This article appeared in a journal published by Elsevier. The attached
copy is furnished to the author for internal non-commercial research
and education use, including for instruction at the authors institution
and sharing with colleagues.
Other uses, including reproduction and distribution, or selling or
licensing copies, or posting to personal, institutional or third party
websites are prohibited.
In most cases authors are permitted to post their version of the
article (e.g. in Word or Tex form) to their personal website or
institutional repository. Authors requiring further information
regarding Elsevier’s archiving and manuscript policies are
encouraged to visit:
http://www.elsevier.com/copyright
Author's personal copy
Fuel 87 (2008) 3322–3330
Contents lists available at ScienceDirect
Fuel
journal homepage: www.elsevier.com/locate/fuel
Abatement of mercury emissions in the coal combustion process equipped
with a Fabric Filter Baghouse
Yan Cao a,*, Chin-Min Cheng a, Chien-Wei Chen a, Mingchong Liu a,b, Chiawei Wang a,b, Wei-Ping Pan a
a
b
Institute for Combustion Science and Environmental Technology (ICSET), Western Kentucky University (WKU), 2413 Nashville Road, Bowling Green, KY 42101, USA
Mingchi University, Taipei, Taiwan
a r t i c l e
i n f o
Article history:
Received 27 November 2007
Received in revised form 10 May 2008
Accepted 16 May 2008
Available online 12 June 2008
Keywords:
CFBC
PC
Mercury emission
Fabric filter baghouse
Fly ash
a b s t r a c t
The purpose of this study was to investigate the dependence of mercury emissions on coal ranks and electric utility boilers equipped with Fabric Filter Baghouses (FF). A comparison of mercury emission rates
and fly ash properties was made between a circulating Fluidized Bed Combustor (CFBC) with FF and a Pulverized Coal (PC) combustor with FF during the burning of all three ranks of American coals. The data
were collected from the Environmental Protection Agency Information Collection Request (EPA ICR)
and WKU ICSET’s mercury testing program. A statistical stepwise regression procedure was used to determine significant factors such as coal rank and types of boilers equipped with FF on mercury emissions
during coal combustion. The higher mercury emission rates were generally found in both CFB and PC
units when lignite was burned. The lower mercury emission rates were generally found in both CFB
equipped with FF and PC units equipped with FF when bituminous coal was burned. There was a statistically significant lower mercury emission in the CFBC equipped with FF than that in the PC units when
sub-bituminous coal was burned. Lower mercury emission rates in electric utility boilers equipped with
FF are due to the active fly ash generated with a larger specific surface area and pore volume. Higher mercury emission rates observed during lignite-fired boilers may be due to their lower specific area of fly ash,
which results from lower LOI, as well as the pore blockage by selenium (Se) for Texas lignite; and sodium
(Na) and potassium (K) for North Dakota lignite. There is no significant mutual benefit for the mercury
captured by the addition of Spray Dry Absorber (SDA) or selective non-catalytic reduction (SNCR) in
the CFBC system.
Ó 2008 Elsevier Ltd. All rights reserved.
1. Introduction
The United States (US) Environmental Protection Agency (EPA)
promulgated the Clear Air Mercury Rule [21] to permanently cap
and reduce mercury emissions from coal-fired electric utilities
boilers, because mercury is a persistent bio-accumulative toxin
that builds up in human body tissue [1]. The US EPA has also recently promulgated the Clean Air Interstate Rule (CAIR) to further
reduce SO2 and NOx. This has led to additional installations of control systems for Particulate Matter (PM), SO2 and NOx, which have
been identified to also reduce mercury emissions without additional cost [13,14,24]. Under Section 111 of the Clean Air Act
(CAA), New Source Performance Standards (NSPS) on mercury have
been established based on Best Demonstrated Technology (BDT)
considering cost, non-air-quality health, environmental impacts,
and energy requirements. However, on February 8, 2008 the US
* Corresponding author. Tel.: +270 7790202; fax: +270 7452221.
E-mail address: yan.cao@wku.edu (Y. Cao).
0016-2361/$ - see front matter Ó 2008 Elsevier Ltd. All rights reserved.
doi:10.1016/j.fuel.2008.05.010
Court of Appeals for the District of Columbia issued a unanimous
decision vacating the US Environmental Protection Agency’s
Clean Air Mercury Rule (CAMR) and the rule ‘‘de-listing” Electric
Generating Units (EGUs) from the list of sources requiring regulation under the Clean Air Act Section 112. There is considerable
uncertainty on the subject of regulators and regulated entities. This
uncertainty is particularly acute when it pertains to the mercury
monitoring provisions of the rule, which will become effective
by January 1, 2009 (No. 05-1097. United States Court of Appeals)
[12].
Based on currently-available EPA Information Collection
Request (ICR) data [8,20], a fabric filter baghouse (FF) can be more
effective for particle-bound mercury capture than an electrostatic
precipitator (ESP). This is due to enhanced heterogeneous oxidation and adsorption of mercury by fly ash in FF. The combination
of Selective Catalytic Reduction (SCR) and Flue Gas Desulfurization
(FGD) is another effective mercury control method. This is due to
the effective control of the oxidized mercury by FGD after the enhanced elemental mercury oxidation by SCR. Thus, BDT of mercury
emissions in coal-fired electric utility boilers is considered to be
Author's personal copy
Y. Cao et al. / Fuel 87 (2008) 3322–3330
either an FF, or an FGD or their combined utilization [6,10,11,18].
However, the use of FF and SCR-FGD at US coal-fired power
plants is limited. There is only a 9.4% installation of FFs in comparison to an 85.8% installation of ESPs and only a 28.2% installation
of FGDs in electric utility boilers for total generating capacity
(MW).
Furthermore, the specific mercury removal performance of suggested BDT is largely varied, depending on coal properties and the
combustion process [6]. Based on the EPA ICR program, the achievable mercury control efficiencies by BDT are 86.7% for bituminous
coal-fired utility boilers. It is only 31.8% for sub-bituminous coalfired boilers (mainly Powder River Basin (PRB) coal) and even as
low as 18.3% for lignite-fired utility boilers. The DOE Energy Information Administration (EIA) estimates that over 50% of the coal-reserve base is bituminous coal, about 30% is sub-bituminous and 9%
is lignite [22]. The most common sub-bituminous coal in the US is
located throughout Montana and Wyoming, and large deposits of
lignite are located near North Dakota and Texas. Based on 1999
survey results, 52% of the total tonnage of coal burned by the electric utility industry is bituminous coal, approximately 36.5% is subbituminous and 6.5% is lignite [23].
There are over 1150 coal-fired electric utility boilers in the US
based on EPA ICR data. Although Pulverized Coal (PC) boilers are
considered part of the older designs, PC boilers account for approximately 86% of the total number of units and make up around 90%
of nationwide generating capacity [24]. Cyclone furnaces account
for approximately 7.6% of both the total units and the nationwide
generating capacity. For comparison, the most advanced coal combustion technologies, such as CFBC and Integrated Gasification
Combined Cycle (IGCC) units, only account for about 3.7% and
0.3% of the total units, respectively. They also account for about
1.3% and less than 0.1% of the nationwide generating capacity,
respectively. The United States strives for energy security by
developing advanced, environmentally-sound technologies and
exploring a range of domestic energy sources. Coal will continue
to prove itself a critical energy resource for the nation [9]. It is
estimated that 159 new coal-fired units with about 70 gigawatts
in total generating capacity will be built in the United States by
2030. Approximately 15% of the new units will be CFBCs (22 new
units) in this resurgence, compared to the 3.7% presently available
CFBCs [23].
CFB combustor development has progressed since the mid1960s because of its lower capital, operational, maintenance, and
electricity generation costs compared to IGCC [2,25]. The US Government also recognized its potential as an efficient and environmentally friendly coal utilization technology. The technology
offers a number of advantages. The long residence time of solid
fuels in the CFBC system results in high combustion efficiency even
with difficult-to-burn solid fuels. Low operating temperatures also
effectively control SOx and NOx emissions. Staging of second air
injection into the CFBC and Selective Non-catalytic Reduction
(SNCR) technologies produces even lower NOx emissions [1,25].
The Spray Dry Absorber (SDA) is followed by an FF for collection
of spent reagent and fly ash. Thus, SNCR and SDA with FF are generally the standard Air Pollution Control Devices (APCDs) for the
CFBC system under the new CAIR.
This study’s purpose was to screen out major factors on mercury emissions in all electric utility boilers by burning all ranks
of coals. The evaluated data were collected from the EPA ICR program and WKU ICSET’s mercury field testing. We compared mercury emission rates and fly ash properties from CFBC boilers and
conventional PC boilers equipped with FF during burning of all
three typical American coals. A statistical stepwise regression procedure used to test the detailed investigations of mercury emission
rates among different set-ups of APCD (SNCR, SDA + FF) in two
commercial CFBC systems is also presented.
3323
2. Methods
2.1. Data collection
Mercury emission data were collected from data bases of the
EPA’s mercury ICR program and WKU ICSET mercury field testing.
The EPA ICR data are from sampling activities, which were obtained based on ASTM D6784 Ontario Hydro Method (OHM). Besides ASTM OHM used, ICSET data were also from sampling
activities by the semi-continuous Mercury Monitor (SCEM) and
EPA Appendix K method. The EPA ICR data were collected upstream of the last air pollutant control device and the stack. The ICSET data were collected from sampling activities, which have been
extended to varied APCD locations and boiler operation conditions.
Thus, it could provide additional information about the dependence of mercury transformation at different locations and mercury emission rates on the boiler performance. Boilers equipped
with FF were selected in this study because FF was predicted to
be more efficient for mercury emission control than ESP. In the collected ICR data, there were 18 PC units, 1 cyclone unit, 9 CFBC
units, 3 Stoker-fired units and 2 IGCC units. Coals burned in these
units included bituminous, sub-bituminous, lignite and their
blends. All ash characterization data were from the ICSET database.
The mercury emission rate, which is commonly expressed as lb/
TBtu, was not used in this study. It does not include information
on mercury input levels so that it is not accurate herein. In this
study, the new mercury emission rate (lg/NM3/(lg/g Btu/lb)) is
defined as mercury emissions in the stack (Hgstack, lg/NM3), per
the mercury content (Hgcoal, lg/g) and also per heating value of
the (BTUcoal, Btu/lb), see Eq. (1). This factor can be used to evaluate
the mercury emission rates, which are dependent on the mercury
content of the coal (Hgcoal) and coal heating value (Btucoal). We
found loss of information in ICR data (for example, a complete
analysis of coal) to calculate F-factor for every case and thus to correlate BTU and flue gas volume to make its unit have a simple
mercury emission rate ¼ Hgstack =½ðHgcoal ÞðBtucoal Þ:
ð1Þ
2.2. Stepwise regression analysis
The collected data were subjected to the stepwise regression
to build up a statistics model of significant analysis of factors
affecting mercury emission rates. Stepwise regression can remove
and add variables into the regression model to identify a useful
subset of the factors. The basic principle in this stepwise regression is to calculate an F-statistic and p-value for each variable
in the model. If the p-value for any variable is greater than Alpha
to remove (0.15), then the variable with the largest p-value is removed from the model. If no variable can be removed, the procedure attempts to add a variable, and the next step begins. After
trial and error calculations, the regression model will supply the
most significant factors that fit the prediction. The selection or
definition of the data’s subset is also dependent upon understanding mercury transformation in the coal-fired combustion process.
In this study, SPSS statistics software was used to fulfill the stepwise regression analysis. We can keep variables in the model
regardless of their p-values. Because analysis procedures require
that factor variables and their corresponding response variables
should have an equal amount in data size, we compiled the data
bank into two groups (mercury emission rate and fly ash). The
factor prediction on the mercury emission rate has 81 sets of data
(54 sets from PC units and 27 sets from CFBC). Nine sets of data
from stoker-fired units and 6 sets from IGCC data were excluded
in the statistical analysis because little data was available and
some of the required information was not collected during tests
in the EPA ICR program for those units that were IGCC and
Author's personal copy
3324
Y. Cao et al. / Fuel 87 (2008) 3322–3330
sion rate of Stoker-fired units was generally low, approximately
1.0 103(lg/NM3/(lg/g Btu/lb)). The PC boilers burning bituminous coals were also low, approximately 9.0 104(lg/NM3/(lg/
g Btu/lb)). The mercury emission rate was increased in PC boilers
when the coal was switched from bituminous coal to a blend of
bituminous coal and petroleum coke or sub-bituminous coal. PC
boilers burning sub-bituminous coal showed an even higher mercury emission rate than those burning bituminous coal. This
increasing trend was at its greatest when lignite coal was burned
in the PC boilers – approximately 2.0 102 (lg/NM3/(lg/g Btu/
lb)). The cyclone-fired boiler burning bituminous coal showed a
very high mercury emission rate, approximately 4.0 102 (lg/
NM3/(lg/g Btu/lb)) among all coal-fired combustion processes.
However, there is only one such unit in the present study. The reason may be due to the higher combustion temperature, causing
less ash to exit the cyclone boiler as fly ash. These conditions likely
result in the lower reactivity and lower concentration of fly ash
available for mercury capture with comparison to that of PC boilers. CFB boilers burning bituminous and sub-bituminous coals
show very low mercury emission rates. Mercury emissions could
also be efficiently controlled in a CFB boiler by burning their blendings. Compared to PC boilers burning lignite, a CFB boiler shows a
lower mercury emission rate of about 1.0 102 (lg/NM3/(lg/
g Btu/lb)). The coal gasification-based IGCC process appears to
have a higher mercury emission rate, around 6.0 103 (lg/
NM3/(lg/g Btu/lb)), than those of coal combustion-based boilers
burning bituminous coal. From analysis, the mercury emission rate
appears to be strongly related to the rank of coal and the type of
boiler.
Similarly shown in the Fig. 1, there is an apparent correlated
trend of Hgash/Hgcoal and mercury emission rates. This may indicate that mercury adsorption by fly ash generated in boilers is a
major method for controlling mercury emissions in coal-fired
Stoker-fired. The fly ash characterization has 38 data sets (30
from PC units and 8 from CFBC units).
2.3. Ash characterization by its physical structure
All fly ash samples in this study were collected in FF hoppers of
selected boilers during ICSET mercury field tests. The surface and
elemental analyses of fly ashes were performed using a JEOL LSM5400 Scanning Electron Microscope (SEM-EDX). The instrument’s
operating parameters were as follows: the electron beam energy
(15 keV), the working distance (30 mm) and the sample tilt angle
(0°). In most cases, three magnifications at 200, 2000 and 10,000
were selected for analysis. Physical surface properties such as specific surface area (BET model or BJH model), pore volume, and average pore size of fly ashes were characterized by a Micromeritics’
ASAP Accelerated Surface Area and Porosimetry instrument
(Micromeritics Instrument Corp.). The specific surface area was calculated by the BET equation from the nitrogen adsorption data in
the relative pressure range of 0.05–0.2. The single point total pore
volume was calculated from the amount of nitrogen adsorbed at a
relative pressure of around 0.95. The micropore (pores 62 nm)
[19] volume was obtained using the t-plots method; and the mesopore (2–50 nm) volume was determined using the BJH method. All
of the calculations were performed with software provided by
Micromeritics. The molecular sieve 13 provided by Micromeritics
was run periodically to check the reliability of this instrument.
3. Results and discussion
3.1. Factors affecting mercury emission rate
The mercury emission rates in boilers with different configurations and different coals are presented in Fig. 1. The mercury emis0.05
10000
mercury mission rate,
Ash LOI,
Hg(0)/Hg(VT)stack,
- ClCoal,
* Hgash/Hgcoal
0.035
100
0.03
0.025
10
0.02
1
0.015
0.01
0.1
0.005
0
Log(Ash LOI), Log(Hg(0)/Hg(VT)stack), Log(ClCoal ),
Log(Hgash/Hgcoal)
1000
0.04
3
mercury emission rate, ( g/NM /( g/g Btu/lb))
0.045
0.01
0
5
B
Stoker-fired
10
B
15
B/P,SB
PC
SB
20
SB/L
25
L
30
A B B/SB
Cyclone
SB
CFBC
35
L
40
B
IGCC
Fig. 1. The dependence of mercury emission rates on boiler types with FF and coal ranks (B: Bituminous coal, P: Petcoke, SB: Sub-Bituminous coal, A: Anthracite coal, and
L: Lignite).
Author's personal copy
Y. Cao et al. / Fuel 87 (2008) 3322–3330
3325
boilers equipped with FF. Mercury is present in the gas phase at
high temperatures during the coal combustion process. Mercury
adsorption by fly ash occurs when the flue gas temperature is decreased downstream of the boiler. In this process, rank-related coal
properties (such as the chlorine, sulfur, moisture and pore structure of fly ash), may influence mercury adsorption on the fly ash.
An apparent decreasing trend of chlorine content in coals is found
when the rank of coals decreases. This is followed by an increasing
trend in mercury emission rates. This may imply that chlorine content in the coal may be the factor affecting the mercury emission
rate. However, there is no significant correlation between the mercury emission rate and mercury speciation in the stack (Hg(0)/
Hg(VT)stack, the ratio of the elemental mercury and the total gaseous mercury), at least by available data shown in Fig. 1. The possible explanation for this could be that the great change of mercury
speciation by the interaction between gaseous mercury and fly ash
occurs after flue gas passes through the FF. The Loss On Ignition
(LOI) content of the fly ash, which is relative to boiler type and coal
rank, seemed to be correlated with the mercury emission rate
based on the limited data available, as shown in Fig. 1.
To more accurately predict the factors affecting mercury emission rates, three trials by a stepwise regression analysis based on
two available data banks were conducted. All three trials investigated the trends in mercury emission rates by different boiler
types burning different ranks of coals. In the first trial, factors included available data on boiler types, coal properties (such as coal
rank, moisture (Mcoal), ash content (Acoal), sulfur content (Scoal),
chlorine content (Clcoal), Hg content (Hgcoal) and heating value
(Btucoal)). Four factors were finally chosen by the built-up regression model based on their importance. The four factors could explain 75.4% of the variation in mercury emission rates, as shown
in Table 1. Among them, the most significant effects on mercury
emission rates are coal rank and boiler type with higher confidence
limits (very low statistical p-value). Other factors, based on a
decreasing sequence of significance (in absolute line coefficient value), were Scoal and Mcoal. According to the affecting trends, four
factors can be categorized into a group of positive factors, which
include Scoal and Mcoal; and a group of negative factors, which include coal rank and boiler type. An increase of Scoal and Mcoal leads
to an increase in the level of mercury emission rates. An increase in
the coal ranks (Level 1: lignite, Level 2: sub-bituminous, and Level
3: bituminous) and an increase in the boiler type level (Level 1: PC
and Level 2: CFBC) lead to a decrease in the level of mercury emission rates. Based on the definition of coal rank and levels of boiler
types, it was found that burning low rank coal or blending it with
higher rank coals in the conventional PC unit result in relatively
higher mercury emission rates.
In order to increase the prediction accuracy by the regression
model, one more factor, mercury speciation in the flue gas
(Hg(0)/Hg(VT)stack), was included in the model build-up in the second trial. All factors were able to explain 81.7% of the variation in
the mercury emission rate, as seen in Table 1. This is a slight
improvement over results achieved in the first trial. The most significant factors affecting mercury emission rates were still coal
rank and boiler type. Other factors, which were found to be less
significant, were Btucoal, Scoal and Mcoal. The same trends of factors
appeared repeatedly in both trials. For the new factor, Hg(0)/
Hg(VT)stack, it appeared that an increase in the Hg(0)/Hg(VT)stack level leads to an increase in mercury emission rates. It is unusual
that the critical factor on mercury speciation, Clcoal, was not a significant factor in the regression model. Nevertheless, Scoal was
found to be a factor in the regression model. It may be implied that
Clcoal, which was found to be a critical factor affecting mercury speciation, did not have a direct effect on mercury adsorption on the
fly ash. Scoal may have a direct effect on mercury adsorption on
the fly ash [3,7,16]. An alternative possibility is that coal rank,
which was positively correlated with Clcoal, may replace the function of Clcoal in the regression model. The third trial by the stepwise
regression procedure was conducted to investigate the most significant factors on the mercury emission rate. Two factors were chosen by the regression model, which are boiler type and coal rank.
These are the most significant factors affecting the variation of
mercury emission rates in this study. Those two factors were able
to explain 71.6% of the variation in mercury emission rates within
the confidence limits. The CFB burning higher rank coal can achieve
the best mercury removal efficiency among all other boilers burning the same rank of coal (temperature factor is not included in this
analysis due to less and incomplete information found in the ICR
database. Discussion on impact of this factor on mercury emission
rates has been included in a reference [20]).
Table 1
Stepwise statistical analysis on factors of mercury emission rates
Fly ash is a key point in explaining the significance of lower
mercury emission rates in CFBC units, as indicated in Fig. 1 and Table 1. The pore structure of fly ash from different coals is presented
in Fig. 2. The specific surface area of fly ash from the CFBC (approximately 15 m2/gram) is generally higher than those from other PC
boilers (generally below 10 m2/gram based on collected data in
this study). The specific surface areas of fly ashes from the PC boilers show a larger scatter, depending on coal ranks. The fly ash from
the lignite-fired PC boiler had the lowest specific surface area
(approximately 1.0 m2/gram) among the three ranks of coal. The
pore volume of fly ash increased when the specific surface area increased. However, the pore size of fly ash did not show any significant trend with different ranks of coals. In order to characterize fly
ash generated by different coal ranks and boiler types, the specific
surface area, pore volume and pore size of fly ash with boiler type
and coal rank were taken into the stepwise regression analysis.
Two factors (boiler type and coal rank) explained the 82.2% variation of the specific surface area of fly ash generated by different
boilers burning different ranks of coals, 75% variation of their pore
volume and 57.9% variation of pore size. This implies that the specific surface area of fly ash is a more significant parameter than
other physical properties of fly ash. An increase in the coal ranks
(Level 1: lignite, Level 2: sub-bituminous, and Level 3: bituminous)
leads to an increase of specific surface area. The same trend is
Code
Factor
Linear coefficient
p-value
1
2
3
4
Coal rank
Boiler type
Scoal
Mcoal
0.005
0.00733
0.00101
0.00016
<0.001
<0.001
0.021
0.063
5
6
7
8
9
10
Coal rank
Boiler type
Hg(0)/Hg(VT)stack
BTUcoal
Scoal
Mcoal
0.0044
0.00864
0.0056
<0.00001
0.0080
0.00013
<0.001
<0.001
<0.001
0.001
0.034
0.083
11
12
Coal rank
Boiler type
0.00738
0.00571
<0.001
<0.001
Adjusted-R2
75.4%
81.7%
71.6%
Level value
Coal rank
Lignite
Sub-bituminous coal
Bituminous coal
1
2
3
Boiler
PC
CFBC
1
2
3.2. The correlation of fly ash properties with mercury emission rates
Author's personal copy
3326
Y. Cao et al. / Fuel 87 (2008) 3322–3330
0.05
25
BET SSA, m2/g
Pore size, nm
Pore Volume, cm³/g
0.04
20
0.025
0.02
10
2
0.03
15
Pore Volume, cm /g
0.035
2
Specific surface area (BET, m /g) or Pore size, nm
0.045
0.015
5
0.01
0.005
0
0
0
5
10
15
20
Bituminous
CFBC
Stocker-fired
25
30
Sub-bituminous
35
40
45
Lignite
PC
Fig. 2. The factors on fly ash properties.
observed in the case of pore volume. The development of pore
structure in the fly ash might play an important part in enhancing
mercury adsorption on the fly ash in the FF, which is correlated to
coal ranks and boilers (Table 2).
Scanning electron microscopy (SEM) has been used directly in
studies of morphological changes during coal combustion in boilers. Figs. 3, 4 and 5 show SEM morphologies of different fly ashes
generated by PC boilers. Texas lignite usually generates a very
smooth and round fly ash in PC units. There are no interconnections between the few pores found within these particles. Higher
Table 2
Stepwise statistical analysis on factors of fly ash properties
Code
Factor
Linear coefficient
p-value
BET
1
2
Boiler type
Coal rank
9.51
1.39
<0.001
<0.001
Adjusted-R2
82.2%
Pore volume
3
4
Boiler type
Coal rank
0.0168
0.0085
<0.001
<0.001
75.2%
Pore size
5
6
Boiler type
Coal rank
4.6
5
<0.001
<0.001
57.9%
Level value
Coal rank
Lignite
Sub-bituminous coal
Bituminous coal
1
2
3
Boiler type
PC
CFBC
1
2
combustion reactivity of low rank lignite results in lower LOI content in its fly ash, and thus, not much carbon is left for developing
the pore structure in fly ash. Alternatively, Selenium (Se) is identified in these particles by using EDX. Selenium may form a coating
layer, which will block the pore structure of fly ash and prevent Hg
from being adsorbed on the inner surface of the ash. This may be
one of the reasons that the Texas lignite-fired boilers show low
mercury capture efficiency. There is also supporting evidence from
two other factors concerning Se plugging under lignite-fired flue
gas atmosphere. First, Se was identified on the surface of the carbon adsorbent of the Appendix K trap (EPA standard method on
mercury measurement by carbon trap), as indicated in Fig. 3-2.
The spot marked with a, b, and c on the surface of the carbon trap
was accumulated with pure Se compounds. This may explain why
the carbon trap loses its mercury capture capability in the flue gas
atmosphere when Texas Lignite is burned. Second, SnCl2 is used as
a commercial solid catalyst in the dry mercury CEM system to reduce the oxidized mercury for mercury measurement purposes. Se
is also on the surface of this catalyst to quickly deactivate under
use in the lignite atmosphere. Both solid samples (activated carbon
and SnCl2) started to lose their adsorption capability or catalyst’s
reactivity after having contacted with Texas lignite-fired flue gas
for just a half day. According to material balance from field testing,
around 60% of the total Se in coal occurs in the gas phase when
burning blended Texas lignite and PRB, while 90% of the total Se
in coal occurs in the flue gas burning Texas lignite only. The majority of coal Se is released in the flue gas during combustion, followed by its condensation on particles, such as fly ash, adsorbent
and catalyst, under a typical temperature of FF in the downstream
of boilers which are burning Texas lignite.
Lignite also is produced in the North Dakota (ND) area in the
United States. A serious deactivation was observed when SCR
catalysts were used when burning ND lignite-fired boilers [4].
Author's personal copy
Y. Cao et al. / Fuel 87 (2008) 3322–3330
3327
Fig. 3-1. SEM pictures of fly ash from Texas Lignite-fired and Sub-bituminous fired PC boilers.
Fig. 3-2. Selenium species on Appendix K trap’s carbon and catalyst from Texas Lignite-fired PC boilers.
Fig. 4. SEM pictures of fly ash from bituminous or sub-bituminous coal-fired PC unit.
The detailed study indicated that the activity loss of SCR catalysts
might be attributed to the pores of the SCR catalyst being plugged
by alkali oxides (Na and K) with a lower melting point [5,15,17]. A
high concentration of Na and K particles attaches to the surface of
the SCR catalyst, filling its pores. Thus, Se for Texas lignite and Na
and K for ND lignite may decrease the surface area of the fly ash.
Together with lower LOI content in fly ash from lignite, a higher
occurrence of Se or Na and K in lignite may be some of the major
reasons for the lower mercury capture capability.
The development of fly ash pore structure from PC boilers burning bituminous coals is presented in Fig. 4(a). Its surface was not as
smooth as those from lower rank coals. Fly ash from PC boilers
burning sub-bituminous coal shows a similar round shape, but a
much smaller particle size. Its irregular particle surface and a less
Author's personal copy
3328
Y. Cao et al. / Fuel 87 (2008) 3322–3330
Fig. 5. SEM pictures of fly ash from Bituminous coal-fired CFBC units.
developed pore structure are presented in Fig. 4(b). In comparison,
the shape of fly ash from CFBC units remains as it was originally in
the coal. The irregular shape is due to the lower temperature in
CFBC and thus less likely to melt, see Fig. 5. The round shape of
fly ash, which was generally found in PC units, is not found in CFBC
units. The pore structure is the same as that found in fly ash from
PC units burning bituminous coals. However, CFBC boilers generate
fly ash with a larger specific surface area and pore volume (see
Fig. 2). The micro-phase structures of fly ash from CFBC units could
be compared to that of the commercially available Hg sorbent
(Darco LH), which is doped with the brominated species to enhance its Hg capture capacity, as presented in Fig. 6. The surface
area of Darco LH (approximately 300 m2/gram) has a much higher
surface area than that of fly ash from the CFBC system (15 m2/
gram). One still could expect good Hg capture performance of CFBC
fly ash due to its higher content in the flue gas if the prevailing
injection rate of commercial Hg adsorbent (Darco LH) in the flue
gas is considered (a 20 times smaller concentration of Draco LH
in the flue gas).
3.3. The correlation of SDA or SNCR with mercury emission rates in
CFBC
The addition of limestone and staged combustion technologies
to CFB boilers could largely control the emissions of SO2 and NOx.
The pursuit of even lower NOx and SO2 emissions under the Clean
Air Act could be achieved by the application of SNCR with the
injection of ammonia (NH3) to reduce NOx to N2. The use of an
SDA with the injection of wet limestone or recirculation of hydrated fly ash also could further reduce the emission of SOx. The
use of SNCR technologies may also result in few ppm (below
5 ppm required by US EPA) of NH3 slipping into the flue gas of a
CFBC, where the NH3 will be adsorbed on the fly ash. A concern
is raised regarding the adsorbed NH3 occupying the pore structure
of fly ash. This condition may prevent Hg adsorption on the fly ash.
In the SDA system, additional wet limestone or fly ash is injected
into the duct, which may benefit enhanced Hg capture. This is
due to the increased content of solid particles in the flue gas.
Two detailed investigations were conducted by the WKU ICSET
team in two selected CFBC units, which demonstrated the effects
of SNCR or SDA on the variation of mercury emission rates. The
data are presented in Figs. 7 and 8.
In Fig. 7 the mercury emission rate from CFBC unit#1 varied
slightly. The mercury emission rate was approximately
1.5 103 (lg/NM3/(lg/g Btu/lb)) on average in the first 3 days
during the stable injection of NH3 in the SNCR system. When the
NH3 injection rate decreased, the mercury emission rate decreased
to approximately 4.0 104 (lg/NM3/(lg/g Btu/lb)) on the fourth
day. However, it remained at approximately 4.0 104 (lg/NM3/
(lg/g Btu/lb)) when NH3 injection returned to a higher level. This
may imply that the NH3 injection does not have any significant correlation between mercury emission rates of a CFBC with equipped
SNCR. Results of mercury emission rates were confirmed by three
methods: SCEM, ASTM OHM and EPA Appendix K. The injection
of NH3 in SNCR in No. 2 CFBC without an SDA unit did not have
any significant effect on the variation of mercury emission rates,
see Fig. 8. The average mercury emission rates were approximately
1.0 104 (lg/NM3/(lg/g Btu/lb)). The comparison of mercury
emission rates at both units indicated that there was no evidence
to show that the use of an SDA on mercury emission rates. Results
were confirmed by both the SCEM and OHM in the stack.
4. Conclusion
Fig. 6. SEM pictures of Hg Sorbent (Darco-LH).
Based on best available data collected from the EPA ICR program and WKU ICSET’s mercury testing program, a statistical stepwise regression procedure was used to determine significant
factors on mercury emissions during coal combustion. Investigations on the dependence of mercury emissions on coal ranks and
electric utility boilers equipped with fabric filter baghouses (FF)
indicate:
Author's personal copy
3329
Y. Cao et al. / Fuel 87 (2008) 3322–3330
0.0025
180
SCEM Hg(VT)
SCEM Hg(0)
OHM-Hg(VT)
OHM-Hg(0)
Appendix K
NH3
0.002
140
120
0.0015
100
80
0.001
60
40
0.0005
20
0
11-27
11-28
11-29
11-30
12-1
12-2
0
12-4
12-3
Date
0.0014
SCEM-Hg(0)
SCEM-Hg(VT)
OHM-Hg(0)
OHM-Hg(VT)
3
mercury emission rate by Hg(0) or Hg(VT) at Stack, ( g/NM /( g/g
Btu/lb))
Fig. 7. Mercury emission rate on stack of selected No. 1 CFBC unit.
0.0012
SNCR on (NH3 Injection)
SNCR off
0.001
0.0008
0.0006
0.0004
0.0002
0
11-29
12-4
12-9
12-14
12-19
12-24
12-29
Date
Fig. 8. Mercury emission rate on stack of selected No. 2 CFBC unit.
1-3
1-8
1-13
NH3 injection rate, lb/hr
mercury emission rate by Hg(VT) or Hg(0),
3
( g/NM /( g/g Btu/lb))
160
Author's personal copy
3330
Y. Cao et al. / Fuel 87 (2008) 3322–3330
(1) Higher mercury emission rates were generally found in both
CFB and PC units when lignite was burned. Higher mercury
emission rates observed during lignite-fired boilers are
likely due to lower specific area of fly ash, which results
from lower LOI, as well as the pore blockage by elements
of Selenium (Se) for Texas lignite; and Sodium (Na) and
Potassium (K) for North Dakota lignite.
(2) Lower mercury emission rates were generally found in both
CFB equipped with FF and PC units equipped with FF when
bituminous coal was burned. There was a statistically significant lower mercury emission in the CFBC equipped with FF
than that in the PC units when sub-bituminous coal was
burned. Lower mercury emission rates in electric utility
boilers equipped with FF are due to their active fly ash generated with a larger specific surface area and pore volume.
(3) There is no significant mutual benefit for the mercury captured by the addition of Spray Dry Absorber (SDA) or Selective Non-Catalytic Reduction (SNCR) in the CFBC system.
Acknowledgements
We gratefully acknowledge mutual financial supports through
projects by the United State Department of Energy (DE-FC2603NT41840), Kentucky Office of Energy Policy (PO2 855
0600002929 1) and USDA-ARS project (No. 6406-12630-002-02S).
References
[1] Anthony EJ. Fluidized bed combustion of alternative solid fuels, status,
successes and problems of the technology. Prog Energy Combust Sci
1995;21:239–68.
[2] Beer JM. Combustion technology developments in power generation in
response to environmental challenges. Prog Energy Combust Sci
2000;26:301–27.
[3] Behr P, Bonnissel-Gissinger P, Alnot M, Revel R, Ehrhardt J. XPS and XAS study
of the sorption of HgO onto Pyrite. Langmui 2001;17:3970–9.
[4] Benson SA, Laumb JD, Croker CR, Pavlish JH. SCR catalyst performance in flue
gases derived from sub-bituminous and lignite coals. Fuel Process Technol
2005;86:577–613.
[5] Crocker CR, Benson SA, Laumb JD. SCR catalyst blinding due to sodium and
calcium sulfate formation. Prep Pap-Am Chem Soc Div Fuel Chem
2004;49(1):169–70.
[6] Eddinger J. Statistical analysis of mercury test data to determine BDT for
mercury emissions. March 15, 2004. <dnr.wi.gov/org/aw/air/reg/mercury/
camr/Hgstatrevision.doc.pdf>.
[7] Kellie S, Cao Y, Duan YF, Li LC, Chu P, Mehta A, et al. Factors affecting mercury
speciation in a 100-MW coal-fired boiler with low-NOx burners. Energ Fuels
2005;19:800–6.
[8] Kilgroe JD, Sedman CB, Srivastava RK, Ryan JV, Lee CW, Thorneloe SA. Control
of Mercury Emissions from Coal-Fired Electric Utility Boilers: Interim Report,
EPA-600/R-01-109, December 2001.
[9] Longwell JP, Rubin ES, Wilson J. Coal: energy for the future. Prog Energ
Combust Sci 1995;21:269–360.
[10] Maxwell WH. Revised new source performance standard (NSPS) statistical
analytical for mercury emissions. May 31, 2006. Available online at <epa.gov/
ttn/atw/utility/NSPS-053106.pdf>.
[11] Maxwell B. Revised new source performance standard (NSPS) statistical
analytical for mercury emissions,” October 21, 2005. Available online at
<dnr.wi.gov/org/aw/air/reg/mercury/camr/Hgstatrevisiondoc.pdf.
[12] No. 05-1097. United States Court of Appeals For The District of Columbia
Circuit On Petitions for Review of the Final Action of the Environmental
Protection Agency. Argued December 6, 2007. Decided February 8, 2008.
[13] Pavlish JH, Scondreal EA, Mann MD, Olson ES, Galbreath KC, Laudal DL, et al.
Status review of mercury control options for coal-fired power plant. Fuel
Process Technol 2003;82:89–165.
[14] Srivastava Ravi K, Hutson N, Martin B, Princiotta F, Staudt J. Control of mercury
emissions from coal-fired electric utility boilers. Environ Sci Technol
2006;40:1385–93.
[15] Rigby K, Johnson R, Neufort R, Gunther P, Hurns E, Klatt A, et al. SCR catalyst
design issues and operation experience: coals with high arsenic
concentrations and coals from power river basin. Presented at the
International Joint Power Generation Conference: UPGC2000-15067, July 2326, 2000.
[16] Rio S, Delebarre A. Removal of mercury in aqueous solution by fluidized bed
plant fly ash. Fuel 2003;82:153–9.
[17] Senior CL, Linjewile T, Denison M, Swensen D, Shino D, Bockelie M, et al. SCR
deactivation mechanisms related to alkali and alkali earth elements.
Conference on Selective Catalystic Reduction (SCR) and Selective NonCatalytic Reduction (SNCR) for NOx Control. Pittsburgh, PA, October 20–30,
2003.
[18] Shea DA, Parker L, McCarthy JE, Chapman T. Mercury emission from electric
generating units: a review of EPA analysis and MACT determination, CRS
Report for Congress. Order Code RL32744. January 21, 2005.
[19] Sing KSW. Adsorption methods for the characterization of porous materials.
Adv Colloid Interfac 1998;76–77:3–11.
[20] Sjostrom S, Bustard J, Durham M, Chang R. Mercury Removal Trends in FullScale ESPs and Fabric Filters, presented at EPRI-EPA-DOE-AWMA Mega
Symposium and Mercury Conference, Chicago, IL, August 21–23, 2001.
[21] US Environmental Protection Agency; Clean Air Mercury Rules, March 2005;
<http://www.epa.gov/mercuryrule/index.htm>. Available online in December,
2006.
[22] US Department of Energy, Energy Information Administration. US Coal
Reserves: 1997 Update. DOE/EIA-0529(97). Office of Coal, Nuclear, Electric
and Alternate Fuels, Office of Integrated Analysis and Forecasting, Washington,
DC. February 1999. Available online at: <http://www.eia.doe.gov/cneaf/coal/
reserves/front-1.html>.
[23] US Department of Energy. Tracking new coal-fired power plants – coal’s
resurgence in electric power generation. January 24, 2007.
[24] United States Environmental Protect ion Agency. Control of mercury emissions
from coal-fired electric utility boilers: Interim Report Including Errate Dated 321-02. EPA- 600/R-01-109. April 2002.
[25] The JEA Large-Scale CFB Combustion Demonstration Project. US DOE Topical
Report Number 22, March 2003. Available online at <http://www.
fossil.energy.gov/programs/powersystems/publications/Clean_Coal_Topical_
Reports/topical22.pdf>.