n
0 : 2
0
Urtbur'v?d C in * v a5^
I |
~
3 4
Ex Oj
Fig 8-2 NOx and unburned C in fly ash
vs. Ex.Oz with LS coal
(Lmestone injecting iretnod : Type (c)
12-18
-------
1500
100C
l*II
ir i i/i cr
2 3 4
Residence time (sec)
Fig 9-1 Relation between gas temperature and residence time in actual boiler
01
*
&
-C —
i_ o; to
< Q £
,'MHI 4 tons/hour combustion
-test furnace at 120% load
Residence time (sec)
01 _
r°
O a
I
v
<0 — *-
c 4! . ® -
u I* o j)
? oSl
Ll o
o a.
J o-
,5 O 3
O o o
c
c
U = U
c _
o a;
-i
Fig. 9-2 Relation between gas temperature and residence time in
MHI 4tons/hour combustion test furnace
12-19
-------
TaDle 1 Analysis of coal used
t«m - Ur t
HS Coal (Hiika Coal)
LS Coai ^Plateau Coal)
HHV (Dry)
Ucil/kg
6.033
6,700
Sjrfdi« mo
MgO
%
0.4
c
flj
SiO?
%
0.1
u
3
A/i,Oa
%
0 1
•A
C
O
U
FeaOa
%
-
Ingition loss
\
%
44.3
Sp»c>ftc gravity
1
1
fCTl3
2.7
Table 3 Items,
methods and locations of measurements
Item j
Method
Location
«n
Oa !
Zircorua cell
«/>
».
i co,
N.D.l.R.
and
•«
\ CO ;
dr4g«r
detector
i *
s
MO*
Chemilu'ninescent analyzer
Hue gas duct
1
*
3
li.
SO*
Ultra violet absorption
analyser and calonmet
analyvs
ric
Flame teirp.
Optical
pyrometer
Observation
I ports at
j furnace
i
Heat flux
Thermo pile heat flow
detector
it
12-20
-------
PERFORMANCE OF SORBENTS WITH AND WITHOUT ADDITIVES,
INJECTED INTO A SMALL INNOVATIVE FURNACE
S. L. Rakes
U.S. Environmental Protection Agency
Office of Research and Development
Industrial Environmental Research Laboratory
Research Triangle Park, NC 27711
6. T. Joseph and J. M. Lorrain*
Northrop Services, Inc.
Research Triangle Park, NC 27709
ABSTRACT
The Environmental Protection Agency (EPA) Innovative Furnace was used to develop
information on the performance of sulfur sorbents with and without additives.
The sorbents were injected at two points in a furnace using a low-N0x type burner
fired at a 14 kW thermal input rate on coal or liquefied petroleum gas (LPG) pro-
pane. The sorbents, a calcitic limestone (Vicron), a pressure-hydrated dolomite,
and pure calcium hydroxide [CafOH^], were tested at Ca/S ratios of 1 and 2. The
additives, NaHC03 and Na2C03, mixed with the sorbent, were tested at 5 weight
percent of the sorbent. Sodium carbonate (Na2C03) appears to be an effective
additive, giving increases in sulfur reduction of 6 to 12 percentage points, de-
pending upon the amount of additive, the sorbent, and injection location.
INTRODUCTION
For the limestone injection multistage burner (LIMB) program to achieve complete
success, sulfur capture greater than that exhibited to date by limestone alone
must be achieved. This project was undertaken to determine the effects of addi-
tives on the utilization or sulfur capture performance of several sorbents. The
additives tested were those found most effective in bench-scale tests and with
high potential for commercial use; I.e., not toxic and relatively low in cost.
The sorbents used were chosen to represent sorbent types being considered; i.e.,
calcium carbonate (Vicron), calcium hydroxide, and hydrated dolomite (Genstar).
Vicron performance has been well characterized by previous LIMB work. The coal
used is one on which an extensive data base exists and which has been used by
several investigators.
This is a status report on the first of a series of investigations.
* Now with Acurex Corporation, Energy and Environmental Division, Southeast
Regional Office, Research Triangle Park, NC 27709
13-1
-------
DISCUSSION OF PROJECT
EPA Innovative Furnace
The EPA innovative furnace is a modified copy of a furnace built by Energy and
Environmental Research Corporation (EERC). The unit is a 6-in. (155-mm) internal
diameter down-fired unit. Three burner configurations may be used: an axial -
fired, a radial-fired, and a premix configuration with provision for tangential
or swirl air. Each of the three configurations may employ the distributed-mixing
burner concept. In the premixed burner, this is accomplished by mixing less than
the total air in the burner and adding the remainder of the air after the burner.
The firing rate of the furnace is adjustable with a nominal rating of 100,000
Btu/hr (29.6 kW) or about 3.2 kg coal/hr. Stable operation at a rate of about
1.4 kg coal/hr (42,000 to 50,000 Btu/hr) is a normal mode for testing. The unit
may be fired on either propane or coal. Normal practice is to fire the unit on
propane, unattended, over weekends, Friday afternoon to Monday morning, and over-
night, 5:30 p.m. to 8:00 a.m. This allows thermal stabilization of the unit and
avoids shock to the refractory by maintaining the unit at the testing temperature.
The unit is of conventional construction with rolled steel rings 30 1n. (762 mm)
in diameter and 1 ft (300 mm), 1.5 ft (450 run), or 2 ft (600 rrcn) high. The steel
rings are insulated by four courses of refractory beginning on the outside with a
good insulating, low strength castable block mix refractory, progressing inward
with castable insulation, following with Castolite 30,* and ending on the inside
with a high strength, erosion resistant "hard face" refractory such as Green Cast
97. After casting the refractory into the steel rings, the rings are "stacked"
using a gasket made of Kaowool or similar material on the outer periphery. A
small amount of uncured refractory 1s applied around the matching faces of the
inner central passage to prevent gas migration to the outer steel shell. Both
the gasket and the uncured inner seal material are formed and spread by the'
weight of the ring and the clamping force of the bolted ring flanges, thus en-
suring a good seal at each joint. The joint surfaces of the combustion passage
are "faired in" and troweled to give a reasonably smooth and uniform combustion
passage.
Inspection or view ports are provided, one a 2-in. (52-mm) circular port at about
7-1/2 in. (190 inn) from the burner tip, and a rectangular port 6 in. x 4 in.
(155 mm x 100 mm), centered 23 in. (585 mm) below the burner tip. Nine thermo-
couples, embedded in the refractory approximately 1/2-in. (12 mm) from the com-
bustion gas central passage, are spaced over the 11-ft (30.48-cm) long furnace
section. There are nine ports, approximately opposite the thermocouple loca-
tions, that are usable for either sampling or injection. Three larger rectangu-
lar ports are available for the insertion of chokes and cooling colls to "stage"
the combustion or divide the furnace into two or three zones with varied stoichio-
metry; the first is usually a "fuel-rich" or "reducing" zone, and the last, an
air-rich or burnout zone. The final furnace section contains cooling water coils
to quench any reactions before the sample is extracted for the continuous moni-
tors. See Figure 1 for diagram.
Solid fuel, such as coal, is metered from a K-Tron"" loss-in-weight feeder. A
similar feeder, but suited to delivery of a smaller amount, meters the sorbent,
such as limestone or lime. The sorbent material is aspirated and transported by
air and injected into the furnace through a water-cooled probe.
(*) Mention of trade names or commercial products does not constitute
endorsement or recommendation for use by the U.S. EPA.
13-2
-------
Operating conditions and emissions are monitored using the following instruments:
Dfgistrip III™ Data Logger
Continuous Monitors:
CO Beckman Model 864 Infrared
CO2 Beckman Model 864 Infrared
O2 Beckman Model 755
SO2 Anacon Model 207 U.V.
MO, NO2, N0X Thermoelectron Model 1 OAR
HC Beckman Model 402
Thermocouples, both B&K types, embedded or sealed
in the refractory.
In addition, gas temperatures are measured using suction pyrometry with stabi-
lized run conditions equal to those 1n the tests. No significant temperature
profile difference was found between gas firing and coal firing at equal heat
input rates, based on fuel heating value and excess air conditions. The actual
gas temperatures for different positions at the test conditions are thus estab-
lished, and no conversion from the thermocouple temperature to the gas tempera-
ture is required. The relationship between thermocouple temperature, at a given
point under known conditions, and the gas temperature is known from the previous
suction pyrometry work. Periodically, suction pyrometry tests are performed dur-
ing emission testing to confirm the gas temperatures found in earlier suction
pyrometry testing.
The SO2 readings and N0X readings from the continuous monitors are compared with
those from EPA Methods 6 and 7 (wet chemistry). It would appear from preliminary
results that the SO2 readings are within 10 percent of the values of EPA Method 6,
but that the N0X readings are more than 10 percent from the values of EPA Method
7. This is being investigated further. Some difficulty (leaking glass-to-glass
seal on the impinger vials) was encountered in collecting Method 7 samples.
There is a provision to inject pure SO2 into the propane immediately before the
burner to give a sulfur input equal to that when coal is fired. The SO2 is
metered into the fuel gas through a rotameter, and the SO2 measured 1n the stack
is compared to the level computed from fuel analysis and measured input of the
fuel and air.
Material Description
The LPG propane used is supplied by FCX, a farmers cooperative. Past experience
indicated that some variance in composition from one tank fill to the other was
possible. This variance, if present, did not manifest itself as changes in the
furnace conditions.
The Vicron limestone is a material designated as 45-3, in reference to the nomi-
nal particle size range of the cut or fraction used. When tested, 99.2 percent
passed through a 60-mesh sieve. This material has been used and reported in a
number of sorbent tests by Energy and Environmental Research Corporation, Acurex
Corporation, and Southern Research Institute. This particular batch was obtained
from Acurex Corporation 1n 1982.
The material referred to as "Genstar Henderson" is MgO-CaO pressure-hydrated to
Mg(OH)2'Ca(0H)p from the Henderson, Nevada, Plant of the Genstar Corporation. It
is a commercial product. Sieve testing gave 97.1 percent passing through a 30-
mesh screen and 66.3 percent passing through a 60-mesh screen.
13-3
-------
The calcium hydroxide is a technical grade laboratory supply item, Fisher number
C-88. Mo special storage precautions, other than closing the bag after removing
the test charge, were used. The sieve test showed 99.5 percent passing through
a 60-mesh sieve.
The sodium bicarbonate used is a laboratory reagent grade, Fisher number S-233.
In a sieve test, 99.7 percent passed through a 60-mesh sieve.
The sodium carbonate used is a laboratory reagent grade, Fisher number S-263. In
a sieve test, 100 percent passed through a 30-mesh sieve and 26.3 percent passed
through a 60-mesh sieve.
The coal used, a Pittsburgh number 8 seam coal, was obtained from Acurex Corpor-
ation in July 1983. The Commercial Testing and Engineering (CTE) analysis fur-
nished with the coal was confirmed by an analysis performed by Pennsylvania Elec-
tric Company (PENELEC) in November 1983. PENELEC analyzed three composite samples
submitted from the coal shipped to EPA, Research Triangle Park, NC, from Acurex
Corporation. The coal is stored in 50-1b (22.7 kg) bags in an unheated, enclosed
storage building. Table 1 presents a summary of coal analyses on an as-received
basis.
PROCEDURES
A standardized test plan for the evaluation of sorbents and sorbent/additive mix-
tures was followed. The plan called for holding the furnace firing conditions
constant while varying the sorbent feed rate, sorbent Injection point, and the
fuel/sulfur type (propane or C3H8/SO-2, Pittsburgh number 8 coal). A test matrix
was devised such that the three sorbents, with and without each of the two addi-
tives, would be evaluated under similar conditions according to the standardized
test plan. Figure 2 illustrates the test matrix used in this study.
Pretest Preparations
Before testing a new sorbent, procedures to ensure correct feeding were performed.
The feed rates required to obtain Ca/S ratios of 1 and 2 in the combustion zone
were calculated from the chemical analysis of the sorbent. The sorbent feeder
was recalibrated for each sorbent of different bulk density. If the new sorbent
material contained an additive, the sorbent and additive were thoroughly mixed in
the turbo-tumbler for 30 minutes.
Routine Testing
Throughout the study, the furnace firing conditions were held constant. The fir-
ing rate was 47,300 Btu/hr (14 kW), the air stoichiometric ratio (S.R.) was
1.25, and the combustion air was preheated to 250°F (121°C).
The testing of a sorbent or sorbent/additlve mixture was divided into two 1-day
test periods. On one day, the sorbent material was injected with the fuel
through the burner. On the other day, the sorbent material was injected through
a water-cooled probe 32 in. (0.8 m) below the burner. The testing routine given
below was followed at each sorbent injection point.
1. Switch from propane (C3H8) to Pittsburgh number 8 fuel, and adjust air
supplies to maintain an air S.R. = 1.25.
13-4
-------
2. Insert the sorbent injection probe (if necessary).
3. Calibrate the continuous emission monitors.
4. Ensure stable baseline combustion conditions.
5. Perform a minimum of two sorbent injections at a Ca/S ratio = 1, gather-
ing [SO2] reduction data.
6. Perform a minimum of two sorbent injections at a Ca/S ratio = 2, gather-
ing [SO23 reduction data.
7. Switch from Pittsburgh number 8 coal to C3H9 fuel, and adjust air
supplies to maintain an S.R. = 1.25.
8. Ensure proper combustion conditions (without SO2 doping).
9. Meter SO2 into the C3H8 fuel to provide sulfur loading equivalent to the
sulfur content of Pittsburgh number 8 coal.
10. Allow SO2 concentration in flue gas to stabilize.
11. Perform a minimum of two sorbent injections at a Ca/S ratio = 1, gather-
ing [SO21 reduction data.
12. Perform a minimum of two sorbent injections at a Ca/S ratio = 2, gather-
ing [SO2] reduction data.
13. Check the zero and span of the continuous monitors.
14. Shut off SO2 supply; purge lines.
15. Remove sorbent injection prcbe (if necessary).
16. Reset combustion air supplies for overnight operation.
Data Collection and Reduction
A Digistrip III data logger was used to collect experimental data from the contin-
uous emission monitors and the thermocouples. The data logger was programmed to
report instantaneous and 2-m1nute averages of flue gas concentrations. In addi-
tion to the data logger, strip chart recorders were used as a visual aid to the
operator to ensure stable furnace conditions.
The baseline CSO23 was taken as a 2-minute average when the strip chart recorder
indicated relatively stable conditions. The sorbent feeder was started exactly on
the data gathering interval. The reduced [SO2] was taken as the first stable 2-
minute average after the initiation of sorbent injection. The sorbent injection
intervals were as brief as possible to avoid possible effects from sorbent build-
up on the furnace wall. To ensure the elimination of a possible bias due to wall
effects, the [SO23 was required to return to baseline conditions within 10 minutes
of the cessation of sorbent feeding for the data to be considered valid. The aver-
age response time of the SO2 analyzer is about 2 minutes.
13-5
-------
In order to correct for slight variations in furnace conditions, all [SCSI's were
corrected to zero percent 02 conditions by Eq. (1).
[S02]
02
= [S02]
obs
21
21
w
(1)
obs
The corrected [SCSI's and the Ca/S (mol/mol) ratio are then used to determine the
percent Ca utilization by Eq. (2).
Ca utilization =
[S02] -[S02]
baseline, 0£ 0? reduced, 0% 0?
[S02J
baseline, 0% O2
TTiTsT^
(2)
Figure 3 is a reproduction of a typical strip chart trace of [S02] through two
sorbent injections. The figure is only for illustrative purposes since experi-
mental data is gathered digitally by the data logger; however, the trace provides
a guide to the eye and an indication of equipment performance.
In this case the sorbent utilization is approximately 15 percent since the A [S02]
is approximately 30 percent and the Ca/S ratio is 2.
Results
Figure 4 illustrates the average performance of each of the three sorbents with
no additives at two injection locations while burning coal. Figure 5 shows the
relative performance of Vicron injected 32 in. (0.8 m) below the burner while
firing with C3H3 and coal with and without additives. Figure 6 shows the rela-
tive performance of Genstar-Henderson injected 32 in. (0.8 m) below the burner
while firing with C3H8 arid coal with and without additives. Figure 7 shows the
relative performance of calcium hydroxide injected 32 in. (0.8 m) below the burn-
er while firing with C3HQ and coal with and without additives.
Each bar of the figures represents a minimum of four data points: two utiliza-
tions determined at a Ca/S of 1, and two at Ca/S of 2. Most bars represent six
or eight data points. The utilizations computed in Eqs. (1) and (2) are aver-
aged for the graphs. Results of sorbent/additive mixture injection through the
burner are not presented because they are still under study.
Discussion of Data
Figure 4 shows the performance of the three sorbents, with no additives, at two
injection locations while burning coal. When Injected through the burner, the
Vicron was only 2 percentage points below the performance when injected 32 in.
(0.8 m) below the burner at a more favorable temperature window. The lower in-
jection point put the material in at a gas temperature of about 2200 to 2300°F
( 1204 to 1260 0C).
The hydroxide and di-hydrate showed a greater increase 1n utilization percentage
when moved to the downsteam injection location.
13-6
-------
The Vicron material, shown in Figure 5, exhibited little change from gas to coal
firing when no additive was used. The effect of the carbonate additive with the
Vicron was very pronounced in gas firing, increasing the utilization from 15 per-
cent to over 28 percent. The effect of the carbonate additive was much less when
used with coal firing. It gave only an increase from just under 14 percent to
just under 21 percent in the coal-fired case. The bicarbonate showed some re-
duction in effectiveness going from gas to coal firing.
From the above, one could say that, whatever the capture process Is for Vicron
alone, it is not significantly sensitive to coal ash interactions. However,
when an effective additive such as the sodium carbonate is used, the capture
process becomes more sensitive to coal/ash interaction.
With the above in mind, one turns to the Genstar, Figure 6. This material is a
pressure-hydrated di-hydrate of the approximate formula CafOH^'MgtOH^. Here
the pattern is reversed. That is, the Genstar shows better utilization under
coal-firing conditions than under gas-firing conditions, both with and without
the additives, sodium carbonate and sodium bicarbonate. Unlike the previous case,
here the bicarbonate is the more effective additive.
One could make the conjecture that coal/ash interactions help the sulfur capture
process in a di-hydrate addition and that less improvement (in percentage points
of utilization) is possible through the use of additives with a di-hydrate com-
pared to the improvement possible with a limestone.
Figure 7 shows that a pure (technical grade) calcium hydroxide shows little
change from gas to coal firing in the percentage utilization either with or with-
out additives. Sodium carbonate is the more effective additive giving about 10
percentage points improvement under coal-fired conditions versus only about 3
percentage points for the bicarbonate. From the above, one can say that the
sulfur capture process for the hydroxide case 1s not greatly affected by the
coal/ash interactions, either with or without additives.
CONCLUSIONS
For Vicron limestone, calcium hydroxide, or dolomite di-hydrate, the better in-
jection point is below or past the burner rather than through the burner for
coal firing.
The sodium additives, sodium bicarbonate and sodium carbonate, provide an en-
hancement of calcium utilization and sulfur capture when used at a level of 5
weight percent of the sorbent with any of the calcium sorbents named above.
The di-hydrate material (Genstar Henderson) gave better utilization with coal
firing than gas firing, both with and without additives.
13-7
-------
AFTER ADDITION OF
NEW SECTIONS
FUEL
(SORBENT)
THERMOCOUPLES
AIR
p. BURNER
r~* *—t SAMPLE
PORTS
.15"
rL
¦ 28"
• 41"
-56'
-61'
¦80'
¦98"
.115"
136" —
29'
41"
57"
66"
80"
100"
116"
31 36"
SORBENT
"INJECTION
WATER
COOL
Figure 1. Diagram of LIMB furnace showing locations of
Type B thermocouples, sampling ports, and sorbent injections.
13-8
-------
Vicron-<
15
Genstar-Henderson
Calcium Hydroxide
No
Additive
5%
NaHCCh
5%
NapCCh
No
Additive
5%
NaHCOi
5%
Na?C0i
No
Additive
5%
NaHCO.i
5%
Na?C03
Coal
Inject
at
Burner
Inject
Below
Burner
C3h8
Inject
at
Burner
Inject
Below
Burner
Figure 2. Experimental test matrix.
-------
3* —»*
a in
e
1/1 -s
o ro
cr oj
ft3 •
C—*« _«f*
fD n
o p»
o i—l
3 C/1
1/1 O
• ro
X3
O
£V
r+
n
Cu
r>
ro
a.
e
"7J
3
S02 CONCENTRATION IS TAKEN DIGITALLY
BY KAYE INSTRUMENT
2 min Reduced Avg
2 min Baseline Avg
D- 2 min Reduced Avg
2 min Baseline Avg
-------
BUSKER INJECTION
DOWNSTREAM ifcJECTlOV
JMESTQNE
Figure 4. Performance of sorbents at two injection points
S
O
•- c
Q £
uj t
g &
UJ _
> S
O ""
sj \fi
Zi o
CQAl FIRtD CAKIONATE
CASHRED CARBONATE
GAS FIRED / 9'CAHft©VATT 1 COAL FIRED B; CARBONATE
NO ADO.'TivE
WITH ADDITIVE
Figure 5. Performance of Vicron and Vicron with additives. Injection
point is below burner.
13-11
-------
GAS HBEC CftOBOMAT^
COAl FIRED CAftaOKAf GAS Fl*ED / Bl-C ARBOATF COAi riqrc / 0I CABBONATT
NO ADClTlVf
With AODftvE
Figure 6. Perfornance of Genstar and Genstar with additives,
point is below burner.
Injection
30
GAS FiRCD CARBO*ATI | COAL ClPED t CA*BONATt | OAS FlMD Bl-CAfl 80NATT i COAL flOFO Bi-CARIOMATE
H ^OADDITlVr WrH AC3ITIVE
Figure 7. Performance of calciurr hydroxide and calcium hydroxide with
additives. Injection point is below burner.
13-12
-------
Table 1
PITTSBURGH #8 COAL ANALYSIS SUMMARY
ON AN AS-RECEIVED BASIS
December
1983
November 1983
Average o
Three
Analysis
CTE
PENELEC
PENELEC
PENELEC
PENELEC
Moisture (1)
1.92
1.42
1.36
1.38
1.39
Ash [%)
7.53
7.50
7.52
7.50
7.51
Volatile {%)
36.73
37.31
37.21
37.16
37.23
Fixed Carbon {%)
53.82
53.77
53.91
53.96
53.88
Pyritic Sulfur (X)
0.85
0.38
0.40
0.39
0.39
Total Sulfur (%)
2.58
2.62
2.60
2.58
2.60
Btu/lb
13,746
13,490
13,478
13,500
13,489
MJ/kg
32
31
31
31
31
13-13
-------
SESSION III: PILOT-SCALE DEVELOPMENT OF FURNACE INJECTION
Chairman, Michael McElroy, EPRI
13-14
-------
PILOT-SCALE CHARACTERIZATION OF A
DRY CALCIUM-BASED SORBENT SO2 CONTROL TECHNIQUE
COMBINED WITH A LOW-NO TANGENTIALLY FIRED SYSTEM
John Kelly, Shigeto Ohmine,* Richard Martin1"
Acurex Corporation
555 Clyde Avenue
P.O. Box 7555
Mountain View, California 94039
and
Dennis C. Drehmel
Industrial Environmental Research Laboratory-RTP
U.S. Environmental Protection Agency
Research Triangle Park, North Carolina 27711
ABSTRACT
A 380-kWt (1.3 million Btu/hr) two-burner-level tangentially fired pilot-scale
facility was used to characterize a dry calcium-based sorbent SO2 capture technique
combined with an offset auxiliary air low-NOx burner with and without overfire air.
Baseline tests using a conventional burner design, with and without overfire air,
showed that the facility properly simulates full-scale uncontrolled and controlled
NO emissions and furnace temperature histories. SO2 test results with dry sorbent
injection and the conventional burner were consistent in level and trend with
limited field data. Efficiency, as determined by carbon monoxide, unburned
hydrocarbon, and carbon in flyash levels, was good for all test conditions and
consistent with field practice.
Dry sorbent SO2 test results showed that SO2 capture is increased by: (1) utilizing
conventional rapid fuel/air mixing burners, (2) injecting sorbent away from the
flame (i.e., not mixing sorbent with the fuel), (3) avoiding contact between ash and
sorbent, and (4) holding sorbent in the sulfation temperature zone (i.e., 1,505
to 1,255K) as long as possible. In addition, fuel and sorbent type had a
significant impact on SO2 capture.
INTRODUCTION
Coal-fired tangential boilers produce 35 percent of the N0X and consume, in Btu's,
45 percent of the fuel used in all coal-fired utility boilers (Reference 1). Since
almost all of the sulfur contained 1n the coal appears in the flue gas as an oxide,
the significant coal fuel usage of tangential boilers also makes them an important
source of S0X emissions.
^Presently affiliated with Imai Marketing, Inc., Sunnyvale, California
Presently at the University of California, Berkeley, California
14-1
-------
The significant contribution of coal-fired tangential boilers to national N0X and
S0X emissions and the expected future increase in the number of these' bo1lers make
them candidates for emission control development. Demonstrating cost-effective N0X
and S0X reduction technology for these boilers will provide regulatory agencies with
additional emissions control options which could be exercised in the future to
prevent environmental degradation.
A pilot-scale low-NOx tangential burner, which incorporates auxiliary air angle
offset relative to the primary fuel jets, has been tested; it shows significant
promise in controlling N0X below current levels. Tests in the EPA pilot-scale test
facility have shown the potential of achieving N0X emissions in the range of 90 to
130 ng NO2 per J (0.2 to 0.3 lb NO2 per million Btu) (Reference 2). This low-N0x
system has gas temperature and oxygen concentration histories which could benefit
the dry sorbent S02 capture relative to conventional burners (Reference 3). With
the low-N0x burner, heat release is spread out in the furnace, thereby lowering peak
temperatures; in conventional burner systems, high peak temperatures can possibly
deactivate sorbent. Also, oxygen deficient zones are present in the low-N0x system.
In these zones, the sorbent could possibly capture sulfur as a sulfide rather than a
sulfate. Sulfide capture could have reaction rate and product volume advantages
over sulfate capture, which is the typical mode of capture in conventional burner
systems.
Given the potential for good N0X and S02 control, low-N0x burner dry sorbent
injection tests were performed to determine: (1) the feasibility of the low-N0x
burner to achieve high 502 capture through dry sorbent injection, (2) the influence
of the injection system and combustor design variables on S02 control, and (3) the
S02 emission performance of several fuels and sorbents.
Description of Conventional Coal-Fired Tangential Systems
Figure 1 illustrates the main features and flow patterns of a tangentially fired
boiler. Fuel and air flow into the furnace through rectangular registers located in
the four corners of the unit. As shown in Figure 1, the burner consists of a
primary air register which introduces the fuel and a small fraction of the
combustion air Into the furnace. Surrounding the primary register is the annular
air register, which contains a fraction of the secondary combustion air. The
auxiliary air registers, above and below the primary and annular registers, contain
the bulk of the secondary combustion air flow. The jets are inherently nonswirling
and fuel/air mixing is slow, relative to front-wal1-fired boilers.
The tangential alignment of the centerlines of the corner jets to the circumference
of a circle in the center of the furnace promotes the formation of a large-scale
vortex within the furnace. Ignition of the fuel is provided by impingement of hot
burnt gases from laterally adjacent burners and large-scale internal recirculation
of combusted gases. Because ignition occurs primarily on the vortex core side of
the fuel jet (see Figure 1), combustion is asymmetric in the near-burner horizontal
plane.
Pilot-Scale Development of a Low-N0x Tangential System
Figure 2 presents a top and side view schematic of the low-N0x system as installed
in the EPA pilot-scale test facility (Reference 2). In this schematic, fuel-rich
and fuel-lean zones in the firebox are identified. The major system features are:
14-2
-------
(1) fuel directed at the conventional burner yaw angle* into the center of the
furnace, (2) some secondary air introduced at the same yaw angle as the fuel, in an
annular space surrounding the fuel jet, and (3) the rest of the secondary air
directed along the wall at and above the fuel jet elevation. During operation, coal
and primary air are injected into the furnace at roughly a 0° to 8° yaw angle with
respect to the firebox diagonal. The primary air flow represents about 15 percent
of the total combustion air. Surrounding the primary fuel jet is an annular passage
in which some of the secondary combustion air is flowing.
The fuel jet initially burns on the jet boundaries in a diffusion flame manner.
Since most of the fuel nitrogen evolves inside the oxygen-deficient jet, the initial
N0X formation is limited for the slow-mix diffusion burning of the jet.
As previously shown in Figure 2, a fraction of the combustion air is directed along
the furnace walls. Since the wall air yaw angle is large (approximatey 45° for a
square firebox), the wall air contributes significantly to the vortex swirl level.
Because the wall jet places the air in a relatively aerodynamically quiescent region
along the wall, 1t remains separate from the fuel 1n the center of the furnace for a
longer period of time than if it was injected at some positive angle away from the
furnace wall. In addition to large wall-a1r yaw angles, Intermediate yaw angles
have been tested which exhibit N0X reduction benefits.
When the wall jets contain a significant fraction of the air flow needed to complete
combustion, the center of the furnace at the fuel entry elevation operates with a
deficiency of oxygen. Under these conditions, N0X (previously formed during the
initial diffusion burning) and any remaining fuel nitrogen evolved in the center of
the furnace are decayed to molecular nitrogen by homogeneous and heterogeneous
chemical processes. Since molecular nitrogen 1s relatively unreactive for N0X
formation in the fuel burnout zone, the decay processes in the center of the furnace
produce low-N0x levels in the stack. Eventually, the air directed along the furnace
walls mixes with the oxygen-deficient vortex core gases and char, completing
combustion.
Besides limiting N0X production to the range 90 to 130 ng NO2 per J (0.2 to 0.3 lb
per million Btu), directing a fraction of the combustion air along the furnace walls
maintains the walls under oxidizing conditions. This eliminates degradation of the
metallic furnace surface by reducing gases. In addition, the Iron In the ash
deposits will be maintained in the ferric rather than ferrous form, thereby raising
the ash fusion temperature and reducing the wall slagging potential.
PILOT-SCALE TEST FACILITY
The pilot-scale facility and instrumentation used 1n the combined low-NO^/SO^^ test
program is a modification of the EPA multifuel furnace pilot-scale facility used on
the prior EPA-sponsored coal-fired tangential system N0X emissions control program
(Reference 2). The facility was modified extensively to upgrade the modeling of
full-scale multiple-burner-level tangential systems and thermal profile effects.
These modifications were necessary to properly simulate N0X emissions and dry
sorbent SO2 control of modern large-scale multiple-burner tangential systems
modified for N0X and SO2 control.
The pilot-scale facility geometry used on this program 1s an approximate 1 to 14
scaling of an existing, modern-design Combustion Engineering tangentially fired
*Yaw angle is defined as the angle the burner port centerline makes with the
firebox diagonal. A typical yaw angle is 6° off the firebox diagonal.
i/i 1
-------
boiler. The overall external height of the furnace is approximately 3.7 m (12 ft)
and it is capable of firing coal or gas at rates up to 587 kWt (2 million Btu/hr).
The main firebox, or radiant section, is a refractory-lined chamber with a
rectangular 1 m by 1.1 m (38 in. by 44 in.) cross section. Two levels of burners
are situated at each corner. Each burner set includes primary air, annular air, and
auxiliary air ports. Overfire air ports are also installed above the pair of
burners at each corner. The primary air and annular air jets have a small degree of
firing angle variability -- up to 8° from the firebox diagonal. All other air jets
can be directed at angles from 6° to 37° away from the diagonal. Three nominal
angles for the auxiliary jets, denoted conventional, intermediate, and a1r-on-wall,
are 6", 21°, and 37° away from the firebox diagonal, respectively. Auxiliary and
overfire air jet velocity can be varied independently of air flowrate through the
burners. Three sizes of auxiliary and overfire air ports can be interchanged to
achieve major variations in combustion air velocity. Overfire air ports are located
0.38 m (15 in.) above the burner centerline.
Volumetric heat release and gas residence time in the radiant section are closely
approximated between pilot and full scale. Realistic furnace-exit gas temperatures
are achieved by low heat loss refractory linings in the main firebox and ash pit.
The facility is fully instrumented with continuous monitors to measure stack NO,
S02, UHC, CO, C02, and 02 gaseous concentrations, and thermocouples are placed at
various locations throughout the furnace to monitor furnace wall and gas
temperatures. In addition, suction pyrometers and gaseous and solid sampling probes
can be inserted into the furnace at various locations in the radiant and convectlve
sections to monitor in situ gas temperatures and gas and solid compositions.
S02 REDUCTION TEST RESULTS AND DISCUSSION
The goal of these tests was to characterize the effect of major parameters
influencing dry sorbent injection S02 reduction in the tangentially fired system.
Emphasis was placed on testing under the low-NOx firing condition (auxiliary air
angle at air-on-wall) to Investigate the effectiveness of simultaneous N0x/S02
control.
The major sorbent injection parameters studied 1n these tests Included sorbent
injection method and location, convective heat exchanger tube arrangement, load,
excess air, flue gas recirculation (FGR), fuel type, and sorbent type. During the
tests, medium velocity air jets were utilized and sorbent transport air flow was
fixed at 22.7 kg/hr (50 lb/hr). All Ca/S ratios reported do not Include the
inherent calcium present in the coal.
*
Baseline S02 Reduction Test Results Summary
The objective of these tests was to determine if the conventional burner pilot-scale
S02 reduction results were comparable in level and trend with past studies. The
tests consisted of varying the major dry sorbent injection parameters, observing
their trends and effects, and then, where appropriate, comparing them with past
studies. All of the baseline S02 reduction tests were carried out with a
conventional burner design in which the auxiliary air offset angle is equal to the
fuel jet angle. The parameters investigated included calcium-to-sulfur ratio,
excess air, load, sorbent injection location, sorbent type, and coal type. Coal
properties and sorbent chemical compositions and sizes are given in Tables 1 and 2,
respecti vely.
14-4
-------
The test results for these parameters showed that the baseline S02 reduction trends
were generally consistent with previous results. Also, based on comparisons with
limited field data, the pilot-scale results reflect field experience. This
comparison is encouraging evidence that the pilot-scale facility SO? reduction
results can be used as a guide in assessing how SO2 reduction performance would vary
with design or operating conditions in a full-scale furnace.
In addition to SO2 comparisons, controlled and uncontrolled NO emissions and
temperature histories compared favorably with field levels. Efficiency, as
determined by carbon monoxide, unburned hydrocarbon, and carbon in flyash levels,
was good for all test conditions and consistent with field practice.
Conventional and Low-N0x/S02 Reduction Test Results
The objective of these tests was to determine the Impact of major N0x-reduction
methods (I.e., offset auxiliary air and overfire air) on the dry sorbent injection
SO2 reduction process. Various combinations of overfire air and auxiliary air
offset angle were used in these tests. Overfire air sorbent injection, where
sorbent was thoroughly mixed with the overfire air prior to injection, was utilized
in these tests.
Figure 3 shows that with the conventional burner, overfire air N0X control reduces
sorbent SO2 capture performance over the whole range of Ca/S values tested. As
indicated on the figure, Illinois coal and El Dorado sorbent were used in these
tests, and load and excess air were set at nominal values (390 kW and 20 percent,
respectively). Next to the percent overfire air designation in Figure 3 is a string
of four numbers separated by slashes which denote the percent of the total
combustion air flowing in the primary, annular, auxiliary, and overfire air ports,
respectively. For subsequent figures, these designations will appear in the figure
conditions table aligned with "air splits." In addition, the string of three
numbers aligned with "convective" in the conditions table on the figure denotes the
number of coolant loops in, respectively, the first three tube banks in the furnace
conventive heat exchanger section. Except where noted, these conditions were
maintained for all of the tests reported.
The test results in Figure 4 show that offsetting the auxiliary air to intermediate
and air-on-wall angles with overfire air, further reduces SO^ capture relative to
the conventional burner with overfire air case. From both Figures 3 and 4 it is
apparent that lowering N0X, through overfire air addition alone and in combination
with offset auxiliary air, reduces sorbent SO2 capture effectiveness.
The varying levels of SO2 reduction are possibly caused by the differences in
fuel-components-air-sorbent contact (or, in general terms, "mixing"). Mixing
pattern differences are a direct result of the three different auxiliary air angles
and overfire air addition. Besides mixing, the different air injection approaches
alter residence time distribution in the furnace. Additionally, the low-N0x systems
control N0X by delaying fuel/air mixing and thereby stretching out the furnace heat
release. The presence of local high-temperature flames in the sorbent injection
zone could be detrimental to sorbent activity and therefore SO2 capture.
Even though sorbent SO^ capture was not optimal with the low-N0x techniques tested,
the offset auxiliary air low-N0x burner was utilized on all further sorbent
injection tests to determine the Impact of design and operating parameters on a
combined N0X and SO2 control system.
14-5
-------
Effect of Sorbent Injection Method and Location
The objective of this test sequence was to determine the effect of sorbent injection
method and location on SO2 reduction with the offset auxiliary air low-NOx burner.
Figure 5 shows the effect on SO2 reduction of injecting sorbent in the overfire,
lower auxiliary (bottom), and primary air ports. In these tests, the sorbent was
thoroughly mixed with the air flow prior to injection through the air ports into the
furnace. The SO2 reduction results show that overfire air sorbent injection
produced SO2 reduction similar to, but slightly better than, lower auxiliary air
sorbent injection. Primary air injection of sorbent consistently gives the lowest
SO2 reduction levels. With primary air injection, the sorbent particle is expected
to be in immediate contact with devolatilizing and burning fuel and therefore may
deactivate due to high peak temperatures and ash contacting. This may be
responsible for the lower SO2 reduction levels for primary air sorbent injection
presented in Figure 5.
SO2 reduction results for probe sorbent injection 0.3 m (12 in.) above and 0.2 m
(8 in.) below the burner zone were also obtained. The sorbent was injected upward
into the center of the furnace through a 13-mm (1/2-in.) diameter nozzle. The SO2
reduction levels for the two probe injection locations were found to be similar to
the overfire air and lower auxiliary air port injection test results. Only primary
air injection of sorbent gave reduced SO2 capture. A possible explanation for such
differences is the sorbent/fuel-components contact. Sorbent injection through
overfire air, lower auxiliary air, and the water-cooled probe all provide a delay in
contact of sorbent particles and fuel components. Such mixing is probably
accomplished downstream of the hot primary combustion zone and outside of the
flames, where gas temperatures are cooler. Sorbent injection through the primary
ports should result in the immediate contact of burning fuel and sorbent in the
high-temperature flame zones. The proximity of the fuel components (including ash)
and sorbent, at high temperatures, may result in sorbent deactivation and therefore
reduced SO2 capture levels.
The similarity of SO2 reduction levels (except for sorbent injection through primary
air) suggests that, regardless of sorbent injection location, the bulk of S02
capture may be taking place farther downstream in the furnace. If sulfur is being
captured as SO2, equilibrium constraints under typical combustion conditions dictate
that sorbent/gas temperatures need to be below 1.500K (2,300°F) for capture to
occur. This would support the postcombustion zone capture of SO2, since radiant
section gas temperatures are typically above 1,500K (2,300°F). To approximately
locate the zones of sulfur/sorbent reaction and test the above postcombustion zone
SO2 capture speculations, sorbent was injected into the furnace's upper radiant and
convective sections using a single-point water-cooled injection probe, as in the
center of the furnace injection tests discussed above.
Injecting sorbent into the convective section from the radiant section or into the
convective section counter to the gas flow gave results similar to all injection
locations, except the primary air injection location and ceiling-down injection
location. The reason ceiling-down injection of sorbent produces I0W-SO2 reduction
levels is not understood. It could be speculated that ceiling-down injection
directs some of the sorbent into flame zones, where deactivation of the sorbent
occurs. However, no data are available to support this conjecture.
The similarity of most of these results, for different injection locations, implies
that stable SO? capture is taking place in the convective section where temperatures
are below 1.533K (2,300°F) and that there is no significant capture occurring in the
radiant section of the furnace. Even if there is sulfur capture in the radiant
section, the calcium-sulfur product may be quickly dissociated.
14-6
-------
All of the previously mentioned SO2 reduction test results using the water-cooled
sorbent injection probe were carried out with a single-point 13-mm (1/2-in.)
diameter outlet nozzle. The effect of using different nozzle sizes was also
investigated. These results showed that nozzle size, and thereby nozzle exit
velocity, had very little effect on SO2 reduction. Apparently, in the relatively
small furnace, sorbent/gas mixing is not a major problem.
Effect of Furnace Load, Convectlve Heat Transfer Tube Arrangements, and FGR
Furnace load, convective heat transfer tube arrangement, and FGR are parameters that
influence dry sorbent injection by affecting primarily the temperature/residence
time profile, and mixing patterns in the cases of furnace load and FGR changes, and
possibly combustion chemistry in the case of FGR. Temperature profiles were
measured to help explain the effect of FGR and load on sorbent SO2 capture.
The test results in Figure 6 show that lowering load significantly improves SO2
reduction. The improved SO2 reduction can be related to a modified thermal history
as load is reduced. The temperature profiles for normal (390-kWjO and low-load
(300-kWf) cases are considerably different. In the temperature zone of 1,255 to
1.505K 11,800* to 2,2509F), the normal-load case provides approximately 1 sec
residence time, while the low-load case provides about 4 sec. Such large
differences in residence time in the sulfation temperature zone may account for the
significant improvement in SO2 reduction as load is reduced.
The effect of FGR on SO? reduction was determined by pumping filtered flue gas out
of the baghouse and back into the furnace through the overfire air ports. Despite
some degree of scatter in the data, the test results in Figure 7 show that the
addition of FGR improves SO2 reduction. This trend was expected because FGR
addition should cool the combustion products to create a more favorable
temperature/residence time profile for sorbent-502 reactions. Testing showed that
the addition of FGR substantially lowers the entire thermal profile of the furnace,
leading to the doubling of residence time in the temperature range of 1,355 to
1.505K (1,800° to 2,250°F), compared to the case with no FGR.
As another check of the effect of temperature history on sorbent capture, three
different convective heat transfer tube arrangements were tested. By altering just
the convective tube arrangements, mixing and heat release patterns in the lower
furnace were not disturbed. Therefore, through this test approach, the significance
of convective section temperature history on SO2 capture could be clearly shown. As
anticipated, lowering the convective section temperature into the sulfation
temperature zone (less than 1,505K (2,250°F)) for a significant length of time
increases SO2 capture. Greater than factors of two in capture were observed between
different heat transfer tube arrangements. These results supported the low load and
FGR tests and clearly showed the importance of convective section temperature
history, independent of radiant section conditions.
To summarize the temperature history results, Figure 8 presents the SO? reduction
obtained for a given amount of residence time in the temperature range between
1,255K and 1.505K (1,800CF and 2,250"F). Temperatures were varied by changes in
load and convective heat exchange tube arrangements. For nominal conditions of
Illinois coal, El Dorado sorbent at a Ca/S of 2, probe injection 0.3 m (12 in.)
above the burners with the baseline low-N0x firing conditions, the SO2 trend shows a
clear increase of capture with increases in residence time in the temperature range
conducive to sulfate formation. It should be noted that FGR test results are also
consistent with these trends. This trend with residence time at temperature is
evident for both El Dorado limestone and dolomite, with medium and coarse dolomite
showing better capture for equivalent residence time.
14-7
-------
Effect of Fuel Type
Four fuels were tested for sulfur capture with El Dorado limestone. Properties of
these fuels are presented in Table 1. Two noteworthy features about the fuels are
their inherent levels of sulfur (ranging from 0.6 percent in North Dakota lignite
and Utah bituminous to 3.7 percent in Illinois bituminous) and their heating values
(ranging from approximately 24,207 J/g [10,407 Btu/lb] for the lignite up to
29,554 J/g [12,760 Btu/lb] for the bituminous coals). SO2 reduction test results
presented 1n Figure 9 for the three bituminous coals fired under nominal conditions
show that high-sulfur Illinois coal gives the highest capture, while the lowest
capture was obtained with the moderately high-sulfur Indiana coal. Low-sulfur Utah
coal gave an intermediate level of capture.
The North Dakota lignite tests were conducted at a reduced load (70 percent of
nominal) consistent with lignite fuels firing intensity practice. Under these
conditions, more SO2 capture is obtained from Illinois coal (42 percent at Ca/S = 2)
than from the lignite (32 percent at Ca/S = 2).
SO2 reduction from a "dirty" fuel, Illinois coal, was compared to that of a "clean"
fuel, natural gas, doped with an equivalent concentration of sulfur 1n the form of
hydrogen sulfide (H2S). Compared to the coal, a factor of two greater sulfur
capture was found for doped gas. It was speculated that either there is some aspect
of homogeneous gas-phase combustion which enhances SO2 reduction or a feature of
heterogeneous coal combustion which inhibits reduction.
A set of tests was performed to determine if the mineral matter in coal was
inhibiting sulfur capture. For these experiments, two mixtures of El Dorado
limestone and Utah coal ash were prepared and injected into the doped-gas fireball.
The mixtures were blended to produce flue-gas-ash concentrations which would
correspond to a high- and a low-ash fuel. At limestone/ash injection rates which
equated to calcium sulfur ratios of 2.0, the low-ash mixture simulated a fuel which
would be 3 weight percent ash, and the high-ash mixture simulated a 13 percent ash
fuel.
In Figure 10, the results of these tests show that, for a calcium/sulfur
stoichiometry of 2.0, the pure limestone captured 19 percent of the SOj, while the
low- and high-ash blends captured 12 and 8 percent, respectively. These results
indicate that Utah coal ash has a detrimental effect on limestone/sulfur reactivity.
It can be speculated that the ash material, through either a gas-to-solid or
solid-to-solid process, combines with the sorbent to reduce its overall activity for
sulfur capture.
Effect of Sorbent Type
Five sorbents were tested under baseline firing conditions with Illinois coal.
Properties of these sorbents are presented in Table 2. Test results presented in
Figure 11 show that sulfur capture varied significantly, from 5 to 29 percent (at
Ca/S = 2), with the sorbents ranked in order of increasing capture as follows:
marl, vicron, chalk, El Dorado, and dolomite.
CONCLUSIONS
Pilot-scale tests showed that delaying the fuel/air mixing, through 0FA and
offset-auxiliary air with 0FA low-N0x approaches, reduces dry sorbent injection SO2
capture effectiveness.
14-8
-------
Extensive dry sorbent Injection tests with the low-NOx offset auxiliary burner
(i.e., air-on-wall) without OFA showed that:
• Temperature history, as affected by changes in load, convective tube
arrangements, and flue gas recirculation, has a very strong impact on
sorbent SO2 capture. Sorbent capture appears to be proportional to
residence time in the 1,505 to 1,255K (2,250°F to 1,800°F) temperature
band.
• Adding ash mineral matter to a doped gas flame significantly reduces
sorbent SO2 capture.
• Injection of sorbent into the primary air port gives consistently lower SO2
reduction, presumably by ash interaction with the sorbent.
• Sorbent injection at all other locations (lower auxiliary air, OFA, probe
center injection above and below the burners, and probe-roof convective and
counter convective), except roof-down probe injection, gave similar and
better SO2 reduction than primary injection.
• Downward sorbent injection from a probe in the center of the roof gives SO2
reductions similar to primary injection.
• For the parameter ranges tested, sorbent injection velocity has an
insignificant effect on SO2 reduction in the relatively small scale
furnace.
• Fuel and sorbent type have a significant effect on S02 reduction.
These test results strongly suggest the following path to optimum sorbent SO2
capture:
• Inject the sorbent downstream in the upper furnace or convective section to
avoid any sorbent deactivation by either high peak temperatures or by fuel
ash.
• Hold sorbent in the active sulfation temperature zone as long as possible
through modification of the convective section heat transfer arrangement,
load or flue gas recirculation.
• Use sorbents which are resistant to deactivation through ash interaction
and high temperatures.
REFERENCES
1. Lim, K. J., L. R. Waterland, C. Castaldini, Z. Chiba, and E. B. Higginbotham.
"Environmental Assessment of Utility Boiler Combustion Modification N0X
Controls, Volume 1." EPA-600/7-80-075a (NTIS PB80-220957), April 1980.
2. Kelly, J. T., R. A. Brown, E. K. Chu, J. B. Wightman, R. L. Pam, E. L. Swenson,
E.B. Merrick, and C.F. Busch. "Pilot-Scale Development of a Low-N0x Coal-Fired
Tangential System," EPA-600/7-81-137 (NTIS PB81-242513), August 1981.
3. Zallen, D. M., R. Gershman, M. P. Heap, and W. H. Nurick. "The Generalization
of Low Emission Coal Burner Technology." In Proceedings of the Third Stationary
Source Combustion Symposium, Vol. II, EPA-600/7-79-050b (NTIS PB292540),
February 1979, pp. 73-109.
14-9
-------
Near burner
Burner air and fuel registers
Vortex Interaction
vorte*
io3 view
buH of
combustion
Side view
Fiaure 1. Schematic of Tangentially Fired System
Secondary _j
Fuel /primary air *
CORS'ER BtHNEH CETAIl
Figure 2. Low-NOx Air-on-Wall Concept Schematic
14-10
-------
Illinois
El Dorado
390 kU
201
Conventional
OFA
20-20-20
Fuel
Sorbent
Firing rate
Excess iir
Ancle
Injection
Convecti ve
a _
o _
a _
LI o_
OFA (Percent)
0 {15/15/7 C/0)
40 (1 5/15/30/40)
~ _
2
3
U
1
5
6
7
4
CALCIUM/SULFUR
Figure 3. Effect of Combustion Air Distribution with Overfire Air
Sorbent Injection on SO2 Reduction
14-11
-------
Fuel
Sorbent
Firing rate
Excess air
Air splits
I n jecti on
Coivecti ve
Illinois
El Dorado
390 kWt
202
15/15/30/40
OFA
20-20-20
~ CONVENTIONAL
A INTERMEDIATE
Q AIR-ON-WALL
1 1 r
3 4 5 6
CALCIUM/SULFUR
Figure 4. Effect of Auxiliary A1r Angle with Overfire Air Sorbent
Injection on S02 Reduction
14-12
-------
o.
O.
s
n 8*
Fuel
SorDent
Firing rate
Excess air
Air splits
Angle
1111noi s
El Dorado
390 kWt
20?
12.5/8.5/79/0
Ai r-on-wal1
~
£
O
PR[MARY
OF A
1_0»ER AUX
calCIum/sui
s
.FUR
Figure 5. Effect of Sorbent Injection Location on SO2 Reduction
Fuel
Sorbent
Excess air
Air splits
Angle
Injection
Convective
11 "1 i noi s
El Dorado
20%
12.5/8.5/79/0
Air-on-wall
0.3 M above
8-0-8
300 KW
300 KW
3 4 5
CALCIUM/SULFUR
Figure 6. Effect of Firing Rate on SO2 Reduction
14-13
-------
Illi noi 5
El Dorado
390 *Wt
Fuel
Sorbent
Firing rate
Excess air
Air splits
Angle
Injection
O _
L2.5/B.5/79/FGR
At r-on-wa11
0.3 M above
O _
o_
Ld o _
ID °-
W ' (M
o_l
2
3
5
6
O
A
7
CALCIUM/SULFUR
Fi'gure 7. Effect of Flue Gas Recirculation on SO2 Reduction
14-14
-------
o.
K
TEMPERATURE RANGE 1505-1255 K
o.
D
K' 8
z
2.J
U
~
~
^ g-
<\J
D
LP 0 —
U» ry
~
A
8
~
~
ILLINOIS ~ ELDORADO
COARSE ELDORAOO
fine dolomite
HE01UH D0LDH1TE
COARSE DOLOMITE
~
o-r
i ~ r i i t • r
l.S 2 2. S 3 3.5 4
RESIDENCE TIME
«. 5
Figure 8. Effect of Residence Time in the 1,505 to 1.255K
(2,250° to 1,800"F) Temperature Band at Ca/S
of 2 on SO2 Reduction
1/1 1
-------
o.
O.
to
Sorbent
firing rate
Excess air
Air splits
AnqTe
In jecti cn
Convecti ve
El Dorado
390 kWt
20%
12.5/fl.5/79/0
Air-on-wal1
0,3 M above
8-0-8
~.
~
ni
O .
<\J
o
~ ILLINOIS #5
A INDIANA #2
O UTAH PSL
-r
2
~r
3
-r
6
5
CALCIUM/SULFUR
Figure 9. Effect of Fuel Type at Nominal Load on SO2 Reduction
14-16
-------
Fuel
Firing rate
Excess air
Air splits
Angle
Injection
Convective
Nat-iral gas +
390 kWt "
20i
12.5/8.5/79/0
Air-on-wal 1
0.3 M above
0-0-20
H2S
EL DORADO lOOX
EL DOR. B8X ASH I 2X
EL DOR. 61 * ASH 39X
T 1 1 r
3 4 5 6
CALCIUM/SULFUR
Figure 10. Effect of Ash Additive on SO? Reduction
14-17
-------
11 nnois
Fuel
Firing rate
Excess air
Air sol its
Injection
Convecti ve
Angle
390 kW
20?
12.5/9.5/79/0
0.3 M above
8-0-8
Ai r-on-wall
EL DORADO
V ICRQN
MARL
DOLOHITE
Chalk
CALCIUM/SULFUR
Figure 11. Effect of Sorbent Type on SO2 Reduction
14-18
-------
Table 1
TEST COAL PROPERTIES
Utah
Power &
North
Ultimate Analysis
Illinois
Light
Dakota
Indiana
(as received)
Bi tuminous
Bi tuminous
Ligni te
Bituminoi
C
62.0
62.5
43.6
70.7
H
4.5
4.6
3.0
4.7
0
7.6
9.8
13.4
8.7
N
1.0
1.1
0.1
1.3
S
3.7
0.6
0.6
1.3
Ash
11.5
18.0
8.8
10.3
HpO
9.6
3.5
30.6
3.1
Heating value (Btu/lb)
Dry
12,325
11,516
10,407
12,760
Wet
11,138
11,116
7,222
Proximate analysis
(as received)
Volatile
36.5
36.9
29.7
34.3
Fixed carbon
42.4
41.7
30.9
52.3
Table 2
SORBENT CHEMICAL COMPOSITION AND SIZES
Sorbent
Properties
CaC03 {%)
MgC03 (%)
Particle
size
El Dorado
Limestone,
El Dorado,
Cali fornia
98.0
0.9
Pfizer
Dolomlte,
Gibsonburg,
Ohio
54.3
45.3
99.81-325 mesh 99J-325 mesh
88J-200 mesh
Vicron (Pfizer)
Limestone,
Lucerne Valley,
California
97.0
1.6
992-325 mesh
Michigan
Marl,
Hopkins,
Michigan
92.2
6.7
Kansas
Chalk,
Jewel 1,
Kansas
82.8
0.9
99J-325 mesh
25J-200 mesh
14-19
-------
BOILER SIMULATOR STUDIES ON SOFBENT UTILIZATION
FOR SO2 CONTROL
B. J. Overmoe, S. L. Chen, L. Ho, W. R. Seeker,
M. P. Heap, and D. W. Pershing
Energy and Environmental Systems Corporation
18 Mason
Irvine, California 92714
ABSTRACT
A 300 kW Boiler Simulator Furnace was used to investigate the influence of com-
bustion and sorbent parameters on the effectiveness of dry sorbent injection
for SO2 control under conditions typical of current utility practice. Extensive
characterization studies were conducted to investigate the role of boiler ther-
mal history on capture effectiveness for limestones, dolomites, and slaked
limes. Data were also obtained for high surface area sorbents [produced by
pressure slaking) and for promoted sorbents [produced by the addition of
appropriate metallic additives) as a function of thermal environment, sorbent
injection location, calcium to sulfur molar ratio, and 502 partial pressure.
In general the results show that captures in excess of 50% at a Ca/S ratio of
2.0 can be achieved by several alternative methods. The experimental studies
were supported by theoretical calculations using grain and pore models which
combined consideration of the heterogeneous chemical reaction and diffusion pro-
cesses.
INTRODUCTION
Sulfur capture by calcium based sorbent has been the subject of much research
over the last two decades(l-4). Furnace injection was tried and abandoned
because of poor performance which was attributed 1) to a combination of poor
sulfur capture because of decreased sorbent activity due ta dead burning at fur-
nace temperatures and inadequate dispersion of the sorbent throughout the com-
bustion gases. Interest in the technique was revived when pilot scale studies
at EER indicated that higher sulfur captures could be achieved under low N0X
combustion conditions. The need to develop a low cost, retrofittable S0X
control technology and reduced performance goals have caused additional interest
in the limestone injection concept. LISG, Limestone Injection Multistaged
Burners, was coined to emphasize that simultaneous N0x/S0x control could be
achieved by using limestone injection in conjunction with a distributed mixing
burner.
When the sorbent is initially injected into the high temperature combustion
zone, calcination occurs and CO2 or H2O is evolved according to the following
reactions:
15-1
-------
CaC03
CaO(s) ~ C02(g)
770
HgC03
m90($) * C02(g)
340
Ca(0H)2
CaO(s) ~ H20(g)
370
In the case of limestones, structural rearrangement takes place as the COj is
being evolved because the crystal structure of the carbonate (rhombohedral) dif-
fers from that of the oxide (cubic). After the calcination process occurs,
sites for recrystallization of the oxide are well dispersed so that localized
crystallites, or grains, are formed with a pore structure between them. Overall
calcination rates have been well documented by Hyatt et al.(5), Borgwardt(6,7),
and Coutant et al.(2). Particle heat-up appears to occur very rapidly (c.a. 50
ms) after the particle is mixed with the hot gas stream and, at typical
industrial combustion temperatures, the overall calcination process appears to
be essentially complete in less than 200 ms. The calcination conditions, par-
ticularly temperature and CO2 partial pressure, influence the ultimate porosity
and grain size. Extended exposure to high temperature causes the grains to
coalesce [sinter)- because the highly mobile oxide units quickly fill in the gaps
where grain boundaries (or contact points) occur. This results in larger
grains, with lower surface area.
Once the sorbent particle is calcined to CaO, it can react with the available
sulfur species to form calcium sulfate. The overall reaction for sulfation is:
but the exact mechanism for this reaction is not known. It may involve an ini-
tial formation of calcium sulfite or result from'a reaction of sulfate ions
with CaO. The overall reaction system has been studied extensively by
Borgwardt(6), Ishihara(8), and others(9,10).
This paper describes the results from a combined experimental and theoretical
study to investigate the influence of combustion and sorbent parameters on the
effectiveness of dry sorbent injection for 502 control under conditions typical
of current utility practice. A 300 kW Boiler Simulator Furnace was used to
investigate the role of boiler thermal history, sorbent injection location,
calcium to sulfur molar ratio, and SO2 partial pressure on capture effectiveness
with limestones, dolomites, and slaked limes with and without metallic promo-
ters. The experimental studies were supported by theoretical calculations using
grain and pore models which considered both the heterogeneous chemical reaction
and the relevant diffusional processes.
S02 + }02 + CaO = CaS04
15-2
-------
EXPERIMENTAL SYSTEMS
All of the data presented in this paper were obtained with the 300 kw Boiler
Simulator Furnace (BFS) which is illustrated in Figure 1. The facility consists
of two main sections; a refractory-lined vertical radiant tower 0.56m in
diameter by 5.8m tall with multiple access ports, and a horizontal convective
section containing air-cooled heat exchangers. The 8SF can be used to simulate
a wide range of time/temperature profiles by positioning water-cooled panels or
rods appropriately in the radiant section.
The facility utilizes a Distributed Mixing Burner (0N6) which down-fires coal or
natural gas. Two preheated air streams, swirl and axial, are supplied to the
burner, and four external staging ports located around the burner allow tertiary
air injection. N0X levels are minimized with the DM3 due to the fuel-rich core
that is produced in the flame zone.
For the natural gas studies, HoS was injected at the burner to simulate the
sulfur content of coal. All of the sorbents were fed from a calibrated, twin-
screw volumetric feeder and injected at the burner with pulverized coal or
natural gas, or injected downstream at section 6 in the radiant section.
Exhaust gas samples were withdrawn at sample port 10B (temperature approximately
1200 K) with a stainless-steel, water-jacketed probe and analyzed using standard
continuous instrumentation for NO, CO, CO2 and O2. SO2 samples were pulled
through the phase discrimination portion of the probe which separated any sor-
bent particles from the gas stream. The SO2 samples were then carried via a
heated sample line to a permeation dryer and on to a Dupont non-dispersive
ultraviolet SO2 analyzer.
MODEL DESCRIPTION
Two sulfur capture models were used for data interpretation and experimental
planning in this work. Both models assume that the sorbent has been fully
calcined prior to the onset of the sulfation reactions; the models differ pri-
marily in the manner in which they view the internal structure of the calcined
sorbent particle. The first, a grain model, was developed by G. D. Silcox and
D. W. Pershing at the University of Utah and is based primarily on the concepts of
Hartman and Coughlin(ll), Pigford and Sliger(12), and Hartman and Trnka(13).
This model treats the sorbent particle as an agglomeration of tiny, non-porous
spherical grains separated by voids. Alternatively, the pore model, developed
by W. Clark and co-workers at EER, treats the sorbent particle as a single unit
penetrated by cylindrical pores of varying sizes and is based on the work of
Bhatia and Perlmutter(lO) and othersC9,14).
In both models the sulfation reaction is assumed to occur at the internal sur-
face of the particle. As the reaction proceeds the product layer of calcium
sulfate is formed. For a molecule of SO2 to be captured by a sorbent particle
it must overcome four resistances in series:
1. Boundary layer diffusion. SO2 must diffuse from the free stream
through the boundary layer around the particle.
2. Diffusion through macropores. SO2 must diffuse through the internal
particle structure.
3. Diffusion through the product layer. An SO2 species must diffuse
through the calcium sulfate product layer to reach unreacted CaO.
4. Surface reaction. Tine heterogeneous reactions forming CaSO^ take
place at the CaO/CaSO^ interface.
15-3
-------
The governing differential equations were obtained from material balances on the
particle and gas stream. Both models assume that the intrinsic kinetics of
CaSO^ formation are zero order with respect to S02» that the CaO grains react by
a shrinking core model, and that the MgO grains do not react. The intrinsic
kinetic and product-layer diffusion rate-constants were obtained by fitting the
fundamental work of Borgwardt(6) in the manner described by Hartman and Coughlin
(11).
SYSTEM OPTIMIZATION
One phase of the work recently completed on the BSF focused on the influence of
the boiler design/sorbent injection parameters on overall SO2 capture. These
studies included consideration of overall excess air, burner stoichiometry,
radiant zone heat removal rate, burner swirl, general sorbent injection location
(burner versus downstream), importance of sorbent premixing with fuel, and
impact of low NOx operation. The results of these studies indicated that the
parameters which were most critical for the optimization of sorbent utilization
were sorbent injection location and the time/temperature history between injec-
tion and 1200 K. These effects are discussed in detail in the following
sections.
SORBENT INJECTION LOCATION
Figure 2 summarizes results obtained on the impact of sorbent injection location
with a typical limestone and slaked lime. These data indicate that the SO2
capture increased approximately linearly with increasing calcium to sulfur
ratio, and higher capture was achieved with downstream injection at approxima-
tely 1500 K (2250"F) for both sorbents. These results agree well with the early
work of Coutant et al.(15) who reported an optimum injection temperature of 1475
K. Figure 2 indicates that the slaked lime was more sensitive to injection
location than the limestone, and that is typical of results with other calcium
hydroxide materials. The desirability of downstream injection shown in Figure 2
is typical of the trends obtained with a wide variety of other sorbents with
both gas-and coal-firing under both favorable and highly quenched thermal con-
ditions.
Figure 3 illustrates the impact of varying the sorbent injection location on the
key parameters controlling SO2 capture. The top half of this figure shows the
gas temperature measured during the experiments reported in Figure 2 as a func-
tion of residence time. These temperature measurements were obtained with a
suction pyrometer and indicate that sorbent particles injected in the burner
zone experienced peak temperatures in excess of 1700 K while the particles
injected downstream cool rapidly from 1500 K. The bottom half of Figure 3 shows
the influence of varying the point of sorbent injection (plotted in terms of
residence time) on the surface area of the calcined sorbent (without SO2
present). Downstream injection greatly enhances the surface area available for
subsequent sulfation reaction due to decreased rate of grain growth at lower
temperatures. The bottom half of Figure 3 also shows the residence time in the
sulfation zone (1500 to 1200 K, 2250 to 1700 F) as a function of the overall
residence time. If the sorbent is injected below 1500 K, the effective resi-
dence time in the sulfation zone decreases rapidly. In addition, a larger por-
tion of the available residence time must be used for in situ calcination of the
stone, and lower temperatures produce reduced diffusion and chemical rates.
Therefore, the overall optimum injection temperature appears to be near the
front of the sulfation window (approximately 1500 K). Injection above this tem-
perature results in decreased sorbent activity due to excessive grain growth
lb-4
-------
(sintering); injection significantly below 1500 K (2250 F) produces higher ini-
tial surface areas but this effect is more than compensated for by the decreased
chemical reaction, product layer diffusion rates, and available residence time.
THERMAL HISTORY
Figure 4-a shows the results of detailed heat transfer calculations (made by
W. Richter at EER) on the detailed gas temperature profile within the sulfation
zone for three full scale utility boilers (solid lines). Time zero was
arbitrarily defined to be the point in the boiler where the gas temperature
reached 1500 K (2250* F) and the profiles have been characterized in terms of an
average quench rate within the sulfation zone (1500 to 1200 K). These results
indicate that the quench rate of commercial boilers may vary by as much as a
factor of 5 and that in typical utility boilers the time available in the sulfa-
tion window may be as little as 1 sec. Figure 4-a also shows measured tem-
peratures (dotted lines) for two thermal profiles produced in the BSF; one,
termed the "American Boiler" profile, which represents current field practice;
and the other, termed "reduced load," which has a more favorable quench rate.
Figure 4-b summarizes the influence of quench rate on SO2 capture with
downstream sorbent injection at a Ca/S ratio of 2.0. Two very different sor-
bents were used in these studies: Vicron limestone which represents a typical
low surface area calcium carbonate, and Genstar pressure slaked lime which is
typical of the high surface area dolomitic hydroxides. The dashed line
represents capture predictions generated by EER's grain model. No model parame-
ters were adjusted to improve the agreement between the experimental results and
the model predictions; the fundamental diffusion and kinetic rates were based on
the data of Borgwardt(6), the surface areas for both sorbents were based on the
bench scale measurements of Slaughter et al.(16), and the thermal profile was
based on the measured temperatures shown in Figure 4-a. Overall the agreement
between the predictions and the data is excellent and, as expected, capture
decreased with increasing quench rate because reaction time decreased. These
results indicate that a particular sorbent injected under identical conditions
in two separate boilers may produce different capture efficiencies because of
variations in the thermal characteristics of the boilers.
IMPACT OF FUEL SULFUR
The influence of fuel sulfur concentration was studied by varying the amount of
H2S doped into the-natural gas flames. SO2 concentrations of 500, 1000, 1800,
and 3000 ppm (dry, 056 O2) were tested with Colton lime and Genstar, pressure
slaked dolomitic lime (type 5). The Ca/S ratio was held constant at 2.0, and
the sorbent was injected at 1500 K (2250* F) for all of the runs. These
results are presented in Figure 5 along with model predictions. Both the data
and the model showed an increase in sulfur capture with increasing SO2 con-
centration. The curvature in the theoretical predictions, which is in agreement
with the experimental results, is due to a trade-off between the zero order che-
mistry and the first order pore and product layer diffusion processes. At low
502 concentrations the product layer diffusion resistance (which is first order
in SO?) exceeds the chemical reaction resistance (zero order with respect to
SO2) for a larger fraction of the sulfation zone residence time; hence, the
overall dependence on gas phase SO^ concentration is much stronger than at high
SO2 concentrations where the chemical resistance is more limiting. The good
agreement between the model predictions and the experimental data is typical of
15-5
-------
that observed with other sorbents and suggests that the model accurately repre-
sents the concentration dependence of the sulfation process.
INFLUENCE OF SORBENT COMPOSITION
Sorbent composition and other physical properties are probably the most impor-
tant factors in determining overall capture performance. With all other
variables held constant, a wide range of sulfur capture can be produced with
different sorbent types. This fact is clearly illustrated in Figure 6: capture
with downstream sorbent injection ranged from 3056 for Vicron to over 75% for
pressure slaked dolomitic sorbents at a calcium to sulfur ratio of 2.0. The
range in capture with high temperature injection (with the fuel) was not as
broad (20 to 40% at Ca/S = 2.0) because the final surface area is less dependent
on the initial sorbent characteristics when the calcination occurs at flame zone
temperatures(15). In general the capture results tend to be grouped primarily
by general sorbent type. On a calcium molar basis, the three pressure slaked
dolomitic sorbents gave the best capture with downstream injection. They were
followed by the natural dolomite, the two normal slaked limes, and finally the
Vicron limestone. For injection with the fuel, both the natural dolomite and
the pressure slaked dolomite gave the highest capture while the calcitic car-
bonate (Vicron) and the calcitic hydroxide gave the lowest capture. (Detailed
sorbent characteristics are given in Table 1.)
Since the sorbents shown in the composite comparison include both calcitic and
dolomitic materials, the total mass feed rate can very substantially for a given
Ca/S molar ratio; hence, both the initial sorbent cost and the amount of ash
which must ultimately be disposed of are variable. In order to compare the dif-
ferent sorbents on a total mass output basis, the abscissa in Figure 6 was con-
verted to the mass parameter, MgO + CaO divided by the total inherent coal ash
(assuming a 1% sulfur and a 10% ash coal) as shown in Figure 7. Compiling the
data on this basis gives an indication of the capture that can be achieved rel-
ative to the amount of additional material that passes through the convective
passes and that must be removed from the particulate collection devices. Even
on this basis the dolomitic pressure slaked limes appear attactive as does the
Warner slaked lime. Again the poorest performance was achieved with the Vicron
limestone. These results suggest that capture in excess of 60% can be achieved
with approximately a 50% increase in the dry ash handling requirements for a 1%
sulfur coal or a 150% increase in solids loading for a 3% sulfur coal.
Since the results shown in both Figures 6 and 7 indicate that dolomitic sorbents
are effective in achieving sulfur capture, the role of magnesium was of par-
ticular interest. Figure 8-a shows the influence of magnesium content for
several slaked sorbents at different quench rates. As noted previously the
overall capture is significantly higher for the thermal conditions with the
reduced quench rate (longer sulfation zone residence time). Capture also
increased significantly as the magnesium content increased. The role of magne-
sium was examined directly by injecting MgC03 al°ne and in a mixture with the
Vicron limestone to achieve a "simulated dolomite." The Ca/tog molar ratio of
the mixture was kept the same as the real dolomite, and the results from these
tests are shown in Figure 8-b. The data from the downstream injection of MgC03
alone suggest that it does not significantly react with SO2 to form magnesium
sulfate prior to the sampling point (1200 K) and this consistent with both
theoretical considerations and previous practical experience. The simulated
dolomite gave essentially identical capture to that measured previously for the
Vicron limestone alone, again confirming that MgSO^ formation was not signifi-
cant. Apparently, the magnesium in the dolomitic materials enhances the capture
by means other than direct capture since mere physical mixing of the two
15-6
-------
materials does not yield the high capture produced with dolomite where the
magnesium is present as an integral part of the sorbent matrix. It has been
previously suggested that the magnesium prevents pore blockage, but the charac-
teristic particle size in the Genstar pressure slaked dolomite is less than 1 ym
ana with sorbent particles that small, the pore diffusion resistance is
insignificant. A more likely explanation appears to be that the Mg enhances
product layer diffusion and/or initial chemisorption of the sulfur species.
SORBENT ENHANCEMENT BY PROMOTER ADDITION
One of the primary aras of study in the BSF has been the addition of various
promoters to enhance sulfur capture. Initially, was found to dramatically
improve capture with Vicron, especially when the promoted sorbent was injected
into the high temperature region at the burner. Many of the transition metal
promoters were evaluated for possible capture enhancement; however, only molyb-
denum and chromium enhanced capture significantly, relative to Vicron.
Subsequently, Borgwardt found that alkali metal compounds (lithium, sodium,
potassium) gave positive results similar to those found with the chromium series
materials. Figure 9 shows that mixing each of the alkali metal carbonates with
Vicron resulted in increased SO2 capture when the sorbent was injected with the
fuel and downstream. In these experiments the ratio of metal ion to calcium was
maintained equal to that for the 5% Cr^ addition. On this basis chromium
proved to be the best promoter in terms of overall capture. The most unusual
thing about the chromium promoted limestone sorbent was that, in contrast to all
previous results, the capture with high temperature injection was equilavent to
that with downstream, low temperature injection. Chromium appears to have the
ability to counterbalance the effect of thermal sintering of the sorbent.
Figure 10 illustrates the influence of 0^03 addition with other types of sor-
bents. ITie open bars represent the capture measured with the sorbents alone,
and the shaded bars indicate the increase in capture that resulted from 5% chro-
mium addition. With all of the sorbents and with both burner zone and
downstream sorbent injection, the capture increases with chromium promotion were
significant. Even the performance of the Genstar pressure slaked dolomite was
improved: 70 to 85X capture for the downstream sorbent injection and 35 to 705K
capture for injection with the fuel. In general the enhancement above the base
line was greater when the promoted sorbents were injected into the high tem-
perature region, although the absolute capture levels were generally higher for
downstream injection.
The exact mechanism for the chromium and sodium enhancement is not clear at the
present time; however, it appears likely that these materials promote capture
by enhancing the product layer diffusion step since the model calculations indi-
cate that product layer diffusion is the primary limitation to increased sulfa-
tion rates. Additional work is needed to optimize the method of promoter
addition and clarify the controlling mechanisms.
CONCLUSIONS
The experimental results and the sulfation model calculations indicate that the
sorbent injection locations and the residence time within the sulfation tem-
perature window can significantly influence the overall sulfur capture for any
particular sorbent. Unless the sorbent is promoted with a metal additive,
downstream injection at approximately 1500 K [2250* F) results in optimum sor-
bent utilization. Increasing the heat removal rate between approximately 1500 K
and 1200 K (2250' to 1700" F) results in decreased sulfur capture. Increasing
15-7
-------
the gas phase SO2 concentration (e.g., due to increased coal sulfur content)
improves sorbent utilization, but the dependence is non-linear due to the com-
bined effects of intrinsic chemistry and diffusion.
In general dolomitic sorbents perform better than calcitic sorbents (per mole of
Ca) and hydroxides are superior to carbonates. The true influence of pressure
slaking is unclear; however, the best sorbents tested (on either a calcium molar
basis or total mass basis) were the pressure slaked dolomites. The magnesium in
the dolomite materials does not react to produce magnesium sulfate; it probably
enhances product layer diffusion. The performance of all sorbents can be
enhanced by the addition of appropriate metallic compounds in relatively small
quantities.
Thus, the results of this study suggest that it is possible to achieve capture
levels significantly above those typical of limestone injection by at least two
alternative methods: use of specially treated sorbents (e.g., pressure slaked
dolomites) or use of promoted limestones. (Clearly these two concepts can be
combined to produce even higher capture levels; however, this may be economi-
cally unattractive.) Figure 11 provides an overall comparison of these alter-
natives under conditions that were designed to approximately simulate current
utility practice (American Boiler profile). In general these data show that,
even under severe thermal conditions (quench rate =330 K/sec = 600* F/sec), it
is possible to achieve 40% capture with approximately a 70% increase in boiler
solids loading (based on a 103> ash, 1% S coal) by injecting either an inexpen-
sive limestone promoted with a metal oxide or a pressure slaked dolomite.
15-8
-------
ACKNOWLEDGMENTS
The authors greatly acknowledge the considerable technical assistance of Dennis
Drehmel and Robert H. Borgwardt of the Uhited States Environmental Protection
Agency and P. L. Case, G. D. Silcox, and R. Payne of the Energy and
Environmental Systems Corporation. The work was funded by the EPA, partially
under EPA contract 68-02-3921.
REFERENCES
1. Gartrell, F. E., "Full Scale Desulfurization of Stack Gas by Dry Limestone
Injection," Volume I, TVA, EPA-650/2-73-019a (NTIS PB 228447), August 1973.
2. Coutant, R. W. et al., "investigation of the Reactivity of Limestone and
Dolomite for Capturing SOj from Flue Gas (Summary Report)," Battelle
Memorial Institute, APTD 0621 (NTIS PB 196749), November 1970.
3. Attig, R. C., and P. Sedor, "Additive Injection for Sulfur Dioxide Control:
A Pilot Plant Study," Babcock & Wilcox Co., APTD 1176 (NTIS PB 226761),
March 1970.
4. Wen, C. V., and M. Ishida, Env. Sci. Tech., 703 (1973).
5. Hyatt, E. P., E. B. Cutler, M. E. Wadsworth, J. Am. Ceramic Soc., 41, 70
(1958).
6. Borgwardt, R. H., Environ. Sci. Technol., 4, 59 (1970).
7. Borgwardt, R. H., A.I.Ch.E.J. (1984).
8. Ishihara, Yoshimi, "Kinetics of the Reaction of Calcined Limestone With
Sulfur Dioxide in Combustion Gases," presented at the Dry Limestone
Injection Process Symposium, Gilbertsville, Kentucky, June 22-26, 1970.
9. Ramachandrin, P. A., and J. M. Smith, A.I.Ch.E.J., 23, 353 (1977).
10. Bhatia, S. K., and D. D. Perlmutter, A.I.Ch.E.J., 27, 226 (1981).
11. Hartman, M., and R. W. Coughlin, A.I.Ch.E.J. 22. 490 (1976).
12. Pigford, R. L., and G. Sliger, Ind. Engnq. Chem. Proc. Pes. Dev., 12, 85
(1973).
13. Hartman, M., and 0. Trnka, Chem. Eng. Sci., 35, 1189 (1980).
14. Simons, G. A., and W. T. Rawlins, I & EC Proc. Pes. Dev., 10,556 (1980).
15. Coutant, R. W. et al., "Investigation of the Reactivity of Limestone and
Dolomite for Capturing SO? from Flue Gas," Battelle Memorial Institute,
APTD 0802 (NTIS PB204385J, October 1971.
-------
Slaughter, D. M., et al., "Influence of Sorbent Composition on SO2 Capture
Efficiency in LIMB Applications," presented at the 1984 Annual AIChE
Meeting, San Francisco, CA, November 1984.
15-10
-------
10B
I
Sample Port
Sample Port 10A
i
Exhaust
Sample Port
Boiler
simulator furi
nace.
-------
o DONNSTREM INJECTION
• FUEL INJECTION
VICRON MARKER LIME
1
i
1
1
2
3
3
2
Ca/S Ca/S
Figure 2. Influence of injection
location on sulfur capture.
1700
2600
2400
U.
O
1477
I-
5
UJ
Si 2000
UJ
»—
3
1800
1255
1600
E
a
oc
UJ
u
u_
oc
ZD
u
u.
u
UJ
a
i/>
Residence H«ie, sec.
Figure 3. Impact of injection location on
residence time and surface area.
-------
Ill (K/sec) 222
t—
D.
U 40
^ 30
- o
20
10
0.5 1.0 1,5 2.0 2.5 3.0 3.5
(a) RESIDENCE TIME, sec
GENSTAR-type S
0 " — • — .
VICRON
_L
_L
200 400 600
(b) QUENCH RATE, °F/sec
Figure 4. The effect of quench rate in sulfation zone.
80
70
60
50
csA'
p'
e«
30
20
10
co^on
O,*
» ¦ i ¦ ¦
tOO 1200 2000 2800
INITIAL S02 CONCENTRATION, ppm (DRY, OZ ty
Figure 5. The effect of Initial SOg concentrations.
"15-13
-------
WITH FUEL DOWNSTREAM
90
80
70
60
50
<10
30
20
10
1 2
3
1
2
3
Ca/S Ca/S
Figure 6. Capture comparison
(reduced load, natural
gas/HpS; see Table 1
for symbol key).
WITH FUEL
DOWNSTREAM
0.?
0.4
0.6
0.8
0.4
0.6
0.2
0.8
ADDED SORBENT HASS (HqO ~ CaO)
INHERENT COAL ASH HASS
Figure 7. Capture comparison - mass basis
(107o ash, 17o S coal).
-------
DOWNSTREAM
?t/*et
£
NAT. GAS/HjS
REDUCED LOAD
O VICRON
• VICRON ~ MsC03
¦ MgC03
A DOLOMITE
Tiqb - 1700OF
t Mq in Slaked Sorbents
(a)
1 2 3
Ca/S or Mg/S
(b)
Figure 8. Influence of magnesium.
DOWNSTREAM
WITH FUEL
I I
/
NAT. GAS/^S
REDUCED LOAD
O VICRON
~ VICRON + 3,51 NA2CO3
O VICRON ~ 51 CR2O3
CiVICRON ~ 4.46Z K2CO3
Q VICRON ~ 2.131 L12CO3
Figure 9. Additive screening results.
15-15
-------
WITH FUEL
DOWNSTREAM
NAT. GAS/H2S
REDUCED LOAD
OPEN BARS:
SHADED BARS;
Ca/Ch - 14.8
Ca/S - 2.0
SORBENT ALONE
INCREASE WITH Cr
V - VICROW
C - COLTON Lift
G - GNESTAR DOLOMITIC(S)
V C G
Figure 10. Capture enhancement with Cr^O^.
WITH FUEL
DOWNSTREAM
1 - 1 I I
/
1 1 1— 1
' / yS
:
1111
jr ;
till
0.2 0.4 0.6 0.8 0.2
AHDiI> ^5.RM.N-L MASS (MqO * CaO)
INHtRENT COAL ASH MASS
0.4
0.6 0.0
American
Boiler Profile
Nat Gas/H^S
O Vliron
O Co) ton Lime
A Dolomltic
a Genstar
v DolomUic-type S
£ Vlcron * 5.OX Cr2Oj
¦ Vlcron ~ 3.5X Na.CO
Figure 11. Overall comparison.
15-16
-------
TABLE 1. PHYSICAL AND CHEMICAL PROPERTIES OF SORBENTS
Sorbent
o
Vlcron
O
Col ton
L1me
0
Mamer
L1me
A
Dolomite
S3
Genstar
Blend (S)
O
Genstar
Dolomltlc
(S)
O
Warner
Dolomltlc
(S)
Type
i
Calclte
Hydrated
L1me
Hydrated
Lime
Dolomite
Hydrated
Lime +
Hydrated
Dolomltlc
Lime
Pressure
Slaked
Dolomltlc
L1me
Pressure
Slaked
Dolomltlc
L1me
Composition
CaCOj
Ca(0H)2
Ca(0H)2
C8CO3 •
HgC03
Ca(0H)2 +
Ca(0H)2 •
Hg(0H)2
Ca(0H)2 •
Mg(0H)2
Ca(0H)2 •
Mg(0H)2
Hean Size
(um)
9.8
4.0
11.8
1.8
1.4
3.5
Density
(g/cm*)
2.7
2.3
2.9
2.3
2.3
2.3
Chemical Comp
(X)
Ca
39.2
51.4
53.9
24.8
45.0
30.3
29.5
0.02
0.21
0.37
0.04
0.26
0.21
SI
0.09
3.30 i
0.51
0.16
0.13
0.94
Fe
0.06
0.10
0.21
0.09
0.07
0.42
Na
0.007
0.007
0.007
Mg
0.33
0.25
0.20
11.3
6.2
16.0
15.9
-------
STUDIES OF SORBENT CALCINATION AND S02-S0RBENT
REACTIONS IN A PILOT-SCALE FURNACE
by
Roderick Beittel
John P. Gooch
Edward-B. Dismukes
Southern Research Institute
2000 Ninth Avenue, South
Birmingham, Alabama 35255
and
Lawrence J. Muzio
Fossil Energy Research Corporation
29541 Vista Plaza Drive
Laguna Niguel, California 92677
ABSTRACT
Furnace injection of calcium-based sorbents for SO? reduction was examined using a
1 x 105 Btu/hr pulverized coal-fired combustor. The effects of sorbent type and
rate, residence time, injection site, staged combustion, and C02-enriched transport
air on overall S02 removal are reported. Properties of calcine were determined for
gas firing at selected conditions. Variations in calcine surface area of 2 to
30 m2/g were found, corresponding to a range of calcium utilization of 12 to 38%.
Sorbent type and temperature at the point of injection were primary determinants of
calcium utilization.
16-1
-------
STUDIES OF SORBENT CALCINATION AND S02-S0RBENT
REACTIONS IN A PILOT-SCALE FURNACE
INTRODUCTION
The Electric Power Research Institute (EPRI) and Southern Company Services (SCS) are
co-funding a project with the objective of developing a dry sorbent emission control
process (DSEC) as a means of reducing sulfur dioxide emissions from coal-fired
boilers. The program consists of an ongoing technology assessment, fundamental
bench-scale studies, small pilot-scale studies, and a conceptual design study for a
40- to 80-MW scale application of DSEC. This paper presents results obtained using
a 1 x 10° Btu/hr combustion system at Southern Research Institute.
A major task in the overall process development effort consists of obtaining an
understanding of the mechanisms which limit sorbent utilization under various condi-
tions. Accordingly, the pilot-scale experiments have been supported by an extensive
analytical effort to provide detailed chemical and physical characterization of
fuels, raw and reacted sorbents, and calcines. The objectives of the pilot-scale
study may be described more specifically as follows:
• Provide a means of evaluating sorbents and process modes under
realistic conditions
• Define the sorbent properties and process conditions which limit
S02 capture and calcine utilization
• Develop a data base for calcium utilization as a function of sorbent
type, firing conditions, fuel, injection mode, and calcine properties
• Evaluate an integrated process which combines furnace injection with
downstream injection
The results obtained to date in the pilot-scale program are primarily concerned with
determinations of calcine properties and calcium utilizations with furnace injection
of high calcium sorbents.
TESTING PROCEDURES AND EQUIPMENT
Combustor
An overall view of the pilot combustor system is shown in Figure 1. The furnace is
a down-fired, refractory-1ined cylinder with a nominal capacity of 1 x 106 Btu/hr.
Dimensions of the furnace and location of sampling and sorbent injection ports are
shown in Figure 2. The furnace is equipped with an adjustable-swirl burner, and
with injection or sampling ports as shown. The baseline firing conditions are given
in Table 1. Figures 3 and 4 show radial temperature profiles for the baseline fir-
ing conditions with 1.0 and 0.6 x 106 Btu/hr fuel rates.
16-2
-------
Note that firing with gas gives much higher temperatures near the burner than does
coal. The difference is especially pronounced with the lower firing rate. Tempera-
tures with either fuel are similar downstream from the burner (not shown).
The combustion gases leaving the furnace pass through a horizontal, refractory-lined
tunnel ("square duct") and then through air-cooled stainless steel piping. The gas
cools to 150 to 300°F before entering a pulse-jet baghouse.
Coal and sorbent are fed gravimetrically into separate eductors and transported to
the furnace. Sorbent, when injected at the burner, passes through the burner core
with the fuel entering a surrounding annulus. Sorbent is injected downstream of the
burner through two opposed 1/2 in. jets perpendicular to the furnace axis.
Residence Time Considerations
Temperature-time profiles showing port locations are given in Figure 5 for baseline
firing conditions. The change in slope after Port S-5 corresponds to the furnace
exit. The 1 x 106 Btu/hr profile is compared in Figure 6 to residence time-
temperature values for four utility boilers. Values for Units 1, 2, and 3 were
calculated from firing rates and temperatures reported for t-fired 300 to 600-MW
boilers operated near rated capacity (1).
It is emphasized that the calculated values are based solely on average flue gas
velocity and average gas temperatures at several elevations within the radiant sec-
tion. The values for Unit 4 are those reported.for a 140-MW, wall-fired unit (2),
and show a much steeper gradient. The lowest temperature for each of the four units
corresponds to the radiant section outlet; the temperature would be expected to drop
much faster on entering the convective passes. The 1 x 106 Btu/hr temperature pro-
file of the pilot unit radiant section is a reasonable approximation of that in
certain full-scale units, and probably provides more time in the 2000 to 2200°F
region than most boilers at rated load. (Reduced load data for Units 1, 2, and 3
were not available.) The pilot unit fired at 0.6 x 106 Btu/hr provides much more
time in the 2000°F region than would be expected in a full-scale boiler.
Continuous Gas Analysis
A sample is withdrawn through an in-stack sintered stainless steel filter at a point
where the gas temperature ranges from 250 to 500°F (depending on firing rate). The
filtered sample passes through a conditioner which removes water by condensation at
35 to 40°F. Equilibrium calculations show that, in the absence of alkaline contami-
nants, the maximum loss of S02 to the condensate from coal or natural gas combustion
would be 1.5 and 3%, respectively. Condensate collected during coal combustion with
injection of hydrated lime contained negligible calcium and the equivalent of 1% of
the gas phase S02- The suitability of this sample system has also been demonstrated
by injecting S02 into the flue gas just ahead of the sample probe. During natural
gas firing, the recovery of S02 in the absence of sorbent is typically 97 to 99% of
the value calculated from the measured feed rates of S02 and fuel. Initiation of
furnace sorbent injection results in a reduction of S02 from 4 to 14% of that which
occurs when the S02 is added with the fuel. That is, of the loss in S02 measured
when sorbent reacts with S02 throughout the combustor system, only 4 to 14% occurs
in the sampler itself. This is an especially severe test of sample system bias in
that the S02 injected at low temperature is exposed to unsulfated sorbent. The
results must be viewed as an upper limit in sample system bias since some of the
reaction may occur between Injection and sampling rather than in the sample system
itself. A less severe test, in which S02 was injected near the sampler during coal
firing, showed no difference in the recovery of added S02 with or without sorbent
injected. Tables 2 and 3 list the results of these tests.
16-3
-------
Fuel and Sorbents
All tests with coal to date have been made with a Jefferson seam, Alabama coal pro-
duced by the Hallmark mine. Coal and ash analyses typical of Lot 1 are shown in
Table 4. Recent tests have been made with a second lot of the same coal which
differs from the first only in sulfur content (2.8 versus 3.4% for the first lot).
The coal is pulverized on site to about 80% minus 200 mesh.
Extensive pilot-scale testing has been completed with two varieties of limestone, a
high-calcium hydrate and a pressure-hydrated dolomitic lime. Table 5 lists complete
analyses for these materials. The St. Genevieve limestone is a relatively pure
carbonate with fine, uniform grains, and with a correspondingly low surface area of
1.5 m2/g. The Marianna limestone, in contrast, features irregular grains, some
fossil remains, sigificantly more impurities, and higher surface area. Both hydrate
materials exhibit characteristically smaller size distributions and higher surface
areas.
Limited testing has been completed with the following sorbents: three additional
grades of the St. Genevieve limestone ranging from 1.5 to 15 um NMD; four additional
grades of Marianna limestone ranging from 1.3 to 35 ym MMD; two grades of pulverized
marble; two precipitated calcium carbonates having mean diameters of 0.7 and
0.07 um; a dolomitic limestone; and a high-calcium, pressure-hydrated lime. Some
properties of these materials are listed in Tables 6 and 7.
Description of Test Procedures
This section describes the two types of tests carried out with the pilot-scale com-
bustor: those in which only the gas-phase reduction of S02 with the introduction of
sorbent during coal-firing was determined; and those in which solids samples were
collected during gas firing for further characterization.
Reduction of gas-phase S09. The standardized test chosen to evaluate a large number
of variables on the basis of overall S02 removal was the following: 1) establish
steady state at the conditions of interest; 2) begin feeding sorbent to the sorbent
eductor; 3) terminate sorbent addition after 10 min. Table 8 gives the definitions
used for reporting calcium-to-sulfur ratio, S02 reduction, and calcium utilization.
The gas sampling system has been shown to make no contribution to the reduction of
S02. That the contribution of deposited sorbent to the overall reduction is minor
is inferred qualitatively from the similar response of the system either to sorbent
initiation or to reduction in S02 feed. Analysis of solid samples has in general
confirmed the apparent utilization calculated from gas composition.
Solids sampling. Solids samples were withdrawn from the furnace through a water-
cooled probe and collected in a glass-fiber thimble at the probe exit. The central
sample tube is separated from the coolant by an annul us through which diluent may be
added to the sample at the probe entrance. Initial experiments showed that in-probe
reaction of calcine sampled either with or without dilution resulted in no more than
1 to 2% each of H20, C02, or S0X in the solid. Thus, the sampling procedure
adopted for this study was to sample undiluted flue gas while maintaining the cool-
ing water exit temperature at 130°F and the collecting filter at 300°F to avoid
condensation.
Subsequent to these initial probe validation tests, much more reactive calcines were
encountered. Injection of high-calcium or dolomitic hydrated lime in the lower
furnace, for example, yields calcines with surface areas of 10 to 15 or 30 to
40 m2/g (compared to 2 to 3 m2/g for the calcine from burner injection of limestones
in the tests cited above). With these calcines, Ca(0H)2 and CaC03 each account for
16-4
-------
5 to 20 tnol % of the total calcium. The reactions to hydroxide and carbonate are
assumed to occur in the probe since the temperature in the furnace is well above the
decomposition temperature of either.
Figure 7 shows the potential for in-probe reaction of S09 with the most reactive
calcine tested. In these tests, calcine was sampled during injection of dolomitic
hydroxide at S-4. One series was completed with S02 added to the gas flame (normal
sampling). A second series was then completed with no S02 added to the flame, but
with S02 added to the inlet of the probe. Any sulfate present in the second series
of samples had to be due to reaction within the probe itself. At the level of
injection, the in-probe reaction is equivalent to the overall reaction when S02 is
present in the furnace (normal sampling). Clearly, this sampling procedure gives
no information on the extent of in-furnace sulfation at the level of injection.
Although it appears that in-probe reaction could account for as much as 30fc of the
utilization measured at S-5 and 5D-2, such is not likely the case. When S02 is
present only in the probe, the reaction is with "fresh" calcine, and represents only
an upper limit in additional in-probe reaction of sulfated calcine. The agreement
between solids analysis and exit S0^ reduction (measured at about 700°F, as
described previously) supports the idea that additional reaction in the probe is
much less.
Neither elevation of the filter temperature to 500°F nor use of air rather than
water as a probe coolant appeared to make a consistent difference in calcine BET
area or composition. While work continues on the development of a better sampling
technique, the present results are thought to be deficient only for the most reac-
tive calcines.
EXPERIMENTAL RE5ULTS
Results of Screening Studies
Ml i
This section discusses the results of tests wherein only the overall reduction in
S02 was measured. Variables examined include: sorbent type, sorbent rate, fuel
rate, injection site, staging of combustion air, and C02 addition to the sorbent
transport air.
Sorbent type and rate. Figure 8 compares the performance of four basic types of
sorbents: limestone, dolomite, calcium hydroxide, and dolomitic hydroxide. The S02
removals shown are the optimum for each sorbent with 1 x 106 Btu/hr firing. The
results shown for Marianna limestone are very similar to those for St. Genevieve
under these conditions. The removals achieved with the pressure-hydrated, high-
calcium lime were the same as with the Longview Ca(0H)2.
Better utilizations have been achieved with injection of Marianna limestone at lower
temperatures and with extremely fine grade's, of calcium carbonates. The results of
Figure 8, however, are representative for each class of sorbent. The enhanced cal-
cium utilization of the dolomitic sorbents is offset by the lower calcium content.
Figure 9 shows the same removal data plotted as a function of the mass feed rate of
sorbent. The collapse of the data for four sorbent types into two distinct lines is
fortuitous, but does illustrate the improved performance of hydrates and the weight-
basis penalty resulting from the presence of the inert magnesium. (Analysis of
sulfated dolomitic samples showed that sulfation of magnesium was negligible.)
Injection site and residence time. The gas temperature at the level of sorbent
injection was the most important variable in determining the utilization of a given
sorbent. Figure 10 shows that the removals achieved with Marianna limestone
injected with different firing rates and fuels lie along a single curve. Note, for
example, that the removal was similar for injection either at S-2 with low-load gas
16-5
-------
firing, or at S-4 with high-load coal firing. The temperatures at the level of
injection are similar for these two cases, whereas the time-temperature profiles are
markedly different.
Additional combinations of injection temperature and residence time were tested with
the following experiment. S02 removal was determined for S-3 injection of Marianna
with natural gas firing rates ranging from 0.55 to 1.0 x 10° Btu/hr. The combustion
air, S02 feed, sorbent, and sorbent transport air, however, were held constant at
the rates appropriate for 1.0 x 106 Btu/hr. The temperature at the injection level
was thus varied from 1500 to 2350°F while maintaining constant geometry and mixing.
Wet flue gas volume and hence, in situ sorbent and S02 concentrations, changed less
than 5% over this range. Gas composition effects other than SO? are assumed to be
of secondary importance. Figure 11 shows that the temperature dependence was nearly
identical with that shown in Figure 10, where injection site, flue gas rate, and
fuel type were variables. The temperature dependence differed only when sorbent was
injected at the furnace exit with the load increased to 1.2 x 106 Btu/hr; that is,
the region where the temperature decay is very rapid. For injection at temperatures
from 1730 to 2000°F, S02 removal was 14 to 38% under these conditions of rapid tem-
perature decay.
For all the tests cited above average residence times were calculated for discrete
temperature regions. Figure 12 shows calcium utilization plotted as a function of
available reaction time between 1600 and 2200°F. This is a rather crude analysis
which assumes that no CaSO^ can be formed above 2200°F, and the reaction rate is
negligible below 1600°F due to chemical kinetics. This plot is presumably consis-
tent with a diffusion-controlled process that produces a negligible increase in
sorbent utilization in a reaction time exceeding two seconds, regardless of tempera-
ture within the stipulated range.
From Figures 10, 11, and 12 it is clear that injection temperature (i.e., calcina-
tion/sintering) determines the upper limit for utilization of this sorbent on
pc-fired boiler time scales. Figure 12 should, in principle, be plotted as a
utilization-versus-residence time curve for each injection temperature (or calcine
surface area, or some other measure of reactivity) but this is not warranted by the
limited number of data points and uncertainties in time-temperature characterization
of the pilot combustor. The time scale in Figure 12 should be viewed as relative,
rather than absolute, but does demonstrate that most of the potential utilization is
achieved rather quickly.
The utilization of St. Genevieve limestone is very similar to that of Marianna over
the 2200 to 2700°F range of injection temperatures. It does not, however, exhibit
the marked increase in performance shown by the Marianna in the 1800 to 2200°F
range.
The performance of high-calcium and dolomitic hydroxides over a range of tempera-
tures is shown in Figures 13 and 14. Both show a peak in S02 removal with injection
between 2000 and 2200°F. A detailed analysis of the relative importance of resi-
dence time has not been made for these sorbents.
Particle size effects. Two grades of St. Genevieve and Marianna limestone were
initially tested to examine particle size effects over a range realistic for full-
scale pulverizer capabilities. The grades thus referred to as coarse and fine
correspond to an WD of 15 and 7 pm for the St. Genevieve, and 35 and 13 pm for the
Marianna limestone. The size range was extended by air jet milling the fine grade
of each to achieve a product WD as small as 1.3 pm. Precipitated calcium carbonate
products having mean diameters of 0.7 and 0.07 um were also tested. Figure 15 shows
the utilization of each sorbent as a function of mean particle size for either
burner or S-4 injection, with baseline firing conditions at 1 x 106 Btu/hr. Utili-
zations of 25 to 30%, comparable to that of hydrated lime, are observed for the
16-6
-------
finest pulverized carbonates and the precipitates injected at S-4 (~2200°F). Very
little increase in utilization with decreasing size is seen for burner injection.
It was expected that since more of the surface area of the very fine particles can
be accounted for by external rather than pore area, these sorbents would be less
sensitive to temperature effects. The results indicate, however, that primary
particles within agglomerates may be sintering to yield a less reactive calcine
before sulfation occurs (i.e., before the temperature decreases below 2200°F). The
Multifex, for example, is extremely difficult to disperse, as evidenced by the
apparent sedimentation diameter (see Table 7).
The correlations shown in Figure 15 are least-square fits to the data, with MMD
considered the only variable. Although no account is taken of other sorbent proper-
ties, these correlations may be useful in a cost benefit analysis of various degrees
of size reduction.
C09 added to transport air. Significant levels of C02 in the sorbent jet entering
the furnace, as would be found if recirculated flue gas were used for transport,
could affect the calcination of carbonate material. Flue gas recirculation was
simulated by adding C02 to the STA (sorbent transport air) to bring the composition
of the transport gas up to 10 to 20% C02 (by volume, sampled from the transport line
after the sorbent aspirator). When C02 was added, the amount of air supplied was
reduced to keep the total volume of transport gas constant. For either S-3 or S-4
injection of St. Genevieve limestone, the presence of up to 20% C02 in the transport
gas had an insignificant effect on the capture of S02. The temperature may be high
enough (at S-4, typically 2200°F) that calcination is little affected by the pres-
ence of C02. In addition, if particle heating is predominantly due to the entrap-
ment of hot flue gas in the transport air (rather than radiation), the gas surround-
ing the particles must contain significant C02 from flue gas by the time calcination
could occur. Hence, the additional C02 brought in with the sorbent would not affect
calcination.
Staged combustion. Tests with staged combustion were limited to St. Genevieve lime-
stone, added during combustion of coal at 1 x 106 Btu/hr, Two basic configurations
were examined: sorbent added at the burner so that, with staging, it sees a fuel-
rich environment-, and sorbent added at the staging level.
Only in the case where the burner stoichiometry was quite low, and the staging air
far from the burner, did staging at all improve the utilization of limestone
injected at the burner. Less severe staging (in geometry and stoichiometry)
resulted in capture equivalent to or lower than no staging.
When sorbent and staging air were both added near the burner, capture was lower than
for unstaged burner injection. Removal with S-3 injection of limestone was improved
slightly by the addition of staging air at the same location, but only when the
staging was fairly extreme (an increase in utilization from 19-20% to 22% with a
reduction of burner stoichiometry to 70%). The small improvement in sorbent effi-
ciency is attributed to temperature effects and enhanced mixing by the staging air.
Further work with staging was not warranted by the results of these exploratory
tests.
Characterization of Solids Samples
All solids samples were taken during firing with natural gas. As a general rule, no
significant differences have been observed in sorbent performance with coal or gas
other than those attributable to temperature at the injection point. Properties of
the calcine are therefore expected to correspond to sorbent performance with coal.
One notable exception is that sorbent injected through the burner is exposed to
16-7
-------
considerably higher temperatures with gas than with coal, the difference being much
more pronounced with low-load firing.
Solids properties reported are BET surface area (by single-point N2 absorption) and
chemical composition. Composition is based on analysis for C, H, and SO4-2 and
reported as CaCO^, Ca(0H)2 and CaSO, . For dolomitic samples, all C, H, and S were
assigned to calcium. Selected samples have been analyzed more extensively with a
wide variety of techniques, but are not discussed in this paper.
St. Genevieve Limestone. Properties of calcine collected during injection of St.
Genevieve limestone either through the burner or at S-4 are shown in Figures 16 and
17. Material sampled less than a foot from the burner, directly below the sorbent
injector, is nearly 30% calcined. Calcination is complete by S-2. In the case of
S-4 injection, material sampled in the plane of injection is 40% calcined, and cal-
cination is complete by S-5. Thus, even with injection at 2200°F, calcination is
very rapid. The surface area is at a maximum for the partially calcined sorbent in
either case. Calcium sulfate cannot form above about 2200°F, by which point (T ~
2200°F at S-4) the surface area for the burner injection case has dropped to 2 m2/g.
The BET area in the reactive zone with S-4 injection is at least 6 m2/g. The sur-
face area of calcine from S-3 injection at 0.6 x 106 Btu/hr was higher at the plane
of injection, about 13 m2/g, but dropped to 8 to 6 m2/g at S-4 and S-5 (not shown).
Calcination was as rapid and as complete as for the cases shown.
Marianna Limestone. Calcine surface areas resulting from three injection modes are
shown in Figure 18. Burner injection again yields a very low-area calcine, but at
lower injection temperatures the area is increased by up to a factor of 7. More
properties of the calcine from S-3 injection at 0.6 x 106 Btu/hr are shown in Figure
19. Figure 20 shows the properties of calcine from an analogous test with S02 pres-
ent. Note the large reduction in surface area apparently due to the formation of
sulfate. The implications of significant amounts of hydroxide and carbonate were
discussed previously under sampling procedures. The St. Genevieve results suggest
that calcination is rapid even at relatively low furnace temperatures. There is no
reason to expect the Marianna limestone or hydroxides (discussed below) to calcine
less completely. Therefore, the presence of uncalcined material is likely to be due
to more extensive in-probe reaction with these more reactive sorbents.
Longview Ca(OH),. Figure 21 shows that the surface area of calcine from burner
injection of Ca(0H)2 is less than 4 m2/g at the point where sulfation can begin.
With injection at a very low temperature, about 1800°F, the calcine area is
increased to 12 m2/g. Sulfation of a calcine from a similar test lowered the area
by about 30%. With S-4 injection at 1 x 106 Btu/hr, the calcine area dropped from
15 m2/g at the injection plane to 10 m2/g at S-5 (not shown). It is notable that in
contrast to the other sorbents, furnace calcine from Longview Ca(0H)2 has never
exhibited a greater surface area than the uncalcined feed.
Genstar dolomitic hydroxide. The surface area of Gerstar injected at the burner
fell from its original 20 mz/g to less than 10 m2/g. Figure 22 illustrates that, in
contrast, injection near 2200°F yielded a calcine area of 30 m2/g which was stable
throughout the region where sulfation could occur. One sample from the plane of
injection, taken with an altered sampling procedure, had an area of 68 m2/g. Sam-
ples subsequently taken in parallel with the two methods did not show a consistent
difference in area or composition. The variations at the injection plane are there-
fore attributed to sensitivity to probe and injector positioning. Properties of the
calcine from S-4 injection either without or with S02 present are shown in Figures
23 and 24. (Note that these results for sulfation also appear in the discussion of
in-probe reaction—see Figure 7.)
16-8
-------
The high calcium utilization of the Genstar is ascribed to the production of surface
area by dehydration of Mg(0H)2, which does not itself react with sulfur in the fur-
nace environment. This idea is supported by the following.
One of the properties of the Genstar Ca(OH)2-Mg(OH) 1s that heating of the sorbent
in the laboratory substantially increases tne specific surface area. The data in
Figure 25 are the results of an experiment in which the Genstar sorbent was subject
to prolonged heating at either 115 or 300°C in a stream of dry N?. Heating at 300°C
for about 11.5 hr raised the surface area from 20 to 114 m2/g and gave a simulta-
neous reduction in weight of about 12%. Heating up to a total of 13.4 hr caused no
further change in either surface area or weight. Results obtained at 115°C for a
total of 2 hr of heating showed virtually no change from the surface area or weight
first observed.
Notations near the right-hand margin of Figure 25 allow one to compare the observed
reduction in weight at 300°C with that which would have occurred if half of the
water of hydration had been removed or if all had been removed to produce in the
first instance either CaO-Mg(OH)2 or Ca(0H)2-Mg0 and in the second instance CaO-MgO.
Obviously the weight loss corresponds much more nearly to loss of half of the water.
Heating at 600°C to complete dehydration yielded a surface area of 50 m2/g.
If only one of the hydroxides lost water, it was more likely to be Mg(0H)2 than
Ca(0H)2. This idea is confirmed by thermodynamic data from the JANAF tables, which
are shown in Figure 26 as the equilibrium water dissociation pressures of Mg(0H)2
and Ca(0H)2 at various temperatures. At 300°C the dissociation pressure of Mg(0H)2
is about 3.0 atm and that of Ca(0H)2 only about 0.003 atm.
Studies of the Longview Ca(0H)2 at 300°C have revealed a marked difference in prop-
erties from these observed for the Genstar Ca(0H)2-Mg(0H)2. Heating at 300°C for
6 hr changed the specific surface area of the Ca(0H)2 only from 17.0 to 17.4 m2/g
and the associated loss in weight was only 0.1%. This observation tends to confirm
the conclusion that only the Mg(0H)2 in the Genstar sorbent is dehydrated at 300°C.
Heating of the Longview to complete dehydration in the laboratory has not resulted
in significant area increase over the raw material.
Surface area and calcium utilization. Table 9 lists the BET areas of calcines from
the tests just discussed. The area given is for calcine sampled below 2200°F in the
case of burner injection, or one level below the injection plane for other cases.
Also listed are the calcium utilizations (based on S02 reduction) observed for the
analogous tests with coal. Undoubtedly the sulfation and calcination processes are
parallel and interdependent when calcination occurs below 2200°F. In addition, the
utilization would be dependent on the nature as well as the magnitude of surface
area (e.g., pore size distribution). Figure 27, however, shows a reasonable corre-
lation between calcine surface area and utilization.
CONCLUSIONS
The applicability of furnace sorbent injection to a given situation depends ulti-
mately on economic factors as well as required and achievable levels of S02 control.
The results of the present study are intended to: a) provide a basis for the evalu-
ation of removals achievable with furnace injection of commercially available sor-
bents, either alone or integrated with post-furnace treatment;, and b) provide a
foundation for understanding and Improving the process. Conclusions from the pres-
ent work are:
• S02 removal is a post-flame process that occurs via the formation of CaS04
in an oxidizing environment.
16-9
-------
• The level of S02 removal is dependent on sorbent type (limestone, dolomite,
hydroxide) with the hydroxides being more effective than the carbonates.
Reductions observed with injection near 2200°F were:
limestone: 40% S02 removal at CaxS = 2
dolomite: 55% S02 removal at Ca/S = 2
calcium hydroxide: 50 to 60% S02 removal at Ca/S = 2
dolomitic hydroxide: 70 to 80% S02 at Ca/S = 2
• The S02 removal with limestone depends on the initial source of the lime-
stone. The more amorphous limestone with moderately high initial surface
area exhibited a higher temperature sensitivity than the more crystalline
carbonate. S02 removals approaching 60% at Ca/S = 2 were achieved with
injection near 1900°F. This phenomenon could be difficult to exploit in
many boilers due to rapid temperature decay in the convective passes.
• Calcine surface area appears to be one of the key parameters in the S02
capture process. The pilot-scale results exhibit a good correlation between
the calcine surface area (at a temperature where CaSO^ can form) and the
level of S02 removal.
• Calcine surface area is highly dependent on the calcination temperature
history; for furnace injection, conditions appear to be primarily controlled
by sintering. Surface areas of the dolomitic, pressure-hydrated lime can
vary from 10 to 35 m2/g depending on the temperature history during the
calcination process.
• Sulfation times appear to be relatively rapid with little additional sulfa-
tion occurring as the residence time in a temperature region between 2200
and 1600°F is increased beyond 1 sec. However, the very low injection tem-
perature required to optimize capture with a sorbent such as Marianna could
pose a dilenma: it is in this region of a boiler that temperature decay
would be rapid, and residence time very short.
• For all sorbents investigated, temperature appears to be a more important
parameter than residence time.
• Optimum temperature for S02 removal is in the range of 1900 to 2150°F, the
lower temperatures corresponding to carbonates and the higher temperatures
corresponding to hydroxides. An optimum occurs because of the interaction
between sintering of the calcine surface and the kinetics of the S02
removal: at high temperatures S02 removal diminishes due to the loss of
reactive surface area caused by sintering, and at lower temperature
decreased reaction rates result in lower S02 removals (either reaction rate
or sol id diffusion).
• Preliminary results using precipitated carbonates (BET surface area
~21 m2/g) yield S02 removals approaching that of the hydroxides.
• Limestone particle size, when varied over the range available with conven-
tional grinding equipment has little effect on overall S0? removal. How-
ever, as noted above, very fine precipitated carbonates yield relatively
high S02 removals.
• Modifying the sorbent injection alrstream to contain C02 levels simulating
the use of flue gas for transport had no effect on the S02 removal.
These results are expected to be useful in an assessment of the role a DSEC process
may have in an overall acid rain control strategy.
16-10
-------
REFERENCES
1. H. R. Hazard, "Influence of Coal Mineral Matter on Slagging of Utility
Boilers," Final Report, EPRI CS-1418, by Battelle Columbus Laboratories to
Electric Power Research Institute, Palo Alto, CA, June 1980.
2. F. E. Gartrell, "Full-Scale Desulfurization of Stack Gas by Dry Limestone
Injection. Vol. I," EPA-650/2-73-019-a,-b, and -c (NTIS PB-228447), U.S.
Environmental Protection Agency, Research Triangle Park, NC, August 1973.
16-11
-------
HEAT EXCHANGER
COOLER
BYPASSES
SECONDARY AND TERTIARY AIR
MAY BE PIPED TO SEVERAL
ALTERNATE INPUT PORTS
COAL
FEED
PREHEATER
BURNER
DOWN-FIRED
FURNACE
WORK PLATFORM
TOTE
BIN
ELECTRIC
HEATER
ASH PIT
FORCED
PRIMARY
AIR & COAL
VENTURI
FLOWMETER
DRAFT
BLOWER
TO BAG
COLLECTOR
ISOTHERMAL
RESIDENCE TIME
CHAMBER
ACCESS PORTS
FOR INSTRUMENTS
WEIGH
FEEDER
VENTURI
MIXER
RESISTIVITY MEASUREMENT
LOCATION
DOTTED LINES
INDICATE WATER
JACKETS
Figure 1. Overall view of the pilot combustor.
-------
SOR BENT THROUGH CORE OR
through annulus with fuel
adjustable swirl
BURNER
0.75 FT
PORTS FOR STAGING
OR SORBENT INJECTION
S1 PORTS FOR
SORBENT INJECTION
OR SAMPLING
30 IN.
5.0 FT
7.5 FT
9 9 FT
SQUARE DUCT
ASH
PIT
Figure 2. 10$ BTU/hr combusTor.
16-13
-------
2800
GAS
1500
COAL
2600
1400
S2
2400
SI— 1300
2200
1200
2000
1000
1800
SD2
900
1600
800
1400
700
1200
0
8 *12
A
+4
INCHES FROM CENTERLINE
Figure 3. Radial temperature profiles, 1 x ICfi BTU/hr.
- - —GAS _
COAL
1500
1400
1300
1200
1100
1000
900
800
700
<
oc
I I I I I ' I 1 I I I I I
8-4 0 4 8 12
INCHES FROM CENTERLINE
Figure 4. Radial temperature profiles, 0.6 x BTU/hr.
-------
2800
1500
2600
106 BTU/hr
1400
S2
O 1.0
~ 0.6
2400
1300
2200
1200
2000 —S1
1100
S01
1000
1800
SD2
S5
900
1600
SD1
800
1400
10
20
3 0
40
50
TIME, seconds
4 3 b 0
Figure 5. Temperature profile showing port locations
2 firing rates with coal.
2800
PILOT UNIT. 106 BTU/hr
O NO. 1
A NO 2
~ NO 3
O NO 4
2600
UJ
CC
3
I-
<
S
UJ
2400
a
2
UJ
I-
Ul
O
<
AC
2200
UJ
>
<
2000
2 0
1.0
30
4.0
TIME. Mcondi
41611II
Figure 6. Temperature profiles of the pilot combustor
and 4 utility boilers. The points for the boilers
show average gas residences time above the top
burner level, calculated from gas velocity.
-------
100
z
o
p
<
N
z>
s
2
u
-i
<
o
80
60
40
20
O BASED ON EXIT S02 REDUCTION
O SOLIDS ANALYSIS. NORMAL SAMPLING
~ SOLIDS ANALYSIS, S02 PRESENT ONLY IN PROBE
SI S2 S3 S4 S5 SD1 SD2
SAMPLE SITE
MT
1 2 5 0 -1 8 5
Figure 7. Catcium utilization for doiomitic time injected at S4 with gas-firing: in-furnace
compared to potential m probe reaction. "MT" is a mass-train sampled
isokinetica/Iy at about 300 °F.
16-16
-------
100
DOLOMITIC
HYDROXIDE
CALCIUM HYDROXIDE
DOLOMITE
60
LIMESTONE
20
MARIANNA (FINE)
DOLCITO
LONGVIEW
GENSTAR
1
2
4
3
Ca/S
6160-1 t 1
Figure 8. Performance of four sorbents. AH with injection at
S4(T = 2200 °F) 1 x 106 BTU/hr Hallmark coal.
100
60
40
O MARIANNA FINE
• DOLCITO
O LONGVIEW C.IOHl
¦ GENSTAR
40
10
20
30
SORBENT FEED RATE. Ib/hr
Figure 9. Comparison of sorbents on a mass feed rate basis. All
with injection at S4, 1 x 10& BTU/hr Hallmark coal.
-------
100
T
80 —
FUEL 106 BTU/hr
O GAS
• COAL
• COAL
06
0.6
1.0
60
40 —
S4 S3 S3
V
,t—!
/
/
K
-/
V
8
\
S4
0
S2
S3
K
B.
1600
1800
2000 2200
Tmax, °F
2400
2600
17S0H1
Figure 10. ASO2 as a function of maximum temperature at injection
for Marianna limestone. Ca/S = S 2.11 - 2.35.
10® BTU/hr
RUNS C770-9 TO «5
CONSTANT f AIR
FEED < S02
RATES V.C«0
A 0 80
O 0.75
(~ 1.2. HIGHER AIR RATESI
.09 "0ȣ
54 X
a
NE-SW
- S02
9 INJECTED
AGAINST FLOW
SD2
W INJECTED
WITH FLOW
1400 1600 1800 2000 2200 2400
MAXIMUM TEMPERATURE AT INJECTION LEVEL. °F nig.int
Figure 11. ASO2 vs temperature of injection of Marianna limestone.
All with S3 injection unless otherwise noted. Ca/S = 1.83
2.27.
-------
MAXIMUM TEMPERATURE AT INJECTION
¦ < 2200°F
O > 2200 °F
30
20
15
10
5
0
0
t, seconds
Figure 12. Utilization vs. bulk gas residence time between 2200 °F and 1600 °F. Marianna
limestone, with Ca/S ~ 2.
16-19
-------
80
0.6 x 10® Btu/hr A
— f°
1 x 106 BTU/hr
60
40
20
S4
S5
8
SD ENTRANCE
TEMP RANGE
S4
8
S3
S2 S3
* 8
S2
~
~
1800
2000
2200
2400
2600
2800
tltlMl!
Figure 13. A SO 2 for Genstar PHDL: Ca/S = 1.05 - 1.16
with Hallmark coal.
100
80
60
20
1
1
1 1
1 1 1
O 0 6 x 10® BTU/hr
r
'
~ 1.0 x 106 BTU/hr
S2
9
-
S4
-
a
S3
S3
a
O
~
S4
- 0
O
-
S2
0
B
-
B
-
1
1 l
1 .
1 I 1
1
-
1800 2000 2200 2400 2600 2800
5)10-111
Figure 14. A SO 2 for Longview Ca(OH)2as a function of maximum
temperature at injection with Hallmark coal. Ca/S =
2.15 - 2.25.
-------
12-91
aso2
UTILIZATION, % =—
Ca/S
ro
©
CO
o
o
cn
o
O)
o
c
5
CJl
£ ^
3 ?
Is-
"r- O
o a
§ 5
»» 3
x 2
Co $
^ £
C
^ S-
^ ft
->
2
2
5-
0
a
01
V)
O
30
CD
m
Z
0
1
O
CO
o
cn
o
CO
o
cn
CO
©
cn
©
o
o
CO
o
cn
o
c/3 rn
t—I—r
MULTIFEX
ALBAGLOS
MARIANNA FPC
ST. GEN. CYCLONE
MARIANNA CYCLONE
MICROWHITE 25
ST. GEN. MILL
ST. GEN. FINE
MARIANNA MILL
MARIANNA FINE
MICROWHITE 100
ST. GEN. COARSE
-------
CALCINATION
SURFACE AREA
FURNACE SECTIONS SQUARE DUCT
Figure 16. Characteristics of calcine sampled during
burner injection of St. Genevieve limestone,
1 x TO6 BTU/hr, 300 swirl.
CALCINATION
AREA
LIMESTONE INJECTED
I I
S4 S5 SD1
SAMPLE SITE
Figure 17. Characteristics of calcine sampled during S4 injection
of St. Genevieve limestone, 1 x 10$ BTU/hr.
-------
20
16
12
<
lii
a
<
1 1 1
INJECTION
106 BTU/hr
1
OB
1.0
_
O S4
1.0
A S3
0.6
—
—
/
0 X
/
0
U
\ -
0
1 1
~
—
/
~O
1
~
-
~
1 1
1 1 1
1 1 1
1
S1 S2 S3 S4 S5
SAMPLE SITE
SD1
SD2
6 2 6 0-1 4 :
Figure 18. Surface area of calcine from injection of Marianna limestone under various
conditions. Natural gas, no SO2 present.
16-23
-------
I—I—I
AREA
Cj(OH)
S3 S4 S5
SAMPLE SITE
ajfto-Mi
e
<
cc
<
Figure 19. Properties of calcine collected during injection
of Marianna limestone at S3. Gas, 0.6 x 10$
BTU/hr, no SO2present.
100
AREA
80
CaO
• 60
o
E
Z
o
«/>
2
| 40
o
S1
S3
S2
S4
S5
SD1
SD2
Figure 20.
Properties of calcine collected during injection
of Marianna limestone at S3. Gas at 0.6 x ICfi
BTU/hr, Ca/S~2.
-------
16
i i—r
CM
E
« 12
LU
a
<
LU
u
<
Li-
ar
=>
W ft
U 8
£
u
UJ
a.
to
S4, 0.6 x 106, NO S02
——S3, 0.6 * 106, S02 PRESENT
.BURNER, 1 x 106, N0S02
BURNER, 1 x 10B, S02 PRESENT
J I I I 1 ...
RP
J
FURNACE SECTIONS
SQUARE DUCT
& 2 6 0 -7 B
Figure 21. Comparison of BET surface areas of samples collected during injection of Ca(0H)2
at different locations and firing rates.
16-25
-------
50
40
10
O S4 INJECTION
~ BURNER INJECTION
SI S2 S3 S4 S5 SD1 S02
SAMPLE SITE
6 2 6 0-138
Figure 22. Surface area of calcine from Genstar PHDL injected during combustion
of natural gas, 1xl(fi BTU/hr, No. SO 2 present.
16-26
-------
AREA
C»(OH)jO'
S3 S4 S5 SD1 SD2
SAMPLE SITE
Figure 23. Properties of calcine collected during injection of
Genstar PHDL at S4. Gas, 1 x 10^ BTU/hr, no
SO 2 present.
so
40
30 —
20
AREA
S2 S3 S4 S5 SD1 SD2
S1
SAMPLE SITE
Figure 24. Properties of calcine collected during injection
of Genstar PHDL at S4. J x JO6 B TU/hr,
SO 2 present.
-------
120 Ac
1.00
WEIGHT AT
t\J300 oc
100
0.96
0.90
s
AREA AT
300 °C
DEHYDRATION TO
MgO + C«(OHI2
OR CjO + Mg(OH)2
E
<
0 85 S
UJ
-------
50
8
U
of
II
CO
3
H
<
S8
z
o
4
N
40
30
20
D 10
O ST. GENEVIEVE LIMESTONE
~ MAR IANNA LIMESTONE
A LONGVIEW Ca(OH)2
V GENSTAR PHDL
0 I I I I I I I I I I
I I I
4 8 12 16 20 24 28 32 36 40
CALCINE BET AREA,m2/g
6 2 6 0-1 81
Figure 27. Calcium utilization with coal firing vs. area of calcine from analogous test
with gas-firing.
16-29
-------
Table 1. Baseline Firing Conditions
Load
Low High
Fuel rate, 106 Btu/hr 0.6 1.0
STA, sorbent transport air, scfm 14 14
STA, as % of total air 10 7
SRq, overall stoichiometry 1.3-1.4 1.2-1.3
Staging None None
Secondary air preheat, °F 600 600
Secondary air swirl, degrees 30 30
Table 2. Comparison of Overall S02 Removal With Removal
Near or in the Sample System, Gas-Firing
Flue gas temperature, °F ppm S0o removed, when added3
Sorbent Pre-sampler^ Gas sampler With fuel Pre-sampler
Limestone 340 270 1530 72
Dolomitic 760 520 940 120
hydrated
1 ime
Hydrated lime 600 400 1280 50
a. During in-furnace sorbent addition, both cases.
b. "Pre-sampler" identifies the point upstream of the gas sampling
probe where S02 is added to the flue gas. In the absence of
sorbent, recovery of S02 injected at this point is typically 99%
of theoretical, based on S02 and fuel feed rates.
Table 3. Recovery of S02 Injected Near Sample Point During
Coal Firing, With or Without Sorbent Addition in Furnace
ppm S02 increment
ppm S02 removed with S02 injected %
Sorbent (sampled at 460°F) at 675°F Theoreticala
None
none
982
102
Hydrated lime
1250
994
103
Dolomitic hydrated
1 ime
658
965
100
Dolomitic hydrated
2430
989
103
1 ime
a. Based on fuel rate, typical fuel analysis, and S02 feed rate.
16-30
-------
Table 4. Analysis of Hallmark Coal
(Jefferson Seam, Alabama, Lot 1)
Coal, as-fi red Ash
% Moisture
1.89
%
Li 20
0.04
% Carbon
74.1
%
Na20
0.25
% Hydrogen
5.21
%
k26
1.5
% Nitrogen
1.73
%
MgO
0.69
% Sulfur
3.44
%
CaO
1.1
% Ash
7.75
%
Fe203
45.9
% Oxygen (diff)
5.89
%
A1203
17.2
%
Si 02
28.7
% Volatiles
37.2
%
Ti 02
1.0
%
sP3°5
0.56
Btu/lb
13,500
%
1.1
Table 5. Chemical and Physical Properties of
Primary Sorbents
Wt % Li 20a
Na20
K20
MgO
CaO
Fe?0n
A1 2°3
St 02
Ti 02
P205
S03
LOI, %b
BET area,
MMD, urn
p, g/cm3
a. Ignited sample.
b. As-received samples.
St. Genevieve Marianna Longview Genstar
limestone limestone hydrated pressure-hydrated
(fine) (fine) 1ime dolomitic lime
0.01
0.02
0.02
0.01
0.07
0.07
0.02
0.05
0.09
0.18
0.08
0.02
1.4
1.2
3.0
39.9
97.6
90.6
93.9
58.2
0.18
1.2
0.30
0.11
0.30
1.7
0.43
<0.2
0.60
4.8
0.51
<0.1
<0.1
<0.1
0.10
<0.3
<0.03
0.25
0.13
<0.03
0.16
0.28
0.27
0.07
43.4
42.2
23.7
27.2
1.5
6.3
21
20.0
6.9
13
2.3
1.2
2.69
2.71
2.24
2.28
16-31
-------
Table 6. Physical Properties of Different Grades of
St. Genevieve and Marianna Limestones
MMD, mtti BET area , m2/q
St. Genevieve limestone, coarse 15 1.1
finea 6.9 1.5
mill 6.0 1.3
cyclone 1.5 3.5
Marianna limestone, coarse 35 6.3
fine3 13 6.3
mill 7.6 3.9
cyclone 1.7 7.5
FPC 1.3 8.3
a. Chemical properties given in Table 5. This grade was
air-milled to produce the finer materials.
Table 7. Properties of Other Sorbents
As-recei ved
Ignited LOI, MMD,3 BET area,
% CaO % MgO wt % urn m2/g
Microwhite 100, 96.1 1.5 43.2 18 0.8
pulverized marble
Microwhite 25, 97.9 1.3 43.9 3.5 3.1
pulverized marble
Albaglos, 95.3 0.44 44.1 0.68 7.1
precipitated CaC03
Multifex, 96.0 0.54 44.2 0.07b 22
precipitated CaC03
Dolcito, 55.8 37.2 45.8 14 2.0
pulverized
dolomite
Western HCPH, 94.2 1.3 25.0 3.7 17
high-calcium,
pressure-hydrated
1 ime
a. By sedimentation except as noted.
b. Manufacturer's value. Sedimentation yielded 1.0 um, appar-
ently due to incomplete dispersion.
16-32
-------
Table 8. Measured Values Used to Calculate SO,
Calculated value
Ca/S
Reduction and Calcium Utilization
Measured values
Baseline S02 ppm
Reduced S02 ppm
S02 Reduction, %
Calcium utilization, %
- sorbent feed rate
- typical sorbent analysis
- molar flue gas rate (from fuel
rate and analysis) corrected to
3% 02
- baseline S02
- average value of S02 in 5-min
period preceding test, corrected
to 3% 02
- average value of S02 over period
2 to 10 min after starting sor-
bent feed, corrected to 3% 02
reduced SO,
100 (1 2_)
baseline S02
SOt reduction
Ca/S
Table 9. Calcine BET Areas and Sorbent Utilizations in Analogous
Gas and Coal-Fired Tests
Gas-fi ring
Sorbent
Injecti on
si te
106 Btu
hr
(No SO, present)
BET area, m27g
for T < 2200°F %
Coal-fi ri ng
Ca/S = 2
Uti1i zatlon
St. Genevieve
B
1.0
2.4
17
St. Genevieve
S-4
1.0
6
19
Mari anna
B
1.0
2
12
Mari anna
S-4
1.0
10
17
Mari anna
S-3
0.6
14
24
Longvi ew
B
1.0
3
18
Longview
S-4
1.0
10
29
Longview
S-3
0.6
12
24a
Genstar
B
1.0
8
24
Genstar
S-4
1.0
30
39
a. aS02 from gas
firing test.
Test not
completed with coal
.
16-33
-------
RECENT IFRF FUNDAMENTAL AND PILOT SCALE STUDIES ON
THE DIRECT SORBENT INJECTION PROCESS
S. Bortz and P. Flament
International Flame Research Foundation
P.O. Box 10.000
Building 3G.25
1970 CA IJMUIDEN
The Netherlands
ABSTRACT
Pilot scale experiments (2-4MW thermal) using a staged combustion air burner have
shown that under optimised conditions, substantial reduction of S02 anissions by
direct injection of calcium based sorbents is possible. Various fuels including
coals, ranging from sub-bituminous to low volatile bituminous, petroleum
residues, and SO2 doped gas flames, have been tested with this technique and SO2
capture levels, with a Ca/S molar ratio of 2 and a calcium hydroxide sorbent, of
between 70 to 80% have been achieved in cases when the peak flame temperatures
were reduced to about 1250 C or lower. Both the temperature field in the furnace
and the sorbent type used have been shown to strongly influence the S02 capture
efficiency with other parameters such as the S02 and 02 concentration playing a
somewhat lesser role.
Some" more fundamental work has also been conducted in plug flow reactors, both
isothermal and non-isothermal, to better defined calcination and sulphation rates
at temperature levels ranging from 700-1300 C. The work which is only partially
complete, will also examine the effect of sorbent characteristics, gas
environment, and initial particle size distribution on the calcination and
sulphation process. The results obtained thus far have shown that at temperatures
above 1000 C, calcination is more than 80% complete in less than 200ms with CaC03
and less than 40ms with Ca(0H)2. Sulphation also appears to be reasonably fast,
reaching an asymptotic value in less than 600ms for CaC03 when the gas
temperature is between 1000 C and 1100 C. The mean particle size, in the range 3-
50 um, has a large influence on the calcium utilization for short particle
residence times.
17-1
-------
LIST OF NOTATION
ERZ - external recirculation zone
IRZ - internal recirculation zone
MC - sorbent mixed with solid fuel
SRI - Primary zone stoichiometry
TA - sorbent injected in the tertiary
t - time
T - temperature
n - molar percent of calciun reacted
t - overall reactor/sample residence
X - overall excess air level
A - % S02 reduction =
100 x (S02 without sorbent - S02
air
with sulphur
time
with sorbent)/S02 without sorbent
1. INTRODUCTION
The concern about anissions of sulphur dioxide and their effect on the
environment is not new. However, the combination of increasing S02 anissions
(in Europe sulphur emitted as S02 has gone up from 12.5 to 25 million tons
per year between 1950 and 1972 [1]) and a much higher interest of the
populations for their environment has led a number of countries to establish
severe legislation for limiting the S02-arussions from fossil fuel fired
systems. Such legislation is already applied or will be applied in the near
future in countries like Germany, Japan, US, Sweden or the Netherlands.
For conventional combustion systans fired with oil or coal, flue gas
desulphurization units are available which can redure the potential S02
anissions by using lime (CaO) or limestone (Ca003), in so-called dry, wet or
wet-dry processes where one essentially ends-up with a mixture of CaS03,
CaS04, CaC03 and CaO, in various proportions depending upon the nature and
efficiency of the process.
Not only the high investment and operational costs of these units, but also
the large space required to install than between an existing boiler and a
chimney is a strong limitation to the generalization of flue gas treatment,
particularly for existing installations.
An attractive alternative is to use the combustion chamber itself as a
reaction vessel for capturing S02, which can be done by injecting calcium
based sorbents, CaC03 or Ca(0H)2, for instance, into the combustion chamber
through the burners or through specially arranged ports.
The IFRF conducted a first series of experiments in 1980 under contract to
the Steinmueller Company [2]. The results were interesting enough to justify
some further work in 1982 as a part of the IFRF members programme (Si-
trials) and also under contract with Charbonnages de France [3] for possible
further applications to a power station fired with a high sulphur coal.
More recently experiments have been carried out in plug flow reactors at the
IFRF, in order to better understand the physical and chemical mechanisms of
calcination and sulphation.
A large quantity of experimental data have been collected during these
complementary experiments and this paper is an attempt to summarize these
17-2
-------
results and to state our present level of understanding of the phenomena
involved in the direct desulphurization.
EXPERIMENTAL SYSTEMS
FURNACES AND BURNERS
Most pilot scale experiments were carried out with an experimental staged
mixing burner previously developed by the Research Station for potential
application to wall-fired boilers and schsnatically represented in figure 2.
The main feature of this burner is the subdivision of the combustion air
into four different streams: primary air, extra primary air, secondary air
and tertiary air (or staging air).
- The tertiary air is injected at the periphery of the quarl through four
discrete ports. Tertiary air velocity can be varied by using variable
diameter inserts in these ports.
- The extra primary air is swirled with a 45 degree fixed vane swirler and
flows inside the annular coal jet. This arrangement has proven to be very
effective for flame stabilisation at high staging rates.
- The secondary air is swirled by means of the standard IFRF movable block
swirl generator. Inserts can also be used in order to maintain sufficient
secondary air velocity when highly staged combustion is considered.
- The primary zone stoichiometric ratio is defined as:
primary air + extra primary air + secondary air
stoichiometric air requirement
and is a simple way to quantify the intensity of staging.
This burner is designed for a nominal throughput of 2.3MW and has been
tested in previous trials where a primary stoichiometry SR 1 = 0.5 could be
achieved with resulting low NOx anissions and maintenance of short and
stable visible flames.
The burner was fitted to the IFRF number 1 furnace which is refractory lined
with internal dimensions of 2 x 2 x 6.25m and horizontally fired (see figure
lc) . For the sorbent injection experiments the furnace was equipped with
eight water-cooled loops which together with heat losses provided a heat
extraction in the furnace of about 50% of the total thermal input giving a
flue gas temperature of 1000 + 50 C and wall temperatures between 800 C and
1000 C.
With the staged mixing burner, the dry sorbents could be injected in three
different modes:
- Mixed with the coal (MC in figures) : by mixing the sorbent and its
transport air with the primary air-coal mixture before the burner.
17-3
-------
- In tertiary air (TA in figures): by injecting the sorbent in the tertiary
air port at the burner. This mode of injection is applicable with staged
combustion only.
- At burner periphery: when non-staged combustion is considered, by means of
four injectors of small diameter (14mn) which were inserted in the
tertiary air ports thanselves (transport air for the sorbents is only 2%
to 4% of combustion air).
Experiments have also been conducted in the IFRF vertical furnace operating
as a non-isothermal plug flow reactor, see figure la. This furnace is 4m
high, has an internal diameter of 0.6m and is refractory lined.
A gas fired burner is set at the top and its function is to generate hot
combustion gases, the temperature and oxygen content of which can be
respectively adjusted within the ranges 1000 C to 1400 C and 1 to 12% by
means of removable cooling pipes and oxygen injection. The essential
function of the upper furnace zone is to create a plug-flow with an average
velocity of lorry's and a well controlled temperature and gas composition.
The material to be studied is injected with an inert carrier gas (Nitrogen
in practice) at the top of the "working section" of the furnace by means of
a solid distributor, which spreads the solid particles across the entire
width of the furnace ensuring, fast radial dispersion of the solids in the
flow. The vertical position of the solid injector can be considered as the
origin of the reacting flow and by inserting sampling probes in the furnace
through accessible ports at variable positions downstream it is possible to
follow the progress of reaction. Residence time measurements have shown that
the furnace is operating as a plug-flow, i.e. without any significant back-
mixing and in such a way that axial distances can be easily converted into
residence times.
Studies with the new IFRF isothermal plug flow reactor have also started,
see figure lb. This reactor tube is 80nm diameter and is divided up into six
electrically heated sections. The hot gases are supplied by a precombustor
in which combustion of a combination of blast furnace and natural gases
produces a gas similar to that achieved with coal combustion. The
temperature of the gases can be varied in the range 700 - 1300 C by changing
inputs and also with cooling pipes which can be inserted into the
precombustor.
The gas concentrations can be altered by changing inputs and by the addition
of N2, C02, H20 or S02. The hot gases are then injected into the reactor
where the electric elements maintain a constant temperature along the
reactor length.
The total gas residence time in the reactor can be varied from 153ms to
about 600ms, by changing the velocity (flow rate) of the gases. By inserting
a water-cooled probe into the reactor from gas and solid samples can be
taken at any height inside the tube. Samples can be taken after residence
times as short as 10ms. The solids are injected at the top of the reactor
tube into a venturi to assure good solid and gas mixing. Solid injection
rates are typically 500g/hr. The solids drawn into the probe are quenched to
about 150 C with an inert gas quench. The quenching time is estimated at
about 2-3ms for particles less than 20 Urn.
17-4
-------
2.2 FUELS AND SORBENTS
Three coals were used in the more recent experiments:
- a German bituminous coal from the Saar area having 1% sulphur;
- a sub-bituminous coal from Gardanne (France) having around 4% sulphur and
30% ash as mined;
- the same Gardanne coal washed down to 8% ash.
Analyses of these coals and of their ashes are given in table 1. The
important feature of the Gardanne raw coal is that its ash contains 50% CaO
giving it potential for significant "natural retention" of S02 in the ash
itself.
High sulphur containing (about 3.5%) solid and liquid petroleum residues
have also been tested with the direct sorbent injection technique.
The analyses of the four main sorbents refered to in this paper are shown in
table 1. The sorbent described as shale is extracted during the mining of
the Gardanne coal and comes from layers adjacent to the coal stream. In
figure 3, the particle size distributions for the sorbents tested, are
shown.
3. SIMPLE BASIC CONSIDERATIONS
When a finely ground CaC03 or Ca(0H)2 is injected into a flame, it undergoes
calcination reactions as follows:
CaC03 -> CaO + C02, or (1)
Ca(OH)2 -> CaO + H20 (2)
Thermodynamic considerations, confirmed by experiments, indicate that
calcination starts around 800 C for (1) and around 500 C for (2), see figure
4. Usual gas temperatures in a pulverized coal flame are significantly
higher and one can expect a "flash" calcination of the sorbent liberating a
freshly calcined lime which can react with S02 according to the overall
chemical equation:
CaO + S02 + 1/2 02 <-> CaS04 (3)
Intermediate steps involving the formation of CaS03 or the direct reaction
of CaO with S03 for instance ought to be considered when a detailed analysis
of the chemical process is required but for the engineer this equation is a
simple description of the overall process.
The reaction described by equation (3) is reversible and at high
temperatures, CaS04 becomes thermally unstable. Data from thermodynamic
calculations and also a few experimental results on thermal decomposition of
CaS04 are available in the literature and they have been plotted in figure 5
[4, 5, 6]. There is a fair agreement between calculations and experiments to
show that with 4% 02 and 1000ppm SO2 in the gas phase, CaS04 is unstable
above 1200 C. It is also seen that the gas phase composition has a strong
influence on the stability of CaS04: in presence of 1% CO and with 1% 02 and
1000ppm S02 in the gas phase (which is typical of what can be found in the
first part of a pulverized coal flame) , the limit of stability of CaS04
17-5
-------
drops to about 1000 C. If carbon is present with CaS04 (in the form of soot
for instance) the limit of thermal stability drops below 900 C for 1000ppm
S02. Although these data are relative to systems at equilibrium, they
indicate clearly that temperature must be a predominant limiting factor in
the capture of S02 by calcium in flames.
RESULTS FROM THE PLUG FLOW FURNACES
3.1.1.Calcination results
In figure 6, calcination rates for CaC03-3 and Ca(OH)2-2 are shown for
temperatures ranging from 700-1300 C. It can be seen that the rate of
calcination for CaC03 is reasonably fast for temperatures greater than 1000
C, while with Ca(0H)2, calcination is extremely fast for all temperatures
examined. It is important to note that when the calcium hydroxide was
calcined at 700 C and 900 C, that significant amounts of CaC03 were quickly
formed. However, at temperatures of 1100 C or higher, very little CaC03 was
found in correspondance with equilibrium considerations, see figure 4.
Also Ca(0H)2 was found in the calcined sample, even after exposure to high
temperatures for long residence times. It is believed that this calcium
hydroxide was reformed in the sampling probe or collection filter.
Calcination rates have also been measured in gas streams with C02
concentrations ranging from 19% to 8.5% and H20 concentrations from 19% to
6.5%. In this range of variation, very little change in the calcination rate
or final level was found.
Measurements of particle size before and after calcination indicate that
some particle breakup does occur during calcination with the percentage
breakup being a function of the gas temperature. With a gas temperature of
1300 C the increase in mass percent of particles less than 10 um increased
frcm 65% to about 85% for the Ca(0H)2 sorbent.
3.1.2 Sulphation results
In figure 7 the sulphation results from the non-isothermal plug flow reactor
are shown. During these experiments the high ash Gardanne coal was injected
into the reactor without additional S02 or calcium (inherent Ca/S = 2.23).
Further the shale mined between the coal seams was injected along with S02
(Ca/S = 2.27). In both cases the potential S02 concentration, whether from
coal combustion or from direct injection was similar. The temperature
distributions shown are gas temperatures as measured by suction pyrometer.
Several conclusions can be drawn from these relatively brief experiments.
At temperature levels where high concentrations of CaS04 are
thermodynamically stable, and high enough for rapid calcination, the rate of
sulphation is reasonably fast reaching an asymptotic value after about 0.5s.
If the shale (Ca003) is injected at a temperature where the formation of
CaS04 is limited by what are probably thermodynamic reasons, curve 3 in
figure 7B, as the tenperature drops below 1000 C, the sulphur capture level
approaches that of curve 4b. However, if the sorbent is exposed to a high
enough temperature, somewhere between 1100 C and 1200 C, then the capture
efficiency seems to be limited to a level considerably below that achieved
17-6
-------
at temperatures where the sorbent is always below 1100 C, see curves la,
2a, 3a, 4a in figure 7.
Another important conclusion is that, especially at temperatures where the
rate and final level of CaS04 formation are maximized (T<1100 C), neither
the mineral matter in the coal ash nor the combustion of the organic
fraction of the coal have a predominant effect on the formation of CaS04.
in figure 8 some initial sulphation results from the isothermal reactor are
shown. These experiments where done with an initial S02 concentration of
2003ppm and a Ca/S ratio ranging from 1.7 to 2.1. These samples have been
collected when the furnace was set up to study calcination and consequently
the residence times are quite short; 150ms for Ca(0H)2 and about 300ms for
CaC03. For both CaC03 and Ca(0H)2 and with the gas composition shown in
figure 8, the optimum temperature for sulphur capture in these short
residence times is between 1000 and 1100 C. At tanperatures below 1000 C
the fraction of calcium present as CaC03 increases, reducing the amount of
calcium available for reaction with sulphur. In both cases the increase of
CaC03 corresponds to a sharp drop in CaS04. The percent CaC03 present at
900 C is much greater with the CaC03 sorbent than with the Ca(0H)2. This
perhaps can be explained partially by the fact that initially the CaC03
reformed after calcination of Ca(0H)2 would be predominately at the
particle surface which is where the reaction with sulphur occurs, while in
the case of the CaC03 sorbent the uncalcined material is probably well
dispersed throughout the particle.
Also the data in figure 8 indicates that at temperatures below 1000 C,
CaS03 begins to be the major calcium-sulphur form present. This again
appears to be related to thermodynamic considerations which show that CaS03
will begin to decompose at about 950 C [7].
The calcium utilization curve with a Ca(0H)2 sorbent shown in figure 8 has
a minimum at between 1000 C and 700 C. The reason for this appears to be
the predominance of CaSQ3 at 700 C, while at above 900 C CaS04 is the major
calcium-sulphur compound found.
Ca(0H)2 sorbents of various mean particle sizes (49, 24 and 3 ym) have also
been injected into the isothermal reactor. With an overall particle
residence time in the reactor of about 150ms, a significant difference in
sulphation level was found at 1100 C and 700 C, see figure 9. Frcrn these
initial tests it is uncertain if the ultimate calciun utilization level is
a function of the particle size, or if only the sulphation rate is changed.
In figure 9, the curves between 700 and 1000 c are drawn as a dashed line
due to uncertainty about the nature of the curve in this range, see figure
8. Also in figure 9 the percentage CaC03 originally present in the Ca(0H)2
sorbent for the different sizes are shown alone with the mass mean particle
size and the measured BET surface areas.
In conclusion, the smaller scale isothermal and non-isothermal work
performed at the IFRF up till now indicates that a relatively narrow
tenperature range exists (1000 C - 1100 C) where the calcination and
sulphation processes are optimized so that high capture efficiencies can be
achieved in short times (< 600ms), and with small Ca(0H)2 particles perhaps
much faster.
17-7
-------
At temperatures above 1100 C the sulphation process is limited by equilibrium
considerations and very likely dead burning, whilst at temperatures below
1000 C the calcination rate with CaC03 or CaC03 formation with Ca(0H)2
sorbents appears to limit sulphation. At temperatures less than 900 C, CaS03
becomes the major sulphur calcium compound formed, at least for short
residence times.
3.2. PILOT PLANT SCALE RESULTS
3.2.1 Effect of flame temperature and injection location
One major controlling parameter for the capture efficiency of sulphur by
direct injection of calcium based sorbents in or near a flame is flame
tanperature. in figure 10 the correlation between peak flame temperature and
the percent S02 reduction is shown as measured when the fuel and sorbent 'were
mixed during various pilot plant scale experiments (2.5 MW thermal input). It
is important to note that most of these data have been generated during
experiments in furnace number 1 where the average residence time is about 6
seconds, and the residence time at temperatures of about 1050 C or less, is
typically about 2 seconds, see figure 12. This again indicates that exposure
of the sorbent to high temperature causes a change (sintering) in the
reactivity of the stone, so that its capture efficiency is reduced when the
gas temperature decreases to temperatures thermodynamically suitable for
sulphation. This appears to hold true for solid fuels without any appreciable
mineral matter, such as petroleum coke, see figure 10.
In figure 11, the effect of peak flame temperature on the sulphur capture
when the sorbent is injected in the staging air stream located as in figure
1. In this case there is no longer a direct relation between peak tanperature
and capture efficiency although a trend of decreasing capture with increasing
peak temperature still exists.
The probable reason for the increased capture values and reduced dependence
of sulphur capture on flame temperature can be seen in figure 13.
When the sorbent is mixed with the staging air, the jet containing the
sorbent is heated by entrainment of both external recirculation and flame
gases and as can be seen in figure 13, the temperature of the bulk of the
staging air is much colder than the flame, in general, the tanperature of the
tertiary air jet for these flames reaches a maximum at approximately the same
tanperature as the external recirculation gases. It is believed that an
indirect relation between peak flame temperature and sulphur capture arises
because a portion of the staging air jet does mix directly with the flame and
also some calcium carried back in the external recirculation zone is
entrained directly into the high temperature flame zone between the staging
air jets.
When the flame temperatures are low as in figure 13c, then the benefit of
injecting the sorbent in the tertiary air is decreased, and high capture
values can be obtained when the coal and calcium are premixed.
3.2.2 Effect of sorbent type
Another important parameter that has been found to control the SO2 capture
efficiency of the calcium sorbents is the sorbent type, in figure 14, the
results of tests with various CaC03 sorbents and a calcium hydroxide are
17-8
-------
shown. Typically Ca(0H)2 gives twice the capture of CaC03 based sorbents. It
is believed that this property can be attributed to a greater active surface
of the calcined hydroxide when compared to the carbonate.
3.2.3 Effect of fuel type
The measured S02 reduction when a calcium hydroxide sorbent was injected into
a staging air stream of various liquid, solid, and gaseous fuels is shown in
figure 15. In general, these results can be explained by the measured peak
flame temperatures and/or the sulphur content of the fuel. With fuels
containing a high sulphur concentration and those giving lower flame
temperatures, the sulphur capture was high. When flame temperatures v^re high
or when the fuel sulphur content was low then the capture efficiency was
reduced.
particularly when the sorbent is injected so that the majority of the calcium
passes around the high temperature flame zone there is no evidence that fuel
mineral matter plays an important role.
3.2.4 Effect of excess air
With Gardanne coal the excess air was varied from 5% (with unwashed coal) up
to 50%. Figure 16 indicates that for the washed coal a high excess air has a
strong beneficial effect on the S02 capture when Ca(0H)2 is used irrespective
of the sorbent injection mode: the S02 capture is increased by a factor 1.4
to almost 2 when the excess air factor is increased from 15 to 50%. Axial
flame temperature measurements were taken for these three excess air levels
in order to ensure that this positive effect was not the result of a
decreased temperature at high excess air. The same beneficial effect of a
high excess air on S02 capture was also observed with "natural retention" as
seen in figure 16. The unwashed coal was fired at very low excess air (5%)
and the natural retention dropped dramatically from 40% at 15% excess air to
20% at 5% excess air. With 50% excess air the natural retention increased to
50%. The same trend was again observed when supplementary injection of
Ca(0H)2 was carried out and figure 16 shows the same effect of excess air for
washed and unwashed coals, for natural or artificial retention and for
different sorbent injection modes.
3.2.5 Effect of SRI
One of the major objectives of the pilot plant scale experimental work was to
investigate the effect of staged combustion upon the efficiency of S02
capture. The results indicate that based on the time-ton perature or chemical
history of the sorbent particles when mixed with the coal, staged flames
generally do not have a direct advantage when compared to non-staged flames.
The advantage to staging the combustion air is that it allows the sorbent to
be injected in the staging air and thus the sorbent can bypass, at least
partially, the high temperature flame zone.
4. EXTENSION TO LARGER SCALE
Although uncertainties about temperatures fields and flame interaction and
mixing in large multi-burner water tube boilers make direct extension of
bench and pilot scale to the large scale difficult, by using some general
information about large boilers and the pilot scale data available from the
17-9
-------
I FRF and other organizations, some conclusions about the best method of
sorbent injection into large boilers can be drawn.
in wall-fired bituminous coal boilers the following general observations are
supported by either experimental data or modelling considerations.
- For large burners (> 30MW), the high temperature flame zone is much longer
and gases/solids are in-flame for a longer time than the 2.5MW flames at
the IFRF scale.
- Some mixing of gases between burners will occur.
These two observations would indicate that most of the gases/solids
travelling through the combustion zone of a boiler are at temperatures above
that at which sulphation will occur and dead burning can reduce the sorbent
reactivity, i.e. T > 1200 C, so that when the temperature drops below 1200 C
the capture efficiency will be reduced. It is uncertain if putting the
sorbent into tertiary air streams around the burner will be as effective on a
large boiler as in the IFRF furnace due to mixing of products between
different burner rows.
The bench scale results have shown that when the sorbent is injected into an
optinum temperature region, 1000 C < T < 1100 C. Both calcination and
sulphation can be fast, being virtually complete in less than 600ms. The
apparent speed of both the calcination and sulphation reactions inside the
narrow tsrsperature range suggests that sorbent injection in the upper furnace
zone, at a position where the temperature is about 1100 C would perhaps be
the most favourable injection position. Hov^ver, if this is to be successful,
good and fast mixing between the sorbent and combustion products must also be
achieved.
5. CONCLUSIONS
- With a CaC03 based sorbent the calcination rate was strongly dependent on
temperature. In order to achieve a 50% calcination level of about 50ms was
required at 1100 C, 80ms at 1000 C and 250ms at 900 C. Above 1200 C the
calcination rate was extremely fast.
- With a Ca(0H)2 sorbent, calcination was virtually complete in about 50ms
for tenperatures ranging from 700-1300 C, but CaC03 was found to form
quickly to an amount in correspondence with thermodynamic equilibrium;
almost 60% at 700 Cf 25% at 900 C and 15% at 1100 C. When S02 was present
there appeared to be a competition between CaO combination with CO2 to form
CaC03 or with sulphur to form CaS03 or CaS04.
- The bench scale tests have thus far shown that sulphation is also a
relatively fast reaction when the gas temperature is between 1000 C and
1100 C approaching an asymtotic value after about 500-600-tis. Calcium
utilization values of at least 20% for CaC03 and 35% for Ca(0H)2 appear
obtainable under the correct temperature conditions and gas environments
within this residence time, work is continuing to determine better the
utilization level potentially obtainable for different type carbonate,
hydroxide and dolomite sorbents.
- When burner and furnace conditions were such to give a peak flame
temperature of only 1250 C or when the sorbent was injected in an air
17-10
-------
stream outside the flame, calcium utilization figures of greater than 25%
for CaC03 and 35%-40% for Ca(0H)2 have been achieved in 2.5MW pilot plant
scale tests.
- Due to the high temperatures expected in large bituminous coal boilers and
the complex mixing between burner rows and the fast potential calcination
and sulphation rate at tanperatures between 1000 C and 1100 C, sorbent
injection in the upper furnace zone in a large boiler appears to be
favourable.
6. REFERENCES
[ 1 ] OECD, 1977
The OECD programme on long range transport of air pollutants;
measurements and findings
Organisation for Economic Co-operation and Development, Paris
( 2 ] G FLAMENT
Direct S02 capture in flames through the injection of sorbents
IFRF Doc. nr. F 09/a/24, IJmuiden, April 1981
[ 3 ] E PARODI and G FLAMENT
Direct SO2 removal fran Gardanne coal flames by the injection of
calcium based sorbent with an IFRF experimental staged mixing burner
IFRF Doc. nr. 3106/2/83
[ 4 ] W M SWIFT et al
Decomposition of calcium sulphate: A review of the literature
Argonne National Laboratory, 76-122, December 1976
[ 5 ] W T REID fct al
Fundamental study of sulphur fixation by lime and magnesia
Final report to Robert A. Taft Engineering Center, PHS, June 30,
1966
[ 6 ] K GOLDSCHMIDT
Versuche zur Entschwefelung von Rauchgases mit Weiskolkhydrat und
Dolcmitkalkhydrat bei Oil- und Kolenstaubfeuerung
Reihe 3, Nr. 22, Oktober 1967, Forschr. Ber, VDI-Z
[ 7 ] J D HATFIELD, Y K KIM, R C MULLINS and G H McCLAELLEN
Investigaton of the reactivities of limestone to remove sulphur
dioxide from flue gas
T.V.A. Report Division of Chsnical Development
17-11
-------
Fuel type
Saar
Gardanne
washed
Heavy
fuel
oil
Delayed
Coke
Proximate
H20
%
1.6
-
—
-
analysis
Volatile
%dry
31.95
44.0
-
9.1
Ash
%dry
7.9
8.4
-
-
LCV
MJ/kg dry
29.5
27.9
39.6
34.6
Ultimate
Total
C
% dry
73.88
71.13
85.5
89.4
analysis
0
10.97
8.55
-
1.79
N
1.64
1.70
0.61
1.46
H
4.72
4.85
10.27
3.34
S
0.90
5.34
3.41
3.38
Ash
CaO
8.51
20.45
_
—
composition
Si02
40.29
22.10
-
-
% of total ash
A1203
22.04
12.5
-
-
Fe203
15.38
14.95
-
-
MgO
2.49
3.20
-
-
Ti02
1.06
0.60
-
-
Na20
0.48
0.40
-
-
K20
1.87
0.70
-
-
P205
-
2.24
-
-
S03
7.25
22.45
—
—
Sorbent type
Ca(OH)2-2
CaC03
1
CaC03
2
Shale
Ccnbustlbles
*7 0
% dry
/. z
CaO %wt.
73.91
55.03
53.2
48.6
MgO
0.47
0.48
1.0
1.75
Si02
0.65
0.56
2.0
1.65
A1203
0.34
0.15
0.1
0.95
Fe203
0.13
0.09
0.3
2.0
Mn304
0.03
0.03
-
0.13
Na20
-
-
-
0.45
K20
-
-
-
0.1
Ti02
-
-
-
0.1
P205
-
-
-
0.16
S03
0.04
-
-
5.1
002
0.81
43.64
43.2
39.0
H20
23.18
-
-
-
Density kg/1
0.360
1.02
0.8
-
Specific area
13-32
3.52
BET M2/g
TABLE 1 - FUEL AND SORBENT ANALYSES
17-12
-------
gGS
refractory
variable
r
0 600
to stack
cooling
solid inj.
—-solid
collection
A. Non-isothermal plug
flow reactor
sorbent
variable
co,. n ,, inj
[ cooling
bypass gas
precombustor
heating elements
comb air
JJFO
NG
reactor tube = 80
= 2000
solid sampling probe
B. Isothermal plug flow
reactor
( all dimensions in mm ]
C. Pilot scale test furnace
Fig: 1 • Experimental facilities used during
sorbent injection trails.
17-13
-------
pilot & non-isothermal plug flow
o o Ca(0H)a-1
a ~ CoCC,- 1
¦ ¦ CaCC,- 2
a a Shale
isothermal piuc flow
0 0 CaCOj -3
e> e CaiOHk-2
Fig:3. Particle size distribution of various sorbents
external recirculation
visible fiame
boundaries
secondary
air
tertiary
air
extra primary
air
natural
gas for
ignition
//Iprimary fuel
/ rich zone „
reverse flow
final mixing zone
fixed
vare
swirler
primary air
& coal
ij movable
block swirier
Fig; 2 . Schematic and principie of the
experimental staged mixing burner for
low NOy combustion of pulverized coal.
17-14
-------
ppm
SO, 50,000
20.000
10000
5000
2C00
1000
500
200
100
Fig:5. Thermal stability of CaSO^ ; Experimental
and theoretical data from literature.
Gas coitid
2C
osition (a)S(b
!
aS0»*C ;2Ca0
) 1% 0l. 1%c
~ 2Sa«CQ~
/
o/.evoCOi. ot
ier curves L0/
4 0,
/
/
/
/
/
r
*
^ /
- ^ /
/
/ /
/ Ss
/?/
CaSO^vCO
;Ca0*Ca-S0f
/
/
/
/
/
/ X'
/ /en mental data
culations
900 1000 1100 1200 1X0 temp.°C.
/
%CaO
CaiOH
CaCOj-MgCO,
temp'C 1500
Fig: ^. Calcination characteristics of various
additives
17-15
-------
700*C
Ca(0H)2
r '3oo'c^"
1 ioo'c
soo'c
*
* * —
25 5C 75 100 125 150 Hms)
\
©
Ca[0H)2
YS
"2C0'C
i k * i i
P ¦
-------
I
T( C)
(A) high ash GARDANNE coal (Ca/S = 2.23)
Gas composition
SOjfpolontialJ 830-970 ppm
yj*0) (B) GARDANNE shale+SC^ injection(Ca/S=2.7)
14)0
1200
Gas composition
0,; 5 - 7 %
S0j(potential),750 ppm
1000
800
rj - CqSOm
TaO
4
2 (sec)
residence time
residence time
Fig: 7- Gas temperature and calcium sulphate formation in the non-isother-
mal piug_ flow_ furnace with shale and high ash coal from GARDANNE
-------
30
"7%
20
Ca%
10
-
Ca(0H)2
t = 160 ms
s
\
/
/
/
\
\
^ H
/
t
10
<3
,03. t = 250 n
ns
/
700
900 J 1100 (c) 1300
900 t 1100 (*c) 1300
CalOHb . t = 160ms
CaCOo . t = 300 ms
Ca COi
900
700 900 T 1100 (*c) 1300
gas composition; 02-5% , C02~1£%.
H20-9.5, SO2 -2000ppm
T 1100 |'c) 1300
Ca/S =1.7-2.1
Fig.-8. Initial sulphation results from the
isothermal reactor
20
*i%
15
10
t = 150
ms.
40*
^ 19.2 rr
^\10.6%Cc
o
'g
iCOj
i
Q.U pm
•—16% C
i
r
)-119rrT/g ^
— °N. \
/
s
' 1
\v
7% CaCOj \
600 800 1000 1200 (*C) 1^00 T
gas composition; 02-5% . C02~U%.
H20-9.5%. S02-2000ppm.
Ca/S=2
Fig:9. Effect of particle size,
Ca(QH)? on sulphation at
various temperatures.
-------
80
Ca(OH)2 mixed with fuel, Flue gas temperatures;
" Ca/S= 2. <^l 950-1050 *C
60 1——
*
£1 Z.0 ; ~
v) o-sub bituminous coal ^^ a
^ n-hvb coal
2^ ~-petroleum coke ;
0
1200 1300 1400 1500 1600 1700 (*C)
Peak flame temp.
Fig:10- Effect of Peak flame temperature on sulphur
capture(sorbent mixed with coal).
Ca
Ca
0H)2 mo
S= 2 °n
-------
Tt C)
1500
flame
, SR 1
uoo -
1.35
0.55
1 SW
0.90
3SW
1.35
1300
°.s
1200 "
1100
1000
0
6 (m) 7 AD
2
3
5
U
Fig; 12- On axis gas temperature along the
furnace length.
17-20
-------
8C
60
¦c
20
0
0
Ca/S
0
~ C0CO3-1
o Ca(OH),,
grange due to
variation of flame
Ca/S
0
(A)Saar coal [1%S d.af! staged (B)Gardanne washed coal(5,8%Sd.a f} (C)Gardanne wcshed coal staged
combustior(SR1.=05l sorbent non staged combustion (SRU1.35) combustion (SR1=05) sorbent
injected with tertiary air. sorbents injected at burner periphery injected with tert.cry air.
Fig: 14. Effect of sorbent type and mass flow on SO2 capture
for various flames with Saar coal and Gardanne coal.
A) Saar coal
BlSaar coal
4 SOt% MC = 3<.%
a 501%TA = 52%
SRI =0.85
TA= 20 ms.
\2O0
1900
external
HP (cm)iO 50 80 100
a SOa% MC - ^0%
SC1%TA = 61%
SR1 = 0.50
TA= 20ms
1300
1400
external
rec ire.
\l f 160O
40 60 8C 100
C) Gardanne coal.
4. SOa%MC = 70% I
SGi'/oTArSO0/®
SR* = 0.65. i
noo I
1050
Fig:13. Staged flames, flame temp, and SO? removal rates
with Ca(QH)? injection (Ca/S = 2)
17-21
-------
100
t SOj \%)
80
60
40
20
CalOHlj injected at pe'iphery
o Ca/S 1
1
c Ca/S 2:1
_
e
y
-
o
1.0 11 12 1.3 1.4
Cc!0H)2 mixed with coal
~ Ca/S 11
B Ca/S 2:1
¦ Cc/S 3.1
Coal
symbol, soroen: ,Ca/S
Gardcnne
(unwashed)
o—o
9 ®
natural
retention
2 21
e---e
Ca(0H)?
1.66
Gardcnne
[washed)
~—~
CafCH)j
1.0
1.0 11 12 13 14 10 1.1 12 1.3 1.4 15
stoichiometric ratio - h
Fig:16- Effect of excess air on sulphur capture
efficiency
4 SO, (%
Cc/S ratio
coal
( daf !
o Gardcnne
5.7% S
~ Blumenlhal
4,0 % S
& Saar
1,0% S
C> Ruhr
1.3% S
petroleum product
• Liquid residue
3,4% S
¦ Delayed coke
3.4 % S
natural gas
~ 1 % S
« 2 % S
¦ 3 % S
Fig.-15. Effect of fuel type on sulphur capture
(Ca(QH>2 external injection.)
17-22
-------
DEMONSTRATION OF BOILER LIMESTONE INJECTION IN AN INDUSTRIAL BOILER
C. E. Fink, N. S. Harding, B. J. Koch,
D. C. McCoy, and R. M. Statnick
Conoco Coal Research Division
Library, Pennsylvania
and
T. J. Hassell
E. I. Du Pont de Nemours and Company
Engineering Service Division
Wilmington, Delaware
ABSTRACT
Consolidation Coal Company (Consol), through its Research & Development arm,
Conoco Coal Research Division, made a commitment to expedite the development of
Boiler Limestone Injection technology (BLI) via demonstration in a Du Pont boiler
(110,000 lb/hr of steam) during the summer and fall of 1984. The goal of the
program was to demonstrate the technical and economic viability of both LIMB and
LI (injection above the burner zone) as low cost retrofit S02 control
technologies, while burning a Consol Northern West Virginia high-sulfur coal.
Technical objectives included S02 removal capability with fifty percent as a
minimum target, boiler operability issues, and ESP impacts. Commercial low-NO
burners purchased from Foster Wheeler Energy Corporation were used in the LIMB
testing. Research-Cottrel1 supplied an electrostatic precipitator and
humidification system. The demonstration program, test equipment,
preliminary results, and BLI economics are described.
INTRODUCTION
With the passage of the 1971 Clean Air Act, stringent environmental regulations
were applied to existing boilers through State Implementation Plans.
Additionally, the current debate concerning federal "acid rain" control further
indicates the need for a low cost, retrofit sulfur dioxide (S0Z) control
technology. In the case of New York State, an acid rain bill was passed into law.
Conoco Coal Research Division (Conoco) is committed to evaluate, develop,
demonstrate, and/or commercialize cost effective S02 compliance technologies. The
strategy is to create a suite of retrofit S02 abatement technologies because no
one process will be the panacea for all coal users. Site-specific economics
dictate appropriate S02 control strategy for each coal application. Boiler
Limestone Injection (BLI) is emerging as one potentially cost effective approach
to S02 control for high-sulfur coal applications.
18-1
-------
In 1982, Conoco conducted a detailed review and economic comparison of retrofit
S02 control technologies which included wet limestone/lime FGD, spray dryer FGD,
coal cleaning, and BL1. Conoco concluded BLI had potentially attractive economics
for cases requiring up to 50% S02 removal, but major technical uncertainties
existed for applications to U.S. coals and to U.S. boilers. The most critical
issues were S02 removal capability in a large utility boiler and possible boiler
operability problems including derate of generation capacity or significant loss
in boiler efficiency. Furthermore, the impact on ESP performance could dictate
additional particulate removal equipment at a cost which could eliminate any
potential economic driving force for BLI. (An economic comparison of BLI to
conventional FGD is described later in this paper.)
The simplicity and the potential low capital costs of BLI were sufficiently
attractive for Conoco to commit to a development and demonstration program
specifically designed for the application of BLI to Consol coals. The overall
goal is to expedite the commercialization path of BLI.
The first phase of the program involved BLI screening tests in the Conoco
1.5 MM Btu/hour pilot-scale combustor located at Library, Pennsylvania. The
objectives of these tests were to determine practical ranges of S02 removal as
well as identify any major radiant-section slagging or convection pass solids
build-up (fouling) problems. A Consol Northern West Virginia coal containing
about 2.5% sulfur, 6.5% ash, and 13,800 Btu/lb was the main coal tested. The
results were:
• Achieved up to 70% S02 removal at a limestone based Ca/S molar ratio
of 3.0,
• Observed no signs of harmful slagging or fouling deposits, and
• Quantified the strong residence time/temperature relationship for
S02 removal.
This work was supported by extensive Conoco laboratory efforts to better
understand the chemistry of limestone injection. (The details of the pilot-scale
combustor results will be presented at the SME/A1ME Conference in New York in
February, 1985.)
With these encouraging results, Consol agreed to support a BLI demonstration at
the Du Pont Martinsville, Virginia plant. The goal was to establish technical and
economic viability of both limestone injection integral with multistage burners
(LIMB) and limestone injection into other locations in the boiler (LI). The term
"LIMB" is used in this paper to generally describe injection integral with the
burners although LIMB originally described a specific method of limestone
injection using the EPA Distributed Mixing Burner.
DEMONSTRATION EQUIPMENT
A simplified schematic of the BLI equipment layout at Martinsville is shown in
Figure 1. The system consists of limestone handling, limestone injection,
particulate collection, and solids disposal.
Limestone Handling
The limestone handling system was designed by Conoco Process Engineering.
Pulverized limestone (>70% minus 200 mesh) was delivered by trucks and
pneumatically transferred into the limestone storage bin. The storage bin
18-2
-------
capacity was 45 tons. (In a commercial operation, it will normally be more cost
effective to pulverize on-site rather than purchase pre-pulverized limestone.)
The limestone was purchased locally on a specification of z90% CaC03 and less than
5% MgC03. Final selection was based on S02 removal capability as measured in the
Conoco pilot-scale combustor and limestone surface areas as determined in the
Conoco laboratory.
Two gravimetric feeders conveyed the limestone from the storage bin to the
pneumatic transfer system. Each feeder operated independently and could feed from
0 to 4000 pph of limestone (equivalent to a Ca/S ratio of 4.0). The limestone
from each of the gravimetric screw feeders passed through a rotary valve into a
4-inch diameter transfer pipe. The transfer lines were both equipped with a
positive displacement, lobe-type air blower. The blowers were designed to
maintain a pick-up velocity of 60 to 75 ft/sec.
The limestone was pneumatically conveyed about 300 feet through two 4-inch lines
from the limestone handling area to the boiler. The limestone was then split to
the various burner injection and LI configurations by a series of splitters.
The limestone handling system was simple to start and shut down. The limestone
addition rate was controlled by setting the desired Ca/S ratio using the Du Pont
powerhouse computer control system. The set point for the limestone gravimetri'c
feeders was then automatically changed based on the boiler steam load.
Limestone Injection
The host boiler was a pulverized coal, wall-fired, 3-steam drum, Sterling-type
boiler built in 1941 by Combustion Engineering. The boiler rating was
110,000 lb/hr steam at 545 psi and 720°F (15 MW electrical equivalent assuming a
10,000 Btu/kWh heat rate). A schematic of the boiler is shown in Figure 2. The
boiler was equipped with steam soot blowers for the superheater and boiler tubes.
Since the radiant section had no soot blowers, it was necessary to hand lance the
screen tubes and the radiant section tubes with 100 psig air. For a permanent BL1
installation, additional soot blowers would have been installed for these tubes as
wel 1.
Although the Du Pont boiler did not have a conventional "nose," a reheat section,
or a Lungstrum regenerative air heater as in a typical utility boiler; the tube
spacings in the convection pass were as tight or tighter than most utility boilers
(Table 1). Consequently, the effect of additional solids loading on the heat
transfer sections gave a good indication of utility boiler operation. Moreover,
since the radiant section temperature profiles of this boiler were representative
of many utility boilers as indicated by the full-load furnace gas exit temperature
of 2100°F, S02 removal capability should simulate performance in a larger boiler.
The only coal tested during the BLI demonstration was the same Consol coal tested
in the Conoco pilot-scale combustor program, i.e., 2.5 ±0.3% sulfur, 6.5% ash, and
13,800 Btu/lb as-burned.
Both LIMB and LI S02 removal techniques were evaluated during the demonstration
program. For LIMB, four new multistage low-NO burners were installed during an
annual boiler turnaround in April and May of 19$4. The new burners were purchased
from Foster Wheeler Energy Corporation and were the commercially available
Controlled Flow/Split-Flame model.1 Each burner had a nominal heat release rating
of 40 MM Btu/hr. Burners for utility installations would be almost identical in
design but 100 to 300 MM Btu/hr capacity.
18-3
-------
Boiler performance with the test coal was evaluated during a coal-alone baseline
run in October of 1983 with the old burners. The new burners were tested with and
without limestone addition in June through September, 1984. Foster Wheeler
assisted Conoco during the baseline and LIMB portions of the testing program.
In the limestone injection (LI) mode, limestone was injected into the boiler at
three different elevations on the wall opposite of the burners and on the side
walls (Figure 2). The different elevations represented different gas temperature
regions and different residence times. During the LI portion of the testing
program (in progress at this time), a parametric test matrix included the
following variables: injection elevation, number of injectors (maximum of six),
injection depth, injection velocity, and injector tip design. The effects of gas
temperature at the injection location, residence time, and limestone distribution
were a function of these independent variables.
The LI approach is an alternative to LIMB, especially for boilers where the burner
heat release rate is sufficiently high that the resulting furnace gas temperatures
exceed the gas temperature application limitations of burner injection, i.e., a
furnace gas exit temperature greater than 2300°F. The application of LI is boiler
specific, and the following two points must be addressed:
1. Is there sufficient residence time available in the optimum
temperature window for S02 capture at cost effective limestone
utilizations?
2. Can a LI system be designed for limestone distribution which results
in intimate mixing with the flue gas in a particular utility boiler?
These issues should be further clarified when the Martinsville data are fully
analyzed and engineering judgement applied to the limestone distribution issue.
Particulate Collection
As shown in Figure 1, a 3000 to 5000 ACFM slipstream portion of the boiler flue
gas was diverted after the induced draft (I.D.) fan to a pilot-scale
electrostatic precipitator (ESP). Normally, the flue gas from the boiler goes to
a baghouse which is common for all Martinsville boilers. The ESP system was
added because most utility coal users have ESP's rather than baghouses. The ESP
was supplied by Research-Cottrel1, a leading manufacturer of particulate control
equipment. The ESP consisted of two fields which yielded a total equivalent 216
SCA at 4000 ACFM. The plate and wire configuration was typical of pre-1971
installations designed for high-sulfur coal applications.
With the addition of limestone, which is converted to lime and calcium sulfate in
the boiler, the ESP inlet particulate grain loading increases substantially along
with the resistivity of the particulates. Both aspects have a negative impact on
the ESP performance because of its constant efficiency characteristic and its
natural dependence on electrical properties of the particulates. ESP emissions
could increase by a factor of ten or more.
For our testing program, Research-Cottrel1 supplied a humidifier as a means to
compensate for both the increased loading and resistivity effects. The
humidification or evaporative cooling approach had three benefits:2
1. Cooling of the flue gas resulted in lower gas volume which means a
higher effective SCA for the precipitator.
18-4
-------
2. Cooling of the flue gas allowed enhanced electrical energization of
the ESP.
3. Cooling and humidification reduced particulate resistivity as
observed in dry scrubbing applications.3
Pulsed energization4 was also studied to address the high resistivity problem
associated with BLI. Since pulsed energization does not compensate for the
increased particulate loading, it must be used in combination with humidification
or some other means of flue gas conditioning to obtain coal-alone base level
emissions.
Sol ids Disposal
The Martinsville plant used a water-hydrovac system for both fly ash and bottom
ash. The Du Pont hydrovac ash removal system was modified to alleviate the
problems experienced during the TVA Shawnee limestone injection test in the early
1970*s.5 When the lime modified fly ash is exposed to water, it has pozzolonic or
cement-like properties. A pneumatically driven ram device was installed in the
hydrovac inlet piping to clear any solids build-up in the hydrovac throat. The
number and diameter of the hydrovac water nozzles were also increased to provide
increased conveying capacity.
The 1imestone/1ime/calcium sulfate/coal ash slurries from l^ie bottom ash, ESP, and
baghouse were then sluiced to a settling/surge Hypalon -lined pond which was
constructed for these tests. The pond had sufficient residence time (minimum of 8
hours) to settle the solids. The supernate from this pond had a pH of greater
than 11 and required neutralization with concentrated sulfuric acid to comply with
plant effluent pH specifications. The neutralized overflow was then channeled to
the existing main ash pond.
The high cost involved with ponding and neutralization would normally dictate dry
solids disposal for large-scale, long-term operation. For these tests,
site-specific requirements made the ponding/neutralization approach more
economical.
PROGRAM OBJECTIVES
The technical objectives of the demonstration program covered S02 removal
capability, boiler operability issues, and ESP impacts.
S02 removal capability was studied over the operating range of the boiler within
reasonable limestone addition rates as dictated by economic constraints. Fifty
percent S02 removal was the target for both LIMB and LI. Operating range of the
boiler included fifty percent load to full load. Testing was limited to a maximum
limestone addition rate equivalent to a 3.5 Ca/S molar ratio. The use of four
burners (two rows of two burners each) permitted the study of burner-to-burner
interaction on S02 removal capability which had not been fully addressed on
pilot-scale combustor units.
Boiler operability issues such as potential derate or substantial loss in boiler
efficiency were major concerns going into the demonstration program. For every
four to five tons of high-sulfur coal burned in a boiler, about one ton of
limestone is required for 50% S02 removal. A big question mark with BLI was "how
would these additional solids affect the heat transfer surface areas in the
radiant and convection pass sections?"
18-5
-------
The impact on ESP performance and the effectiveness of particulate emission
mitigation techniques are critical to the process economics of BL1. The addition
of a new ESP or baghouse could double the capital cost requirements of BLI.6
PRELIMINARY RESULTS
Since the demonstration program is still ongoing at this time, only preliminary
data and conclusions are presented here. Further reduction of the data is
required before detailed analysis can be reported. Full material balances will be
presented in a future paper.
LIMB Testing
Baseline boiler testing on the high-sulfur Consol coal without limestone addition
was performed with the new burners during June and July, 1984. During this
period, the burners were tuned for optimum low-NO and burner flame conditions. A
NO level of 0.4 lb/MM Btu was achieved and subsequently maintained throughout the
LIMB testing program.7
LIMB testing began in mid-August and was completed at the end of September.
Testing was around-the-clock on a seven-days-a-week basis. The testing was made
in two phases: screening/optimization tests and longer term tests.
a. SO, Reduction
After several weeks of screening and optimization tests, 50% S02
reduction was demonstrated at a Ca/S molar ratio of 2.5 to 2.75
during a three-day run. Equivalent calcium utilization was about
18% to 20%. During the LIMB screening and optimization tests, best
calcium utilization resulted while injecting limestone into only
the top two burners. This mode of injection was then used during
the long-term runs.
S02 reduction (measured by S02 analyzer and modified EPA Method 5)
was a strong function of Ca/S ratio and boiler temperatures, which
vary both with boiler steam load and cleanliness of the boiler.
S02 removal data were collected between Ca/S ratios of 1.5 and 3.5.
The calcium utilization decreased slightly as the Ca/S ratio was
increased in agreement with work by others.6'9'10 An exact
correlation was difficult to determine because of the background
interference caused by the change in load and the cleanliness of the
boiler as measured by the solids build-up rate on the heat transfer
surface areas. For example, as boiler load was decreased from
110,000 Ib/hr to 50,000 lb/hr (half load), S02 reduction increased
to over 60% at a Ca/S ratio of 2.5.
The impact of boiler cleanliness on S02 removal is illustrated in
Figures 3 and 4. In Figure 3, S0Z reduction is plotted versus
time. With clean tubes at time zero, S02 removal was 55%. As
illustrated in Figure 4, as the heat transfer surface areas were
covered with deposits, the flue gas temperature increased. The
boiler gas exit temperature increased until after four hours soot
blowing was required as dictated by the steam superheat temperature
alarm point. With these higher boiler gas temperatures, S02
removal dropped off to about 50%. When soot blowing and hand
lancing were not performed for a 14-hour period, as shown in
Figure 3, S02 removal dropped off to 42%.
18-6
-------
b. Boiler Operability
The addition of limestone did not cause any boiler derate as
110,000 lbs/hr steam was achievable while removing 50% S02. Also,
no significant loss in boiler efficiency was measured during these
conditions.
With the addition of limestone, the inert solids loading in the
boiler increased by a factor of about three compared to coal-alone
baseline tests. (At full boiler load, about 2200 lb/hr of
limestone was required to achieve 50% S02 removal.) Based on
visual observations, the solids deposition rate on the radiant and
screen tube sections of the boiler also increased by a factor of
three when limestone was added at a 2.5 Ca/S ratio. The tube
deposits were very friable and easily soot blown, and increased
boiler cleaning frequency maintained steam cycle and boiler exit
temperatures within control limits as illustrated in Figure 4.
Comparisons of frequency of convection pass soot blowing and boiler
hand lancing plus ash pulling duration are shown in Table 2 for
operation with and without limestone addition. The radiant section
of the boiler required the most dramatic impact in cleaning cycles,
which increased from once per day to four times per day.
LI
Limestone injection testing began on October 1 and was limited to a 30-day testing
period. Preliminary results supported the earlier pilot-scale observations that
S02 removal is a strong function of injection location and injection methods.
About 50% S02 removal was achieved at a Ca/S ratio of 2.5 while operating at about
80% of full boiler load. For the Martinsville boiler, preliminary data also
indicated that calcium utilization was slightly lower during best LI conditions
compared to LIMB operation. Further testing is in progress to determine if either
residence time and/or limestone distribution is the cause for the difference in
results.
Boiler operability impacts appear to be less with LI operation compared to LIMB
operation because of a reduced solids deposition rate in the boiler radiant
section. Other impacts are similar to LIMB operation.
Boiler Inspection
After two months of semi-continuous limestone addition, the boiler was shut down
in mid-October and thoroughly inspected for any erosion, corrosion, or pluggage.
No unusual solids deposits, corrosion or erosion were found. Even though no
catastrophic short-term effects were evident, a much longer period of operation,
i.e., six months to one year, is required to demonstrate lack of problems in a
utility boiler.
Particulate Collection
The additional lime and calcium sulfate material from the test boiler presented no
problems to the Martinsville baghouse. Some plugging problems were experienced,
however, when the lime-modified fly ash from the baghouse ash hoppers was
inadvertently wetted and allowed to deposit in the ash removal piping. Other
plugging problems were experienced in the vacuum side of the hydrovac system
indicating that additional conveying capacity may be required.
18-7
-------
As previously discussed, limestone addition creates problems with ESP operation,
and mitigation control techniques are required to avoid installing costly
additional ESP plate area or a new baghouse. While operating at a 2.5 Ca/S ratio,
the particulate loading increased by a factor of three and the resistivity
increased from a satisfactory level of 1090hm-cm to a more troublesome level of
2x1011 Ohm-cm. As a result, the ESP particulate emissions increased by a factor
of 8.4 when compared to operation with only the coal fly ash. This emission is
approximately equivalent to burning a coal having an ash content of 37%.
The results of the pilot-scale ESP tests are summarized in Table 3. ESP
performance is expressed in two ways:
1. The relative emission rate expressed as a fraction of the
particulate emissions experienced during the boiler limestone
injection case with the ESP alone.
2. The equivalent coal ash content which relates the emission to a
hypothetical high-sulfur coal only, unaltered ESP case.
Baseline tests with coal-alone showed particulate collection efficiencies in the
98% to 99% range, levels typical for a full-scale unit.
Two potentially cost effective technologies, pulsed energization and
humidification, were tested as control strategies to improve the particulate
emission level. These technologies were tested independently and in conjunction
with one another.
Humidification in combination with the use of low ash coal was the most effective
mitigation strategy, reducing the boiler limestone injection base emissions by a
factor of four (relative emissions rate of 0.24). The emissions for this case
were equivalent to burning a 12% ash coal in the boiler. Further emissions
reduction is expected for full-scale units where the existence of additional ESP
electrical fields should reduce the impact of gas (and solids) bypassing and
particle re-entrainment. For example, a six percent bypass would account for the
difference in emissions between the coal-only and the BLI plus humidification
cases.
Pulsed energization in combination with a low ash coal overcame the penalty
resulting from high particulate resistivity improving the baseline boiler
limestone injection emission by a factor of 1.7 (relative emission rate of 0.58).
This emission is equivalent to feeding high-sulfur coal having an ash content of
25%.
These preliminary ESP results are based on screening tests. Longer term
confirmation runs were made during the long-term LI testing program in late
October.
Conoco also evaluated S02 reduction potential across the ESP by operating the
humidification process at a closer approach to dew point than would be required
for particulate removal alone. (A la spray dryer FGD.) Results are too
preliminary to report at this time.
Sol ids Disposal
The solids wastes (lime/calcium sulfat^modified fly ash) generated during this
test program were settled in the Hypalon -lined pond as discussed previously. The
solids will be covered with fly ash and landfill, and the area will be reclaimed.
18-8
-------
The issue of solids disposal in a commercial installation must be addressed on a
site-by-site basis. The preferred and probable method will be dry disposal and
would be similar to handling waste solids from a FBC or spray dryer FGD system.
Conoco is actively studying this area. Several drums of lime/calcium
sulfate/modified fly ash were provided to the Electric Power Research Institute
(EPRI), who plan to evaluate the disposal implications of BLI created wastes in
1985.
BLI ECONOMICS
Conoco has completed a 500 MW economic comparison of BLI and conventional wet lime
or limestone FGD for utility retrofit applications using a 2.5% sulfur coal
(Figure 5). The costs are plotted on a dollars per ton of S02 removal basis to
compare processes which have different S02 removal efficiencies. Both BLI and FGD
economics were determined on a consistent basis so relative costs are comparable.
Absolute dollars were intentionally left off the figure to avoid debate over the
assumptions and bases of economic analysis. Capital costs are represented by the
crosshatched portion of the bars. O&M costs are the remaining portion.
Two bars are shown for the wet FGD approach. The main difference between the FGD
bars reflects the degree of difficulty in retrofitting an FGD system into an
existing power plant. The "easy" retrofit represented by the left bar would
reflect ample space available and no major equipment relocation. The right bar
would reflect a "difficult" retrofit installation representing a congested limited
space scenario. The difference between these two cases could be a factor of two
in capital costs.
The main assumptions for the BLI case were:
• 50% SO2 removal at a Ca/S ratio of 2.5.
• No major boiler modifications required other than new burners or LI
injection equipment.
• No major ESP modifications; that is, humidification plus new
rappers, ESP controls, etc. are sufficient.
• Limestone at $12.50/ton delivered and waste disposal at $7.50/ton.
The main cost driving force of BLI compared to conventional scrubbing techniques
is the potentially lower capital requirements. The "easy" and "difficult" FGD
cases require about 70% and 220% more capital, respectively, compared to the BLI
scenario. On a total cost basis, BLI is about 25% to 50% lower than the FGD
cases.
As with all S0Z abatement strategies, BLI costs are very site specific. Since
limestone and disposal costs account for about 45% of the total costs, the
attractiveness of BLI is a strong function of these site specific factors.
In summary, BLI looks extremely attractive compared to conventional scrubbing if a
plant has a favorable limestone supply and solids disposal situation or if the
plant has a difficult FGD retrofit problem. Obviously, the boiler must have a
satisfactory temperature/residence time profile for BLI to be effective before
these other site-specific factors can even be considered.
18-9
-------
CONCLUSIONS
Based on the analysis of preliminary data from the boiler limestone injection
demonstration at Martinsville, it is concluded:
1. 50% S02 removal is achieved at a Ca/S molar ratio of 2.5 to 2.75
while operating the boiler at full load. S02 removal increases at
lower loads.
2. LIMB appears to yield slightly better limestone utilization than LI.
It may be preferable in a commercial installation unless high burner
heat release rates make LIMB non-practical.
3. S02 removal with BLI is very sensitive to radiant section
temperature profiles caused by either changes in boiler load or
cleanliness of the heat transfer surface areas.
4. BLI, both LIMB and LI, did not adversely impact boiler operability
in terms of boiler derate or efficiency.
5. BLI did require changes in operating procedures, i.e., increased
soot blowing, lancing, and ash hopper pulling duration. For a
permanent installation, lancing could probably be avoided by the
installation of additional soot blowers.
6. Humidification of the flue gas from BLI operations reduced the
emissions from the pilot-scale ESP to an acceptably low level
(equivalent to burning.12% ash coal without BLI). Lower relative
emissions, approaching those corresponding to burning 6.5% ash coal
without BLI, are anticipated in full scale ESPs which have more
electric fields. Without mitigation techniques, BLI caused ESP
particulate emissions to increase by a factor of eight, due to both
the increased resistivity and the increased particulate loading.
7. Although it was successfully demonstrated that settling and
neutralization could be implemented to comply with local water
quality standards, dry disposal of the lime/calcium sulfate/modified
fly ash is recommended because of economic considerations.
8. Projected BLI costs are attractive compared to conventional wet lime
or limestone FGD if site specific limestone supply and solids
disposal costs are favorable or if the FGD retrofit is difficult.
Acknowledgement
The authors express sincere gratitude to Robert Chesney and other Martinsville
plant power house staff and to John De Ruyter of the Du Pont Engineering Service
Division for their cooperation during the entire demonstration program.
18-10
-------
REFERENCES
1. Vatsky, J., "Foster Wheeler's Low NO Combustion Program: Status and
Developments," 1982 Joint (EPA/EPRI) Symposium on Stationary Combustion NO
Control, Dallas, November 1-4, 1982.
2. MacDonald, J. R. and Dean, A. H., "Electrostatic Precipitator Manual
Southern Research Institute, EPA Contract, 1982.
3. Wilkinson, J. M., "Baghouse vs. Precipitator for Dry Scrubbing System-Pilot
Study Results," 4th International Coal Utilization Conference, 1981.
4. Puille, W. (EPRI); Landham, E. C. and Dubard, J. L. (Southern Research
Institute); and Sparks, L. E. (EPA), "Estimate ESP Efficiency Gains from
Pulse Energization" Power, May, 1984.
5. Gartrell, F. E., "Full Scale Desulfurization of Stack Gas by Dry Limestone
Injection," PB-230-384, August 1973.
6. Andes, G. M., Becker, D. F., and Klett, M. G., "Capital and Operating Costs
for Retrofitting LIMB Equipment to Coal-Fired Power Plants," ASME/IEEE Joint
Power Conference, Toronto, October 1-5, 1984.
7. Vatsky, J. and Schindler, E., "Limestone Injection with an Internally Staged
Low-NO Burner," First Joint (EPA/EPRI) Symposium on Dry S02 and Simultaneous
S0z/N0* Control Technologies, San Diego, November 13-16, 1984.
8. Flament, G., "Simultaneous Reduction of NO and SO Emissions from Turbulent
Diffusion Flames by Application of Staged Combustion and Direct Injection of
Calcium-Based Sorbents," in "Strategies and Methods to Control Emissions of
Sulfur and Nitrogen Oxides," National Swedish Environment Protection Board,
Report PM1637 (1983), IFRF.
9. Doutant, R. W.t Simon, R., Campbell, B. and Barrett, R. E. (Battelle Columbus
Labs), "Investigation of the Reactivity of Limestone and Dolomite for
Capturing S02 From Flue Gas," Final Report, October 1971.
10-. Chughtai, M. Y., Michelfelder, S., and Leikert, K., "Operation and
Performance Report of the Steinmuller Low-NO Control via Sorbent Injection,"
EPA/EPRI Joint Symposium on Stationary Combustion NO Control, Dallas,
November 1-4, 1982. x
18-11
-------
Figure 1
BOILER LIMESTONE INJECTION DEMO.
Gaa To
Stick
Flue Qas
Limestone
8torage
Bin
Feeders 1
DU PONT
BOILER
LI
Humldlfer
Baghouse
XX
J
Air.
P"01 Scaled
X~Y/W
LIMB
Limestone
i jy
Acid
Neutraln.
System
Settling
Pond
•Water To Plant
18-12
-------
FLUE OA8 TO
PARTICULATE
COLLECTION
Figure 2
SCHEMATIC OF DU PONT BOILER
18 FT DEPTH
^EOEND
0 LIMESTONE INJECTION LOCATIONS
(5) FURNACE OAS EXIT-FLUE OAS TEMP 7 8100* F AT FULL LOAD
0 SCREEN TUBES yf
(5) SUPERHEATER TUBES J lURNERS
0 BOILER TUBE8
0 STEAM DRUMS
0 CONTINUOUS MONITORING OF FLUE QAS (80 2 ,NOx ,0g ,C0. *C02 )
0 ECONOMIZER
0 AIR HEATER
18-13
-------
Figure 3
EFFECT OF BOILER CONDITIONS ON SO2 REMOVAL
( Whin Soot Blowing And Hand Lancing Wara Dalayad )
60
55
Clean Tub**
ho
©
S02 50
REMOVAL
45
%
40
0 o
©
L
i.
¦ ' ' ' '
0 2 4 6 8 10 12 14
TIME - HRS.
-AFTER SOOT BLOWING AND HAND LANCING -
Figure 4
EFFECT OF LIMESTONE ADDITION
ON STEAM AND BOILER EXIT GAS TEMPERATURE
STEAM
TEMP.
°F
800
750
700
650 -
BOOT BLOW A
HAND LANCE TUBES
BOILER
725 EX,T
GAS
700 TEMP.
®F
675
650
2 3 4 5 6
TIME - HRS.
— AFTER LIMESTONE ADDITION —
18-14
-------
Figure 5
ECONOMIC COMPARISON OF BLI vs
CONVENTIONAL WET FGD
$/TON
OF
S02
REMOVED
0 & M
Costs
Capital
Costs
BOILER LIMESTONE
INJECTION
FGD
18-15
-------
TABLE 1
DEMONSTRATION BOILER TUBE SPACINGS
CLEARANCE BETWEEN
BOILER TUBE BANK TUBES, INCHES
Screen Tubes
3.75
Superheater Tubes
1.25-2.25
Boiler Tubes
3
Economi zer Tubes
1
Air Preheater
2.25 O.D.*
*Flue gas flows
inside the tubes:
air outside.
TABLE 2
IMPACT OF BLI ON BOILER 0PERABILITY
Frequency of Soot Blowing and Radiant Wall/Screen Tubes
Cleaning Cycles and Ash Hopper Pulling Duration
OPERATION
COAL-ALONE
DURING BLI
Soot Blowing Frequency
once/shi ft
twice/shift
Radiant Wall/Screen Tubes
Lancing Frequency
once/day
four times/day
Ash Hopper Pulling Duration
base time
three times base
TABLE 3
PARTICULATE
COLLECTION PERFORMANCE
TECHNOLOGY
ASH TYPE1
EQUIVALENT
RELATIVE HIGH-SULFUR
EMISSION COAL ASH
RATE2 CONTENT (%)
ESP
Coal Alone
0.12 6.5
ESP
Coal + BLI
1.00 37
ESP + Pulsed Energization
Coal + BLI
0.58 25
ESP + Humidification
Coal + BLI
0.24 12
LTest coal contained 6.5% ash.
2Base case of 1.00 reflects ESP particulate emission level while
operating with boiler limestone injection.
18-16
-------
PILOT-SCALE STUDIES OF IN-FURNACE HYDRATED LIME
INJECTION FOR FLUE GAS S02 EMISSION CONTROL
G.F. Weber, M.H. Bobman, and G.L. Schelkoph
University of North Dakota Energy Research Center
Combustion and Environmental Research Division
P.O. Box 8213, University Station
Grand Forks, North Dakota 58202
ABSTRACT
Simultaneous control of S0^/N0x emissions, derived from the combustion of low-rank
coal, is under investigation at the University of North Dakota Energy Research
Center (UNDERC). Process development work has been performed on both bench-scale
and pilot-scale systems. Direct furnace injection of calcium-based S0£ sorbent
materials is the S0£ control technique under evaluation.
Furnace Injection tests have focused on the direct injection of pressure-hydrated
lime at flue gas temperatures ranging from 1500° to 3000°F (815° to 1650°C)
followed by collection in a baghouse operated at temperatures up to 1000°F
(540°C). Sorbent utilization values up to 55% at a Ca/S mole , ratio of 1.0 have
been observed. Eighty percent S0£ reduction has been observed at Ca/S mole ratios
of <_2.0. Sorbent utilization in the baghouse has never exceeded 10% for baghouse
temperatures ranging from 800° to 1000°F (425° to 540°C). Residence time and the
temperature regime of the sorbent injection location appear to be the critical
parameters controlling SO^ reduction and sorbent utilization.
INTRODUCTION
Recent national attention has focused on acid rain and its potential threat to the
environment. Scientific groups and individuals across the country generally agree
that acid deposition problems exist, but no consensus has been reached as to the
severity of the problems, the deposition mechanisms, or the appropriate
remedies. Proposals range from immediate enactment of laws to reduce SO2 and N0„
emissions from combustion sources to additional research studies of acid
deposition phenomena.
The University of North Dakota Energy Research Center (UNDERC), formerly the Grand
Forks Energy Technology Center (GFETC), has been conducting flue gas
desulfurization (FGD) studies since the early 1970's. The first major FGD program
pioneered the use of alkaline fly ash in a wet scrubber system to control SOg
emissions from western coals. Fly ash alkali FGD studies were conducted both on a
pilot-scale process development unit and a full-scale utility boiler (1,2,3).
Approximately 4200-MW of western U.S. generating capacity currently employ the fly
ash alkali process to reduce SO? emissions. A subsequent spray dryer FGD test
program evaluated spray dryer/fabric filter FGD using twenty fly ashes fran
lignite, subbituminous, and bituminous coals in a pilot-scale system (4). Direct
injection of sodium-based S0£ sorbents into combustion systems was also
investigated. Nahcolite and trona were evaluated in a pilot-scale combustion
19-1
-------
system using a fabric filter baghouse (5). Both materials were capable of
retaining 80% S02 with overall tests results similar to those reported by Muzio,
et al (6).
Pilot-scale evaluation of direct furnace injection of calcitic and dolomitic
limestone and quicklime began at the former GFETC in early 1978 (7). Sulfur
dioxide reduction and sorbent utilization values did not exceed 16% for the
limestone injection tests. The relatively low utilization of the limestone was
probably due to an inadequate time/temperature profile necessary for reaction with
SO2 and poor development of the sorbent surface area. Maximum quicklime
utilization was 32% at a Ca/S ratio of -v.1.0. A DOE-sponsored field test on a 50-
MW utility boiler examined direct injection of limestone into boiler flue gas and
into a burner operated in a low-N0x, staged-combustion mode (8). Limestone
utilization up to about 16% and SO2 removal up to 47% were achieved. Samples of
the resulting fly ash/calcined limestone mixture were subsequently found to be
reactive in a spray dryer FGD process.
Recent work at the UNDERC has focused on the development of a dry technique for
simultaneous control of S0x/N0^ emissions from low-rank coal derived flue gas.
The approach combines direct furnace injection of a calcium-based sorbent with
particulate collection in a baghouse. Specifically, direct furnace injection of
calcitic pressure-hydrated lime and additive-enhanced calcitic pressure-hydrated
lime are under evaluation for SO2 control. Three N0^ control techniques have been
or are currently under evaluation: 1) selective catalytic reduction (SCR) using a
commercial NHo/SCR reactor system installed downstream of a high-temperature
baghouse (800 -1000°F, 425°-540°C), 2) collection of potential throwaway N0X
reduction catalysts in a high-temperature baghouse with upstream NH3 injection,
and 3) additive addition to calcitic pressure-hydrated lime with subsequent N0X
control occurring in a baghouse operated at conventional temperatures. This paper
will focus on results from SO2 control experiments on pilot-scale combustion
systems.
EQUIPMENT AND EXPERIMENTAL PROCEDURE
Three combustion systems were used during the pilot-scale studies: a propane-fired
system and two systems with natural gas- or pulverized coal-firing capabilities.
The propane-fired (PF) combustor and the ash fouling (AF) combustor were both
operated using the high-temperature baghouse. The particulate test combustor
(PTC) system used a baghouse operated at conventional temperatures. A significant
portion of the injection tests performed 1n the past year were conducted using
calcitic pressure hydrated lime produced in a bench-scale batch pressure hydrator
built and operated at the UNDERC. Equipment descriptions and general experimental
procedures are presented in the following paragraphs.
Propane-Fired Combustor/High Temperature Baghouse System
The propane-fired combustor system has two operating configurations, high-
temperature and 1 ow-temperature. This paper discusses only the high-temperature
(PF/HTB) configuration.
The high-temperature configuration consists of: a propane-fired combustor, 5
annular heat exchangers, a l2-1nch (30.5 cm) ID test section, a baghouse, and gas
sampling instrumentation. Figure 1 provides a simplified diagram of the unit.
The propane-fired combustor is a water-cooled refractory-lined system. Four
stainless steel coils are embedded in the 6 inch (15.2 cm) refractory walls to
provide limited control of combustor outlet temperature. The cooling coils can be
19-2
-------
operated in either a parallel or series configuration. The outer shell of the
combustor is 3/16 inch (0.48 cm) carbon steel. A sight port is located in the
combustor transition section. Indicated combustor outlet temperature is nominally
2700°F (1480°C) at a combustion air flow rate of 130 ± 10 scfm (3.7 ± 0.3 nr/min)
and 4% oxygen in the flue gas (on a dry basis).
The bagiiouse is a pulse-jet fabric-fiIter type designed to operate at 130 scfm
(3.7 m /min), and a temperature range of about 250° up to 1000°F (120° to
540°C). A total of 12 cage mounted bags, each 4.5 inches (11.4 cm) in diameter by
8 feet (2.44 ra) in lerwth, are hung from 2 tube sheets providing a total filter
area of 113 ft (10.5 m ).
The flue gas sampling system is capable of extracting the flue gas from three
locations in the system; post-combustor (SB No. 1), pre-baghouse (SB No. 2) and
post-baghouse (SB No. 3). Flue gas is analyzed and data are recorded for oxygen
(O2), carbon dioxide (CO2), sulfur dioxide (SO2), and oxides of nitrogen (N0X).
Ash Fouling Combustor/High-Temperature Baghouse System
The ash fouling combustor/high-temperature baghouse system combines a combustor
with natural gas- or pc-firing capabilities with the high-temperature baghouse
described previously. Figure 2 illustrates some basic system components. The
combustion chamber is approximately 30 inches (76 cm) in diameter, 8 feet (2.44 m)
high, and is refractory lined. Combustion air is preheated with an electric air
heater. Preheat temperatures to the combustor normally run from 700° to 900°F
(370° to 480°C) depending on coal moisture.
Utility grind (80% less than 200 mesh, (74 ym)) coal is produced in a hammer
mill. The coal is then screw-fed frcm a hopper into the throat of a venturi
section on the primary air line to the burner at a rate of about 75 lbs/hr (34
kg/hr). Numerous thermocouples and sorbent injection ports are located in the
flue gas duct downstream of the combustor. A more detailed system description can
be found in a previous publication (9).
Particulate Test Combustor System
The particulate test combustor system combines a combustor with natural gas- or
pc-firing capabilities with a baghouse or ESP operated at conventional
temperatures. This 500,000 Btu/hr (126,000 Kg-cals/hr) combustor was designed
specifically to generate fly ash representative of that produced in a full-scale
utility unit. The PTC is illustrated schematically in Figure 3. The combustion
chamber is 24 inches (61 cm) in diameter, 8 feet (2.44 m) high, and is refractory
lined. Vertical orientation of the combustor minimizes wall deposits and
refractory lining helps to ensure adequate flame temperature for complete
combustion and prevents rapid quenching of coalescing or condensing fly ash. The
mean residence time of a particle in the combustor is approximately 3 seconds
based on superficial velocity. Combustion efficiencies of at least 99%, based
upon 1 oss-on-ignition of the fly ash and absence of CO in the flue gas, are
evidence that incomplete combustion is not a problem with this test combustor.
Comparison of PTC-generated fly ash with fly ash collected from a full scale plant
burning the same coal , shows no major differences.
Three separate modes of operation are available with the PTC baghouse, which is
illustrated in Figure 4. The first is the shaker chamber which has 3 compartments
of 3 bags each for a total of 9 bags. Both shaker amplitude and frequency are
adjustable. Bag cleaning is accomplished by taking 3 bags at a time off-line for
19-3
-------
shaking. The air-to-cloth ratio for the shaker mode is approximately 3 ft/min
(0.9 m/min).
The other 2 modes of operation that are available both use cage-mounted bags. One
of these is the pulse jet chamber which has 3 bags for an air-to-cloth ratio of 9
ft/min (2.7 m/min). In spite of the high air-to-cloth ratio, available data
indicates that it is possible to operate this chamber at high efficiencies and
reasonable pressure drops. The other chamber has 6 cage-mounted bags with an air-
to-cloth' ratio of 4.5 ft/min (1.4 m/min). This chamber uses low pressure
expansion (reverse air) for cleaning. Cleaning is done on-Hne with 2 bags at a
time being cleaned.
The PTC baghouse is designed to operate at temperatures up to 800°F (427°C).
However, for this work the operating temperature was limited to 300°F (150°C).
Each compartment of the baghouse is equipped with heaters so that the baghouse may
be preheated to prevent condensation during startup. A more detailed description
of the PTC system can be found in a previous publication (10).
Experimental Method
After startup and stabilization of any of these canbustion systems, tests begin
with pneumatic injection of the sorbent into the flue gas stream at a specified
location with an adjustable screw feeder. Data collected during each test include
system temperatures, static and differential pressures, flow rates, sorbent feed
rate, and flue gas constituent concentrations (S02, 0?, N0„, and CO^). Samples
collected during and after each test include sorbent feed, fly ash, baghouse
hopper material, and coal (during pc-firing). Analyses of samples include surface
area, particle size distribution, scanning electron microscopy, x-ray diffraction
(XRD), x-ray fluorescence (XRF), and routine coal analysis.
Bench-Scale Batch Pressure Hydrator
The pressure hydrator currently in use is a bench-scale batch system capable of
producing 5 lbs (2270 g) of pressure hydrated lime per run. A schematic of the
pressure hydrator is presented in Figure 5. The hictfi pressure air line
pressurizes the water storage tank to 150 psig (10.3 x 10^ N/m ). The storage
tank is connected to the reactor chamber by a 0.5 inch diameter (1.3 cm) stainless
steel pipe. A ball valve between the reactor and the storage tank controls the
rate of flow of the water Into the reaction chamber. The reactor consists of an 8
inch (20.3 cm) schedule 80 carbon steel pipe that is 12 inches (30.5 cm) long with
carbon steel flanges capping both ends. A paddle-wheel stirrer was installed in
the reactor to mix the lime and water thoroughly. The paddle wheel rotates at 2
revolutions per minute. Ejection of the hydrated product is accomplished through
a 1 inch (2.5 cm) stainless steel pipe and ball valve. An orifice plate,
downstream of the ball valve, measures the gaseous flow rate of the reactor
effluent. A nylon bag capable of collecting submicron size particles is clamped
onto the pipe downstream of the orifice plate.
For the batch operation of the pressure hydrator, a known amount of lime (-140
mesh, 110 pm) is introduced into the reactor. The internal wall of the reactor is
heated to 330°F (170°C). When the reactor is at the proper temperature, water is
carefully added under pressure to the lime in the reactor. The water not reacted
with the lime i,s vaporized, increasing the reactor pressure to approximately 100
psig (6.9 x 10 N/m ). Due to the exothermic hydration reaction, the reactor
temperature rises to about 410°F (210°C). The pressure and temperature of the
reactor are monitored on a strip chart recorder.
19-4
-------
About 20 minutes after starting water injection, the reactor temperature levels
off and begins to drop. At this point, it is assumed that all of the lime has
reacted to the hydrated form. The hydrated lime is then ejected from the reactor
through a ball valve. The rapid expansion of the water vapor trapped in the pores
of the hydrated lime during ejection is responsible for the reduction in product
particle size. The gaseous flow rate is maintained at about 10 scfm (0.3
irr/min). At this point the submicron pressure hydrated lime may be collected in a
nylon bag, or injected directly into a pilot-scale combustion system.
DISCUSSION OF RESULTS
Propane-Fired Combustor/High-Temperature Baghouse System
Pilot-scale evaluation of direct in-furnace injection of calcium-based sorbents
began in 1978. Results of those early tests indicated that better SO? reduction
and sorbent utilization could be achieved with a hydrated lime than witn limestone
or quicklime.
Hydrated lime injection tests were performed to investigate the effects of flue
gas injection point temperature, baghouse temperature, and Ca/S ratio on SOp
reduction. A commercial slaked calcitic hydrated lime was the primary sorbent
used with a few tests conducted with a commercial dolomitic hydrated lime. All
the hydrated lime used during the test series was air classified to -325 mesh
(<44 um).
Baghouse temperature, air-to-cloth ratio, and SO? levels were nominally constant
at 900°F (480°C), 6.3 ft/min (1.9 m/min), and 65(J ppm, respectively. Fly ash was
continuously injected into the flue gas at a rate of approximately 2.0 gr/scf (4.6
g/m ) to provide a filter cake on the ceramic bags. X-ray fluorescence data for
the commercial slaked calcitic hydrated lime showed calcium oxide (CaO) content
was approximately 58% by weight. Surface area prior to injection was
approximately 10.0 m /g. X-ray fluorescence data for the dolonitic hydrated lime
showed calcium oxide (CaO) and magnesium oxide (MgO) content were approximately
40% and 30%, respectively. Surface area of the dolcmitic hydrated lime was about
15.0 nr/g prior to injection. Stoichiometric ratios were based on the CaO content
of the respective sorbents.
Sulfur dioxide reduction and sorbent utilization data for the commercial slaked
calcitic hydrated lime are presented as a function of flue gas injection point
temperature in Figures 6 and 7, respectively. Figure 6 shows that SO^ reduction
for Ca/S ratios of 0.8 to 1.33 increased from about 15% to 40% as injection
temperature increased from 1650° to 2600°F (900° to 1425°C). The curve begins to
level off at injection temperatures ranging from 2300° to 2800°F (1260° to 1540°C)
indicating a maximum SO2 reduction of 40% to 50% for the range of Ca/S ratios
stated above. The SOg reduction profile for the Ca/S range of 1.5 to 2.2 was
similar, although the curve was slightly higher. Improved SOg reduction with
increasing injection temperature may be attributed to an increase in the particle
residence time in the temperature regime where S02-Ca0 reaction kinetics are
significant (1600°-2200°F, 870°-1200°C). These results were achieved at
temperatures higher than those reported by other investigators and are probably
attributable to the short particle residence times at high temperatures (2200°-
2800°F, 1200°-1540°C) in the propane-fired combustor system.
Hydrated lime utilization increased with increasing flue gas injection point
temperature for two Ca/S ranges, as shown in Figure 7. Utilization values
increased from about 13% to 40% as injection temperature increased from 1650° to
2820°F (900° to 1550°C) for Ca/S ratios ranging from 0.8 to 1.2. Higher Ca/S
19-5
-------
ratios, 1.5 to 2.2, showed the same trend but, the curve was slightly lower at
temperatures above 2000°F (1095°C).
The effect of baghouse temperature on SOo reduction and hydrated lime utilization
is presented in Figure 8. Injection point temperature and air-to-cloth ratio were
nominally constant at 2800°F (1540°C) and 5.8 ft/min (1.7 m/min), respectively.
Both SO2 reduction and hydrated lime utilization increased slightly, -v.5%, as the
baghouse temperature increased from 700° to 900°F (370° to 480°C). Figure 9
illustrates total hydrated lime utilization, and hydrated lime utilization within
the baghouse, as a function of baghouse temperature. Injection point temperature
and air-to-cloth ratio were nominally constant at 2850°F (1570°C) and 5.8 ft/min
(1.7 m/min), respectively. Total hydrated lime utilization increased from 30% to
35% as the baghouse temperature increased from 700° to 900°F (370° to 480°C). For
the same temperature range, utilization within the baghouse increased from 5% to
10%.
Sulfur dioxide reduction and sorbent utilization data for the commercial dolomitic
hydrated lime are presented as a function of injection point temperature in Figure
10. Ca/S ratios ranged from 1.8 to 2.9. Both SO2 reduction and sorbent
utilization show a 50% increase as injection temperature increased from 1800° to
2075°F (980° to 1135°C). Sulfur dioxide reduction also increased significantly as
injection temperature increased from 2075° to 2460°F (1135° to 1350°C), but this
change is more a result of the variable Ca/S ratio than injection temperature.
Sorbent utilization values show essentially no change for injection temperatures
above 2075°F (1135°C). In this case the variability of the Ca/S ratios is masking
small increases in utilization which would be expected. Direct comparison of
results indicate the commercial slaked calcitic hydrate may be as good or a better
candidate for S0^ control even though the commercial dolomitic hydrate was found
to have a higher initial surface area.
Ash Fouling Combustor/High-Temperature Baghouse System
As a result of the initial injection tests on the propane-fired system, subsequent
experiments for the simultaneous S0x/N0x control program have focused on the use
of hydrated lime. Specifically the use of a pressure hydrator 1s being
eval uated.
The hydrated product from the bench-scale batch pressure hydrator developed at
UNDERC was found to be dry, and x-ray diffraction and scanning electron microscopy
showed the product to be 100% hydrated with an average particle size of less than
1.0 um. Figures 11 and 12 illustrate that complete hydration of CaO to Ca(0H)2
occurs in the pressure hydrator.
Figure 13 presents SO2 reduction as a function of flue gas injection point
temperature for a series of 5 tests performed on the ash fouling combustor/high-
temperature baghouse system firing pulverized coal. Baghouse temperature, air-to-
cloth ratio, SO2 levels, and Ca/S ratio were nominally constant at 940°F (505°C),
3.0 ft/min (0.9 m/min), 700 ppm, and 1.0, respectively. The sorbent used was a
calcitic pressure-hydrated lime produced in the bench-scale pressure hydrator.
Two sets of data are presented in Figure 13. The upper curve depicts total SO2
reduction and the lower curve SO2 reduction within the high-temperature baghouse.
Sulfur dioxide reduction in the baghouse never exceeded 7% for the six tests
performed. Low SO2 reduction in the baghouse is consistent with previous
results. Operation of a baghouse at temperatures ranging from 800° to 1000°F
(425° to 540°C) may not increase utilization of calcium-based sorbents enough to
warrant use of a high-temperature baghouse only for SO2 control. But in
19-6
-------
combination with a N0X control technique, the 5% to 10% increase in sorbent
utilization may be significant.
Maximum total SO? reduction was approximately 55% at an injection point
temperature of about 1900°F (1040°C). Lov«r SO? reduction at injection
temperatures below 1900°F (1040°C) was probably a result of insufficient residence
time in the flue gas temperature region where SC^-CaO reaction kinetics are most
favorable. As injection temperature increased from 1900°F (1040°C), SO? reduction
again decreased. This decrease may have been caused by sintering of the sorbent
particles (reduced surface area development) or some Interaction with the fly
ash. Although fly ash interference is possible, injection tests performed at
UNDERC have not shown fly ash interference to be a problem. Figure 14 presents
SO2 reduction as a function of injection temperature for a series of natural gas-
fired injection tests. Although the curve is similar to that in Figure 13, actual
SO2 reduction is slightly lower for tests performed during natural gas-firing.
A single injection test was conducted at conditions similar to those presented in
Figure 13 except the Ca/S ratio was increased from 1.0 to 2.0. The flue gas
injection point temperature was xl900°F (1040°C). Sulfur dioxide reduction and
calcium utilization were about 80% and 40%, respectively. Figure 15 illustrates
SO2 concentration as a function of time for the two-hour injection test. The
peaks observed at 65, 110, and 145 minutes were caused by inconsistencies in the
sorbent feed rate. Frcm 60 to 180 minutes, average S0£ concentration decreased
approximately 100 ppm. This decrease is believed to be partially a function of
SOg reduction within the baghouse and sorbent fallout in the refractory-1 ined flue
ducts of the AF/HTB system.
Two injection tests were performed at Ca/S ratios of 1.0 but at SO2 levels of 1300
and 3000 ppm. Results from these tests, presented in Table 1, showed SO2
reduction remained in the 45 to 50% range for the higher SO2 levels. Injection
tests have not yet been performed at Ca/S ratios exceeding 1.0 for SO2 levels
greater than 1000 ppm.
A single injection test was also performed using a commercial dolonitic hydrated
lime. The specific sample used for this test was collected from a commercial
hydration facility prior to ball milling. Test conditions and SO? reduction data
are presented in Table 1 (Test No. 0484.9). The resulting S0p reduction for this
test was about 40%, as compared to the 55% value obtainecl with the calcitic
pressure hydrated lime produced in the UNDERC bench-scale pressure hydrator.
Without further testing, it is not clear whether this difference is due to the
composition of the dolomitic hydrate (presence of MgO), or to the difference 1n
pressure hydration conditions.
Test results from the AF/HTB system were different from those obtained from the
propane-fired system in that maximum SO2 reduction and hydrated lime utilization
were observed when injecting the hydrated lime into a temperature regime of about
1900°F (1040°C) rather than 2800°F (1540°C). This indicates the overall
time/temperature profile of the AF/HTB system was more favorable at lower flue gas
temperature than those reported for the propane-fired system. Hydrated lime
utilization values of 40% to 55% were observed in both systems for Ca/S ratios
ranging from 0.8 to 2.0.
19-7
-------
Particulate Test Combustor System
A review of the data generated on the AF/HTB system resulted in a decision to
perform subsequent pilot-scale injection tests on the particulate test combustor
(PTC) rather than the AF/HTB system. Reasons for this decision included: 1) the
availability of the system (fewer projects sharing time on same pilot-scale
system), 2) the PTC is better suited for evaluation of additive-enhanced pressure
hydrated lime sorbents, and 3) the high-temperature baghouse showed no significant
advantage with respect to SO^ control alone (the PTC baghouse typically operates
at more conventional temperatures, 200° to 500°F (95° to 260°C)).
The purpose of the initial series of in-furnace injection tests on the PTC system
was to generate morphological (surface area and particle size distribution) and
$C>2 reduction data for calcitic and dolomitic pressure-hydrated lime. Calcitic
pressure-hydrated lime used in this test series was produced in the bench-scale
pressure hydrator. The dolomitic hydrate was obtained from a commercial source.
Injection tests performed to generate morphological data were conducted firing
natural gas to produce an SC^-free flue gas stream. A Beulah lignite was used
during pc-fired injection tests.
Surface area data were not all available for inclusion in this paper. Evaluation
of particle size distribution data (as determined from multicyclone sampling) from
dehydrated sorbent samples collected during injection tests using calcitic and
dolomitic pressure-hydrated lime indicates 90% of the particle mass to be larger
than 2 ym. It is believed this observation is a result of the agglomeration of
submicron particles. Multicyclone samples have been submitted for analysis using
SEM techniques.
Direct comparison of SO2 reduction data from the PTC and AF/HTB systems was made
at similar combustor operating conditions. The most significant differences in
the two systems were baghouse operating temperature and flue gas duct
configuration downstream of the combustor. Typically the PTC baghouse operates at
x300°F (%150°C) compared to 900° to 1000°F (480° to 540°C) for the AF/HTB
system. Flue gas ducting downstream of the PTC combustor was designed to minimize
particulate fallout whereas the AF flue gas ducting has 5 90°-bends resulting in
particulate fallout.
A comparison of SO? reduction data from the two systems as a function of injection
point temperature during natural gas-firing is illustrated in Figure 16. Although
the data from the two systems result in similar curves, SO2 reduction in the PTC
system appears to average about 10% higher for the range of injection temperatures
evaluated, 1300° to 2700°F (700° to 1480°C). Table 2 presents SO2 reduction data
for the PTC and AF/HTB systems generated during pc-fired injection tests. Again,
direct comparison of SO2 reduction data generated at similar test conditions shows
slightly better results were achieved in the PTC system. These data might
therefore indicate the time/ temperature profile of the PTC system is better than
the AF/HT system with respect to S0o reduction. Reduced sorbent fallout may also
have contributed to better SO2 reduction observed in the PTC system.
CONCLUSIONS
Direct furnace injection of calcitic pressure-hydrated lime is capable of 80% SO2
reduction at a Ca/S ratio of 2.0 in the UNDERC pilot-scale combustion systems.
Calcitic pressure hydrated lime produced in a bench-scale pressure hydrator was
found to be a better sorbent for SO2 reduction than the commercial dolomitic
pressure-hydrated lime.
19-8
-------
Operation of a baghouse at temperatures up to 1000°F (540°C) does not increase
utilization of calcium-based sorbents enough to warrant use of a high-temperature
baghouse only for SO2 control. But in combination with a N0X control technique,
the 5% to 10% increase in sorbent utilization may be considered significant.
As expected, residence time and the temperature regime of the sorbent injection
location appear to be the critical parameters controlling SO2 reduction and
sorbent utilization.
REFERENCES
1. H.M. Ness, et al . "Pilot Plant Scrubbing of SO? With Fly Ash Alkali From
North Dakota Lignite." Presented at the Lignite symposium, Grand Forks, ND,
May 18-19, 1977.
2. H.M. Ness and S.J. Selle. "Control of Western Power Plant Sulfur Dioxide
Emissions: Development of the Ash-Alkli FGD Process and Dry Adsorption
Techniques at the Grand Forks Energy Technology Center." DOE Symposium on
Environmental Control Activities, Washington, DC, November 1978.
3. H.M. Ness, et al. "Power Plant Flue Gas Desulfurization for Low-Rank Western
Coals." Presented at the Lignite Symposium, Grand Forks, ND, May 30-31, 1979.
4. W.T. Davis and G.D. Reed. "Reactivity of Fly Ashes in a Spray Dryer FGD
Process." DOE Contract No. AC18-81FC10492, May 1983.
5. H.M. Ness, et al. "Combined Flue Gas Cleanup/Simultaneous S0x/N0x Control
Quarterly Technical Progress Report." U.S. Department of Energy, Grand Forks,
North Dakota:
DOE/GFETC/QTR-79/4, July-September 1979
D0E/GFETC/QTR-80/1, October-December 1979
DOE/GFETC/QTR-80/2, January-March 1980
D0E/GFETC/QTR-80/3, April-June 1980
DOE/GFETC/QTR-80/4, July-September 1980
D0E/GFETC/QTR-81/1, October-December 1980
DOE/GFETC/QTR-81/2, January-March 1981
DOE/GFETC/QTR-81/3-4, April-September? 1981
DOE/GFETC/QTR-82/1, October-December 1981
DOE/GFETC/QTR-82/2, January-March 1982
6. L.J. Muzio, et al. "22 MW Coal-Fired Demonstration of Dry SO? Scrubbing with
Sodium Sorbent Compounds." Presented at the 76th annual meeting of the Air
Pollution Control Association, Atlanta, GA, June 19-24, 1983.
7. H.M. Ness, et al. "Combined Flue Gas Cleanup/Simultaneous S0x/N0x Control
Quarterly Technical Progress Report." U.S. Department of Energy, Grand Forks,
North Dakota:
DOE/GFETC/QTR-82/3, April-June 1982
DOE/GFETC/QTR-82/4, July-September 1982
DOE/GFETC/QTR-83/1, October-December 1982
8. G.M. Blythe. "Dry Limestone Injection Test at a Low-Rank Coal-Fired Power
Plant." DOE Contract No. AC18-80FC10200, November 1982.
19-9
-------
9. Honea, F.I. Studies of Ash Fouling Potential and Deposit Strength in the
GFETC Pilot Plant Test Furnace. Proceedings of the International Conference
on Ash Deposits and Corrosion Due to Impurities in Combustion Gases, Richard
W. Bryers, Ed., July 12-17, 1981.
10. Sears, D.R. and S.J. Miller. Impact of Fly Ash Composition Upon Shaker
Baghouse Efficiency. Paper 84-56.6 presented at the 77th Annual Meeting of
the Air Pollution Control Association, San Francisco, California, June 24-29,
1984.
19-10
-------
Tlme/Tempera1ure Profile
SAMPLE BOX
2250" F
to Slack
SAMPLE
BOX
O
SAMPLE|
BOX
I. D FAN
F.D. FAN
Figure 1. Propane-fired combustor/high-temperature baghouse system.
BAGhOjSE
CuUE"
BAGHOUSE
inleT VALVE
T Taiu'ft aro<
Ce^bus'or 3 3$s
?0CC°- - 1500CF 1 44s
1 i30°F - ^SO^F, 0 7Ls
14t»0°F - 95CrF. C D?s
F L J £ GAS
SAMPLE BO*
BAOHCtSE
ByPaSS va„vE
8C1
HUC'(V-
BAGhOUSE
;j4VP; L 13 O >
// PRIMAPT Alft
' AND COAL
COAL \ .
»EfcDEB_^
Figure 2. Ash fouling combustor/high-temperature baghouse system.
19-11
-------
700 F
31
I
1l
1325°P
Hsot \
Etctianga
F3 Fon
Prehaol
Cool
1 825 F
Cofnbuitof
S«C. Air
Pr-m. Air
L i- -i
r*~ i Cyclone
n
r
j-
¦/ ^ (X}
0 For it
&
Time/Temperature Profile
Gombustor,~3.0s
1B25°F-1325°F, 0.25a
1325°F-700°F. 0.24s
70O°F-300oF. 1.32s
ESP
Boghouse
e=4X3=
r-—»>r—
Puis*
Rtv*1t
Shaker
Air
_J I
-_J 1
J #-
[ J
M 1 -
L=-
Lr-^-- -
300 F
Figure 3. The particulate test combustor, a 500,000 Btu/hr pc-fired furnace.
Pulsing
Control
Shaking
Mechanism
Back Wash
Fan
Vl\
Shaker Bag
Compartment
9 Bags
Bag Support
Case
Inlet
Hoaaers and Knite Valves
Figure 4. Three-node experimental baghouse, a 200 ACFM pilot facility
operating in shaker, low pressure expansion (cage-mounted reverse
air) or pulse mode.
19-12
-------
«—DO-
Air Line
Recorder
Solution
!Storage
Tank
Thermocouple
Digital Indicator
f Rupture
_L Disk
Static
Pressure
Sensor
T.C.
Direct Injection
into Pilot-Scale
Combustion
System
ji Paddle Wheel Stirrer
Ball
Valve
-1X1-
Reactor
Orifice
Filter Bag
Figure 5. Basic components of the bench-scale batch pressure hydrator.
SB-t
10-
502 - 650 ppn
-A Ca/S - 0.8 - 1.2
-o C«/S ~ l.'S - 2.2
Baghousa Tarapara'tura - BS0°F
Rlr-To-Cl o"th R*+ I o - 6.3 f't/nln
1400
1600
1600 2000 2200 2*00 2600
INJECTION TEMPERRTlIRE,DEGREES F
1800
3000
Figure 6. Sulfur dioxide retention versus flue gas injection point temperature.
19-13
-------
60H
70-
60
50i
40"
30"
U
T
I
L
I
Z
P
T
I
0
N 20"
10-
SO2 ~ 650 ppm
Cn/5 - 0.0 - 1.2
Co/5 - J.5 - 2.2
Boghoucs Tanpara^jri
Alr-To-Clo+h Rn+lo ¦
- B80°F
6.3 f-t/mlp
.--ft
1400
Figure 7.
1600 1600 2000 2200 2400 2600
INJECTION TEMPERATURE.DEGREES F
2600
3000
Utilization of slaked calcitic hydrated lime versus flue gas
injection point temperature.
B0-j a 50, REDUCTION
1 ° HYDRATED LIME UTILIZATION
2 Ca/s ~ 0.6 - 1.:
70-i
¦! SO2 ~ 650 ppm
- Injoc+lon Tomporo+ura - 28fJ0°F
60-^ AIr-To-C1oHh Rb+Io ~ 5.8 f"t/roln
P 50-^
E
R
X 20-;
10-
600 700 800 900 1000
SOGhOUSE TEMPERATURE,DEGREES F
Figure 8. Sulfur dioxide reduction and utilization of slaked calcitic
hydrated lime versus baghouse temperature.
19-14
-------
80"
H 70-
Y
C
R
fl 60"
T
E
D
50~
7
A
E 40-
--A TOTAL UTILIZATION
-o UTILIZATION UIThrN ThE BPGrtGUSE
Ca/S - 0.8 - J.3
SO? - 6S0 ppi»
!njoc+lon Tamper e"tur« -¦ 2853°F
P!i—To-Clo+h Ro't! o -¦ S. 8 f-t/m'n
! 30-
P 20~
T -
f "
N 10*^
i :
^ •
600
700 800 900
BPGHOUSE temperpture,DEGREES F
J 000
Figure 9. Utilization of slaked calcitic hydrated lime versus baghouse
temperature.
80-^
i
701
60-1
50-
¦i
4
-1
40 H
P
E
R
C
E
N
T 30-
a S02 REDUCTION
° mydrpted LIME UTILIZATION
SO? ~ 6S0 PPJ"
Cb/S - 1.8 -2.8
Baghousa Tomporo"tur« 900°F
Plr-To-C1o+h Re+lo 6.3
20"
10-
B~W-
1600
1800 2000 2200 2400 2600
INJECTION TEMPERATURE-DEGREES F
2800
¦ • • r
3000
Figure 10. Sulfur dioxide reduction and dolomitic hydrated lime utilization
versus flue gas injection point temperature.
-------
5.0
4.0
3.0
Figure 11. X-ray diffraction scan of quicklime feed to pressure hydrator.
3.0
2.4
1.8
1.2
0.8
^
47
50
65
20
38
29
— 100
80
60
Ca'OHU
40
£
4-733
20
: . 1 r
1 1 1— 1
I' — I 1 » ¦ r— t
20 29 38 47 56 65
°20
Figure 12. X-ray diffraction scan of hydrated product from pressure hydrator.
19-16
-------
90-:
s
-
0
80~:
2
-
70^
R
E
p
D
U
50~
C
-*
T
40^
T
-
c
30-:
N
20^
10~
%
-i
0n
Co/S ~ 1.0
SO^ - 700 PPf11
Baghouse Temp. - 940°F
P/C Ret I o - 3.0 f-t/mln
7\—i
11 1 1 1 urn nn
rrrr ttttttti
& Total 502 Reduction
0 SOj Reduction In the Baghouse
J 300 1500 1700 1900 2100 2300 2500 2700 2900
INJECTION TEMPERATURE, DEGREES F
Figure 13. Sulfur dioxide reduction versus injection temperature for the ash
fouling combustor/high-temperature baghouse system.
100d
90-:
5
-
0
80-:
2
-
70~:
R
-
E
60-:
D
-
U
50*:
C
-
T
s
40-^
I
-
0
30T
N
-
20~
10-:
-A
0"!,.,, , (11 M T j , M „ M ,l| t! M I lit l| M 1 'I I I' 1 |« 11 n I Itlftl " 1 I M I | * M ' T
I300 1500 1700 J 300 2100 2300 2500 2700 2900
INJECTION TEMPERPTURE (DEGREES r~:
Figure 14. Sulfur dioxide reduction versus injection temperature for natural
gas-fired injection tests.
19-17
-------
1000
900
800
a 700
E
a
a
Ca/S - 2.0
Inject. Temp. ~ 1900 °F
Baghouse Temp. - 940 °F
A/C Ratio 3.0 ft/min
\
2
O 600
'V
1
z
HI
o
z
o
o
Cd
o
CO
500
400
300
200
100
0
/
Vi
y\
/ \
K
J
"X v
» » ' » » « 1 * » * » ' ' 1 ' ' « * 1 1 ' ' 1 » ¦ ' ' » »
r
N
v/
¦ I I I I I I 1 I I
60
120
TIME (Min.)
180
240
Figure 15. Sulfur dioxide concentration versus time for a calcitic pressure
hydrated lime injection test on the pc-fired combustor/high-
temperature baghouse system.
100-
90-:
5
0
80^
2
¦
70-
R
-
E
60-1
D
-
U
50-
C
-
T
40-
I
-
0
30-
N
-
20-
10-
Z
0-
Ca/S-1.0
S02~7°0pPm
A/C Ratio - 3.0 ft/min
Calcitic Pressure Hydrated Lime
1825/48
O PTC
a AF/HTB
2300/37
335/ 32
2700/32
1300 1500 1700 1900 2100 2300 2500 2700
INJECTION TEMPERATURE (DEGREES F1
2900
Figure 16. Sulfur dioxide reduction versus injection temperature for natural
gas-fired injection tests.
19-18
-------
TABLE 1
PRESSURE HYDRATED LIME S02 REDUCTION DATA3
Test.
No.
Total S02
Reduction, %
so2
Cone, ppm
Ca/S
Sorbent
Utilization, %
0584.1
55
725
1.0
55
0584.5
• 45
1320
1.0
45
0584.7
48
3000
1.0
48
0484.9
40
740
1.0
40
aInjection temperature, baghouse temperature, and air-to-cloth ratio were
nominally 1900°F, 940°F, and 3.0 ft/min, respectively.
bThe sorbent used in tests 0584.1, 0584.5, and 0584.7 was a calcitic pressure
hydrated lime (75% CaO) produced in the bench-scale pressure hydrator. The
sorbent used in test 0484.9 was a commercial dolomitic pressure hydrate
(40% CaO and 30% MgO).
TABLE 2
S02 REDUCTION DATA3
Test
Pilot Scale
so2
Total S02
Sorbent
No.
System
Cone, ppm
Reduction, %
Ca/S
Util ization, %
0584.1
AF/HTB
72 5
55
1.0
55
1184.1
PTC
920
45
0.7
64
0484.8
AF/HTB
700
80
2.0
40
1284.1
PTC
1000
74
1.7
44
0484.9
AF/HTB
740
40
1.0
40
1084.1
PTC
870
40
0.8
50
aInjection temperature and air-to-cloth ratio were nominally 1900°F and 3.0
ft/min, respectively.
^The sorbent used in tests 0584.1, 1184.1, 0484.8, and 1284.1 was a calcitic
pressure hydrated lime (75% CaO) produced in the bench-scale pressure
hydrator. The sorbent used in tests 0484.9 and 1084.1 was a commercial
dolomitic pressure hydrate (40% CaO and 30% MgO).
19-19
-------
BENCH SCALE PROCESS EVALUATION OF IN-FURNACE NOx AND SOx REDUCTION
BY REBURNING AND SORBENT INJECTION
S. B. Greene, S. L. Chen, D. W. Pershing,
M. P. Heap, and W. R. Seeker
Energy and Environmental Research Corporation
18 Mason
Irvine, CA 92718-2707
ABSTRACT
Reburning involves the injection of a secondary fuel above the main firing zone
of a combustion to produce a reducing zone which acts to reduce N0X to molecular
nitrogen. Overfire air is added above the reburn reducing zone to complete the
combustion. The reburning process has been combined with the injection of
calcium-based sorbents (e.g., limestone) to investigate the potential for com-
bined N0X and S0X reduction. Bench scale evaluations of the process carried out
in a plug flow furnace at 23 kW^ have indicated that N0X reductions of up to 70
percent and sulfur captures of up to 50 percent (at Ca/S = 2) can be achieved
depending on a number of process variables. The dominant variables include the
Initial N0X level that is to be reduced, the reburning fuel type (pulverized coal
or natural gas), and the residence time and temperature in the reducing zone.
For sulfur control, the dominant parameters are the amount of sorbent added, the
sorbent type, and the injection temperature.
INTRODUCTION
This paper addresses the reburning technology which removes NO from combustion
products using fuel as the reducing agent. It has been found to involve similar
kinetic processes to those involved in combustion modification by staged com-
bustion. This technology is variously referred to as:
• In-furnace N0X reduction
• Reburning
• Staged fuel injection
• Mitsubishi Advanced Combustion Technology (MACT)
Reburning can be considered as the process which allows in-furnace N0X reduction
to take place.
The concept of NO reduction by flames has been known for over a decade. A flue
gas N0X incinerator was developed by the John Zinc Company (1), and Wendt, Stern-
ling, and Matovich (2) found that NO could be reduced in laboratory flat flames
by injecting methane into the combustion products. Recently, Japanese investi-
gators have reported the application of reburning to large test furnaces (3).
The MACT (Mitsubishi Advanced Combustion Technology) in-furnace N0X removal pro-
cess applies the concept of reburning to a boiler. Part of the fuel bypasses the
main heat release zone and is injected above the main burners to provide the fuel
20-1
-------
for reburning. It is claimed that the N0X produced by the main firing system can
be reduced "to half at any level of concentration" (3). Hitachi Shipbuilding and
Engineering has a U.S. patent on multistage fuel injection for N0X control (4).
EER under contract to the U.S. Environmental Protection Agency has been investi-
gating the reburning process as it might be applied to U.S.-designed pulverized
coal utility boilers for the last few years. The initial activity has involved
bench scale testing of the impact of process variables on the N0X removal effi-
ciency. The studies have been performed in tower furnaces that allow control of
process parameters over the range of interest in utility boiler furnaces. The
reburning process can be divided into three zones:
• Primary Zone: This main heat release zone accounts for approximately
80 percent of the total heat input to the system. The zone is operated
under overall fuel-lean conditions, although the burners might be low-
N0X distributed mixing burners. The level of N0X exiting from this
zone is the level to be reduced in the reburning process.
• Reburning Zone: The reburning fuel (normally about 20 percent of the
total fuel requirements) is injected downstream of the primary zone to
create a fuel-rich reduction zone. The reactive nitrogen entering this
zone comes from two sources: The primary NO level and the fuel nitro-
gen in the reburning fuel. These fuel nitrogen species apparently
react with the hydrocarbon fragments from the reburning fuel to produce
intermediate species such as NH3 and HCN while some is converted to N£
and some is retained as NO. The products of this reduction zone are
the reactive nitrogen species such as: NO, char nitrogen, NH3, and
HCN, which will be referred to as total fixed nitrogen (TFN). In order
to optimize the NO reduction by reburning, it is necessary to minimize
the total reactive nitrogen exiting the reburning zone.
• Burnout Zone: In the burnout zone, air is added to produce overall
lean conditions which oxidizes all the remaining fuel and converts the
total reactive nitrogen either to NO or N£.
The reburning process can be combined with the injection of calcium containing
sorbents such as limestone, to achieve simultaneous N0X and S0X control. Calcium
oxides, which are formed when calcium carbonate decomposes, can react with gas-
eous SO2 to form calcium sulfate (CaS04)* "^e calcium sulfate particles can then
be removed with the ash particles using the normal particulate removal systems.
The sorbent injection process is very compatible with reburning since the addi-
tion of the reburning fuel and overfire air provides excellent media for trans-
porting the sorbent in the upper zone of the furnace. There are similar require-
ments to disperse the sorbent and the reburning fuel and burnout air.
The processes that take place in each of these zones have been recently evaluated
in terms of the reduction of total reactive nitrogen and gaseous sulfur (5).
This paper will summarize the major findings of that study. Based on these
results, a process model is currently under development which will allow an
assessment to be made of the effectiveness of applying reburning/sorbent tech-
nology for different fuel types and applications.
EXPERIMENTAL
The process studies were carried out in the refractory lined Control Temperature
Tower (CTT) which is shown schematically in Figure 1. The CTT has a total firing
rate of between 18 and 24 kW (60,000-80,000 Btu/hr) in the main combustion
chamber. The main combustion chamber is 20.3 cm in diameter and includes a long
quarl entry to promote flame stabilization and to provide for one-dimensional
on ?
-------
plug flow. The time/temperature profile along the furnace could be manipulated
by using back-fired heating sections. The back-fired sections consist of natural
gas burners fired into refractory channels in the direction opposite to the main
chamber. The high-temperature gases pass through the channels surrounding the
main chamber (see the radial cross-sectional view in Figure 1) and minimize the
temperature decay along the furnace. A more rapid temperature decline can be
achieved by leaving the back-fired channels off or by inserting cooling coils
around the main chamber. The tower is equipped with numerous ports located along
the axis of the reactor that allow the installation of zone separation chokes,
fuel and air injectors, cooling coils, and sampling probes.
The CTT was configured into three zones: (1) the primary zone was formed using a
premixed burner firing pulverized coal or propane doped with various levels of
H2S and NO, under lean conditions (typically 10 percent excess air); (2) the
reburning zone formed by injecting the reburning fuel (either coal or doped gas)
at various flow rates to control the reburning zone stoichiometry and; (3) the
burnout zone in which air was injected to bring the overall stoichiometry to
typically 25 percent excess air. The parameters in each of these zones were
examined separately in terms of how they influenced the exhaust level of N0X.
The test series were performed by establishing the level of N0X from the primary
and then increasing the amount of reburning fuel addition and burnout air cor-
respondingly to decrease the reburning zone stoichiometry and maintain the
overall burnout zone stoichiometry. In this manner, the residence time and
temperature in the reburning zone were maintained relatively constant while the
reburning zone stoichiometry was varied. Sorbent injection was carried out for a
variety of different sorbents injected with the reburning fuel and with the
burnout air.
Both flue gas and in-combustor measurements were made of N0X, O2, CO/CO2, SO2,
HCN, and NH3 by techniques presented in Table 1. In addition, the gas tempera-
ture was measured throughout the reactor by using a suction pyrometer (Type B
thermocouple). Details of sampling and analysis procedures and test conditions
are available elsewhere (Greene et al., 1985).
The bench scale testing has provided fundamental insight into the chemical pro-
cesses that control N0X reduction and sulfur capture and the impact of the key
process variables. The full range of parameters in each zone was investigated:
Primary Zone
- Stoi chi ometry
- Fuel type
- NO level
- SO^ level
Reburning Zone
- Stoichiometry
- Mixing rate of reburning fuel
- Reburning fuel type (propane,
- Nitrogen content of reburning
- Temperature
- Residence time
- Transport media for reburning
Burnout Zone
- Temperature
- Excess air
- Air mixing rate
hydrogen,
fuel
CO, and coals)
fuel (air or inert)
in
-------
For the SO2 capture studies, the additional variables included:
• Sorbent type
• Injection location
• S0X concentration
• Temperature profile
• Additives
RESULTS—N0X
Although most of the parameters investigated had some effect on the reduction
level achieved by the reburning process, the dominant parameters were found to be
those associated with the reburning zone condition and the primary zone NO level.
Figure 2 shows the effect of three parameters: reburning fuel type (propane or
Utah bituminous coal), primary zone N0X level [(N0X)p], and reburning zone
stoichiometry (SR2) • These data were taken at the baseline conditions shown in
Table 2 and are expected to be typical of the times and temperatures that would
exist for applications to pulverized coal (p.c.) fired boilers. The optimum N0X
reduction occurred when the reburning zone was fuel-rich at an overall stoich-
iometry of 0.9. For high levels of primary N0X, 630 ppm, the exhaust level of NO
was reduced to 200 and 250 ppm for propane and coal as reburning fuels, respect-
ively. For low levels of primary NO the reduction levels were not as signifi-
cant, dropping from 190 ppm to 100 ppm for propane reburning and only to 180 ppm
for coal. The effectiveness of propane over p.c. as a reburning fuel can be
attributed to the fuel nitrogen difference. As propane was doped with ammonia to
the same nitrogen content of the coal, similar N0X levels were produced.
A wide variety of reburning fuels were investigated including hydrocarbon and
nonhydrocarbon gaseous fuels and coals of varying rank and nitrogen content. The
properties of these fuels are supplied in Table 3. A comparison of the effect-
iveness of different reburning fuels, drawn from Figure 3, indicates that most
fuels are similar. The nonhydrocarbon fuels are generally less effective than
those containing hydrocarbons (particularly at longer reburning zone residence
times). The Yallourn brown coal was the most effective reburning coal chiefly
due to its low fuel nitrogen content and high volatility, and there was a general
decrease in effectiveness with fuel nitrogen. Some problems were encountered
under fuel-rich conditions to complete the burnout of lower volatile coals such
as anthracite and the low-volatile Rosa coal; however, similar reduction levels
were achieved under fuel-lean conditions where burnout was adequate.
The predominant effect of fuel type was found to be the nitrogen content of the
reburning fuel. The detrimental effect of the nitrogen becomes more apparent at
the lower levels of primary NO. As shown in Figure 4 at high primary NO levels,
the level of reduction achievable by all reburning fuels tested was similar at
the optimum stoichiometry and was in the range of 60-70 percent reduction. It is
more difficult to achieve the same reduction at lower levels of N0X with any fuel
type; however, fuels containing fuel nitrogen exacerbate the limitation. Below
an initial level of 200 ppm of NO, gaseous fuels containing no fuel nitrogen are
required to achieve an overall reduction by reburning.
These studies have indicated that N0X reduction by reburning is a kinetically
controlled process with features similar to the staged combustion processes that
have been extensively investigated. In the rich reburning zone the temperature,
reaction time, and reactant concentration, all influence the ultimate reduction
of NO that can be achieved. The data presented above was at the baseline reburn-
ing conditions of T^ (at the entrance of the reburning zone) of 1700 K (2600°F)
and a total reburning zone residence time of 400 msec. The effectiveness of the
process is increased at longer residence times as shown in Figure 5 in the range
-------
of 140 to 750 msec in the rich zones for all hydrocarbon fuels; nonhydrocarbon
fuels (Hg and CO) had no residence time effects. Detailed species analysis
within the reburning zone has indicated that the mechanisms suggested by Glass
and Wendt (6) for the rich postflame decay of nitrogenous species are consistent
with these results. Although the effectiveness of all coals tested increased
with residence time, the magnitude of changes were coal dependent. The bitum-
inous coals demonstrated the largest effect of time similar to gaseous hydro-
carbon, while the lower ranked lignite and brown coals were less influenced by
residence time.
The influence of the reburning zone temperature was also dependent on the reburn-
ing fuel type (Figure 6). As the entrance temperature to the reburning zone was
increased from 1700 K to 1833 K (2600°F to 2840°F) the exhaust N0X levels
decreased for all reburning fuels. The largest effect occurred for gaseous
fuels, while the impact with pulverized coal was less dramatic. These data
suggest placing the reburning jets as close to the main burner zone as is feas-
ible to increase the temperature, and having as large a reburning zone as
possible by separating the air injectors away from the reburning fuel jets.
RESULTS—S0X STUDIES
The primary objective of the sorbent injection studies was to determine the
sulfur capture that could be achieved for conditions that were optimal for N0X
control by reburning. As indicated in the previous section, these conditions
were a reburning zone stoichiometry of 0.9 and high temperature and long resi-
dence times in the reburning zone. The baseline conditions for the sorbent
studies are also provided in Table 2. The baseline sorbent chosen was a high-
purity calcitic limestone with a trade name of Vicron 45-3. This sorbent has
been investigated extensively in other EPA programs.
In Table 4, a list of the sorbents investigated in this study and their prop-
erties is presented. The sorbents, all calcium containing material s, include
limestones, dolomitic limestones, and hydrates of limestones. The primary
differences between these sorbents are the presence of magnesium, whether or not
they are hydrates and particle size.
The sorbents were injected either with the reburning fuel or at the same location
as the burnout air. In Figure 7 is shown the sulfur capture when the Vicron
sorbent was injected in these locations as a function of stoichiometry in the
reburning zone. Capture is relatively insensitive to stoichiometry and the
differences that do occur can be attributed to slight differences in the furnace
temperature profiles (Figure 7b). Significantly better captures were observed
for injection in the burnout zones. The reburning zone is significantly hotter
than the burnout zone, resulting in less reactive stones (7). Specific surface
area measurements were made for sorbents injected into the furnace without SO2
present. These data indicate a direct (inverse) dependence of specific
surface area, and therefore reactivity, with injection and calcination
temperature (Figure 8).
The other sorbents have a similar dependence on injection location (Figure 9). In
every case, injection with the burnout air was preferred with capture increasing
with calcium/sulfur ratio. Dolomite was the most reactive sorbent tested in
these experiments. However, more dolomite by weight must be added to achieve the
same Ca/S ratio due to the Mg component. The capture difference between sorbents
and injection temperatures was found to be directly related to the specific sur-
face area after calcination (see Figure 8). The higher surface areas of dolomite
are likely due to presence of the magnesium which prevents the calcium oxide par-
ticles from recrystal1izing into closed forms with low surface area. The causes
?n-5
-------
for the higher surface area of hydroxides are uncertain but may be due to a tran-
sition crystal state.
The results presented above are for gas reburning and sorbent injection.
When coal was used as the reburning fuel, the results were similar but somewhat
increased over the gas results (Figure 10). These improved capture levels cannot
be attributed to changes in furnace temperature profiles since only slight
changes were measured. The discrepancy is likely due to mineral matter inter-
action which can enhance sorbent reactivity (8).
SUMMARY AND CONCLUSIONS
The process chemistry of the reburning/sorbent injection technology has been
examined in some detail. N0X reduction by reburning has been found to be similar
to staged combustion processes. The influence of individual parameters for each
zone of the process on the effectiveness of the N0X control achieved has been
examined. Table 5 presents a summary of the determined influences of these pro-
cess variables. The impact refers to the direct effect on the chemistry of the
process. Under conditions of actual applications, there were a number of sec-
ondary influences which will influence the actual level of N0X achieved. For
example, to maintain total load, the firing rate to the main combustion zone will
be lowered by an amount corresponding to the reburning fuel addition. This
reduction in load can have a variety of effects (both positive and negative) on
the NO level exiting this zone and therefore the exhaust emission of N0X.
When combined with sorbent injection, the reburning process has the potential for
simultaneous N0x/S0x control. Sulfur capture by calcium containing materials is
very sorbent- and temperature-dependent. The important process variables are
also provided in Table 5. The highest sulfur captures were achieved with dolo-
mitic sorbents injected at lower temperatures; e.g., with burnout air. The
dominant physical parameter of the sorbent is the specific surface area after
calcination. Specific surface area appeared to directly correlate with capture
for the sorbents and injection locations investigated in this study. However, it
is uncertain what determines the specific surface area of different sorbents.
These studies have concentrated on the chemistry of the reburning/sorbent injec-
tion process under ideal conditions; i.e., rapid mixing and distinct zones.
Activity is now underway to investigate the impact of scale and finite rate
mixing. These tests are being carried out at a firing rate of 3 MW (10 x 10^
Btu/hr).
ACKNOWLEDGEMENTS
The authors wish to express their appreciation to the U.S. Environmental Protec-
tion Agency which supported this work under Contract 68-02-3925. In particular,
the project officers R. E. Hall and D. B. Henschel, along with the Reburning
Program advisory panel, contributed significantly to the program direction. We
would also like to thank our colleague Brian Jacobs for his technical assistance.
REFERENCES
1. Reed, R. D. Process for the Disposal of Nitrogen Oxide. John Zinc Company,
U.S. Patent 1274637, 1969.
2. Wendt, J. 0. L., C. V. Sternling, and M. A. Matovich. Fourteenth Symposium
(International) on Combustion, The Combustion Institute, 1973, p. 897.
?n-fi
-------
3. Takahashi, Y., et al . Development of Mitsubishi "MACT" In-Furnace N0X
Removal Process. Presented at U.S.-Japan N0X Information Exchange, Tokyo,
Japan, May 25-30, 1981.
4. U.S. Patent 4,395,223. "Multistage Combustion Method for Inhibiting
Formation of Nitrogen Oxides." Okigarni, N¦, et al., 1983.
5. Greene, S. B., S- L. Chen, W. D. Clark, M. P. Heap, D. W. Pershing, W. R.
Seeker. "Bench-Scale Process Evaluation of Reburning and Sorbent Injection
for In-Furnace N0x/S0x Reduction." EPA IERL-RTP-1698, January 1985.
6. Glass, J. W. and J. 0. L. Wendt. "Mechanisms Governing the Destruction of
Nitrogeneous Species During the Fuel Rich Combustion of Pulverized Coal."
Nineteenth Symposium (International) on Combustion, The Combustion Institute,
1982, p. 1243.
7. Cole, J. A., J. C. Kramlich, G. S. Samuelsen, W. R. Seeker, and G. D. Silcox.
"Reactivity of Calcium-Based Sorbents for SO2 Control." First EPA/EPRI
Symposium on Dry SO2 Control, San Diego, 1984. Also EPA Final Report
(Contract 68-02-3633), in preparation.
8. Overmoe, B. J., S. L. Chen, M. P. Heap, D. W. Pershing, and W. R. Seeker.
"Boiler Simulator Studies of Limestone Injection. "Simulator Studies of
Limestone Injection." First EPA/EPRI Symposium on Dry SO2 Control, San
Diego, 1984. Also EPA Final Report (Contract 68-02-3633), 1984.
?n-7
-------
Observation
Port
Removdbl e_
Choke
Back-Fired
Heating
Channel
Removable
Cooling
Co i 1
Back-Fi reel
Heati ng
Channel
Removabl e
Cooli ng •
Coil
b 4
b 4
F'ue Gas
Sa-npl i ng
Location
3ack-Fi red
Heatino Channel
Reburni ng
Zone
Tz Stack
Figure 1. Cross-sectional views of the control
temperature tower.
(N0x)3 = 190
(NO j = 630
* r
>>
k.
•o
CL
X
o
Z
600
Coal
Propane
215
Coal
129
Coal
rcpane
o.:
0.8
l_
:.s
1.1 0.7 0.8
:.s
i.i
36.4 27.3 10.2 9.1 0
Heat, indue a<; rphurnirig fuel ^Dercentl
SR.
Figure 2. Influence of process parameters on
reburning effectiveness.
20-8
-------
(NC
x P
190
~i r
-i r
700
6C0
a
¦o 4 OC
\ 300
20C
100
• C3H8
O Utah
~ Beulah
0 Colstrip
b Yallourn
Ci Rosa
A 42
CO
J L
J L_
0.7 0.8 0.9 1.0 1.1
SR,
(NO )„ = 630
x p
Figure 3. Comparison of different reburning fuels.
u:
"i 22
4->
S 1QD
u
c
c.
^ 30
©
^ 6C
?.
23
SP>2 = 0.9
Indiana Coal
100 2C0 300 400 500 60C 700 800 900 1000 1100 1200
(N0x)p, ppm (dry, 0S 0?)
Figure 4. Impact of primary N0X on effectiveness of
reburning,
20-9
-------
C3H8 C°al
630
63C
500
140 ITS
140 ms
« 4Q0 -
300
400 ms
400 n5
200
750 ms
100
750 ms
750 ms
1.0
0.7
0.8
1.0
0,8 0.9
1.0
0.7
" e
0.9
0.7
Figure 5. Influence of residence time.
C3Hg C3Hg+NH3 'Jtah Coal
(NO
630
700
- Open Symbols: T.^1700
Solid Symbols: ='833
600
o
O
u
•o
, 300
X
z
200
100
0.7
0.8
0.9
1.0 0.7
O.B
1.0 0.7
O.B
0.9
1 .0
sr2 sr2 sr2
Figure 6. Influence of reburning zone temperature.
20-10
-------
REBURNING WITH C3H8
50
Ca/S - 2
(S02)p = 1940 PPM
£
z>
H
• VICRON WITH AIR
Q.
<
O
A VICRON WITH FUEL
30
l-
z
UJ
o
DC
UJ
Q.
20
10
0.8
0.9
1.0
0.7
1.1
sr2
1 " 1
- i - i ¦ —i i r
"1 I 1 " 1
2600
-
2400
-V
-
2200
- \rk.
SR2 = 0.7
-
2000
>
-
1800
| 1J^
11
cm J
5
1600
1400
REBURN-
ING
" ZONE
I 0.7\
| BURNOUT ZONE
-
0
« i
_l. 1 _ L... _
K
1700
1477
1255
1033
0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.6 1.8 2.0 2.2 2.4
RESIDENCE TIME, sec
(b)
Figure 7. Sorbent injection—impact of injection location
and furnace characteristics on sulfur capture
for reburning conditions.
20-11
-------
o>
tM
<
UJ
tr
<
UJ
o
<
u.
GC
z>
co
o
u.
o
UJ
a.
CO
30 h
25
20
15
10
5
1366
T
1588
1811 K
(78)
^DOLOMITE
(84) , LIMESTONE
(VICRON) "V^(82)
5.(92)
HYDRATED
LIME
Ca(OH)2
_L
_L
X
1800 2000 2200 2400 2600 2800
PEAK CALCINATION TEMPERATURE, 0 F
Figure 8. Sorbent injectiori--sorbent type. Impact of
injection temperature on specific surface area without
sulfur present for three sorbents. Numbers in paren-
thesis are extents of calcination after approximately
0.5 sec.
VICRON
DOLOMITE HYDRATED LIME
IT
u
H
Q.
<
O
O
DC
LU
a.
I 1
1 1
1
I
50
... .
• WITH A
BURNOUT/
AIR /
40
— "
//
—
30
WITH
+/~
BURNOUT
/
/WITH
y<5
20
AIR
pREBURN
A
/ -
/-
FUEL
/
10
/www ¦
/o
-•reburn
0
, FUFL
i 1
,
1
1
Ca/S MOLAR RATIO
Figure 9. Impact of injection location, sorbent
type, and Ca/S ratio on sulfur capture under reburn-
ing conditions (C3HQ reburning, SR2 = 0.9).
20-12
-------
Vlcron
Dolomi te
1 1
i 1
-
/ °
: /
V
y
' i
i i
SR2 « 0.9
O c3h8/c3hb
# Coal/Coal
Injection With
Reburnlng Fuel
Ca/S
Figure 10. Comparison of coal versus gas for sulfur
capture when sorbent is injected with the reburning
fuel.
20-13
-------
Table 1
CONTINUOUS GAS ANALYSIS INSTRUMENTS
Gas Measured
Detection Principle
Manufacturer
Model No.
Range
M0, N02
Cnemi1uminescent
Tnermo Electron Corp. (TEC0)
10 AR
0-1000 ppm
CO
Nondispersive Infrared
(NDIR)
Anarad, Inc.
AR500R
0-500 ppm
C02
Nondispersive Infrared
(NDIR)
Anarad, Inc.
AR500R
0-25%
°2
Paramagnetic
Taylor Servomex
QA-272
0-10S
S02
Nondispersive
Ultraviolet (NDUV)
DuPont
400
0-5000 ppm
Hxuy
Flame Ionization
Becicman Instruments, Inc.
402
500 ppm (C3Hg)
(Healed)
Table 2
BASELINE OPERATING CONDITIONS
Primary Zone
• Propane fired at 17.6 kW (60 x 10^ Btu/hr)
• SRi = 1.10
• (N0x)p » 190 ppm (dry, 0% O2)
• (SOgjp = 1940 ppm
Reburning Zone
• Tz - 400 ms
t T1 (reburning fuel injection temperature)
= 1700 K (2600°F) lower auxiliary burners off
Burnout Zone
• SR3 = 1.25
• T2 (burnout air injection temperature) = 1505 K
= (2250°F) lower auxiliary burners off
• Sorbent - Vicron 45-3; Ca/S = 2
Refractory chokes were placed at the reburning fuel
and burnout air injection locations (see Figure 1).
The chokes separated the zones by preventing any
backmixing between the zones. Tne primary fuel (pro-
pane) was doped with NO to the desired concentra-
tions .
20-14
-------
Table 3
FUEL ANALYSIS
Fuel
Yal1ourn,
Beulab, ND
Colstrip, M7
Indiana
Pennsylvania
Rosa,
Ulan
Australla
AL
Property
Rank
Brown
Lig A
Sub B
HVB
Antnracite
MV
HVB
Proximate Analysis
[Percent, as
recei ved)
Hoi sture
13.97
33.10
21.27
4.54
5.13
8.02
6.39
Ash
1.26
7.12
9.58
8.96
5.74
6.79
7.40
Volatile Matter
45.20
28.65
30.82
37.73
4.39
21.81
38.89
Fixed Carbon
39.57
31.13
38.33
48.77
84.74
63.38
47.32
Calorific Value
MJ/kg
21.9
16.9
21.3
28.4
30.5
31.2
28.7
Sulfur
0.20
0.76
0.50
1.87
0.44
0.96
0.64
Ultimate Analysis
(Percent, dry)
C
66.03
65.29
67.52
71.17
88.45
81.23
73.17
H
4.55
3.96
4.36
4.75
2.14
4.73
5.55
N
0.55
0.99
1.38
1.44
0.79
1.74
1.54
S
0.23
1.14
0.63
1.96
0.47
1.04
0.66
Asn
1.46
10.64
12.17
9.39
6.05
7.38
7.90
0
27.IB
17.98
13.94
11.29
2.10
3.88
11.18
20-15
-------
Table 4
SORBENT PROPERTIES
Sorbent
Vicron 45-3
Dol omi te
Hydrated Lime
Petrographic Char-
acterization:
Type
Calcite
Dolomite
Hydrated Lime
Composition
CaCOo
CaC03+MgC03
Ca{OH)2
i Grain Size,
500-700
<100
30
Top Size, Mm
40-45
Mean Size, mn
11-12
34-40
11-12
Bottom Size,
0.3-0.8
1.0
0.3
Density, g/cm^
2.757
2.855
2-350
Oil Absorption
14
l Uncalcined
' S. A., m2/g
0.6-0.9
0.6
12.1-13.5
Ignition Weight
Loss, % (1000°C)
41.0-44.0
44.2-47.1
25.3
Hydration Weight
Gain (1000°C)
33.08
67.5
Chemical Composi-
tion, %
Ca
39.2
21-0
51.0
Al
0.07
0.09
0.26
Si
0.07
0.37
0.80
Fe
0.037
0.102
0.22
Na
0.022
0.033
Mg
0.485
12.1
K
0.011
0.039
0.51
Cr
0.006
0.27
20-16
-------
Table 5
INFLUENCE OF PROCESS VARIABLES ON REBURNING EFFECTIVENESS
Parameter
Primary Zone
Stoichi ometry
Fuel Type
NO Level
Reburning Zone
Stoichiometry
Mixing Rate of Reburning
Fuel
Fuel Type
Temperature
Residence Time
Transport Media
Burnout Zone
Excess Air
Air Mixing Rate
Temperature
Sorbent Injection
Sorbent Type
Injection Location
Sorbent Rate
Temperature Profile
Impact
• No effect except will require more re-
burning fuel for burner operation.
• No direct effect. Can influence
through temperature and NO level
entering reburning zone.
• Strong effect. More difficult to
reduce lower levels of primary NO.
• Optimum at overall stoicniometry
of 0.9.
• Faster mixing preferred.
• Hydrocarbon fuels more effective;
fuel nitrogen content detrimental
at lower primary NO level.
• Reduction increases with increasing
temperature (2400-2900°F).
• Strong impact, increasing with time
(100-750 msec).
¦ Inert (oxygen free) transport meaia are
desirable since less reburning
fuel is required to attain opti-
mum stoichiometry.
• Not important except for burnout.
• Not Important.
• Not important unless temperature
1s dropped to 1200 K where selective
NHj-N0 reactions can take place to
enhance reduction.
• Dolomite ~Hydroxide ~Limestone.
• Lower reactivity at higher temperatures.
• Capture determined by Ca/S ratio.
• Downstream temperature profile influ-
enced capture.
-------
SESSION IV: BURNERS FOR SIMULTANEOUS S02/N0X CONTROL
Chairman, G. Blair Martin, EPA, 1ERL/RTP
20-18
-------
EVALUATION OF LOW-NOx BURNERS FOR S02 CONTROL
R. Payne and A. R. Abele
Energy and Environmental Research Corporation
18 Mason
Irvine, California 92718-2706
ABSTRACT
Limestone injection with multistage burners, termed "LIMB," was conceived as a
process in which sulfur capture by sorbent material injected through burner
passages may be combined with conditions which produce low N0X emissions in
staged, pulverized coal flames. The potential for simultaneous N0x/S0x control
by low-NOx distributed mixing burners was evaluated in pilot scale research
furnaces. Experimental and conuiercial burner designs, ranging in capacity from
10 x 10^ to 100 x 10^ Btu/hr, were considered. Sulfur capture by injected
sorbents was relatively insensitive to burner design. Sulfur capture with
limestone was generally in the range of 35-40 percent at Ca/S = 2, and up to
40-55 percent for hydrated lime. The key factors in SO2 removal were reactivity
of the sorbent and the temperature history to which the sorbent particles were
exposed.
INTRODUCTION
The emphasis of U.S. energy industry has been on the expanded use of coal in
utility and industrial applications. Because of the characteristics of coal and
its combustion, expanded coal use may result in the increase of pollutant
emissions, including N0X and S0X. The reduction of N0X, S0X, and particulate
emissions from fossi1-fuel-f1red boilers has been a major objective for the U.S.
EPA and all of the major boiler burner manufacturers for several years. This is
evidenced by a number of unrelated concurrent efforts that have been and are
being conducted to develop low-N0x burners. Various EPA programs have
demonstrated the principle of staged combustion as a means of controlling N0X
emissions from coal-fired combustion systems, and have defined design guidelines
for the Distributed Mixing Burner (DMB) concept as a viable control technology
for new and retrofit applications.
For the last several years EER has been working with the EPA on the development
of a low-N0x pulverized coal burner for wall-fired applications. This DMB
consists of a circular burner surrounded by outboard tertiary air ports. The DMB
concept involves staging the combustion process to minimize N0X emissions while
maintaining an overall oxidizing atmosphere 1n the furnace so as to have minimal
impact on furnace slagging and corrosion. N0X production from fuel nitrogen
compounds is minimized by driving a majority of the compounds into the gas phase
under fuel-rich conditions and providing a stoichiometry/temperature history
which maximizes the decay of the evolved nitrogen compounds to N2 (!)¦ Thermal
?1-1
-------
N0X production is also minimized from the fuel-rich zone because of the reduced
temperature in a low-NOx burner.
"LIMB," Limestone Injection with Multistage Burners, was coined to describe the
potential simultaneous N0x/S0x control using limestone injection with a DMB. It
was originally thought that the conditions by which N0X emissions were reduced
with the DMB concept might also enhance the capture of sulfur species with
calcium-based sorbents. The lower temperatures within the low-NOx DMB flame were
expected to reduce the degree of deactivation of sorbent materials and thus
enhance sorbent reactivity. Injection in the burner would also allow for
improved dispersion and mixing of the sorbent particles in the furnace.
This paper summarizes results from an ongoing series of EPA-sponsored inves-
tigations designed to develop and evaluate low-NOx burner designs for combined
N0x/S0x control with sorbent injection. The burners evaluated included conmer-
cial second-generation low-NOx burners and commercial pre-New Source Performance
Standard (NSPS) burners, as well as experimental configurations, at firing rates
up to 100 x 10^ Btu/hr. These different burner designs provided a comparative
evaluation of commercial burners in the test furnaces with field operation. The
tests, which were conducted in pilot scale facilities at EER, evaluated effects
of burner design and operation, fuel type, sorbent type, and sorbent injection
configuration. Although a large portion of the work conducted to date has been
related to the characterization of N0x emissions, this paper will be concerned
exclusively with SO2 reduction by sorbent injection in, or close to, the burner.
EXPERIMENTAL SYSTEMS
Test Burners
The potential for SO2 reduction with sorbent injection has been evaluated for the
burner designs listed in Table 1. The burners included several externally staged
designs, with tertiary air ports located outside the burner exit, a coranercial
internally staged burner, and a commercial pre-NSPS burner. The experimental DMB
designs, based on EPA DMB design criteria (2) and fabricated for EPA Contract
68-02-2667, utilized generic burner hardware to provide flexibility in operation.
The two designs of the prototype DMB for application to industrial boilers,
evaluated under EPA Contract 68-02-3127, incorporated Foster Wheeler commercial
burner hardware. One of the designs, prototype DMB II, utilized a proprietary
Foster Wheeler exit geometry in conjunction with tertiary air ports, while the
geometry of prototype DMB III was based on EPA design criteria. The two
Steinmuller Staged Mixing (SM) burner designs, evaluated under EPA Contract
68-02-3916, included the SM burner design in operation at Weiher Unit 3, to
provide a basis for extrapolating the research furnace test results to an
operating boiler, and the modified SM II burner, that incorporated advanced
design concepts. Two commercially available Riley Stoker (RSC) burners were
evaluated at two firing capacities to establish the performance characteristics
in the test facility with which to compare the performance of the RSC DMB to an
operating utility boiler as part of EPA Contract 68-02-3913.
The design and performance characteristics of the burners are unique to each
manufacturer. The experimental DMBs are all equipped with two parallel secondary
air passages with individual swirl and flow control, outboard tertiary air ports,
and a central coal passage with a center body, impeller type coal spreader. The
industrial prototype DMBs and the RSC DMB also have two parallel secondary air
-------
passages and tertiary air ports, but the industrial prototype DMBs utilize the
tangential inlet, annular coal passage characteristic of Foster Wheeler equipment
while the RSC DMB uses a central coal passage with a venturi nozzle tip and cen-
ter body impeller. The two Steinmuller SM burners have only one secondary air
passage and incorporate an annular coal nozzle much like the Foster Wheeler
design. Both annular coal nozzle designs have a large-diameter central passage
which accommodates an igniter as well as producing a bluff body recirculation
zone to help stabi 1 ize the flame. The two commercial Riley Stoker designs have
single registers with the difference between the two being the coal nozzle and
impeller designs. The RSC Controlled Combustion Venturi (CCV) burner was also
equipped with tertiary air ports to permit evaluation of the burner under staged
conditions. In the discussion of experimental results, conuiercial burners are
designated by letter code rather than by name.
Research Furnaces
Testing has been conducted in the following research furnaces:
• Large Watertube Simulator (LWS)
• Medium Tunnel (MT)
• Small Watertube Simulator (SWS)
The LWS, shown in Figure 1, was designed to simulate the geometry of a small
front-wall-fired boiler, with a hopper bottom and a nose above the firing zone.
The furnace sidewalls are partially insulated with refractory. The MT and SWS,
shown schematically in Figure 2, are both partially insulated, horizontal tunnel
furnaces with burners mounted on one end and with the exhaust exiting from the
opposite end. The LWS and MT are both cooled externally with water spray while
the SWS is cooled with a water jacket. Nominal firing rates for these furnaces
are 100 x 10^, 50 x 10^, and 10 x 10^ Btu/hr for the LWS, MT, and SWS,
respecti vely.
Fuels and Sorbents
A wide range of coals and sorbents have been evaluated in these pilot scale
burner tests. A total of nine different coals have been tested with sulfur con-
tent ranging from 0.72 to 3.97 percent on a dry basis. The principal coals of
interest, listed in Table 2, are all high-volatile bituminous coals. Utah coal,
a high-volatile B bituminous with nominally 0.75 percent sulfur content, is used
at EER as a baseline coal.
The sorbents used have been calcium-based materials, including limestone,
hydrated lime, dolomite, and pressure-hydrated dolomitic lime. Typical sorbent
characteristics are listed in Table 3. The baseline sorbent is Vicron 45-3, a
pulverized high-purity limestone chosen because of the exceptional quality
control which ensured repeatability among batches.
?i-3
-------
RESULTS
Burner Tests in Large Watertube Simulator
The sulfur reduction potentials achieved by burners tested in the LWS with Utah
and Indiana coals are compared in Figures 3 and 4. The data shown were obtained
at optimum full-load design point conditions for each burner. The DMB data shown
represent data from tests of the 70 x 10® Btu/hr industrial prototype DMB III.
The other burners were tested at a nominal 100 x 10® Btu/hr. The limestone
material was injected both with the coal and through nozzles located on the axis
of each tertiary air port. SO2 capture was highest for the industrial prototype
DMB III for both coals and each sorbent/injection location pair. The other con-
sistent trend was that hydrated lime injected through the tertiary air ports pro-
duced greater SO2 reduction than did the limestone material. The SO2 removal
achieved at Ca/S = 2 was in the range of 35-40 percent for the limestone and
40-55 percent for the hydrated lime. None of the 100 x 10® Btu/hr burners had a
consistent advantage in terms of SO2 reduction potential.
The effect of firing rate on SO2 reduction of burners tested in the LWS is shown
in Figure 5. Burners C and E, and burners B and G represent scaled pairs of
burners. Burners C and B have a design capacity of 100 x 10® Btu/hr, while
burners E and G are scaled for 50 x 10® Btu/hr. Reducing the thermal input from
100 x 10® Btu/hr to 50 x 10® Btu/hr resulted in an improvement in sulfur capture
for burners C and B. Capture for the scaled-down burners, E and G, was better
still than the full-scale burners operated at half load. Because of the LWS
characteristics, a reduction in load increases mean residence time but also
reduces the mean and peak furnace temperatures. The result is that the time
available in the sulfation temperature window, 1500-2200°F, is relatively
insensitive to load. The aerodynamics of the full-scale burners operated at half
load are considerably different than those of a burner designed for reduced
loads; thus, interaction of the sorbent particles in the region around the burner
also di ffers.
The effect of coal composition on sulfur capture is shown in Figure 6. The
trends for all three fuels were similar, with the highest sulfur capture achieved
by injecting hydrated lime through the tertiary air ports, and with little dif-
ference in the effectiveness of injection locations with limestone. Intuitively,
an effect of composition, sulfur content in particular, would be anticipated.
The sulfation reaction would be thought to be driven in part by the concentration
of sulfur species. The SO2 capture data presented, however, show no discernible
effect of coal composition.
The effect of sorbent type on SO2 reduction is shown in Figure 7. The test
results are shown for staged conditions with sorbent injected through the ter-
tiary air ports and unstaged conditions with sorbent injected into the coal
stream. Dolomite achieved the highest capture for both cases. Overall, these
results indicate that dolomite is a better sorbent for comparable Ca/S molar
ratios than either limestone or hydrated lime. However, approximately 60 percent
more dolomite by weight is required to achieve comparable Ca/S ratios. Such
results are also consistent with data obtained in smaller furnaces and laboratory
scale equipment.
The effect of sorbent injection velocity is shown in Figure 8. Doubling the
velocity of the sorbent when injected through the tertiary air ports as a double
concentric jet Increased capture from 37 to 44 percent at Ca/S = 2. The higher
speed jets are characterized by significantly increased entrainment of hot
21-4
-------
combustion gases from the flame zone. The sorbent particles may also be
bal1istically thrown past peak temperature regions of the flame. In either case,
sulfur capture was improved.
Thermal Environment and Furnace Characteristics
The two primary processes by which sorbent particles remove gaseous SOg, sorbent
calcination and sulfation, have been shown to be very sensitive to the
temperature history experienced by the sorbent particles (3). The reactivity of
a given sorbent is strongly dependent on the peak temperature seen by the sorbent
particles, and sulfation is determined by the amount of time available in the
most favorable temperature regime of approximately 2250-1800°F (4). The thermal
characteristics through the sulfation temperature window are shown in Figure 9
for the experimental furnaces of interest to this study. In Figure 10 these
temperature profiles have been coupled with a sulfation model to predict overall
SO2 capture for a CaC03 sorbent of differing reactivity. Sorbent reactivity is
dictated by the peak temperature (and time) experienced by the sorbent, and may
be characterized by measurable parameters such as specific surface area (3).
The characteristic curves for the LWS furnace indicate that, compared to most
boilers, this furnace has a long time available in the sulfation window, and that
overall SO? capture is not limited by sulfation. In this case SO2 capture will
be determined by the reactivity of the sorbent material, and hence by peak tem-
perature. For the LWS experimental data, the spread in SOg capture by the CaC03
(for different burners, burner parameters, and injection methods) can be
explained by differences in peak temperature seen by the sorbent of no more than
200°F. Such temperature differences can easily be achieved by relatively small
changes in burner and sorbent injection aerodynamics.
In order to further investigate the impact of furnace thermal environment, lim-
ited data are presented for tests in the MT and SWS furnaces. Sulfur captures
for a 50 x 106 Btu/hr experimental DMB and the industrial prototype DMB II in the
MT furnace are compared in Figure 11. Both burners were fired at a nominal 50 x
10*> Btu/hr to fit the confines of the furnace. These data indicate large
differences 1n sulfur capture for injection of limestone with the coal compared
with injection through the tertiary air ports. Although capture by sorbent
injected with the coal for the industrial prototype DMB II in the MT is
comparable to results from the LWS, injection through the tertiary air ports is
much lower in the MT. The high firing densities and flame confinement in the MT
apparently produced an unfavorable thermal environment for the particles injected
through the tertiary ports. The central core of the flame, however, apparently
was unaffected by the MT cooling profile and thus yielded sorbent particles whose
reactivity was similar to those produced in the LWS.
The effect of furnace cooling on sulfur capture was also evaluated in the SWS, as
shown in Figure 12. The thermal environment, designated by the furnace tempera-
tures, was varied by changing the insulation distribution within the furnace.
Again, conditions with the lowest bulk temperatures yielded the highest SO2
removal. In fact, the configuration with an exit temperature of about 2300°F
produced an essentially unreactive sorbent.
Extrapolation to Boiler Temperature Profile
Experimental data can be translated into predictions for anticipated boiler SO2
capture in much the same way that Figure 9 was used in the generation of Fig-
ure 10. One example for an assumed boiler temperature profile is presented in
Figure 13. This figure shows the effect of injecting a CaC03 sorbent at differ-
-------
ent elevations in the boiler. When sorbent is injected near the burner zone, the
peak temperatures experienced limit sorbent reactivity and hence overall SO2
removal. As the sorbent injection location is moved to lower temperatures at
higher elevations, reactivity and SO2 capture improve. Ultimately there is a
trade-off between increased reactivity and reduced residence time in the sulfa-
tion temperature window. In the example shown, for the specific boiler and
sorbent combination of Figure 13, the optimum injection temperature is between
2100 and 2200OF.
This example is consistent with the experimental furnace data, and would suggest
that for conventional sorbents there is little advantage to sorbent injection in
the burner zone, and that upper furnace injection 1s to be preferred. Moreover,
the reactivity vs. temperature characteristics of normal limestone materials are
such that high SO2 removal levels are not expected. High SO2 capture can, how-
ever, be achieved with very reactive sorbent materials, even in adverse tempera-
ture fields. An example is shown in Figure 14 for the SWS furnace operating with
a temperature gradient not atypical of many U.S. boilers. This figure illus-
trates that, as the injection location 1s moved away from the burner zone to
downstream locations, SO2 removal is improved. Downstream injection of a
pressure hydrated dolomite results in greater than 70 percent SO2 removal at a
Ca/S = 2. This particular material is available commercially and has shown
consistently good performance in both small and pilot scale tests.
CONCLUSION
The results of these pilot scale tests indicate that burner design has only lim-
ited effect on sulfur capture. Capture depends on the interaction of the sorbent
with the flame and can be optimized for all burners. However, data indicate
that, for limestone at Ca/S = 2, only 35 to 40 percent SO2 removal can be
expected. Hydrated lime, when injected away from the flame zone, can be expected
to achieve 40 to 55 percent capture at similar stoichiometry. Differences in SO2
removal rates In the LWS furnace are consistent with differences of no more than
200°F in the peak temperature experienced by the sorbent.
The experimental data can be rationalized in terms of sorbent reactivity, and by
the temperature/time profile through the optimum sulfation temperature window.
Both parameters are influenced by sorbent characteristics and by specific furnace
thermal environment. A knowledge of both parameters is required to interpolate
data between experimental furnaces, and to extrapolate to operating boilers.
By avoiding high temperatures within the flame and exposing the sorbent to a
favorable temperature history, sorbent reactivity can be maximized. However,
each sorbent material has unique characteristics. Data indicate that the sulfur
capture potential of conventional limestone is limited and that other materials
are becoming available which offer the potential for high SO2 removal rates.
ACKNOWLEDGMENTS
This paper was based upon work conducted under several United States Environ-
mental Protection Agency contracts, including 68-02-2667, 68-02-3127, 68-02-3913,
and 68-02-3916. The authors would like to acknowledge the contributions of
Foster Wheeler Energy Corporation, Riley Stoker Corporation, and L. and C.
Steinmuller GmbH. The authors also wish to express their appreciation to the
staff of EER intimately involved in the conduct of the pilot scale tests.
o i -
-------
REFERENCES
1. R. Gershman, M. P. Heap, and T. J. Tyson. "Design and Scale-Up of Low
Emission Burners for Industrial and Utility Boilers." Proceedings of the
Second Stationary Source Combustion Symposium, Volume V, Addendum,
EPA-600/7-77-073e, (NTIS PB274-897), July 1977.
2. D. M. Zallen, R. Gershman, M. P. Heap, and W. H. Nurick. "The Generalization
of Low Emission Coal Burner Technology." Proceedings of the Third Stationary
Source Combustion Symposium, Volume II, Advanced Processes and Special
Topics, EPA-600/7-79-050b (NTIS PB292-540), February 1979.
3. J. A. Cole, W. D. Clark, M. P. Heap, J. C. Kramlich/G. S. Samuelsen, and W.
R. Seeker. "Fundamental Studies of Sorbent Calcination and Sulfation for SO2
Control from Coal Fired Boilers." Draft Task Final Report, Energy and Envi-
ronmental Research Corporation, EPA Contract 68-02-3633, October 1983.
4. P. Case, L. L. Ho, W. D. Clark, E. Kau, D. W. Pershing, R. Payne, and M. P.
Heap. "Testing and Evaluation of Experimental Wall-Fired Furnaces to Deter-
mine Optimum Means to Reduce Emissions of Nitrogen and Sulfur Oxides." Draft
Final Report, Energy and Environmental Research Corporation, EPA Contract
68-02-3921, January 1984.
?1-7
-------
SCRUBBER
SAFETY VENT
PURLINS
OBSER-
VATION/
SAMPLE
PORT
SLAGGING
PANELS
ACCESS
DOOR
\
BURNER
MOUNTING
PLATE
FIRING DEPTH
22 FEET
WIDTH
16 FEET
OVERALL HEIGHT
50.5 FEET
WALL FIRED
PARTIAL INSULATION
SPRAY COOLED
Figure 1. Large Watertube Simulator Furnace
MEDIUM TUNNEL FURNACE
DIAMETER: 14 Feet
LENGTH: 20 Feet
SPRAY COOLED, PARTIAL
REFRACTORY LINING
50 x 10® Btu/hr Nominal
SMALL WATERTUBE SIMULATOR
DIAMETER: 6 Feet
LENGTH: 17 Feet
WATER JACKET, PARTIAL/VARIABLE
" REFRACTORY LINING
10 x 10® Btu/hr Nominal
Figure 2. Schematics of Medium Tunnel and Small Watertube Simulator
21-8
-------
TERTIARY
T^_r_r
TERTIARY
UJ
DC
D
H
Q.
<
O
CM
o
(0
*
COAL
Ca(OH)
CaCO
CaCO
Ca/S Molar Ratio
Figure 3. Comparison of Sulfur Capture Potentials for Burners
Tested in the LWS with Utah Coal
TERTIARY . COAL TERTIARY
70
DMB
eo
DMB
»-
0.
<
O
40
CM
O 30
20
Ca(OH)
CaCO
CaCO
10
o
1
4 o
2
3
1
2
3
4
4 0
1
3
2
Ca/S Molar Ratio
Figure 4. Comparison of Sulfur Capture Potentials for Burners
Tested in the LWS with Indiana Coal
21-9
-------
E(50) /
'M
CaCOg
WITH COAL
G(50) r /
B(50)
(100)
TERTIARY
4 0 1 2 3 4
( ) LOAD 10® Btu/hr Ca/S Molar Ratio
Figure 5. Effect of Firing Rate on SO2 Reduction
G(50)
BC60)
B(100)
Cb(0H>2
TERTIARY
O 0.74% S COAL
70
2.73% S COAL
T
A 3.97% S COAL
60
Ul
cc
D
I-
o.
<
O
cP
-------
STAGED
UNSTAQED
TERTIARY
AIR
COAL
0.75% S COAL
0
100 x 10 Btu/hr
O CaCO^
~ Ca(OH)2
O DOLOMITE
Ca/S Molar Ratio
Figure 7. Effect of Sorbent on SOj Capture with Burner C
o 30
CaCO
• 70 x 106 Btu/hr
• 2.6% S COAL
• SORBENT INJECTION
" A WITH COAL
- TERTIARY AIR AS
DOUBLE CONCENTRIC
JET
O LOW VELOCITY
0 HIGH VELOCITY
Ca/S Molar Ratio
Figure 8. Effect of Sorbent Injection Velocity on SO2
Capture with Industrial Prototype DMB II
21-11
-------
FURNACE
EXIT PLANE
SW
2400
LL
O
2200
LLI
DC
D
h-
<
CC
III
Q.
2
III
h-
sws
COLD
LWS
2000
1800
- V —
1.0
2.0
1.0
TIME (SEC)
Figure 9. Experimental Furnace Temperature Characteristics
at Nominal Load
50
<
>
O
2
HI
DC
eg
O
CO
40
30
20
10
LWS
MT
SWS (HOT)
• SO REMOVAL VS SORBENT
2
REACTIVITY (CaCOg)
• SULFATION MODEL GIVES S02
REMOVAL FOR FURNACE dT/dt
• REACTIVITY RELATED TO PEAK
CALCINATION TEMPERATURE
5 10 15
SORBENT REACTIVITY
Figure 10. Predicted Sorbent Reactivity Based on Furnace
Characteristics
21-12
-------
4
£
• 50 x 10® Btu/hr
A INDUSTRIAL PROTO-
TYPE DMB II
~ DMB IV (V-50)
CaCOg INJECTION
SOLID - WITH COAL
OPEN - TERTIARY
LU
IT
H
Q.
<
O
CM
o
CO
60
50
40
30
20
10
2 3
Ca/S Molar Ratio
Figure 11. Sulfur Capture with Sorbent Injection in the MT
EFFECT OF FURNACE
COOLING
10 x 108 Btu/hr
DMB
V cold-t.ex|T:
O INTER - Te
1 1
i
1
CaCO TERTIARY PORTS
3
-
-
-
/o
/
" //
-
- /W
•D"d
-
i
i
•
1900° F
2150 F
~ HOT - Te = 2300 F
1 2 3
Ca/S Molar Ratio
Figure 12. Effect of Furnace Cooling on Sulfur Capture in the SWS
21-13
-------
T
Ca/S - 2
2000 ppm
BOILER
TEMPERATURE
PROFILE
CAPTURE
0.6 1.0 1.S 2.0 2.5
RESIDENCE TIME (SEC.)
3.0
BO
40 -0
m
7}
o
30 ™
20
10
O
>
"O
-t
c
3J
m
S02 CAPTURE VS
SORBENT INJECTION
AT DIFFERENT BOILER
LOCATIONS
Figure 13. Experimental Data Translation to Boiler Temperature Profile
downstSeam
(2300°F)
TERTIARY
PORTS
WITH
COAL
X
12 3 4
Ca/S Molar Ratio
6
10 x 10 Btu/hr
DMB
T 2100°F
C A
~ DOLOMITE
o PRESSURE SLAKED
DOLOMITE
Figure 14. Effect of Sorbent Injection Location on
Sulfur Capture in the SWS
21-14
-------
Table 1
TEST BURNERS
Burner/Component Firing Capaci ty
Manufacturer Burner Desl9n (10? Btu/nr)
Experimental Modified DMBs Externally Staged 10, 50, 100
Foster Wneeler Industrial Prototype Df€ II Externally Staged 70
Industrial Prototype DMB III Externally Staged 70
Stelnmuller SM Burner (Werner) Externally Staged 50, 100
SM II Burner Externally Staged 100
Riley Stoker RSC DhB Externally Staged 100
Controlled Combustion Ven- Internal and
tun Burner External Staging 50. 100
Flare Burner Pre-NSPS 50, 100
Coal
Utan
Indiana
II1inois
Saar
(W. Germany)
Table 2
PRINCIPAL COALS
Rank
HV3 Bituminous
HVB Bituminous
HVC Bituminous
MVA Bituminous
Sulfur Content
(Wt I. Dry)
0.73 - 0.96
1.35 - 2.73
3.55 - 3.97
0.72 - C.87
Table 3
TYPICAL S0RBENT CHARACTERISTICS
Median Density Chemical Analysis
Sorbeni Composition Diam. (g/ci*3) (Wt %)
(>iin) Ca Mg
Vicron 45-3
CaC03
9.8
2-71
39.0
C.49
Col ton Hydrated Lime
Ca(0H)2
4.0
2.28
51.9
0.25
RUK Hydrated Lime
Ca(OH )2
4.0
2.25
52.9
0.43
Dolomite
CaC03'MgC03
12.0
2.87
34.7
11.3
Pressure Slaked
Dolomitic Line
CaC OH >2 *
Mg(OH )2
2.2
2.28
28.6
16.0
21-15
-------
LIMESTONE INJECTION WITH AN
INTERNALLY STAGED LOW-NOx BURNER
J. Vatsky and E. S. Schlndler
Foster Wheeler Energy Corporation
110 South Orange Avenue
Livingston, New Jersey 07039
ABSTRACT
Foster Wheeler's Internally staged, Controlled Flow/Spl1t-Flame low-NOx burner has
been in utility field service since 1979. The ease of retrofit to existing steam
generators would make the burner a cost-effective means of Implementing the
Limestone Injection Multistage Burner concept if satsifactory sulfur capture can be
achieved. Two methods of limestone injection internal to the burner are available:
pre-mixed with the coal and separate from the coal. The former technique was
evaluated in a joint program with the U.S. EPA, while the latter was evaluated
independently by Foster Wheeler using a novel Injection method. Over 60 percent
greater sulfur capture was obtained using the Foster Wheeler technique on a
50 million Btu/hr single burner test furnace.
21-16
-------
DEVELOPMENT OF INTERNALLY STAGED
BURNERS FOR LIMB
G. C. England, R. Payne, and J. Clough
Energy and Environmental Research Corporation
18 Mason
Irvine, California 92718
ABSTRACT
This paper discusses an experimental investigation directed toward the
development of retrofit technology to achieve simultaneous in-furnace control of
N0X and S0X emissions from coal-fired boilers. Results obtained in a 2.9 MW^
(10 x 10® Btu/hr) test furnace show that N0X reductions of 50 percent can De
obtained from high-velocity pre-NSPS Durners by modification of both the coal
nozzle and the air distribution without the use of external air ports. The
effectiveness of calcium-based sorbent injection for SO2 control is limited by
soroent reactivity and time/temperature characteristics of the Doilers.
Experimental results suggest that sorbent injection can be optimized using
coaxial air/soroent jets to achieve maximum sorbent reactivity and dispersion,
and thereby optimize SO2 removal.
INTRODUCTION
Coal-fired electric utility boilers are a major source of N0X and SO2 emissions
to the atmosphere. These pollutants are considered to be the primary precursers
to acid rain. Approximately 85 percent of N0X emissions and almost all SO2
emissions from coal-fired utility plants are from boilers constructed prior to
1971, when EPA's New Source Performance Standards (NSPS) for large (greater than
73 MWt or 250 x 10^ Btu/hr) boilers first became effective. In order to control
acid rain it 1s desirable to have cost-effective N0X and S0X control
technologies which can be implemented on a retrofit basis to existing pre-NSPS
boilers.
The objective of retrofit N0x/S0x control technologies should be to maximize the
potential emission reduction within the retrofit constraints of specific units
or classes of boilers. Many current low-N0x coal burner technologies capable of
meeting NSPS N0X emission limits utilize external (staging or tertiary) air
ports or reduced secondary air velocity (enlarged burner throat diameter) to
effectively delay fuel/air mixing. For many existing boilers, however, it may
be impractical to enlarge the burner throat or install external air ports on a
retrofit oasis due to structural or other considerations. Thus there 1s a need
to determine the extent to which N0X emissions can be reduced as a function of
(*) There are no 22~ pages in this volume.
23-1*
-------
the range of retrofit constraints imposed by pre-NSPS Doiler designs. The
parallel application of dry sorbent injection into the boiler furnace for
reduction of S0X emissions offers the potential for significantly reducing the
emission of acid rain precursers.
This program is directed toward the Investigation of internally staged low-NOx
coal Durners for wall-fired boilers, and the utilization of calcium based
sorbent injection for S0X control. The process by which SO2 is captured by
calcium based sorbents has Deen described in other work (1, 2). Key to the
success of the combined N0x/S0x control concept is to provide the sorbent with
the optimum conditions for maximizing reactivity, achieving good dispersion into
the combustion products, and providing adequate residence time within the
optimum temperature range (900-1200°C) for SO2 capture to occur (2,3).
Experimental studies are currently underway in 3 MWt (10 x 10^ Btu/hr) and 29
MWt (100 x 10^ Btu/hr) facilities to investigate combined N0x/S0x control via
application of internally staged burners with sorbent injection. Specific goals
of these investigations are to determine the impact of burner design parameters
on the level of N0X control achievable without the use of air ports external to
the burner throat, particularly under conditions typical of pre-NSPS boilers.
Recent evidence suggests that sorbent injection near the burner is not the most
attractive means for achieving high SO2 captures. Therefore, emphasis has been
placed on sorbent injection in the upper furnace. Experiments under this
program have investigated the effect of sorbent injection parameters on SO2
capture in order to establish guidelines for optimizing sorbent injection
methods. This paper presents selected results from these experimental studies
conducted in the 3 MWt (10 x 10& Btu/hr) experimental test furnace.
EXPERIMENTAL
Burners
Tests have been conducted with burners designed to achieve both high and low
secondary air velocity at the throat section (Figure 1). The high velocity
burner was designed to be representative of typical pre-NSPS Durners with a
secondary air velocity of approximately 58 m/sec (190 ft/sec). Pre-NSPS burners
are typically designed to maximize combustion intensity in the boiler furnace.
In many existing boilers it is impractical to perform extensive modifications to
the burner throat, and burner modifications are limited to the coal nozzle or to
air distribution within the existing burner throat. These tests investigated
the potential N0X reduction achievable by modifications to the coal nozzle
without changing the existing burner throat. The effect of adding a baffle to
the secondary air passage in order to split the secondary air stream (see Figure
1) was also investigated.
The low velocity burner was designed to produce a secondary air velocity of 24
m/sec (80 ft/sec) and represents a more flexible retrofit situation where throat
enlargement to reduce air velocity is possiDle. Both the high and low velocity
burners utilized the same air register which supplied the secondary air through
two annular passages at the throat of the burner. Air flow and swirl level
through each secondary air passage was independently controlled to investigate
the impact of these parameters on N0X control.
23-2
-------
Coal Nozzles
Tests were conducted with axial coal nozzles with a variety of spreader designs
in both burner configurations. The spreaders consisted of two types: axial vane
swirlers with vane angles ranging from 15 to 45°; and splitters which divided
the primary air/coal into four streams and injected the fuel without swirl at
angles from 30 to 60° (included angle).
Coal injection with reduced primary air flow (dense-phase transport) was also
investigated. Primary air was reduced from approximately 1.9 to 0.2 kg air/kg
coal. The diameter of the dense-phase coal nozzles was reduced in order to
maintain the primary injection velocity at approximately 18 m/sec (60 ft/sec).
Three injector types were used with dense-phase coal transport: a coirmercial
spray nozzle; a conical spreader with a 30° included angle; and an axial vane
swirler with a 30° vane angle.
Small Watertuoe Simulator (SHS) Furnace
The SWS facility is illustrated in Figure 2. The cylindrical test furnace has
an inside diameter of 1.8 m (6 ft) and is 5.2 m (17 ft) in length. Burners are
fired axially from one end of the cylinder, and the furnace is completely water
jacketed with a partial refractory lining to control heat extraction. Gas
temperature at the exit of the furnace is typically Detween 1000 and 1200°C
(1800 and 2200°F). Nominal firing capacity is 2.9 MW (10 x 10^ Btu/hr).
Fuel s
A Utah high volatile bituminous coal was used for all coal tests. Analysis of
the Utah coal is given in Table 1. Sorbent injection tests were conducted Doth
with coal and with natural gas flames aopeo with SO?- The sorbent used in these
tests was commercially available limestone (CaC03) with a particle size
distribution of between 3 and 45 m.
N0X EMISSIONS
Figure 3 shows N0X emissions obtained with various coal nozzles tested in the
high velocity burner configuration. N0X emissions are corrected to 0% stack 02-
Excess air was constant at approximately 20%, firing rate was 2-9 MW (10 x 10®
Btu/hr), and combustion air preheat temperature was approximately 315°C (600°F).
The data are presented as a function of gas temperature at the exit of the
furnace. The variation in exit temperature is due primarily to the thermal
inertia of the furnace cooling system and due to ash deposition in the furnace,
which changes the extent of furnace heat extraction. The lines shown on Figure
3 and the following figures represent the "best fit" through data points
obtained under similar burner operating conditions for each burner/nozzle
combination. It can be seen that N0X emissions generally increase with
increasing furnace exit temperature for similar burner conditions, and that the
dependence on temperature is unique to the specific nozzle/burner combination.
23-3
-------
N0X emissions as high as 925 ppm ana as low as 350 ppm coula De achieved through
variation of coal nozzle design and burner adjustments. Most of the N0X
emissions data range between 500 and 750 ppm which is within the range
encountered in the field for pre-NSPS boilers. It was difficult to reduce N0X
emissions below 450 ppm while maintaining satisfactory flame stability. High
N0X flames were generally characterized by low length-to-diameter ratios and by
flame standoff between 0.1 and 0.6 burner exit diameters. The shaded area
represents the range of emissions achieved with a 15° axial swirl vane nozzle by
variations in burner adjustments, including adjustments to coal nozzle setback,
coal spreader setback, secondary air distribution, and swirl distribution
between the secondary air channels. These data show that N0X can vary by as
much as a factor of three depending only on burner adjustments. Data are also
shown in Figure 3 for a splitter type nozzle which produced emissions in the
range of 600-700 ppm. In this minimum flexibility Durner configuration there
did not appear to be a distinct advantage of utilizing the splitter type
nozzle. It should be noted that, for all data presented in this paper, CO
emissions were below 60 ppm and flames were stable.
Figure 4 shows the effect of adding a conical baffle to the inner secondary air
sleeve of the high velocity burner (as shown in Figure 1) in order to divert
some of tne combustion air away from the main body of the flame, thereby
reducing the local stoichiometric ratio in the initial regions of the flame.
Adding the baffle did not appear to reduce N0X emissions substantially for the
splitter type nozzle. However, with the 15° axial vane swirler, N0X emissions
were reduced to between 225 and 500 ppm by adding the baffle. The baffle
enhanced stability of the flames, and most of those conditions which produced
low-NOx emissions were associated with flames established well within the burner
quarl . Comparison of these results with those shown In Figure 3 indicates that
the air baffle substantial ly reduced the level of N0X emissions from the high
velocity burner configurations and dramatically increased the range of burner
adjustment for which N0X was oelow 400 ppm. The same range of conditions
without the baffle typically resulted in N0X between 700 and 900 ppm.
Figure 5 shows N0X for several nozzles tested in the low velocity burner
configuration. The shaded areas represent the range of N0X emissions achieved
with each nozzle design by variation of burner adjustments. The effect of
increasing the burner throat diameter was most pronounced for one of the
splitter coal nozzles (30° included angle) and for the 15° axial vane swirler.
For equivalent burner and furnace conditions, N0X reduction achieved with the
low velocity burner was 51 percent for the splitter nozzle and 46 percent for
the 15° axial vane swirler compared to results obtained in the high velocity
burner.
N0X formation in practical flames is determined to a large extent by mixing
processes in the flame and heat transfer characteristies of the furnace. Flame
shape is important because it reflects the mixing characteristics of the burner.
Figure 6a shows a preliminary attempt at correlating N0X emissions data with the
observed flame dimensions for various coal nozzles in the low velocity burner.
The flame dimensions are represented in Figure 6a by the observed flame length-
to-diameter ratio (L/D). The shaded area on the figure represents the overall
data trend for axial swirl vane nozzles.
-------
There is a general trend for decreasing N0X emissions with increasing flame
1 ength-to-diameter ratio, although there is noticeaole scatter especially for
low L/D (short flames). Data appear to fall above the shaded area when the
flames are detached from the burner and fall slightly below the shaded area when
the flame is sharply divided (e.g., with coal splitter nozzles).
Previous work by Payne, et al . (4) showed that N0X emissions from oil flames
produced by many different burners fired in different furnaces could be
correlated by a factor which takes into account flame shape, furnace heat
extraction, and entrainment characteristics of the flame. Since the bulk
entrainment characteristics (defined by the Thring-Newby parameter) are constant
for constant burner velocity and diameter, a similar correlation was checked for
the low velocity burner configuration which accounts only for flame residence
time (in terms of firing rate and flame volume) and furnace heat extraction.
These parameters are expressed as (Q/Yf) x (1/He) where Q is the gross heat
input in megawatts, Vf is the observed flame volume, and He is the heat
extracted in the furnace in megawatts. As shown in Figure 6b the correlation
appears to work well for a single coal nozzle, but yields different N0X values
for different nozzles. This indicates a need for another parameter which takes
into account differences in near-burner mixing as a result of coal nozzle
design. The figure does seem to suggest that, for equivalent flame and heat
extraction conditions (i.e., at a given value for the aocissa), N0X emissions
are lower for the two splitter nozzles than for the axial vane swlrler nozzle,
and are lowest when dense-phase coal transport is utilized. Since the coal
injectors are located axially in the center of the burner, the reduction in coal
nozzle diameter associated with dense-phase coal transport (to maintain primary
injection velocity) results 1n Increased area availaDle for secondary air flow.
As a result the baseline secondary air velocity of 58 m/sec (190 ft/sec) was
reauced to 44 m/sec (145 ft/sec) when dense-phase coal nozzles were used in the
high velocity burner. Similarly, secondary air velocity was reduced from 24 m/
sec (80 ft/sec) to 21 m/sec (71 ft/sec) when dense-phase coal nozzles were used
in the low velocity burner. The reduction in N0X achieved with dense-phase coal
transport probably results from changes in mixing and flame characteristics
brought about by the combined effect of reduced secondary air velocity and
altered coal/air distribution.
S02 EMISSIONS
The injection of dry calcium-based sorbents such as limestone and dolomite into
boiler furnaces offers a potentially less expensive means of reducing SO2
emissions compared to post-boiler scrubbing. The practical limitations to
achieving high SOg removal rates appear to be related to sorbent reactivity and
residence time within the temperature range for which sulfation of the sorbent
is possible. Favorable temperatures are usually encountered downstream of the
burner zone in the upper furnace and superheater area of most U.S. boilers.
Sorbent dispersion Into the combustion products 1s easiest to achieve if the
sorbent is injected through the burner itself; however, sorbent reactivity is
sharply reduced when exposed to flame temperatures and coal mineral matter.
Therefore, injection of the sorbent away from the burner has been suggested,
either through ports near the burner throat or in the upper furnace, downstream
from the burners.
23-5
-------
Figure 7 shows SO2 removal versus calc1um-to-sulfur molar ratio (Ca/S) for
various sorDent injection locations near the high velocity Durner. The aata
snown in Figure 7 were oDtained using natural gas fuel doped with pure SC^. The
coal spreader support pipe located on the axis of the coal nozzle (see Figure 1)
was replaced with a natural gas injector. As shown in Figure 7, SO2 removal is
approximately the same regardless of sorDent injection location, which suggests
that sorDent reactivity is equally limited for all near-burner injection
locations. This indicates that it could be difficult to meet SO? capture
oDjectives (50 percent SO2 removal at Ca/S = 2-0) if the sorDent is injected
near the Durner.
Based on these results and results of other related EPA research programs,
recent efforts under this program have concentrated on topics related to sorDent
injection in the upper regions of a Doiler furnace remote from the Durner zone.
In these upper regions, injection will typically be by means of jets, where some
carrier medium (air or flue gas) will De required to ensure adequate penetration
and dispersion of the sorDent. A particular emphasis has therefore Deen placed
on the investigation of sorbent injection techniques, and whether available jet
parameters offer any secondary means Influencing the reactivity of sorbent
materials. Interest has centered particularly on heating rate, and the
potential use of double concentric jets to provide thermal shielding, either for
conventional soroents or for high reactivity materials (5,6) which may be very
temperature sensitive.
The simplest injection scheme involves a single jet of sorDent material
dispersed in carrier air. Figure 8a shows SO2 removal for CaC03 injection using
a single smal1-diameter (2.54 cm) sorDent jet injector along the axis of the SWS
furnace. An array of small gas-fired burners was arranged around the sorbent
jet to produce simulated combustion products at 1200°C (2200°F) flowing parallel
to the soroent jet. Natural gas doped with pure SO2 was fired through these
small gas burners, and the furnace walls were completely insulated to minimize
heat extraction. Furnace gas temperature (background gas temperature) was
approximately constant (within 15 percent) throughout the furnace. The velocity
of tne sorDent jet was varied Dy changing the transport air flow rate. SO2
removal was similar for injection velocity of 30 and 15 m/sec (100 and 50 ft/
sec) but was much lower for low injection velocity of 7.6 m/sec (25 ft/sec).
Since the peak temperature experienced by the sorbent jet is believed to be
approximately.the same for all three injection velocities, it is believed that
the heating rate in the initial regions of the Jet may affect the development of
surface area as the raw sorbent undergoes calcination to CaO. This implies that
some degree of control over sorbent reactivity may be achieved by controlling
Injection parameters. However in this experiment, complete mixing of the
sorbent and combustion products does not occur until they enter the exhaust
duct, and slight differences in dispersion may account for the effect of sorbent
velocity on capture. Further experiments are planned to evaluate this effect.
Figure 8b shows the effect of gas temperature near the point of sorbent
injection (background gas temperature) for the 30 m/sec (100 ft/sec) sorbent
jet. SO2 removal decreases sharply as gas temperature 1s Increased from 1200°C
(2200°F) to 1540°C (2800°F). This suggests that the reactivity of the sorbent
decreases as the background gas temperature increases. Thus, it is evident that
-------
in practical systems high sorbent temperatures must be avoided in order to
prevent deactivation of the sorbent.
Sorbent Injection through single small jets is the simplest way to introduce
sorbent into the upper furnace of the boiler and may result in high sorbent
surface areas if high peak temperatures can be avoided since rapid heating rates
are easily achieved; however, it is difficult to achieve adequate mixing between
the sorbent and furnace gases since jet penetration 1s limited with small single
jets. Combining the single sorbent jet with a larger annular air jet provides a
means for increasing total jet momentum, thereby increasing jet penetration to
achieve adequate sorbent dispersion. A coaxial air/soroent jet can also be
designed to screen the sorbent from high temperature gases as well as control
sorbent heating rate. This would allow significantly greater control over the
ultimate reactivity of the sorbent. Figure 9 illustrates the effect of jet
parameters on SO2 removal with coaxial air/sorbent jets injected into 1425°C
(2600°F) combustion products. Comparing these results with Figure 8b at 1425°C
indicates that SO2 removal at Ca/S ratio of 2 has been increased from 22 percent
for the single jet to 29-35 percent with the coaxial jets. The results shown in
Figure 9 indicate that a 20 percent relative variation in SO2 removal can be
achieved by varying velocity and diameter of the annular air jet. Although the
absolute level of capture is low for CaC03 because the ultimate surface area is
low, the relative enhancement of SO2 removal achieved using the annular jet may
be very significant for other sorbents which are more reactive. Experimental
work is currently under way to investigate the impact of jet parameters on SO2
removal efficiency with other sorbents and to further develop the relationship
between sorbent injection parameters and SO2 removal.
DISCUSSION
These results suggest that N0X reduction on the order of 50 percent may be
possible through modifications to the coal nozzle and/or secondary air
distribution without modifying the burner throat, especially where uncontrolled
N0X emissions are high Initially. For instance, it may be possible to reduce
N0X emissions from 500-750 ppm to 225-500 ppm by adding a baffle to the
secondary air passage on a typical pre-NSPS burner. The use of dense-phase
coal transport and secondary air baffles appears to be particularly effective
with "minimum flexibility" high velocity burner configurations. The coal
injector used with dense-phase coal transport can be smaller in diameter which
decreases secondary air velocity significantly. Since dense-phase transport
also tends to concentrate the coal near the axis of the flame, stabilizing the
flame within the burner quarl results in low-M0x emissions without excessively
long or poorly defined flames. However, since most coal mills are designed to
operate with a fixed primary air flow, dense-phase coal transport may not be
feasiDle without extensive modifications to existing coal delivery systems.
Secondary air baffles used with high velocity burner configurations tend to
promote stability of the ignition front within the burner quarl as well as
divide the air flow for delayed mixing and are, therefore, effective in reducing
N0X emissions, particularly for detached flames.
These data tend to support the conclusion that, for high secondary air velocity,
coal nozzles which concentrate the coal in a narrow region on the flame axis
-------
will tend to produce low-NOx emissions if the flame is staDilized within the
burner quarl . Coal nozzles which divide ana disperse the coal stream to wide
angles result in relatively faster mixing and therefore do not achieve
significant N0X reductions. For burners with low secondary air velocity, flames
generally tend to be Digger than for burners with high secondary air velocity,
and burner adjustments have a more significant impact on N0X. Coal nozzles in
low velocity burners which divide the coal to produce smaller separate flames
produce lower N0X emissions if the separation between the smaller flames is not
clearly defined; however, if the injection angle is too wide or the flame is
detached, coal and air mix more rapidly and smaller N0X reductions are achieved.
Although sorDent injection near the burners is desirable to achieve mixing of
the sorbent with the combustion products, thermal deactivation of the sorbent
appears to be unavoidable and injection away from the high temperature burner
region is more attractive, particularly for very reactive (high surface area)
sorbents. Coaxial air/sorbent jets can be used in upper furnace sorbent
injection to control sorbent dispersion by proper jet design. Coaxial jets also
appear to offer a means for injecting sorbent into the upper furnace which
allows additional control over those parameters which affect sorbent reactivity.
The annular air jet can be particularly effective in reducing the extent of
thermal sorDent deactivation when injection into high temperature regions of the
furnace is necessary.
These results have also indicated that the heating rate of the sorbent in the
initial region of the jet may influence the development of surface area during
calcination. Since heating rate is affected by the entrainment characteristies
of the jet, it may be possible to optimize jet design to achieve maximum sorbent
reactivity. Further information is required to define the exact nature of the
relationships between jet design and the overall SO2 removal process for
sorbents which exhibit differing characteristies.
ACKNOWLEDGEMENTS
This work was supported under EPA Contract 68-02-3692. The authors wish to
acknowledge the guidance and technical contributions of D. B. Henschel, the EPA
Project Officer.
REFERENCES
1. Borgwardt, R. H., "Kinetics of the Reaction of SO2 with Calcined Limestone,"
Environ. Sci. and Technology, 1970, 4 (1 ), 59-63.
2. Cole, J. A., W. D. Clark, M. P. Heap, J. C. Kramlich, G. S. Samuelsen, and
W. R. Seeker, "Fundamental Studies of Sorbent Calcination and Sulfation for
SO2 Control from Coal-f1red Boilers," Task Final Report, Energy and
Environmental Research Corporation, EPA Contract 68-02-3633, in preparation.
3. Case, P. L., L. Ho, W. D. Clark, E. Kau, D. W. Pershing, R. Payne, ana M. P.
Heap, "Testing ana Evaluation of Experimental Wall-Fired Furnaces to
Determine Optimum Means to Reduce Emissions of Nitrogen and Sulfur Oxides,"
23-8
-------
Final Report, Energy ana Environmental Research Corporation, EPA Contract
68-02-3921, in preparation.
4. Payne, R., T. Akiyama, ana J. Witkamp, "Stuaies on the Pollutant Emissions
Characteristics of a Nunrner of Oil-Firea Conaitions," American Flame
Research Comnittee Symposium on N0X Reauction, Newport Beach, CA, 1980.
5. Overmoe, B. J., S. L. Chen, L. Ho, W. R. Seeker, M. P. Heap, ana D. W.
Pershing, "Boiler Simulator Stuaies On Sorbent Utilization for SO2 Control,"
First Joiat Symposium on Dry SO2 ana Simultaneous S02/N0X Control
Technologies, San Diego, CA, November 13-16, 1984.
6. J. A. Cole, J. C. Kramlich, G. S. Samuelsen, W. R. Seeker, ana G. D. Silcox,
"Reactivity of Calcium-Basea Sorbents for SO2 Control." Ibia.
23-9
-------
Secondary Air
Coal
Nozzle
Coal &
Primary Air
IZI
m
m
si
si
12)
HIGH VELOCITY BURNER
58 m/s (190 ft/sec)
LOW VELOCITY BURNER
24 m/s (80 ft/sec)
Baff1e
SWIRLERS
&
SPLITTERS
DENSE PHASE
TRANSPORT
SECONDARY AIR BAFFLES
COAL NOZZLES
Figure 1. Experimental burners.
23-10
-------
Heat Exchanger
Multiclone
Pulverize
Coal
System
F.D. Fan
Water Jacket
Measurement Ports
Figure 2. Small watertube simulator facility.
CHARACTERISTICS
Shape
Diameter
Length
Refractory
Insulation
Volumetric
Heat Release
Firing Rate
Heat Release
Per Cooled
Surface
T Exit (Typical)
Cylinder
6 Ft (1.8m)
17 Ft (5.2 m)
50%
20,000 Btu/Hr-Ft3
(0.21 MWt/m3)
10 x 106 Btu/Hr
(2.9 MWt)
60,000 Btu/Hr-Ft2
(0.19 MWt/m2)
1800°F - 2200°F
(980°C - 1200°C)
-------
1000
CM
o
o
£
-o
e
CL
Q.
800
600
400
200
RANGE ACHIEVED THROUGH
BURNER ADJUSTMENTS
(15° SWIRLER)
(lines indicate best fit for similar burner'
adjustments) | t
1800
2000
i
2200 °(
*
1000 1100
Furnace Exit Temperature
1200 °C
Figure 3.
N0x emissions achieved with high
velocity burner.
SWIRLER NOZZLES
O 15 degree
# 30 degree
O 45 degree
SPLITTER NOZZLE
+ 15 degree
23-12
-------
'.\i
o
>>
i.
T3
E
Q-
O.
800
600
400
200
RANGE ACHIEVED THROUGH
^/"BURNER ADJUSTMENTS
(15° SWIRLER)
(lines indicate best fit for similar
burner adjustments)
O SWIRLER (15°)
+ SPLITTER (15°)
1800
¦ ¦
2000
*
2200 °F
1000 1100 1200 °C
Furnace Exit Temperature
Figure 4. NO emissions with secondary air baffle
ad3ed to high velocity burner.
23-13
-------
1000
800
C\J
O
TD
£
CL
CL
600 -
400 -
200
RANGE ACHIEVED THROUGH
BURNER ADJUSTMENTS^
(15° SWIRLER) \
f 8°°°
(lines indicate best fit for similar
burner adjustments)
i . I i |_
1000
1
J.
2000
I
SVIIRLERS
O 15 degree
% 30 degree
SPLITTERS
+ 15 degree
x 30 degree
2200 °F
_L
1000 1100 1200 °C
Furnace Exit Temperature
Figure 5. NO emissions with low velocity burner.
23-14
-------
ro
CO
i
100C-
O
C\l
O
o*5:
o
>>
T3
Q.
CL
T 1 1 1
LOW VELOCITY BURNER'
xDetachfid
Flames^
O SWIRLERS
8 SPLITTERS
# DENSE PHASE
9
6°d-
a
Data Trend-
swirlers
40C
'a
a
s )
•>
. . v
¦ tl
¦ - -iv
X
I
2 3 4
FLAME L/D
(a)
O -
1000
OJ
o
5-Q
800
O
>>
S-
600
T>
E
400
a.
o.
»
X
200
NO
T r
0 1
T
T
SWIRLERS
SPLITTERS 1
^.ONE
(DENSE PHASE.
X
Q 1 ,m-3,
Vf h;
(b)
Figure 6. Relationship between NO and (a) flame shape,
and (b) flame/furnace parameters.
-------
CaCO^ Sorbent
Natural gas with SO2 added
50
40
•>
ro 30
>
0
1 20
cm
o
1/1 10
0 12 3 4
Ca/S
SORBENT INJECTION LOCATION
A PRIMARY AIR
§ SECONDARY AIR (THROAT)
FRONT WALL (LOW INJECTION VELOCITY)
FRONT WALL (HIGH INJECTION VELOCITY)
Figure 7. Near-burner sorbent injection—effect of in-
jection location (2.9 MW, high velocity burner).
23-16
-------
rO
>
O
E
Cd
CNJ
o
ul
| | l
- 030 m/sec
~ 15 m/sec
O7.6 m/sec
- O 1200°C
~ 1425'C
~ 1540"C
Figure 8. SOg capture in single sorbent jets for (a) 1200*C
background gas temperature and (b) 30 m/sec sorbent injection
velocity.
SORBENT JET VELOCITY:
O 15 m/sec
~ 30 m/sec
AIR JET VELOCITY:
18 m/sec
9 m/sec
50 - 15 an Air Jet ¦¦ 20. cm Air Jet --25 cm Air Je
40
30
20
CM
O
CO
-O
10
0
Ca/S
Figure 9. Effect of jet velocity and diameter on S02
capture for sorbent injection in coaxial
air/sorbent jets (background = 1425°C).
23-17
-------
TABLE 1. ANALYSIS OF UTAH COAL
As Received
Ultimate Analysis:
C (%) 67.42
H {%) 5.02
N [%) 1.38
S {%) 0.67
0 (%) 12.71
Moisture
(%)' 3.14
Ash
{%) 9.67
Gross Heating Value (kj/g) —
Gross Heating Value (Btu/lb) —
23-18
-------
SESSION V: POST-FURNACE SO2 REMOVAL
Chairman, Dan Giovanni, Electric Power Technologies, Inc.
a
24-i
-------
CHARACTERIZATION OF ALTERNATE SODIUM SORBENTS FOR
FABRIC FILTER S02 CAPTURE
R. Hooper
Electric Power Research Institute
Arapahoe Test Facility
P.O. Box 10577
University Park Station
Denver, Colorado 80210
ABSTRACT
Injection of dry-sod1um powders Into the Inlet plenum of fabric filters (baghouses)
has been demonstrated to be an effective flue gas desulfurization (FGD) technique.
On a full-scale utility boiler, injected sodium products have been shown to meet New
Source Performance Standards (NSPS) of 702 removal for low-sulfur coal applications.
Higher levels of SO2 removal have also been demonstrated with the injection of
greater quantities of sodium reagent. The amount of reagent required is shown to be
sensitive to the reagent particle diameter. Results are given for commercially
available sodiumblcarbonate, trona, and light soda ash (sodium carbonate). Supply
of sodium reagents is becoming increasingly more abundant as several chemical
commodity suppliers expend both research and venture capital dollars to enter the
potential utility market. To date, sodium bicarbonate reagents have been shown to
provide better sodium utilization than sodium carbonate and trona.
24-1
-------
INTRODUCTION
Background
The process of combining SO2 and particulate control using dry-sodium
reagents is gaining increased utility interest. This paper reviews and
updates events relevant to the use of dry-sodium-based injection into
the inlet of a fabric filter.
The Electric Power Research Institute (EPRI) is interested in evaluating
the advantages and disadvantages of dry-sodium injection because the
technology has the potential to greatly simplify hardware requirements,
and to significantly reduce the capital requirements for both new and
retrofit SO^ control applications. Evaluating the importance of simplicity
is qualitative, but may be weighted heavily by utilities concerned about
operation and maintenance of emission control equipment.
EPRI began studying dry-sodium injection in 1977. To date, it has published
feasibility study results (1J_, bench-scale Investigation results (2J, and
full-scale demonstration results obtained in testing at the 22 KW Cameo
station of the Public Service Co. of Colorado (2). An economic evaluation
of dry-sodium injection is scheduled to be published soon. Work is now being
conducted at EPRI1s Arapahoe Test Facility in Denver to characterize and
assess potential new dry-injection reagents for utility applications. Results
from this pilot scale work and their implication to the full scale application
are the subjects of this paper.
Figure 1 Is a simplified flow schematic of a coal-fired power plant with dry-
sodium injection for SO^ control. Unit processes are much the same as those
required for coal handling: transportation, storage, pulverization, and
injection.
With dry-sodium injection, the sodium reagent is fed into the flue gas stream
(nominally at a temperature of 300°F) ahead of a baghouse and downstream of the
air heater. In the ductwork, the sodium bicarbonate in the reagent particles
decomposes to sodium carbonate (Na^CO^) in a "popcorn" fashion, forming an open
porous microstructure and exposing more particle surface area. The Na^CO^
reacts with the S0£ in the flue gas and subsequently collects along with fly
24-2
-------
ash on the bags in the baghouse as part of the dustcake, further removing SO^
Both the spent reagent and the filtered fly ash on the bags are then removed
in the normal course of bag cleaning and collected for disposal. Typically,
70-80% of total SO^ removal occurs in the baghouse, and the remaining 20-30%
in the ductwork.
Dry-sodium injection offers a number of advantages over alternate SO^ collec-
tion options. These include:
• Capital costs are significantly lower because of the comparative
simplicity of the process.
• Only equipment already in conrnon use at coal-fired power plants
is required.
• No wet slurry or sludge is required.
• The systems are easily retrofitted to boilers equipped with
baghouses.
• Power costs are low.
• Scaling and corrosion are minimal.
• Flue gas reheating is not necessary.
• Pli4 OAS Pfc. COv
-------
Reagent Availability
Sodium reagents that have received the most utility and supplier attention
include nahcolite (naturally occurring sodium bicarbonate, NaHCO^ ). trona
(naturally occurring NaHCO^ 'Na^CO^ ' 2^0). Huge supplies of both
nahcolite and trona are estimated to exist in the United States: over 30
billion tons of nahcolite, and in excess of 85 billion tons of trona.
Nahcolite is the preferred reagnet because of its high bicarbonate content
and, in tests conducted,to date, its high sodium utilization. Permits were
filed in October 1984 to begin conmercial solution mining of nahcolite in
the Piceance Basin in Colorado. Trona is commercially available in large
quantities. Both have been shown to easily achieve 70%-plus SO2 removals.
Given the promise of dry-sodium injection for SO2 control, several companies
have recently expressed interest and/or begun offering processed reagents for
use of utilities. A list of these companies is given below in Table 1. This
list demonstrates the dramatic turnaround in available suppliers of reagent.
Only a few years ago, no firm suppliers could be identified.
This potential for improved product availability has increased the level of
utility interest in a more detailed analysis of dry-injection. Also, and
significantly, these suppliers are developing improved processed reagents
which have substantially lower fractions of inert material. This means less
inert material to be transported and stored by the utility, thereby lowering
overall reagent costs.
TABLE 1
SUPPLIES OF SODIUM - BASED REAGENTS - 1984
Allied Chemical Corporation
Cominco American
Church & Dwight Company, Inc
Industrial Resources, Inc
FMC Corporation
Kerr-McGee Chemical Corporation
Natrona Ind
Stauffer Chemical Company
Tenneco
Texas Gulf Chemicals Company
24-4
-------
PILOT PLANT DESCRIPTION
The dry sodium injection pilot plant (Unit 3A) 1s located at the EPRI Arapahoe
Emissions Test Facility (Denver, Colorado). Flue gas for the pilot plant is
withdrawn at the rate of 1250 scfm from Public Service Company of Colorado
boiler number 4 (100MW nominal load). In the pilot plant, injection of reagent
materials is accomplished by continuously moving sorbent from a small hopper
via a screw feeder into the suction port of an eductor (Figure 2). Approximately
5 cfm of carrier air transports the reagent materials Into the duct at a point
6 ft upstream of the baghouse inlet. Flue gas velocity at the point of injection
is approximately 60 fps. A scale continuously determines feeder system weight
during testing. This parameter is recorded so that a loss of weight over time
may be deteremined. Gas concentrations of S02 and 02 are continuously monitored
at both the system inlet and outlet. Near constant inlet S0£ concentrations are
maintained by an S02 dosing system. NOx levels are determined by using a single
NOx analyzer continuously switched between the system inlet and outlet for
discrete sampling periods. The pilot plant is in full time operation using a
three hour reverse gas fabric filter cleaning cycle. Numerous controls and test
parameters, including gas flow, temperature, and baghouse pressure drop are
continuously monitored.
plue. o^vnoo acf'*.
sc -
Figure 2. Ill us tra t (on of Unit JA pilot plan dry injection
feed System at Arapahoe.
24-5
-------
DISCUSSION OF RESULTS
Sodium Bicarbonate
The technical aspects of dry-sodium injection have been studied by several
researchers over the past few years. The most important is the utilization
of sodium in the reagent. Figure 3 illustrates SO2 removal versus NSR
(normalized stoichimetric ration) for low-sulfur (0.5X), western coal
obtained with nahcolite, at the bench-scale, and at the full scale Cameo
demonstration. S02 removal is shown to improve substantially from the bench-
scale to the full-scale. A major reason for this improved utilization is now
believed to be associated with reagent particle size distribution. The reagent
at Cameo was pulverized much finer than those for the bench-scale research.
Results published by other researchers are similar to those reported on the
bench-scale; the results from the Cameo demonstration reported the highest
sodium utilization of any previous work.
normalized stoichiometric ratio (nsr 1
Figure 3. SO2 removal vs. NSR for oahcolice illustrating improvement in
results reported from the beach-scale to the full-scale.
Tests conducted with lov-sulfur (O.SX), western coal.
» • full-Sole _
Oeoonttrttion
Bench-Set) c
10
0
0.1 o.« a.« 0.1 j.o 1.1 i.« 1.4 i.« 1.0 t.i x.«
24-6
-------
EPRI is currently conducting pilot-scale research to evaluate the effectiveness
of the numerous new reagents now being offered. This work is being conducted
by KVB, Inc at Arapahoe.
Figure 4 illustrates SO2 removal versus NSR (normalized soichiometric ratio)
when injecting sodium bicarbonate of five (5) distinctly differing character-
istic particle sizes. These results have many implications to the application
of dry injection technology.
S02 REMOVAL vs. NSR
SODIUM BICARBONATE
100
MHD-llum
F.G. Temp." 300°F
0 0.5 1.0 1.5 2.0 2.5
Normalized Stoichiometric Ratio , NSR
Figure 4. SO^ renoval vs. NSR for sodium bicarbonate Illustrating
the dependence of SO^ to the HMD (mass mean diameter)
injected into the flue gas duct.
First, SO2 removal is shown to significantly improve with decreasing reagent
particle size and SO2 removal greater than 90% is illustrated for the finest
particle sizes. Note, however, that a significant amount of energy may be
required to pulverize reagents to a size necessary to optimize sodium utili-
zation (SOj removal divided by NSR). Fortunately, these sodium bicarbonate
reagents should be easily pulverized in comparison to the conventionally mined
ores of trona and nahcolite.
24-7
-------
Second, it should be noted that the sodium bicarbonate for these tests was
provided by three reagent suppliers each using different methods to process
their reagent. Even using bicarbonates mined and processed by different
methods, the results were consistent—sodium utilization with bicarbonate
injection is apparently a function of particle size, not manufacturing method.
Die sodium bicarbonate reagents used to produce this data were provided by
Church and Dwight, Kerr-McGee, and Industrial Resources, Inc.
Third, the advantage of smaller reagent size does not appear to be caused by
the increasing specific surface area of the reagent (specific surface is
inversely proportional to particle diameter) but more likely because of better
reagent distribution throughout the baghouse. Figure 5 is an illustration of
the baghouse compartment of the Unit 3 pilot plant. It can be seen that the
required path for a reagent particle to reach the bag thimble in the corners
nearest the inlet is much more treacherous than to reach the bags directly
opposite the inlet duct. This is even more dramatic for larger reagent
particles. To optimize SO^ removal, particles must be small enough to follow
the gas stream and distribute evenly across the tubesheet to each bag inlet.
The result of maldistribution is that some bags become enriched with reagent
and some are starved. This appears to be the mechanism that increasingly
degrades SO2 removal as reagent particle diameter increases (as illustrated
in Figure 4).
Defuse*, PLATES
Figure 5. Illustration of the baghouse compartment of
the Unit 3A pilot plant.
24-8
-------
These findings are important when considering injection design criteria, reagent
pulverization requirements and techniques, and possibly baghouse inlet transition
and hopper designs. For example, the requirements on a pulverizer for dry-
injection might be slightly more demanding (i.e., finer particles required) when
the boiler load is reduced and the gas flows are lessened. The carrying velocity
into a baghouse compartment hopper needs to be sufficient to keep a reagent
particle from falling out; and the reagent distribution needs to remain somewhat
uniform across the tubesheet. This underlines the importance of a good aero-
dynamic modeling effort relating to the injection system and baghouse designs,
especially at lower boiler steam loads where flue gas flow rates are substantially
lower than design specifications.
Flue gas temperature has been demonstrated to affect reagent utilization of sodium
bicarbonate as temperatures range between 260°F and 340°F. This is shown in
Figure 6 where SO2 removal is plotted versus NSR for a reagent injected at 300°F
and 330°F (Figure 6a) and for another reagent injected at 260°F and 300°F
(Figure 6b). These temperatures generally span normal operating temperatures for
utility baghouses. These curves also show that reagent particle specification
can be matched with baghouse inlet temperature specifications as a trade-off in
system design considerations.
SOj RErtDVAL vj. NSR
SODIUM BICARBONATE, mo-Mua
SOj REMOVAL v». NSR
SOOIIX BICARBONATE, mo-llum
130 F
300 F
?60 F
0 0.5 1.0 i.S Z.O 0 J-" *-0
Normalized S to left trie Ratio, NSR Nomillitd Stoichiometric Ratio. NSR
Figure i (i I b). SOj renttval v». NSR for Uo sodlun bicarbonate r«4genti llliutrating the dependence of S0?
removal lo Hue gai temperature.
24-9
-------
Trona
Trona was acquired for this project from Wyoming by TenneCo and from California
by Kerr-McGee. The materials were pulverized into size fractions and treated
in the same manner as done previously with the bicarbonate reagents. Although
characteristic size dimension measurements have not been completed, the reagents
were provided 1n distinct size fractions allowing tests to be performed that
generated the data shown in Figure 7. These results follow the same trend as
previously shown. Note, however, the significant differences in sodium utili-
zation between the bicarbonate (Figure 4) and trona (Figure 7) reagents. Again,
utilization is calculated as SO2 removal divided by NSR. Until more complete
characteristic size information is available, the size fractions indicated will
suffice as relative indicators of reagent size.
Figure 8 is a graphic of the affect that flue gas temperature has on SO^ removal
when injecting trona. Although a more complete series of tests are needed to
identify a perferred operating temperature, it is important to note that Increas
ing the flue gas temperature decreases the effective SOg removal of trona. This
is directly opposite the trend shown for sodium bicarbonate.
S02 REMOVAL vs. NSR
TRONA
100 -
1
O
CO
-400
5 60 <
/X /
1
J' s "20° Mesh#
OL
w s
N
O . ^
-325 Mesh* yT
ist 40 *
/
' F.G. Temp-* Z65°F
ro
O O
i i
•Estimate of particle sije
901 thru stated mesh
0 0.5 1.0 1.5 2.0 2.5
Normalized Stoichiometric Ratio, NSR
Figure 7. SO^ removal vs. NSR for Trona Illustrating the
dependence of SO^ removal to the HMD of the reagent.
24-10
-------
100
80 "
60
40 -
20 -
S02 REMOVAL vs. NSR
TRONA
F.G. Temp-* 265 F
F.G. Temp, ¦ 300 F
Note: Estimate of particle size
90% -400 Mesh
o 0.5 1.0 1.5 2.0
Normal lied Stoichiometric Ratio, NSR
Figure 6. SOj removal vs. NSR for Trona Illustrating the dependence
of SO^ on flue gas temperature.
Soda Ash
Previously reported results from tests that directly inject soda ash (sodium
carbonate) into the flue gas demonstrate S09 removal below 20% for NSR values
greater than two (NSR > 2) . However, at Arapahoe, sodium carbonate injection
has demonstrated substantially more S02 removal than previously reported. The
results shown in Figure 9 were produced by injecting a "light ash" provided by
Kerr-McGee. "Light ash" reagents produced by other suppliers are also scheduled
to be tested. Shown in Figure 10 is the collage of results from the injection
of sodium bicarbonate, sodium carbonate, and trona. The reagents were provided
by a single supplier, pulverized to a similar particle size, and each injected
at 300°F. It is of interest to observe that mathematical addition of S0£
removal from a one-to-one molar mixture of sodium bicarbonate and sodium carbonate
(as in the chemical formula for trona) produces an SO2 removal nearly equal to
that empirically obtained when injecting trona.
24-11
-------
100
80
- 60
§
oc
CM
5» 4o
20
0
Figure 9. SO^ removal vs. NSR for "light" sodium carbonate.
100
BO
™ 60
>
cc
eg
S 40
20
0
Figure 10. S02 retiwval vs. NSR for three reagents produced from
a single feed stock by Kerr-McGee.
S0? REMOVAL vs. NSR
"LIGHT" SODIUM CARBONATE INJECTION
Note: Estimate of particle size
90X -400 Hesh
T
0-5 1.0 1.5 2.0
Normalized Stoichiometric Ratio. NSR
S02 REMOVAL vs. NSR
f F.G.Temp.- 300°F
A j
/ S ~ Bicarbonate
/ ¦ Trona
• Soda Ash (Light)
Note: Estimate of particle size
90% -400 Hesh
0 0.5 1.0 ' 1.5 2.0
Normalized Stoichiometric Ratio, NSR
24-12
-------
SUMMARY
The few and simple unit processes required to successfully operate a dry-
injection system make the technology an attractive, economic alternate for
SO2 control on coal-fired power plants. The interest of several large,
potential suppliers in entering the market is especially promising in terms
of making available better performing and lower cost reagents. Dry-sodium
Injection systems designed to optimize reagent utilization will provide
benefits in all areas of the technology, minimizing reagent and solids hand-
ling requirements, hopefully lowering levelized costs, and reducing the sodium
levels in the fly ash waste product.
24-13
-------
REFERENCES
1. Bechtel Corporation, "Evaluation of Dry Alkalis for Removing Sulfur
Dioxide from Boiler Flue Gases," FP 207, October, 1976, Electric
Power Research Institute, Palo Alto, CA.
2. L. J. Muzio, J. K. Arand, "Bench-Scale Study of the Dry Removal of
SO2 with Nahcolite and Trona," CS-1744, March 1984, Electric Power
Research Institute, Palo Alto, CA.
3. L. J. Muzio, T. W. Sonnichsen, "Demonstration of SO- Removal on a 22
MW Coal-Fired Utility Boiler by Dry Injection of Nancolite," CS-2894,
Vol. 1, March, 1983; Vol. 2, June 1984, Electric Power Research
Institute, Palo Alto, CA.
4. R. W. Scheck, D. J. Naulty, A.E.E. Gallagher, R. P. Grimri, D. A.
McDowell, R. J. Keeth, J. E. Miranda, "Economic Evaluation of Dry
Injection FGD Technology," RP-1682-1, Electric Power Research
Institute, Palo Alto, CA.
-------
DRY INJECTION SCRUBBING OF
FLUE GASES BY THE SHU PROCESS
M. Schu'tz, J. Schumacher, M. Esche, and H. Igelbiischer
Saarberg-Holter Umwelttechnlk GmbH (SHU)
Hafenstrasse 6
D-6600 Saarbriicken, West Germany
ABSTRACT
The SHU dry sorbent injection process has been installed at the Ruhr recycling
center refuse incinerator to remove stack gas S0X, HC1, and HF. The incinerator
produces 60,000 to 142,000 m^/h of flue gas with typical or design concentrations of
70 to 275 ppm S0X, 300 to 1,080 ppm HC1, and 20 ppm HF. Prior to gaseous pollutant
cleanup, particulates are removed in an electrostatic precipitator, and the
combustion gases are cooled to 200 to 220°C by heat exchangers. Dry scrubbing
occurs In a contact section fed by calcium hydroxide and steam at a feed ratio of
0.5 kg steam/kg Ca(0H)2- The feedrate 1s metered based on the pollutant level in
the treated gas. The flue gas, Ca(0H)2 and steam are swirled to enhance contact and
absorption. Following absorption, spent sorbent is removed by a fabric filter. The
unit entered service in 1982. Test results show SO? removal efficiencies of 55 to
85 percent, depending on inlet concentration, and Hcl removal efficiency usually
over 90 percent. Selected solids analyses showed approximately 37 percent calcium
hydroxide consumption corresponding to a utilization ratio of 1:2.7.
INTRODUCTION
The SHU dry injection gas scrubbing process has been developed and applied by
Saarberg-Holter Umwelttechnik GmbH which was formed in 1974 as a joint property of
the power plant operator Saarberg AG and the engineering firm Hoiter GmbH, Gladbeck.
Saarberg AG is an integrated energy firm operated by the Federal Republic of Germany
and the Saarland. They are the second largest producer of hard coal in West Germany
and their power plants account for 50 percent of electricity generated in the
Saarland. Holter GmbH, Gladbeck is an engineering and construction company
supplying environmental control systems, coking plants, and foundries. The SHU
process 1s marketed by four licensees or partners:
• S-H-L, Saarberg-Hol ter Lurgl GmbH, Saarbriicken, West Germany
• Davy McKee Corporation, Lakeland, USA
• Simmering-Graz-Pauker AG, Austria
25-1
-------
• BV Koninklijke Maatschapplje "de Schelde", Netherlands
• Groupement with Lurgi S.A., France
Table 1 lists current and planned Installations of SHU systems. This paper
summarizes the application to the Ruhr recycling center (RZR) which is the first
large-scale incineration installation of the dry scrubbing process. The process was
previously tested in a pilot plant at the waste incineration center at Ebenhausen in
Bavaria.
FACILITY
RZR produces electric power and district heating by burning approximately
100,000 tonnes/year of domestic and bulk refuse in a stoker-fired furnace. Incoming
bulk refuse is shredded by two cylindrical cutters and stored in a bunker from which
it is transferred by crane, together with domestic refuse, to the stoker-fired
furnace. The refuse is extremely heterogeneous with wide fluctuations in heat
content and composition. In the furnace, these fuel fluctuations produce variations
in furnace temperature (held below 800°C), flue gas concentration, and particulate
loading. Surplus heat in the stoker combustion gases is used in a heat recovery
boiler to produce up to 57 tonnes/hour of steam (I). Solid particles are removed
from the boiler stack gas in an electrostatic precipitator and the gas is then
passed to the dry scrubbing process.
The dry scrubing system is designed in two sections, each with a cooler, contact
section, and fabric filter (Figure I). The design specifications are listed in
Table 2. Upstream of the contact section, the boiler stack gas is cooled to
approximately 200°C by a secondary cooling system to protect downstream filters.
Calcium hydroxide and steam are then injected into the stack gas in the contact
sections in proportion to the quantities of gaseous compounds to be removed, which
are mainly sulfur dioxide, hydrogen chloride, and small amounts of hydrogen
fluoride. The spent sorbent from the reaction is collected by a fabric filter and
sluiced out of the system.
Contact Section
Each contact section consists of three contact tubes each fitted with two flow
nozzles, venturi throats, and an impactor grate. Dry calcium hydroxide and steam
are injected into the cooled stack gas via binary jets. The calcium hydroxide is
introduced on a stoichiometric basis via a metering device which is described in
greater detail below. Table 3 shows an average analysis of the sorbent. The volume
of steam was set at 0.5 kg steam/kg sorbent based on parametric tests. In the
subsequent venturi tubes and impactor grate, the gas, calcium, and steam are swirled
so that most of the gaseous components to be removed are bonded to the partially
condensing steam and the surface of the calcium particles.
Steam Input. The addition of steam via the two flow nozzles improves the
distribution of the calcium in the stack gas flow and also raises the level of
abatement for the gaseous components that are to be removed.
The water/steam cycle produces superheated steam at 320°C (608°F) and 32 bar
(463 psi). Before it enters the two flow nozzles the steam is reduced to 16 bar
(232 psi) and approximately 240°C (464"F). Total steam consumption is restricted to
524 kg (1,153 lb) per hour. During unit startup, steam for stack gas scrubbing is
taken from the oil-fired auxiliary steam generator.
25-2
-------
Sorbent Metering. The calcium hydroxide silo with a capacity of 150 m3 (5,300 ft3)
is filled by dumper trucks using a pneumatic conveyor system. The silo extractor
unit Is designed as a three-part unit in which each section has its own independent,
mechanically operated vibrating extractor. A pneumatic metering and conveyor unit
is installed beneath the rotary valve of each extractor. The feed pipes to the
binary jets are positioned downstream of each metering and conveyor unit. The
metering precision 1s percent.
To avoid plugging as a result of condensation during pneumatic transport, the pipes
are Insulated and can be heated electrically. The surfaces of the rotary valves
have been specially treated at the points where deposits may form.
Viton B or rubber is used for the gaskets. The straight pipe sections for the
pneumatic calcium conveyor are made of mild steel and the hinge fittings of
thick-walled plastic. This was the only way to meet the requirements specified for
operational life. To avoid static charge buildup 1n the plastic, ground wires have
been inserted which are connected to the adjoining steel pipes.
Filter Pockets
The two pressure shock filters each consist of two vertically aligned rows of filter
pockets made of Teflon. The filter pockets are stretched over aperture frames of
1.5 by 1.4m. Table 4 gives the dimensions of the filter assembly.
Filter Cleaning. An eight-bar (116 ps1) compressed air system is provided to clean
the filters. Only four filter pockets are cleaned at any given time, two from each
row. There must be at least four filter pockets operating normally between the two
1n the row being cleaned to maintain a steady filtering process.
Cleaning with compressed air blasts is controlled in such a way that the pressure
drop loss remains constant within a certain operational range.
The pressure drop 1n a filter depends of course, among other things, on the strength
of the filter cake (adhesive calcium layer) and the face velocity of the fabric or
gas flow. In this instance, filtering at a pressure loss of 15 mbar (0.22 ps1) has
proved optimal.
Extraction System. The particles removed from the pressure shock filter are carried
by a scraper chain conveyor to a rotary valve and thence removed from the pressure
area. The reacted lime is transferred to a s11o via a screw mounted traversly to
both filters.
OPERATION
Before the chemisorption unit startup, both filters, including both the upstream
multiple contact sections and a part of the downstream treated gas duct, are
preheated to working temperature. This preheating 1s achieved by keeping the air 1n
the unit 1n circulation by a fan and heating 1t up 1n a steam-heated heat-exchanger.
Measuring points in each of the two filters monitor the temperature and show when
the working temperature has been reached.
25-3
-------
Sorbent Metering
The concentrations of hydrogen chloride and sulfur dioxide are measured in the
treated and untreated gas. Normally, the level of pollution in the untreated gas,
the volume of the gas flow, and the stoichiometric ratio determine the amount of
sorbent needed.
The measuring devices currently available for gaseous components react sluggishly
and cannot therefore detect rapid fluctuations 1n level. Changes 1n the flow of
stack gas and deviations in the measurements of the metering devices make it even
more difficult to adjust to an optimal supply of sorbent without exceeding the
permitted concentrations in the treated gas stream. Consequently, values for the
treated gas which fall below the maximum permitted levels are superimposed on the
sorbent metering. If the permitted levels are exceeded, additional calcium
hydroxide is briefly injected into the absorption system.
Operating Results
Since the unit entered into service 1n 1982, various test series have been taken by
SHU and other independent groups. The temperature of the untreated gas normally
ranges from 200 to 220°C (390 to 430°F) after the cooler and the level of moisture
is 40 to 70 mg/m^ (STP). The concentrations of SO? recorded in the untreated gas
have been between 200 and 800 mg/m^ {70 to 275 ppm). The operating results cover a
relatively wide range of 55 to 85 percent removal efficiency as shown in Figure 2.
All gas outlet concentrations fell within the range of 90 to 160 mg/m^ (30 to
55 ppm).
Experimental results at other waste Incinerators showed (2) that SO3 comprises about
1 percent of total sulfur oxide emissions. Comparing the process described here
with other dry Hme-based processes for sulfur removal, it can be seen that
injecting steam in addition to calcium hydroxide considerably increases the SO2
removal efficiency. In some applications, this process is competitive with
wet-scrubbing.
There was no discernable Influence on SO2 removal efficiency due to fluctuations of
hydrogen chloride between 500 and 1,750 mg/m^ (300 to 1,080 ppm) in the untreated
gas. The removal efficiency for hydrogen chloride 1s usually over 90 percent so
that levels in excess of 85 mg/m^ (52 ppm) were rarely found in the treated gas.
The hydrogen fluoride in the treated gas was well below 1 mg/m^ (1 ppm).
In addition to obtaining high removal efficiencies, the program also sought to
obtain a high level of sorbent utilization. This Is generally defined by the ratio
of absorbed mole to Ca mole compared to stoichiometric. Since the present system
exhibits fluctuations 1n the volume of gas and concentrations of pollutants, 1t 1s
particularly difficult to specify a stoichiometric ratio from the abatement level of
the gaseous components.
Accordingly, some analyses were made on the solid matter and the quantity of calcium
hydroxide consumed. These analyses showed that approximately 37 percent of the
calcium hydroxide had been consumed, which represents a utilization ratio of 1:2.7.
Domestic refuse contains heavy metals such as cadmium, lead, mercury, zinc,
vanadium, chrome, etc. which are present In the flue gas and flyash. It Is
therefore advisable to use a fabric filter for fine particle removal. By cooling
down the gas to 200"C (392°F) most of the gaseous heavy metal compounds condense
onto the flyash and sorbent which are removed In the fabric filter. The particulate
concentration behind the fabric filter is less than 5 mg/m^.
25-4
-------
Cost of Operating Materials
To meet design specifications, the gas scrubbing unit requires 400 kg calcium
hydroxide and 200 kg steam per hour. Including transport, the price is DM 150/tonne
($50/tonne) for calcium hydroxide and DM 28/tonne ($9.30/tonne) for steam for power
producers. The corresponding costs per hour are DM 60 ($20/tonne) for the calcium
hydroxide and only DM 5.60 ($18.70/tonne) for steam which means that the proportion
of the costs accounted for by the steam is relatively low (8.5 percent).
Storage of Used Lime
Given the heavy metals and chlorine compounds in the used lime, the operators opted
to store it in a special dump near the recycling center. In general, in Germany
there are local bylaws covering the dumping of waste products. The main criteria
are the proportion of water-soluble components, the tendency of the dumped material
to become airborne, the rate of seepage, metal content, and hygroscopicity.
REFERENCES
1. B. Neukirchen, Energie aus Abfall, Inbetriebnahme der beiden Verbrennungsanlagen
im RZR YGB Kraftwerkstechnik 64, Heft 5, 1984, Selte 448/452.
2. P. Davids, K. Gerhards, and W. Brocke, "Die derzeitige und zukunftige
Luftverunreinigung durch Mul1verbrennungsanlagen — Emission und
Emisslonsverminderung — Staub-Relnhalt, Luft 33 (1973), Nr. 12, Selte 483-489.
25-5
-------
gas inlet
steam
condensate
1 cooler
pressure air
©
H><]®
2 adsorber
3 fabric filter
4 fan
5 silo calcium hydrat
6 dosing system
7 screw conveyer
8 pneumatic
9 byproduct silo
10 preheater
Figure 1. Schematic of Dry Gas Cleaning at Ruhr Recycling Center
% 100
90
80
70
60
50
40
c -30
£ 20
(0
>
o
£
£ 10
o
en n
/i
>1
0,5 >
-------
1939
1988
1987
1987
1967
1587
1985
1985
1985
1984
1984
•983
1983
Table 1
FLUE GAS DESULFURIZATION PLANTS WITH THE SHU PROCESS
client
Application
FT33 capacity
Gyp suit use
RWE
Power-station Neurath
.EEWAG, West-Berlin
Pcver-Staticn Reuter
EWE
Pcver-Stdtion Neurath
Umlar*Jverba«d Frankfurt
AVA Osthafen
Preussische Elektrizitats-AG
Hannover,
Power Station Heyden IV
3EWAG, West-Berlin
Po-er Station Reuter
City Frankfurt
KVA Frankfurt
tvordweststadt
S tactreinigung
Numberg
Von Roll AG
Switzerland
Kraftwerk Sexbach Verwaitungs-
aesellschaft mbH
Ilse Bayemwerke Energie-
anlager. OrbH
City Frankfurt
MVA Frankfurt
NonJweststadt
BEWftG, West-Gerlin
Pc^er Station Lichterfelde
Saarberguerke AG
Saarbriicken
Pwer Station Weiher III
12oo MW
lignite-fixed boiler 12oo MW
3 oo W
coal-fixed boiler
9oo MW
lignite-fired boiler
municipal waste
incinerator
800
coal-fixed
boiler
3 00 fW
coal-fixed boiler
runic ipal
waste
incinerator
monicipal waste
incinerator
Ofenlinie 4
municipal waste
incinerator
7 So MW
coal-fixed
boiler
municipal
waste
incinerator
15o w
oil-fired boiler
7o7 m
coal-fired boiler
3 00 MW
9 00 MW
3 x
71,5co sefm
yes
yes
yes
800 KW
3oo W
2 x
74,6oo scfrr,
71 ,Soo scfcrt
yes
37,5oo scfm (partly absorp-
tion)
2 60 MW
2 x
74,600 scfm
150 w
125 MW
yes
yes
yes
25-7
-------
Start
1982
1982
1962
1982
1982
1979
1979
1976
1974
1972
Table 1 (Concluded)
FUEL GAS DESULFURIZATION PLANTS WITH THE SHU PROCESS
Client
Application
FCD capacity Gypsum use
Widmer 1 Ernst AG
.Switzerland
KVA Schwandorf
irjiicipal
waste
incinerator
58, Soo scfm
BEWAG, West-Berlin
Pcwer-Station Lichterfelde
1 5o MW
oil-fired boiler
15o W
yes
Widrer 4 Ernst AG
Switzerland
RZR Her tan/Germany
Widrer & Ernst AG
Switzerland
RZR Herten/Gemany
Saarbergwerke AG,
Saarbrucken
Power-Station Vblklingen
BayT.Landesamt fur
Unveltschutz
MVA Ebenhausen
Saarbergwerke AC
Saarbriicken
Power Station Weiher III
waste
product
incinerator
municipal
waste
me inerator
23o
coal-fired boiler
[fluidized bed)
special waste
inc meratcr
(dry prototype
scrubbing)
7o7 MW
coal-fired
boiier
77,000 scfm
1 5, Soo scfm
23o W
6,2oo scfm
225 fW
(mainly CaCl,,
CaFj )
(mainly CaCl,",
CaF, )
yes
(mainly CaCl, ,
CaF, )
yes
Siidwestdeutsche Femwarme QnfcH
Saarbrucxen
Saarbergwerke AG
Saarbriicken
Power Station Weiher IX
5tadtv«rke Solingen
MVA Solingen
5 t/h
municipal
incinera tor
15o W
coal-fired
boiler
5 t/h
municipal
waste incinerator
19,ooo scfm
78,ooo scfm
1,9oo acfm
(mainly CaCl,,
CaF, )
yes
(mainly CaCl,,
CaF, )
25-8
-------
Table 2
DRY ABSORPTION SYSTEM SPECIFICATIONS
Stack gas flowrate
Temperature of stack gas prior to cooler
Temperature of stack gas in filter
Concentrations in untreated gas
SO2 + SO3
HC1
HF
60,000 to 142,000 m3/h
(2.12 x 106 to 5.02 x 106 ft3/hr)
483 to 573 K
(410 to 572°F)
443 to 513 K
(338 to 464°F)
(342 ppm) 1,000 mg/m3 (as SO2)
(920 ppm) 1,500 mg/m3
(20 ppm) 20 mg/m3
Table 3
AVERAGE ANALYSIS OF CALCIUM HYDROXIDE
Ca(0H)2
Mg(OH)2
C02
Si02
A1203
Fe203
Mn304
S03
H20
Of which H20-free
Combustion loss
Particle size 98 percent less
than 0.09 mm (about 170 mesh)
Bulk density
90 to 94 percent
1.4 to 2.2 percent
Maximum 2.0 percent
1.0 to 1.7 percent
0.45 to 0.65 percent
0.25 to 0.40 percent
0.05 percent
0.05 to 0.2 percent
23.0 to 24.0 percent
0.5 to 1.0 percent
23.0 to 24.0 percent
0.4 kg/dm^
25-9
-------
Table 4
DIMENSIONS OF THE FILTER ASSEMBLY
Length
Width
Height (excluding
treated gas duct)
Number of filter
pockets (total)
Number of filter
pockets per double
row
Number of filter
pockets per row
Total filter surface,
approximate
16m (52.5 ft)
8m (26.2 ft)
6m (19.7 ft)
816
408
204
3,000 m3 (32,300
25-10
-------
FLUE GAS DESULFURIZATION BY COMBINED FURNACE
LIMESTONE INJECTION AND DRY SCRUBBING
L. E. Sawyers and P. V. Smith
Babcock A Wilcox Company
Research S Development Division
Alliance, Ohio 44601
and
T. B. Hurst
Babcock 4 Wilcox Company
Fossil Power Divisi on
Barberton, Ohio 44203
ABSTRACT
Furnace limestone injection with dry scrubbing offers a viable economic alternative
to wet scrubber systems for flue gas desulfurization. The combined technology is
most promising from a technical and economic standpoint 1n application to eastern
high-sulfur coals, where S0? reduction requirements are most stringent. Combining
the two systems can offer required SO- reduction in excess of the sole use of either
system. Also, combined use of the two systems can represent significant savings in
reagent costs over more expensive lime reagent- This is because low-cost limestone
is injected into a furnace for calcination to lime and collected lime and ash
materials are recycled and employed as the principal reagent in the dry scrubber
system.
In pilot plant tests sponsored by the Department of Energy, which investigated the
combined process as applied to eastern high-sulfur coals, various furnace injection
methods, calcium-to-sulfur stoichiometric ratios (Ca/S), furnace load, and rear-
furnace temperatures were studied. Results indicated potentially high SO- removal
and a cost-effective process with a combined optimized system. These test results,
in addition to Babcock & Wilcox research and development experience with the two
technologies (separately and in combination), are reviewed in this paper.
INTRODUCTION
Many processes have been developed to remove sulfur dioxide from utility boiler flue
gases. One of the more successful processes is the wet scrubber system. Furnace
limestone injection and dry scrubbing are also two viable processes for flue gas
desulfurization. Like wet scrubbers, their development has grown in response to
legislative measures, including the new source performance standards (NSPS) imposed
for emissions in 1971. Interest in both systems is increasing in an attempt to
develop alternatives to the economic and process-related problems of wet scrubbers.
26-1
-------
The limestone injection process injects limestone into the furnace of a utility
boiler. Flash calcination occurs to produce lime that, in turn, undergoes reaction
with sulfur species in the flue gases. Dry scrubber processes handle flue gases
exiting the boiler air heaters through spraying of an alkaline slurry.
Though applied to a wide variety of coals, furnace limestone injection has not yet
been proven to reduce SO- emissions to NSPS requirements. Dry scrubber systems have
successfully been used on western low-sulfur coal for SCU removal at NSPS require-
ments. However, their application to eastern high-sulfur coals have not been
proven. Combining the two technologies has the potential to produce flue gas desul-
furization at and above the requirements for SC^ emissions for a complete range of
coals. Both systems enhance and complement one another. An optimized combined sys-
tem with use of recycled ash material as dry scrubber slurry could be both a tech-
nical and economic challenger to wet scrubber systems for flue gas desulfurization.
Though similar in some respects to the other, limestone injection and dry scrubbing
have developed independently. Babcock S Wilcox (BAW) has been involved extensively
in developing both technologies, as well as their combination. A review of B&W
research and development programs with results are presented in this paper.
TECHNOLOGICAL REVIEW
Furnace Limestone Injection
The technology of injecting limestone in a furnace for flue gas desulfurization has
been under development at BSW for nearly 20 years. Work began in the late 1960s
with pilot plant studies followed by full-scale testing in the early 1970s. With
the growth of the Environmental Protection Agency's (EPA's) limestone injection
multistage burner (LIMB) technology in the early 1980s, SAW again became involved
with limestone injection through in-house and contract studies. Before reviewing
this development work and discussing technical findings relative to limestone
injection technology, an overview of the limestone injection mechanism for furnace
limestone injection is presented.
Furnace limestone injection for SO- reduction involves limestone (CaCO.,) calcination
to lime accompanied by lime reaction with sulfur dioxide (SO-) in an oxidizing atmos-
phere to form calcium sulfate (CaSO^). This series of reactions is represented as:
CaCO^ --> CaO + CO^ (calcination) (1)
CaO + SO^ + 1/2 0^ CaSO^ (sulfation) (2)
In a reducing atmosphere produced through use of multistage low-NO burners,
following limestone calcination, sulfur capture proceeds through rlaction of
hydrogen sulfide (H-S) and/or carbonyl sulfide (COS) to produce calcium sulfide
(CaS) by the equations:
CaO + H£S —> CaS +¦ H£0 (sulfide formation) (3)
CaO + COS —> CaS + C0£ (sulfide formation) (4)
All of these reactions, their rate, degree of completion, and stability of the re-
action products, are influenced by stoichiometry (Ca/S molar), limestone composi-
tion, burner design, burner operation, temperature at point of injection, flue gas
residence time in the furnace, degree of mixing of limestone with flue gases,
particle surface area, and CO^, Og, and SOg gas partial pressure, among others.
?fi-7
-------
Probably the most influential factor For SC^ adsorption by limestone injection is
time and temperature history of the reagent in the flue gases of the furnace. The
theoretical dissociation temperature of limestone is 1652"F for ideal conditions,
according to Schwarzkopf U). Limestone calcination at boiler conditions may range
up to 22Q0°F. During reaction, limestone particles expand before decomposition.
Calcination begins first at the surface and continues inward, toward the center of
the particles, increasing pore volume and surface area but leaving particle volume
unchanged. At the completion of calcination, particles achieve the largest pore
volume and surface area with unchanged particle volume. This characteristic of lime
particles is termed "soft burned" and is the most reactive state of lime (2^. Lime
particles having a sufficiently small particle size distribution will have the
greatest reactivity. As temperature and/or residence time is increased, sintering
of the particles begins, also termed "hard burned" or "dead burned," and particle
volume, pore volume, and surface area are decreased. At. this state, lime particles
are least reactive.
Calcined lime reactions with sulfur compounds are also largely dependent on time and
temperature. In an oxidizing atmosphere, calcium oxide and sulfur dioxide are most
reactive at about 1800°F U), but reactivity occurs throughout a temperature range
from 1400s to 23Q0eF. At utility boiler furnace conditions, the reaction may occur
appreciably between 1800° and 230Q"F at 3i excess 0„ and with SO- concentrations in
the range of 100 to 2500 ppm (£). The oxidizing product, CaSO., decomposes rapidly
as temperature is increased above 2300"F. The cause of this decomposition has been
studied by many. Baker and Attar (5J and Dewing and Richardson (6J reported a
transformation in crystalline structure that is less stable than the initial struc-
ture. It was also determined by Baker and Attar (6) and Besmann, et al. (7) that
CaSO^ is more reactive with metal oxides and reducTng agent impurities at high temp-
eratures, which serve to hasten decomposition. These impurities are components
within the coal fly ash. For reducing atmospheres, lime is most reactive with H?S
and COS at about 2300°F, according to Borgwardt, et al. (8^. However, Turkdogan, et
al. (9) determined the temperature to be 2000°F for the maximum limiting sulfide
solubility of lime in equilibrium with CaS. Like CaSO., the CaS reaction rate may
decrease below 1800°F, and products may decompose above 2300°F.
B&W has studied many of the factors that influence limestone injection; the factors
include: additive properties, stoichiometry, injection point, particle size, S0„
concentration, flue gas oxygen level, particle residence time, catalyst, and
additive recycling. They were studied in the late 1960s under contract to the
National Air Pollution Control Administration (1£). A total of 415 tests using
seven coals and 129 different sorbents were run employing a 2- to 20-lb/hr
pulverized-coal pilot plant. The results are summarized:
• SO- reductions as high as 71$ at a 3.8 Ca/S molar stoichiometric ratio
were measured. Average values at Ca/S stoichiometric ratios of 1.0 and
2.0 were 21% and 34%, respectively.
• Marl was the most reactive additive at a 3.8 Ca/S stoichiometric ratio,
2300°F injection temperature and 2-second residence time.
• Injection in the mid to upper furnace at about 2500°F was the most
effective location.
2
• Increasing particle surface area up to 2000 cm /g improved performance.
¦ Additive performance increased with increasing SO^ ppm concentration.
• Decreasing combustion excess, air decreased SOg removal.
26-3
-------
• Furnace slagging decreased with increasing stoichiometric ratio above
2.0 Ca/S.
0 Sintering strengths of fly ash mixed with the additive were reduced
below those of the fly ash alone at initial tube bank deposition;
however, strengths increased with time.
• Precipitator performance was projected to decrease about 20% with
additive injection, as observed from high-dust-resistivity
measurements.
These results were applied to demonstrate the technology on a full-scale utility
boiler in the early 1970s. The program was sponsored by the EPA and involved B&W in
the design and installation of a commercial limestone injection system for a 150-MW
unit at TVA's Shawnee Station (U.). Results of the five-phase effort were discour-
aging; only about 261 S0? removal at a Ca/S stoichiometric ratio of about 2 was
achieved at upper furnace injection. In addition, process problems, such as plug-
ging of the boiler reheater and ash hydroveyor, decrease in electrostatic precip-
itator performance, and changes in the characteristics of waste products handled by
the waste disposal system, further complicated efforts. Development of limestone
injection technology ceased at B&W following this program.
In the early 1980s, B&W again began studying furnace limestone injection through an
in-house program. The process was now being investigated for combined NO and S0X
reduction utilizing B&W 1ow-NO burners. A B&W in-house study (12) investigated
limestone injection locations for SO2 reduction through the use oT the B&W Alliance
(Ohio) Research Center (ARC) 6 mi 11ion-3tu/hr combustion and fuel preparation facil-
ity with four dual-register, 1ow-NO burners. Test results showed that injection in
the upper furnace region provided tne greatest sulfur capture.
Presently, B&W is engaged in other furnace limestone injection work. These efforts
include pilot plant studies for application to large-scale utility boilers, small
(less than 150-MW) utility boiler applications and demonstration, and in-house
studies. Along with the in-house programs, B&W is investigating application of the
technology to utility boiler units with cyclone burners and to pulverized-coal, 1ow-
NO units equipped with dry scrubber systems.
Dry Scrubber Systems
The technology of dry scrubbing for flue gas desulfurization has been progressing
since the late 1970s. Most investigations have been applied to low-sulfur western
coals. B&W's development efforts have advanced since this time on all levels, from
pilot plant studies and large-scale demonstration programs to design, engineering,
and construction of commercial units. Two commercial units are presently in startup
operati ons.
The dry scrubbing process is fairly straightforward. Process equipment includes a
dry scrubber reactor and a particulate collector. Sulfur dioxide removal is facil-
itated through dispersion of a finely atomized alkaline slurry in the reactor
chamber, where it is mixed with a stream of flue gases flowing from the air heaters
of a coal-fired utility boiler. As particles in the slurry dry from the sensible
heat of the gases, S0~ molecules diffuse into the thinning water layer of the part-
icles, and sulfur capture occurs. The reaction product is collected or disposed of
through the particulate collector, completing the process.
25-4
-------
Numerous reagents have been employed In dry scrubbing, including lime, soda ash,
trona, and nahcolite. B&W's dry scrubbing experience has centered on lime due, in
part, to its availability. Lime (CaO) is used as a reagent in a water slurry, which
produces calcium hydroxide CalOH)^ through an exothermic reaction. The product
reacts with SCL in the flue gas to yield calcium sulfite (CaSt^) as the final
principal product:
CaO + H20 --> Ca(0H)2
Ca(OH )2 + S02 --> CaS03 + H^O
As the lime particles in the flue gas flow through the dry scrubber reactor, the
rate of reaction decreases as the particles dry.
Following the drying process, when the particles are dried to their equilibrium
moisture content, the rate of reaction is greatly decreased. Following the dry
scrubber reactor, lime particles in the flue gases are collected in the particulate
collector. Additional reactions occur in the particulate collector through direct
contact as flue gas flows through ash or near the unreacted lime deposits.
A number of variables influence the dry scrubbing process performance technically
and economically. These include CaO/SO- stoichiometric ratio, use of recycled
materials, reactor approach to saturation temperature, S0? concentration, reagent
type, and reactor inlet temperature, among others. All or these variables play
important roles in optimizing the system for application to low- and high-sulfur
coals.
Stoichiometric ratio (the quantity of reagent used in the system) largely influences
S0„ removal. High stoichiometric ratio aids S0? capture. In addition, operating
the dry scrubber at a high relative humidity ennances sulfur capture. Sulfur cap-
ture may also be enhanced through use of ash recycle, which increases reagent utili-
zation by making use of unreacted calcium hydroxide that has passed through the
reactor; it also increases the utilization of coal ash alkalinity.
B&W became involved with the dry scrubbing concept in late 1977. At that time,
Basin Electric Power Corporation was soliciting bids for a dry scrubber system at
its Antelope Valley Unit No. 1. The fuel for the plant was a low-sulfur coal that
had a highly alkaline fly ash.
To qualify as a .bidder for the Basin project, BiW had to build and operate a pilot
facility to demonstrate the technology. The pilot unit was located at Basin Elec-
tric's Neal Station in Velva, North Dakota and was sized to handle 8000 ACFM of flue
gas. The pilot featured a horizontal-flow reactor. Lime slurry was introduced into
the dry scrubber through a Y-jet pneumatic atomizer. Operation of this pilot began
in June 1978 and continued for approximately one year.
The results from this pilot work were very encouraging and resulted in B£W obtaining
a contract from Basin Electric for its new Laramie River No. 3 Station. During the
Velva test program, two important system parameters were observed to have a profound
influence upon the dry scrubber performance (1J). First, as the exit gas tempera-
ture from the dry scrubber was made to approach the adiabatic saturation temperature
of the gas, SO2 removal improved substantially. Secondly, the influence of alkaline
fly ash on SO2 removal was observed.
After completing the tests at Neal Station, a large-scale demonstration unit
capable of handling 120,000 ACFM of flue gas was constructed at Pacific Power and
Light's Jim Bridger Station in Rock Springs, Wyoming. This unit initially began
operation in the summer of 1979 and operated until December 1980. The unit was de-
signed with completely automated controls and included a baghouse and precipitator.
26-5
-------
The main function of this unit was to demonstrate the scaleup and controllability of
the process in a utility environment. This demonstration included long-duration
runs, including automatic load following (14). The most significant result of the
tests was that the high alkalinity of Laramie River coal (sub-bituminous) enabled
capture of at least 65% SO^ with water spray.
A second pilot facility that has the flexibility for changing fuels and testing a
wide range of variables was constructed at BSW's Alliance Research Center. This
1,500-AFCM unit began operation in July 1979. The unit is connected to a
5 x lO°-Btu/hr combustor that is designed to burn coal, oil, or gas. The pilot has
been used for testing many different coals, with sulfur ranges from 0.4% to 2.2%.
These coals also represented a wide range of ash alkal inities. Testing also has
been conducted using oil and gas (15).
B&W has correlated the results of its two pilot units and large-scale demonstration
unit. This gave B&W the ability to perform tests in a tightly controlled manner and
then to scale up the results based on the 1,500 ACFM pilot and the 120,000 ACFM
demonstration unit.
As a result of these development efforts, B&W sold two commercial systems to the
utility market. The first unit (580 MW) was sold to the Basin Electric Power
Corporation for its Laramie River No. 3 Station. The second unit (447 MW) was sold
to the Colorado Ute Electric Association for its Craig No. 3 Station. The first
unit is equipped with an electrostatic precipitator, and the second unit with a
baghouse. Both systems burn low-sulfur western coal. A schematic of the dry
scrubber design of these units (with plenum arrangement) is shown in Figure 1.
Presently, B&W is engaged in pilot and full-scale testing to optimize the dry
scrubbing process. A recent study (16) characterized dry scrubber technical and
economical potential for eastern higTnsulfur coals. The pilot test program
sponsored by the Department of Energy (DOE) under Contract DE-AC22-81FE-17056
("Eastern Coal Spray Dryer Evaluation") investigated the various dry scrubber system
variables as affected by high-sulfur coals. It was concluded from the program that
satisfactory S0? removal for these coals required use of a fairly high quantity of
reagent, dry scrubber operation at low approach to saturation, and use of recycled
ash material. The study suggests that the use of limestone injection with ash
recycle would increase the economic attractiveness of dry scrubber systems compared
to wet scrubber systems for flue gas desulfurization of these high-sulfur coals.
DRY SCRUBBING AND LIMESTONE INJECTION FOR EASTERN HIGH-SULFUR COAL
0vervi ew
The potential drawbacks in using the dry scrubber for units burning high-sulfur coal
are both technical and economical. The high-sulfur coals generally have very low-
alkaline fly ash, requiring more reagent to be used in the dry scrubber. Also, the
sulfur dioxide removal requirements are more stringent (90% removal required). The
dry scrubbing process has a built-in limitation for the amount of reagent slurry
that can be sprayed into the flue gas stream. This spraying capacity is controlled
by the available heat in the incoming flue gas that can be used for evaporation of
the moisture in the slurry. Also, as the sulfur loading in the gas Increases, 1t
becomes necessary to increase the solids concentration of the slurry. Limits exist
regarding the slaking, pumping, and atomizing of viscous slurries.
26-6
-------
From an economic standpoint, the cost of reagent material can be a very important
parameter; lime is considerably more expensive than limestone. For low-sulfur
coals, the quantity of reagent required for dry scrubbing is low. For high-sulfur
coals, however, where the reagent cost can be a dominant factor, the relative dif-
ference in costs between lime and limestone can be the dominant factor in present
application of dry scrubber technology to eastern coals.
Furnace limestone injection in combination with dry scrubbing of eastern bituminous
coal is one means of increasing the economic attractiveness of dry scrubber systems.
This approach was examined by B&W. The first study (17) was1 an exploratory examina-
tion of this concept. A 2.31-sulfur coal was fired; TTmestone was injected into the
furnace at various stoichiometric ratios (Ca/S). Recycled material was used as
slurry in the dry scrubber at various stoichiometric ratios (Ca/SCL). Overall SO^
removal was as high as 861.
The second study was completed by B&W under DOE Contract DE-AC22-81FE 17056 (U>J and
was geared toward investigating dry scrubber system variables and system design op-
tions to expand effectiveness for eastern coals. Furnace limestone injection was
evaluated as a design option to determine whether reagent requirements would be
decreased by furnace calcination of limestone to lime. The lime was then entrained
in the flue gases flowing to the dry scrubber. We anticipated that reagent require-
ments at the dry scrubber would be reduced because of SO- capture in the furnace.
A 3.1%-sulfur coal was used. Various locations, stoichiometric ratios, and flue gas
temperatures were studied in the furnace. An atomized water spray was used in the
dry scrubber.
Test Apparatus
BAW's 1500-ACFM dry scrubber pilot (Figure 2) was used for the DOE and in-house
tests. Three main components make up the system: the basic combustion test unit
(BCTU), the dry scrubber, and the baghouse particulate collector.
Pulverized coal stored in a 5-ton hopper was screw fed into the primary air line,
where it was transported to the 5 mi 11ion-Btu/hr BCTU, as shown in Figure 3. Coal
was fired using a 1ow-NO burner operated in a 1ow-NO mode, which produced a long,
lazy flame. Limestone was metered to the furnace at § rate corresponding to the
particular stoichiometric ratio desired via a 160-1b/hr Vibra screw feeder. An
eductor was used to aspirate limestone flow for furnace injection. In the furnace,
limestone rapidly calcined to lime as it mixed with the flue gases. S02 capture was
facilitated as flue gas, lime, and ash matter flowed through the BCTU furnace. The
gases encountered an approximate 1-second residence time in the furnace at tempera-
tures ranging from about 2500"F at the hottest portion of the flame to near 1500°F
at the furnace exit.
Flue gases exiting the furnace were routed to the dr7 scrubber (Figure 4). A por-
tion of gases was sent through a water-cooled, counter-current heat exchanger to
maintain a 300°F dry scrubber inlet temperature. Cooled flue gases entered the dry
scrubber through a windbox plenum, where a perforated plate distributed the gas for
even flow through the registers of the Turbo-Diffuser slurry atomizer assembly.
The Turbo-Diffuser, a B&W trade name for the slurry atomizer assembly, features a
pneumatic atomizer that disperses slurry flowing through an inner tube by pressure
of compressed air flowing through an outer tube. Flue gas was directed around the
atomizer, where tangentially located vanes in a register imparted a swirl to provide
adequate mixing of gas with atomized slurry.
26-7
-------
Slurry was produced by a 1000-1b/hr-capac1ty paste slaker equipped with a 20-mesh
screen to remove large unslaked lime particles and impurities. Lime from the paste
slaker was discharged into the main product tank, where water was added to adjust
slurry to the appropriate percentage of solids. For ash recycle slurries, ash col-
lected from the dry scrubber and baghouse was manually dumped into the tank. Slurry
was pumped to the Turbo-Diffuser, where supplemental water was added to the Turbo-
Diffuser to maintain appropriate dry scrubber approach to saturation temperature.
Approach to saturation temperature is defined as the difference in temperature be-
tween the dry scrubber outlet temperature and the adiabatic saturation temperature
of the gas. Inside the dry scrubber, flue gas SO2 diffused into the lime particles,
resulting in S0„ capture. Heavy ash matter fell out into the dry scrubber hoppers
for disposal or recycle, and lighter ash matter was carried with the flue gases to
the baghouse and cyclone.
The 500-ACFM baghouse removes particulate matter from one-third of the flue gas and
is equipped with an in-line heater for reheating flue gas before entering the bag-
house. This ash is collected on the inside of 16, 10-foot-long x 4-1/2-inch-diam-
eter woven bags with three antlcollapse rings. Additional SO2 capture occurred as
flue gas S0~ flowed through the inside of the bags contacting accumulated unreacted
lime and asn matter. A cyclone separator with the capability of handling 100% of
the flue gases collects particulates from the remaining two-thirds of the flue
gases. Flue gases from both particulate collectors were then directed to the
500Q-ACFM induced-draft fan and then discharged to the atmosphere through the stack.
Test Description
Two separate studies were conducted: one funded by DOE (16), and one funded intern-
ally oy 3&W (1_7). The DOE limestone injection/dry scrubBTng tests examined two
furnace injection schemes (primary and secondary) and three stoichiometric ratios
(1.0, 2.0 and 3.0). Rear-furnace temperatures of 1800", 2000°, and 2200°F were
studied by varying the firing rate at an approximate 3$ excess 0?. Note that the
8CTU furnace has no provision for measuring furnace exit gas temperature; therefore,
temperatures were measured through a viewport at the rear of the furnace. Simultan-
eous to the 3CTU limestone injection, water spray was used in the dry scrubber, with
an approach to saturation temperature averaging 37°F. Baghouse performance was not
monitored in the tests.
The in-house study considered only primary injection mode, with a 3.0 and 4.0 fur-
nace stoichiometric ratio. A 1.6 and 2.8 stoichiometric ratio, respectlvely, was
studied in the dry scrubber employing ash recycle at a 25"F approach to saturation
temperature.
In the primary injection mode, limestone was mixed with coal and primary air in the
feed pipe. The mixture was injected into the furnace through the low-NOx burner.
Injecting limestone in this manner created two conditions relative to reagent
reactivi ty:
• The greatest degree of mixing of calcined lime and flue gas SOg
• The longest time period in the highest temperature zones of the
furnace.
The main concern expressed for this injection scheme is decreased reagent utiliza-
tion as a result of the latter. Since sulfur capture reaction is time and tempera-
ture sensitive (though mixing may be adequate), lime reactivity may be reduced due
to dead burning (loss of lime particle surface area) and dissociation of product.
26-8
-------
Secondary injection for the DOE tests introduced limestone through a side viewport
downstream from the burners and in a cooler temperature zone. The cooler tempera-
tures reduce the opportunity of calcined lime dead burning and product dissociation.
Although flue gas furnace residence time was appropriate, mixing of reactant SOp and
lime may have been reduced. The proper injection location was arrived at through
preliminary shakedown tests, including high-velocity traverse (HYT).
These tests were conducted to determine a limestone injection location along the
length of the BCTU furnace that would correspond to a 2000°F injection temperature.
A typical furnace temperature profile for 4 million Btu/hr at 31 furnace excess 0?
and a 2QQ0°F rear-furnace temperature is shown in Figure 5. Viewport No. 2 on both
sides of the BCTU furnace was designated as the secondary injection site. The sec-
ondary injection ports are located on both sides of the periphery of the flame, far
enough upstream to facilitate mixing and produce an appropriate residence time. This
injection point is about 2.5 feet from the burner and 14 inches into the furnace and
affords S0£ and CaO a 1.3-second residence time in the furnace, assuming plug flow.
Of interest in this figure is the temperature and shape of the flame produced
through the low-NO burner. The flame is fairly long and narrow, exhibiting cooler
temperatures of 23009F or less. The core of the flame occurs downstream of the
burner along Port No. 2. This configuration of the flame reduces NO formation,
while the cooler temperatures benefit limestone calcination and sulfur capture.
Test Results and Conclusions
The results and operating conditions of the DOE tests are tabulated in Table 1. A
representation of the furnace SC^ removal results (Figure 6) indicates a general
increase in SOp removal with stoichiometric ratio. The dry scrubber results exhibit
a generally unchanging trend with furnace stoichiometric ratio. Sulfur dioxide re-
moval in both unit operations was much lower than anticipated. Overall SO- results
are shown in Figure 7. The in-house study results (tabulated in Table 2 and shown
in Figure 8) indicate increased dry scrubber SCL removal with the use of recycle
siurry.
Primary injection tests of the DOE study, all run at 4 million Btu/hr furnace load,
2000°F rear-furnace temperature, and 3% excess Op, produced a high of 20% SOp re-
moval in the furnace at the high 3.0 Ca/S stoichiometric ratio. With water spray in
the dry scrubber at a 37°F approach to saturation temperature, S02 removal indicated
a slight decrease with stoichiometric ratio, averaging 8%. The difference in dry
scrubber S02 removal for these tests can be considered negligible. In the overall
system, a high of 24* SO^ removal was achieved at the 3.0 stoichiometric ratio.
The secondary injection tests run at the various furnace operating conditions pro-
duced somewhat more representative results. The highest furnace S0? removal of 261
was attained at reduced load (3 million Btu/hr), low rear-furnace temperature
(1800°F), a high furnace stoichiometric ratio (3.0 Ca/S), and 3.0% excess furnace
Op- The best dry scrubber removal of 18% was obtained using water spray for the
reduced-load tests. At full load and a 2000°F rear-furnace temperature, furnace S0~
removal increased with stoichiometric ratio (as in the primary injection test) to
22% at 3.0 stoichiometric ratio. An average dry scrubber SO, removal of 13% was de-
termined for these tests with water spray. The higher 2200°F rear-furnace tempera-
ture tests produced an average 20% furnace removal at all furnace stoichiometric
ratios with no apparent trend. Sulfur dioxide removal in the dry scrubber for these
tests also exhibited no apparent trend, averaging 10% SOp removal with use of a
water spray. For the overall system, the reduced-load tests produced a high (39%)
S02 removal, and the full-load tests averaged near 28% removal.
26-9
-------
The in-house study considered primary limestone injection for use with either ash
recycle or water spray in the dry scrubber. Furnace SOp removal did not exhibit a
dramatic increase with stoichiometric ratio but averagea 24% removal. In these tests
using 3.0 Ca/S furnace stoichiometric ratio and full load, dry scrubber SOo removal
did not increase above 12% with water spray, even with a low 25°F dry scruDber ap-
proach to saturation temperature. With use of recycle slurry (rather than water
spray) at a 1.6 CaO/SOp stoichiometric ratio, dry scrubber SOo removal increased to
22%, producing 40% removal overall. At a higher furnace stoichiometric ratio of 4.6
Ca/S, dry scrubber removal with water spray increased to 43% SOp removal. With the
use of recycle slurry at 2.8 Ca/S stoichiometric ratio, however, the dry scrubber
achieved an 82% removal with an 86% removal for the system. This study illustrates
the influence of the use of ash recycle in the dry scrubber on SO2 removal.
Overall furnace and dry scrubber removal employing water spray were low. With the
use of water in the dry scrubber, the alkalinity for SOp removal was obtained from
entrained particulates in the flue gases. These particulates require slaking to
fully activate the alkalinity. Furnace removals were low for both studies, appar-
ently due to a larger-than-optimal limestone grind size, 70% - 200 mesh. This large
grind size of the limestone produced a reduction in particle surface area on calcin-
ation. With exposure to furnace heat, additional surface area is lost, limiting
calcined lime reactivity. For the primary injection test, in addition, the long
residence time in the high-temperature region of the furnace further reduced lime
reactivity. The problem of low SOo removal for the secondary tests was most likely
due to insufficient mixing of calcined lime and flue gas SO- in addition to large
grind si2e. In the dry scrubber, SO2 removals were low largely due to ash dropout
in the piping and heat exchanger between the furnace and the dry scrubber. The near
70% loss of furnace carry-over ash with unreacted lime eliminated much of the
reagent required for SOp capture. As a result, dry scrubber S0? removal did not
rise above 18%.
In conclusion, the two studies (OOE and in house) together revealed a number of
technical advantages of the combination of both systems (limestone injection with
dry scrubbing for flue gas desulfurization) over either system separately.
• The DOE study indicated limestone injection away from the burners
downstream of the furnace produced somewhat higher SOp removal in the
furnace than limestone injection with the coal througn the burners.
System optimization is required to increase SOg removal.
• Also shown through the DOE tests 1s a direct dependence of furnace SOp
removal on furnace stoichiometry. As the furnace stoichiometric ratio
is increased, furnace removal increases.
• The B4W in-house study revealed an increase in dry scrubber SOp
removal, with recycle slurry at high stoichiometric ratios.
• The in-house study also indicated that the combination of both systems
(limestone injection employing a high stoichiometric ratio with dry
scrubbing using recycled slurry) improves overall SOp removal above the
use of either system separately.
• Both studies indicate an optimized furnace limestone injection and dry
scrubber system can produce SOo removal for eastern high-sulfur coals
approaching -- if not surpassing -- the 1979 EPA new source performance
standards (NSPS) for high-sulfur coals. This system would include use
of appropriate grind size and a suitable mechanism for mixing limestone
with flue gases away from the burners and a dry scrubber employing re-
cycle slurry. Neither system separately can achieve the 70% to 90% S0?
reduction requirement without use of a large quantity of reagent.
26-10
-------
Use of recycled slurry produced from inexpensive limestone used in furnace injection
decreases — if not eliminates -- use of lime or other expensive dry scrubber re-
agents. In economic terms, we anticipate that this use of limestone as a reagent
decreases reagent costs for eastern coals and equates -- if not reduces -- capital
and operating cost compared to that of wet scrubber systems for these coals (15).
FUTURE WORK
The combination of limestone injection with dry scrubbing for flue gas desulfuriza-
tion holds much promise as an alternative to wet scrubber systems. B&W realizes the
potential as well as the implications of the combined system. Therefore, BAW has
planned in-house studies and will solicit funding for programs designed to investi-
gate the intricacies of the technology from the laboratory scale to the utility
boiler scale. Of particular interest to B&W is application to units equipped with
cyclone burners. Presently, B&W is engaged in a test program to study the combined
process in a large-scale utility boiler.
REFERENCES
1. Schwarzkopf, F., Lime Burning Technology, Precision, Millville, Pa., 1974.
2. Eades, J. I., and Sandberg, P. A., "Characterization of the Properties of
Commercial Lime by Surface Area Measurements and Scanning Electron Microscopy,"
The Reaction Parameters of Lime, ASTM STP 472, American Society for Testing and
Materials, 1970.
3. Ottennan, I., Mitt. Lebor. Geo!. Dirnstes NFHJ, Berlin, Akademie-Verlag, 1951.
4. Kelley, K. K., U.S. Bureau of Mines Bulletin 406, 1937.
5. Baker, D. C., and Attar, A., "Sulfur Pollution from Coal Combustion. Effect of
the Mineral Components of Coal on the Thermal Stabilities of Sulfated Ash and
Calcium Sulfate," Env. Sci. Tech., (15) 1981.
6. Dewing, E. W., and Richardson, F. D., Trans. Faraday Soc., Vol. 55, 1959.
7. Besmann, T. M., et al., "Thermodynamic Calculations on Coal Ash and CaO
Interactions," Ceramic Bulletin, Vol. 59, No. 4, 1980.
8. Borgwardt, R. H., et al., "Surface Area of Calcium Oxide and Kinetics of
Calcium Sulfide Formation," Environmental Progress, Vol. 3, No. 2, 1984.
9. Turkdogan, E. T., et al., "Sulfide and Sulfate Solid Solubility in Lime,
Magnesia, and Calcined Dolomite: Part 1 CaS and CaSO^ Solubility in CaO,"
Metallurgical Transactions, Vol. 5, 1974.
10. Attig, R. C., and Sedor, P., "Additive Injection for Sulfur Dioxide Control - A
Pilot Plant Study," HEW Order 4078-01, March 27, 1970.
11. "Full Scale Desulfurization of Stack by Limestone Injection," EPA Report
650/2-73-019-a.
12. LaRue, A. D., and Liang, A. D., "Furnace Limstone Injection Technologies to
Reduce S0£ Emissions," ASME Joint Power Conference, Toronto, October, 1984.
13. Hurst, T. B., "Dry Scrubbing Eliminates Wet Sludge," Joint Power Generation
Conference, Charlotte, N.C., October 7 - 11, 1979.
26-11
-------
14. Hurst, T. B., and Bielawski, G. T., "Dry Scruober Demonstration Plate --
Operating Results," EPA Symposium on Flue Gas Desulfurization, Houston, Texas,
October 28 - 31, 1980.
15. Downs, W., et al., "Control of S02 Emissions by Dry Scrubbing," American Power
Conference, Chicago, 111., April zl - 23, 1980.
16. Sawyers, L. E., et al., "Eastern Coal Spray Dryer Evaluation," U.S. Department
of Energy, DE-AC22-81FE-17056, to be published.
17. Doyle, J. B., and Jankura, B. J., "Furnace Limestone Injection with Dry Scrub-
bing of Exhaust Gases," 1982 Spring Technical Meeting of the Central States
Section of the Combustion Institute, Columbus, Ohio, March 22 - 23, 1982.
Y-JET ATOMIZER
REHEAT
AIR
TO STACK
SLURRY
SPRAY
C'
GAS
INLET
GAS FLOW
Figure 1. Dry scrubber reactor plenum arrangement
26-12
-------
pulverised
coal
HOPPER
PULVC RIZEO
I IM| S TONE
HOPPER
PRf HE ATEO
SECONDARY
COMBUSTION AIR
ORV
ASH
INJECTOR
WATER
paste uO
S UAHER
BASKET
STRAINER
STEAM
SLURRY
PRODUCT
TANK
OBIT
Strainer
TEMPERATURE
CONTROL
*ATE R
DRTE R
STACK
BAGHOUSE
5 PR A T
IN UNE
NtACER
WATER
COOLED
Hf a r
t* CHANGER
Figure 2. 1500-ACFM dry scrubber pilot
25-13
-------
STACK
STEAM DRUM
RISERS
DAMPERS
GAS/OIL/COAL
i TUBE BAN K5
DOWNCOMERS
WATER JACKET
DOWNCOMER
BLOWER
HOT GAS
ORIFICE
HOT AIR
NATURAL GAS
TWO AIR HEATERS
BLOWER
HOT GAS
STACK
DAMPER
OUTSIDE AIR
ORIFICE
PRESSURE
BLOWER
Figure 3. Basic combustion test unit (BCTU)
26-14
-------
PLENUM
REGISTER ARRANGEMENT
DSR REACTOR
COMPARTMENT NO. 1 HOPPER
COMPARTMENT NO. 2 HOPPER
COMPARTMENT NO. 3 HOWER
DUPONT S02 ANALYZER
INLIT
»1
Figure 4. Dry scrubber dimensions
PORT 1 PORT 2 PORT 3 PORT 4
BURNER
POSITION
PORT 1 PORT 2 PORT 3 PORT 4
Figure 5. Furnace temperature profile; 4 mi 11ion-Btu/hr load,
31 0^, 2000°F FEGT, low-N0x> dual-register burner
HVT
TRAVERSE POINT
INJECTION
PORTS
ZOOO'F
FURNACE
AlOO'F \
O 2000CF
ISOTHERMS, deg f
O 1900
~ 1950
o 2000
A 2050
2100
0 2150
<3 2200
¦ 2250
• 2300
26-15
-------
40
35
30
. 25
Z
o
£
u
3
Q
u.
K 20
(N
O
if!
U.
u
<
z
tr. 15
3
10
O SEC. INJEC; 1B00-F FURNACE REAR TEMP
3 4 MILLION BTU/HR LOAD
A SEC INJEC; 2000-F FURNACE REAR TEMP
4 MILLION BTU/HR LOAD
~ 5EC. INJEC; 2200;F FURNACE REAR TEMP
4 MILLION BTU HR LOAD
O PRIM INJEC; 2000:F FURNACE REAR TEMP.
4 MILLION BTU HR LOAD
J 1 L
1 2 3
FURNACE STOICHIOMETRIC RATIO. Ca S MOLAR
Figure 6. Furnace SO reduction results -- DOE furnace
limestone injection test
26-16
-------
40
35
30
Z
o
U
2
a
25
!N
o
(/»
20
Z
'jj
>
o
1 5
10
o
A
~
o —
SEC. INJEC; 1 BOO^F FURNACE REAR TEMP.
3 - 4 MILLION BTU HR LOAD
SEC. INJEC; 2000;F FURNACE REAR TEMP
4 MILLION BTU, HR LOAD
SEC INJEC; 2200CF FURNACE REAR TEMP
4 MILLION BTU 'HR LOAD
PRIM INJEC. 2000"F FURNACE REAR TEMP,
4 MILLION BTU/ HR LOAD
I
_L
I
1 2 3
FURNACE STOICHIOMETRIC RATIO. Ca/S MOLAR
Figure 7. Overall S0_ reduction results -- DOE furnace
limestone injection test
26-17
-------
100
80
E
3
~-
Q.
4
U
IN
o
l/l
_l
4
H
o
60
40
20
FURNACE
DRY SCRUBBER
SYSTEM CAPTURE
SYSTEM AND
SAGHOUSE
1
i
t
I
P
/
/A
i
%
1
i
f
t
'//
f
I
3 0 1 0 1 6 1.0 4 6 2.B 28
INLET Ca/S
TEST
SERIES
1
TEST
SERIES
2
'/¦'
I
I
I
//,
I
Y',
V/a
2.8
Figure 8. SO,, removal -- B4W in house study
26-18
-------
Table 1
RESULTS OF DOE LIMESTONE INJECTION TESTS
Furnace
SO, Removal U)
Run No.
Injection
Method
Load
(MkB/hr)
o, m
Port 4
T emp 1*F)
Stolchi ometry
(Ca/S)
Furnace
£
Dry Scrubber
Overall
212'1)
Sec.
3
3.0
1800
3
26.0
17.3
38.8
213
Sec.
2
4.0
1800
2
13.6
17.8
20.4
215121
Sec.
4
3.0
2000
3
22.4
10.5
30.6
216
Sec.
4
3.0
2000
2
13.3
15.9
27.1
217
Sec.
4
3.0
2000
1
7.3
11.9
23.6
218(3)
Sec.
4
3.0
2200
3
19.3
10.0
27.4
219
Sec.
4
4.3
2200
2
20.6
11.8
30.0
220
Sec.
4
3.7
2200
1
19.9
8.1
26.4
221{41
Prim.
4
2.8
2000
3
20.0
5.1
24.1
222
Prim.
4
2.75
2000
2
13.6
a.s
21.2
223
P rim.
4
3.0
2000
1
4.1
10.6
15.0
NOTES
FOR DRY SCRU8BER OPERATING CONDITIONS:
Avg. operating conditions, Runs 212 and 213:
Approach temp -- 37"F
Inlet temp. -- 308*F
Residence t1«e -- 9.79 sec
Avg operating conditions, Runs 215
Appraoch teap — 38*F
Inlet teap -- 304*F
Residence time -- 11.04 sec
217:
Avg. operating conditions, Runs 218 - 220:
Approach teap -- 35"F
Inlet temp. -- 309"F
Residence tine -- 10.50 sec
Avg operating conditions, Runs 221 - 223:
Appraoch temp -- 37"F
Inlet temp -- 309*F
Residence Time -- 10.69 sec
26-19
-------
Table 2
OPERATING CONDITIONS AND TEST RESULTS -- BSW IN-HOUSE STUDY
TEST SERIES 1 TEST SERIES 2
Operating Conditions
Coal Feed Rate (lb/hr) 325 325
Sulfur in Coal (1) 2.3 2.3
Furnace;
- Temperature (*F) 2550 2530
- Ca/S Ratio 3.0 4.6
Dry Scrubber:
- Inlet Temperature ("F) 314 425
- Outlet Temperature (*F) 151 160
- Approach-to-Saturatlon
Temperature (*F) 25 25
- Ca/S Ratio 1.6 2.8
- Inlet Oust Loading
(lb/acfm) 2.41 3.13
(i Theoretical) (30.5) (27.5)
Slurry Solids (1) 21.1 19.5
Test Results
SO^ Removal (1)
- Furnace 23 24
- Dry Scrubber
(With Water Only) 12 43
- Dry Scrubber
(With Recycle Slurry) 22 82
- Systea
(With Recycle Slurry) 40 86
26-20
-------
PILOT EVALUATION CF COMBINED S02 AND PARTICULATE REMOVAL ON A FABRIC FILTER
Franz G. Pohl
Southern Research Institute
P.O. Box 55305
Birmingham, Alabama 35255-5305
Michael Mctlroy and Richard Rnudy
Electric Power Research Institute
P.O. Box 10412
Palo Alto, California 94303
ABSTRACT
The injection of calcium compounds into the flue gas upstream of a fabric filter is
under evaluation as a means of simultaneous S02 and particulate emission control.
Pilot tests are being conducted on a 100 acfm slipstream from a pulverized coal
fired boiler burning 3% sulfur coal. Water injection, steam injection, and heat
extraction are used to condition the flue gas to desired temperatures and
nunidities. S02 removal occurs in the flue gas ductwork, in tne expansion plenum
upstream of the fabric filter, and across the fabric dustcake. Tests with pressure
hydrated dolomitic lime indicated higher S07 removal as tne flue gas temperature
approached the saturation temperature. Fifty percent S02 removal was observed at
an approach temperature of 25°F when the sorbent was injected at a
calcium-to-sulfur ratio of 1:1 and up to 80% SO removal at a calcium-to-sulfur
ratio of 2.9:1.
OBJECTIVE
A pilot test facility was built at tne Scholz power station of Gulf Power Company
in Sneads, Florida, to perform proof-of-concept tests of a combined S0„ and
particulate control process involving injection of calcium-based alkali compounds
upstream of a fabric filter. The 100 acfm facility operated on a flue gas
slipstream from a pulverized coal-fired boiler burning 33» sulfur coal. Calcium
based sorbents, including pressure hydrated dolomitic lime (the majority of tests
used tnis lime), pressure hydrated high calcium lime, conventionally slaked
hydrated lime and quick lime were evaluated. In addition to performing comparative
tests with the different sorbent materials,"an assessment of flue gas parameters
with respect to promotion of S02 removal efficiency was performed . These
parameters included moisture content, flue gas temperature, temperature difference
between dew point and actual flue gas temperature (commonly referred to as approach
temperature), and relative humidity. A further objective was to quantify S02
removal in the various portions of the pilot plant, including the ductwork, the
inlet plenum to the fabric filter, and the fabric dustcake.
27-1
-------
PROCESS DESCRIPTION
A scnematic diagram of the all-dry SO, removal process is shown in Figure 1. In
this process pulverized sorbent material is fed into the flue gas Detween the air
preneater and baghouse. The reagent can react with SO in the short residence time
(1 sec) ductwork, the long residence time (1 min) inlet plenum to the fabric
filter, or on the fabric filter dustcake where intimate contacting between S02 and
reagent can occur. The flue gas was conditioned through steam injection, water
injection and/or heat extraction upstream of the fabric filter. By injecting steam
at flue gas temperature the moisture content of the flue gas was increased without
affecting approach temperature or relative humidity significantly. Heat extraction
allowed reduction of the approach temperature and increased the relative humidity
without changing the absolute moisture content. The injection of water served two
purposes: it increased the absolute moisture content, and dropped the flue gas
temperature as a result of latent heat conversion in the evaporation process. This
lowered the approach temperature and increased the relative humidity.
TEST FACILITY
Figure 2 shows a schematic diagram of the 100 acfm slipstream facility. "The
sorbent feed system consisted of a screwfeeder mounted on an electric scale for
feedrate determination. Sorbent was fed into a prototype deagglomerator, from
which it was carried out by a 1 cfm gas flow and injected through a dispersion
nozzle into the 8-in. diameter duct. The water injected into the flue gas was
metered through a rotameter, preheated to enhance evaporation (thus minimizing
problems associated with wetting the duct walls), and sprayed through an ultrasonic
atomizer into the flue gas. Steam was electrically generated, superheated to the
desired temperature, and metered through an orifice. One acfm slipstream off of
the 100 acfm duct was routed through a heated filter (Fabric Filter Sampling
System) to simulate tne fabric filter.
TESTS AND RESULTS
In Figure 3 an S02 profile throughout the system for a typical test is shown. The
time scale on the abscissa represents the residence time of the sorbent after
injection. The SO, concentration in ppm at any given location is shown as the
ordinate. In this~example, the SO concentration upstream of the sorbent injection
was 2400 ppm. Sorbent was injected at a reagent molar ratio of Ca/S = 1:1. Water
was injected into the flue gas 8 feet (2 sec) downstream of the sorbent injection
location. This dropped the temperature from 300 to 210°F. The SO, concentration
was reduced to 2000 ppm. Using heat extraction the temperature upstream of the
filter dropped to 135°F. The SO, concentration upstream of the fabric was 1500
ppm; downstream of the fabric 1100 ppm of SO was measured. The overall S02
removal through the entire system was about 54 percent.
Several of these injection tests were conducted at different approach
temperatures; an example of some of these results is illustrated in Figure 4. In
each of the vertical bars, which symbolize the overall S02 removal, partitions
indicate the contributions of duct, plenum and dustcake-fabric reactions. The
results indicate a trend of lower removal efficiencies at higher approach
temperatures. Figure 5 shows the total SO removal rates vs. relative humidity for
most of the tests performed with pressure nydrated dolomitic lime for a
stoichiometry range from 0.75 to 1.34. The graph again indicates the trend of
higher removal rates at higher relative humidities. In some of these tests
(circles and triangles) no water was injected; only steam injection and heat
27-2
-------
extraction (circles) or neat extraction only (triangles) were utilized to raise the
relative humidity.
Throughout each test the SCL concentration was monitored at the 4 sample locations
previously mentioned. The 4 locations were scanned sequentially. This allowed
variations in S02 concentration levels to be monitored as the test progressed,
figure 6 illustrates normal changes in the S02 concentrations at tne 4 sample
locations throughout a typical test. These results could be used to predict also
how SO, removal rates would develop in a baghouse facility with periodic cleaning
cycles^
Five different sorbent materials were investigated: quick lime, pressure hydrated
high calcium line, conventionally slaked hydrated lime, pressure hydrated dolomitic
lime, and precalcined pressure hydrated dolomitic lime. A comparison of results of
these tests at a reagent ratio between 2 and 3 are summarized in Figure 7. The
The hignest removal rate (78%) was achieved by precalcined pressure hydrated
dolomitic lime, followed by pressure nydrated dolomitic lime (60%), conventionally
slaked hydrated lime (47%), pressure hydrated high calcium lime (45%), and quick
1 iTie (2 3%).
SUMMARY AND CONCLUSIONS
These tests were intended to be proof of concept for all-dry sorbent injection of
calcium sorbents for S02 emission control. Analysis of the data have led to the
following conclusions:
• S02 and particulate removal via calciun sorbent injection upstream
of a fabric filter is technically feasible.
• Removal rates of 50% were accomplished at -eagent ratios of 1:1, and
up to 80% at reagent ratios of 3:1.
• The relative humidity and reagent ratio appear to be the most
important variables governing the removal efficiency.
• S02 removal occurred in the ductwork, fabric filter plenum and in
the filter dustcake.
• Precalcining pressure hydrated dolomitic lime, significantly improved
its reactivity (higher surface area) and SO,, removal.
ACKNOWLEDGEMENTS
This project is sponsored by Electric Power Research Institute under Contract No.
RP2533-2. Michael McElroy and Richard Rhudy are the EPRI Project Officers. Their
guidance in this study is gratefully acknowledged.
27-3
-------
SCRBENT N.FC.m.%
DUST CAKE AND FABRIC
HEAT EXTRACT ON
vVA'ER S"EAiM A.EC" CN
Figure 1. Schematic of SO Removal Process
S02 REV,OVAL ZONES
I DUCT
ez 2 PLENUM
E3 3 DUST CAKE
27-4
-------
SORBENT INJECTION
WATER INJECTION
STEAM INJECTION
FILTER
VENTURI
M
1 acfm
100-150 acfm
drop out BOX
fyj—' FLLE \
fOf \ GAS DJCT /
30,000 acfm
FLUE GAS
30CJF
2000 ppm SO2
ICO—150 acf-n
fabric filter
SAMPLING SYSTEM
Figure 2. Dry Injection Test Facility
27-5
-------
WATER
2500
300°F
DUCT
2000
PLENUM
E
a
a
FABRIC
O
VI
1000 —
500
10
10
30
40
0
20
50
60
SORBENT RESIDENCE TIME, sec
Figure 3. System 50, Profile with Pressure Hyclratea Do'cmitic
Lirre at a'Stoichicmetric Ratio of Ca/'S = 1:1
-------
1 I 1 I 1 I 1 I 1 I
O STEAM AND COOLING
O WATER AND COOLING
~ A COOLING ONLY a
PRESSURE HYDRATED DOLOMITIC LIME Q
a
A
A
A
a
o
o
J I I 1 I I I I I I 1 I I
0 10 20 30 40 50 60 70 80
RELATIVE HUMIDITY, %
Figure 5. Total SCL Removal vs. Relative Humidity
27-7
-------
2000
SLIPSTREAM
INLET
FF PLENUM
INLET
1500
UPSTREAM
OF DUSTCAKE
DOWNSTREAM
OF DUSTCAKE
500
33°F APPROACH TEMPERATURE
SCHOLZ PILOT
0
3
1
2
TIME FROM START OF SORBENT INJECTION, houri
Figure 6. SCL Removal vs. Time From Start of Sorbent Injection
with Pressure Hydrated High Calcium Lime
27-8
-------
TOTAL S02 REMOVAL. %
ro
i
VD
UZ>
C
ro
•-vj
o
rt-
o>
GO
O
^3
0>
3
O
<
0/
o
-5
O
O)
-5
(t>
3
r-r
C/0
O
-5
cr
rr>
o>
rt-
(D
ro
o
o
<71
O
00
O
O
o
QUICK
LIME
PRESSURE HYDRATED
HIGH CALCIUM LIME
CONVENTIONALLY SLAKED
HYDRATED LIME
PRESSURE HYDRATED
DOLOMITIC LIME
PRESSURE HYDRATED
DOLOMITIC LIME
PRE CALCINED
Ca/S
= 2 11
Ca/S = 2 3 1
Ca/S =2 9 1
Ca/S — 2 7 1
Ca/S = 2 9 1
tj
o
4*
O
o>
o
00
o
o
o
l/»
-------
TECHNICAL REPORT DATA
(Please read Imiruciions on the reverse before completing)
1 RE'OflT Nfl
, " EPA/600/9-85/020
2
3. RECIPIENT'S ACCESSION NO.
m 5 23 23'5 3/iS
4 title and subtitle proceedings: First Joint Symposium on
Dry SOo and Simultaneous SC>2/NOx Control Technol-
ogies! Volume 1. Fundamental Research and Process
Development
5. REPORT DATE
July 1985
6. PERFORMING ORGANIZATION CODE
7 AUThORiS)
P. Jeff Chappell, Compile
r
8. PERFORMING ORGANIZATION REPORT NO.
9. PERFORMING OP0ANI2ATION NAME AND ADDRESS
Acurex Corporation
10. PROGRAM ELEMENT NO.
555 Clyde Avenue
Mountain View, California
94039
11. c6ntract/grant NO.
68-02-3993, Task 1
,*v
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
Proceedings; 10/84-4/85
14. SPONSORING AGENCY CODE
EPA/600/13
is. supplementary notes project officer is P. Jeff Chappell, Mail Drop 63, 919/
541-3738. Volume 2 is power plant integration, economics, and full-scale exper-
ience.
i6 abstract y^e proceedings document the First Joint Symposium on Dry S02 and
Simultaneous S02/NOx Control Technologies, held November 13-16, 1984, in San
Diego, CA. The symposium, sponsored jointly by EPRI and EPA, was the first
meeting of its kind devoted solely to the discussion of emissions control processes
based on dry injection of calcium or sodium sorbents to meet S02 and NOx regula-
tions for coal-fired power plants. Processes that were discussed included: direct
furnace injection of calcium-based sorbents, sorbent injection combined with low-
NOx burners for simultaneous S02/N0x control, and post-furnace injection of cal-
cium and sodium sorbents. The symposium provided a timely forum for the exchange
of data and information on the current status and plans for these emerging technol-
ogies. The presented papers began with a keynote address on acid rain strategies
and control technology implications, followed by overviews of EPRI, EPA, and
Canadian programs and the utility perspective for dry control technologies. Other
papers focused on the latest advances in fundamental research and process design,
power plant integration and economics, field applications, and full-scale testing.
In addition to the U.S. and Canada, attendees represented West Germany, France,
The Netherlands, Austria, and Japan.
17.
KEY WORDS AND DOCUMENT ANALYSIS
a. DESCRIPTORS
b.lDENTIFiERS/OPEN ENDED TERMS
c. COSATl Field/Croup
Pollution Electric Power Plants
Nitrogen Oxides Coal
Sulfur Dioxide Combustion
Sorbents Economics
Calcium
Sodium
Pollution Control
Stationary Sources
Dry SO2 Control
Simultaneous SO2/NOX
Control
13B 10B
07B 21D
21B
11G 05C
13. D'STRiBUTlON STATEMENT
19 SECURITY CLASS (This Report)
Unclassified
21. NO OP PAGES
< 43a' "¦ ;
Release to Public
20 SECURITY CLASS (This page)
Unclassified
22. PRICE
EPA Form 2220-1 19-73)
------- |