EPA/600/R-06/156
September 2006
EVALUATION AND MITIGATION OF VISIBLE ACIDIC
AEROSOL PLUMES FROM COAL FIRED POWER
BOILERS
FINAL REPORT
EPA Contract No. EP-C-04-056
Work Assignment Numbers 0-3 & 1-4
Southern Research Institute Project No. 11402
November 14, 2006
By:
Peter M. Walsh, Joseph D. McCain, and Kenneth M. Gushing
Southern Research Institute
2000 Ninth Ave. South
P.O. Box55305
Birmingham, AL 35255-5305
For:
C. Andrew Miller, Project Officer
U.S. Environmental Protection Agency
Office of Research and Development
National Risk Management Research Laboratory
Air Pollution and Prevention Control Division
Research Triangle Park, NC 27711
-------
Notice
The U.S. Environmental Protection Agency through its Office of Research and Development
funded and managed the research described here. It has been subjected to the Agency's review
and has been approved for publication as an EPA document. Mention of trade names, products,
or services does not convey, and should not be interpreted as conveying, official EPA approval,
endorsement or recommendation.
-------
Foreword
The U.S. Environmental Protection Agency (EPA) is charged by Congress with protecting the
Nation's land, air, and water resources. Under a mandate of national environmental laws, the
Agency strives to formulate and implement actions leading to a compatible balance between
human activities and the ability of natural systems to support and nurture life. To meet this
mandate, EPA's research program is providing data and technical support for solving
environmental problems today and building a science knowledge base necessary to manage our
ecological resources wisely, understand how pollutants affect our health, and prevent or reduce
environmental risks in the future.
The National Risk Management Research Laboratory (NRMRL) is the Agency's center for
investigation of technological and management approaches for preventing and reducing risks
from pollution that threaten human health and the environment. The focus of the Laboratory's
research program is on methods and their cost-effectiveness for prevention and control of
pollution to air, land, water, and subsurface resources; protection of water quality in public water
systems; remediation of contaminated sites, sediments and ground water; prevention and control
of indoor air pollution; and restoration of ecosystems. NRMRL collaborates with both public and
private sector partners to foster technologies that reduce the cost of compliance and to anticipate
emerging problems. NRMRL's research provides solutions to environmental problems by:
developing and promoting technologies that protect and improve the environment; advancing
scientific and engineering information to support regulatory and policy decisions; and providing
the technical support and information transfer to ensure implementation of environmental
regulations and strategies at the national, state, and community levels.
This publication has been produced as part of the Laboratory's strategic long-term research plan.
It is published and made available by EPA's Office of Research and Development to assist the
user community and to link researchers with their clients.
Sally Gutierrez, Director
National Risk Management Research Laboratory
in
-------
TABLE OF CONTENTS
List of Figures v
List of Tables viii
Nomenclature ix
Abstract xivv
1. Introduction 1
2. Evaluation of the Possible Extent of the SOs Emissions Problem in the
Electric Utility Industry 5
3. Improvements in SOs Measurement Technologies 13
4. Exploratory Study of 863 Adsorption by Coal Fly Ash 20
5. Formation of Sulfur Tri oxide in the Convection Section of Coal-Fired
Electric Utility Boilers 30
6. Development of a Estimator for Removal of SO3/H2SO4 Across Air Preheaters 47
7. Evolution of Acid Mist in Wet Flue Gas Desulfurization Units and Stacks 57
8. References 84
Appendix A. Correlations for sulfuric acid density, sulfuric acid surface energy,
and the vapor pressure of water over sulfuric acid at 333.15 K (140 °F) A-l
Appendix B. Computer code for the calculation of acid mist properties in the
scrubber and stack B-l
IV
-------
LIST OF FIGURES
Figure Page
Figure 2. 1 . Measured SOs concentrations versus temperature at the exit of an
air heater and cold-side electrostatic precipitator [[[ 10
Figure 2.2. 863 emissions for states in the Mississippi Valley and Eastern U.S.
emitting > 1% of total projected SOs emissions for the region .............................. 12
Figure 3.1. Schematic diagram of the Apogee QSIS probe modified for
use with the SOs monitor [[[ 14
Figure 3.2. Comparison of SO3 concentrations measured by conventional controlled-
condensation methods (CCM) with those measured using Apogee
QSIS probes and a heated hose to deliver samples to a controlled-
condensation condenser [[[ 16
Figure 3.3. Comparisons of 863 concentrations measured by the conventional controlled-
condensation method (CCM) with those measured using the modified
Apogee QSIS probe incorporating the hot gas dilution approach to
evaporating condensed H2SO4 [[[ 17
Figure 3.4. Modified Apogee QSIS sampling probe [[[ 19
Figure 4.1. Layout of ash samples in the exposure chamber as seen from above ...................... 26
Figure 4.2 Typical concentration versus time measurements at the inlet and outlet
of the SOs exposure chamber [[[ 27
Figure 4.3. Total soluble sulfate found in ash samples after exposure to 863 at the
conditions indicated versus ash base:acid ratios [[[ 28
Figure 4.4. Plot of SO4 on ash as predicted by the regression equation and the
-------
List of Figures (Continued)
Figure Page
Figure 5.3. Surface temperature and SOs formation in the superheater under the
assumption of moderate fouling on the steam side, with tube scale
exposed to flue gas 34
Figure 5.4. Comparison of the calculated SOs formation to measurements for the
Hickling Station modeling 35
Figure 5.5. Equilibrium and kinetic constraints on SOs formation ............................................ 39
Figure 5.6. Modeled temperature profiles through a 1300 MW pulverized
coal-fired utility boiler [[[ 43
Figure 5.7 SOs concentration profile through a 1300 MW pulverized coal-fired
utility boiler as predicted by the model [[[ 43
Figure 6. 1 . Estimated SO3/H2SO4 losses across combustion air preheaters versus
average air preheater exit temperature for a temperature offset of 35 °F ................ 48
Figure 6.2. Estimated air preheater exit SO3/H2SO4 concentration versus average
air preheater exit temperature for a temperature offset of 3 5 °F .............................. 49
Figure 6.3 . Effect of flue gas moisture content on the estimated air preheater exit
SO3/H2SO4 concentration for a temperature offset of 35 °F .................................... 50
Figure 6.4. Estimated SOs/H^SC^ losses across combustion air preheaters versus
average air preheater exit temperature for a temperature offset of 30 °F ................ 51
Figure 6.5. Estimated air preheater exit SOs/H^SC^ concentration versus average
air preheater exit temperature for a temperature offset of 3 0 °F .............................. 52
Figure 6.6. Comparison of predicted and measured 863 concentrations at the exit
of rotary air heater No. 1 for a temperature offset of 30 °F .................................... 53
Figure 6.7. Comparison of predicted and measured SO3 concentrations at the exit
of rotary air heater No. 2 for a temperature offset of 30 °F ..................................... 54
Figure 6.8. Comparison of predicted and measured SOs concentrations at the exit
of rotary air heater No. 3 for a temperature offset of 33 °F ..................................... 55
-------
List of Figures (Continued)
Figure Page
Figure 7.1. Properties of acid mist as a function of time in the SC>2 scrubber
and stack—base case 68
Figure 7.2. Properties of acid mist as a function of time in the SC>2 scrubber
and stack—increased H^SC^ case 71
Figure 7.3. Properties of acid mist as a function of time in the 862 scrubber
and stack—increased scrubber droplet size case 73
Figure 7.4. Properties of acid mist as a function of time in the SO2 scrubber
and stack—decreased scrubber droplet size case 74
Figure 7.5. Properties of acid mist as a function of time in the SC>2 scrubber
and stack—increased size of droplets passing the mist eliminator 76
Figure 7.6. Measured size distributions of acid mist and other particulate components
downstream of scrubber module "A." 79
Figure 7.7. Measured size distributions of acid mist and other particulate
components in the stack downstream of a scrubber system 80
Figure 7.8. Measured size distributions of acid mist in the stack as shown in
Figure 7.7 after replacement of the scrubber system with one of
a different type 81
Figure 7.9. Comparison of droplet diameters extrapolated from measurements
with those predicted by the model 82
Figure 7.10. Comparison of predicted plume opacities versus
concentration with those measured by a certified "smoke reader"
for a 1300 MW unit with a pollution control system consisting of
an SCR followed by a cold-side ESP and an SC>2 scrubber 83
Vll
-------
LIST OF TABLES
Table
Table 2.1. Ratio of SO3/SO2 based on 16 field tests
Table 2.2. Number of units predicted to exceed selected values of 863 concentration
under scenario assumptions 7
Table 2.3. Sulfur contents of coals fired in plants in the Mississippi Valley and Eastern U.S 8
Table 2.4. SOs removal efficiencies for control system components 9
Table 2.5. Comparisons of observed and predicted SO3 concentrations 9
Table 2.6. State-by-state tabulation of estimated SOs emissions from individual
power plant units in the Mississippi Valley and Eastern U. S 11
Table 3.1 Comparison of measurements made using a conventional CCM probe
with those made with the hot-gas dilution probe downstream of a
full-scale utility scrubber 18
Table 4.1. Exposure conditions and results 22
Table 4.2. Properties of ashes used in adsorption study 23
Table 4.3. Atomic concentrations of cations for ashes used in adsorption study 24
Table 4.4. Results of multiple regression to obtain predictive equation for uptake of SOs 25
Table 5.1 Conditions for the SOs model run whose results are shown in
Figures 5.6 and 5.7 42
Table 6.1. Measured and predicted average exit SO3 concentrations for four
rotary air heaters 47
Table 7.1. Values of the parameters used in the calculations for the base case 70
Vlll
-------
NOMENCLATURE
cad mean speed of acid mist droplets, m/s
Cfg mean speed of flue gas molecules, m/s
cw mean speed of water vapor molecules, m/s
Cad concentration of acid mist droplets, kg/m3
Cado initial concentration of acid mist droplets, kg/m3
Cbi average gas concentration in the boundary layer of a tube or particle, kmol/m3
Cc Cunningham correction factor, Eq. (7.9), dimensionless
CD drag coefficient for scrubber spray droplets, dimensionless
Cg total gas concentration, kmol/m3
Cu2so4 concentration of H^SO/t, kg/m3
Cs total gas concentration at a deposit or particle surface, kmol/m3
Csd concentration of scrubber spray droplets, kg/m3
Cw,ad concentration of water vapor in equilibrium with the surfaces of acid mist droplets,
kg/m3
Cw,sd concentration of water vapor in equilibrium with the surfaces of scrubber spray
droplets, kg/m3
CW;00 concentration of water vapor in the free stream, far from droplet surfaces, kg/m3
d characteristic length, diameter, m
dad diameter of acid mist droplets, m
dado initial diameter of acid condensation nuclei, m
dp mean particle size, m
dsd diameter of scrubber spray droplets, m
dt tube diameter, m
Dad diffusion coefficient for acid mist droplets, m2/s
IX
-------
NOMENCLATURE (CONTINUED)
DW:fg pseudo-binary molecular diffusion coefficient for a mixture of water vapor with the
other combustion products, m2/s
DSOS molecular diffusion coefficient of 863 in combustion products, m2/s
ESP electrostatic precipitator
fcap fraction of initial H2SO4 captured by scrubber spray droplets, dimensionless
FGD flue gas desulfurization
kB Boltzmann constant, = 1.38 x 10"23 J/K
k effective rate coefficient for the heterogeneous reaction between SO2 and O2, m/s
kt global reaction rate constant for compound /'
kad rate coefficient for coagulation, m3/s
kt effective rate coefficient for the heterogeneous reaction between SO2 and O2 at tube
or ash deposit surfaces, m/s
kp effective rate coefficient for the heterogeneous reaction between 862 and 62 at
particle surfaces, m/s
ksa effective overall rate coefficient for transport of water vapor between scrubber
spray droplets and acid mist droplets, Eq. (7.24), dimensionless
kj rate coefficient for Reaction (5.4), m/s
&2 rate coefficient for Reaction (5.5), m/s
Knadiad Knudsen number for diffusion of droplets toward each other, dimensionless
Krifg,ad Knudsen number for acid mist droplets in flue gas, dimensionless
KnWtad Knudsen number for water vapor diffusing to and from acid mist droplets in flue
gas, dimensionless
LG liquid to gas feed ratio in the scrubber, gal/1000 scf or dimensionless
mad mass of acid mist droplets, kg
Nad number concentration of acid mist droplets, m"3
initial number concentration of acid mist droplets, m"3
-------
NOMENCLATURE (CONTINUED)
P\v,ad partial pressure of water vapor in equilibrium with the surfaces of acid mist
droplets, Pa
Pw.af partial pressure of water vapor in equilibrium with a flat sulfuric acid surface, Pa
Pw.sd partial pressure of water vapor in equilibrium with the surfaces of scrubber spray
droplets, Pa
rso3 rate of sulfur trioxide formation, kmol/m3-s
R universal gas constant, = 8314.51 J/(kmol-K)
Re Reynolds number, dimensionless
S effective area of catalytic surface per unit of gas volume, nr1
Sad specific surface area of acid mist droplets, m2/kg
Sp external surface area of fly ash particles per unit of gas volume, nr1
Ssd specific surface area of scrubber spray droplets, m2/kg
St surface area of tube or ash deposits on a tube, per unit of gas volume, m'1
Sc Schmidt number, dimensionless
SCR selective catalytic reduction
Sh Sherwood number for mass transfer, dimensionless
Shp Sherwood number for mass transfer through the external boundary layer of a
particle, = 2, dimensionless
Shsd Sherwood number for mass transfer to and from scrubber spray droplets,
dimensionless
ShWtad Sherwood number for transport of water vapor to and from acid mist droplets,
dimensionless
Sht average Sherwood number for mass transfer between tube or deposit surface and
the free stream, dimensionless^ time, s
t residence time, s
t0 residence time at the entrance to a row of tubes, s
XI
-------
NOMENCLATURE (CONTINUED)
T absolute temperature, K
Ufg velocity of flue gas in scrubber, m/s
usd terminal fall velocity of scrubber spray droplets relative to the scrubber, m/s
Wfg average molecular weight of flue gas, kg/kmol
WH2so4 molecular weight of H2SO4, kg/kmol
WH2so4-Hp molecular weight of H2SO4-H2O, kg/kmol
Ww molecular weight of water, kg/kmol
Xi mole fraction of species i, dimensionless
X$o3 mole fraction of SOs in the free stream, dimensionless
Xso3,eq mole fraction of SOs at equilibrium, dimensionless
Xso3,eq,p mole fraction of 863 at equilibrium at particle surfaces, dimensionless
Xso3,eq,t mole fraction of SOs at equilibrium at tube or ash deposit surfaces, dimensionless
X$o3,s mole fraction of SOs at catalyst surface, dimensionless
Xso3,o initial mole fraction of SOs at the entrance to a row of tubes, dimensionless
Ya mass fraction of H2SO4 in sulfuric acid, dimensionless
Yao initial mass fraction of H2SO4 in acid condensation nuclei, dimensionless
Ype2o3 mass fraction of all iron oxides in deposit or particles, from ash analysis,
dimensionless
YFeox mass fraction of higher iron oxide in deposit or ash particles, dimensionless
Ypeo mass fraction of reduced iron oxide in deposit or ash particles, dimensionless
X-l
Xll
-------
NOMENCLATURE (CONTINUED)
Greek Symbols
ft factor matching droplet diffusion behavior in the continuum and free molecule
regimes, Eq. (7.3), dimensionless
At small increment of time, s
Xad mean free path of acid mist droplets, m
Xfg mean free path of flue gas molecules, m
}iw mean free path of water vapor molecules in flue gas, m
]Ufg viscosity of flue gas, kg/(m-s)
pa density of sulfuric acid, kg/m3
psd density of scrubber spray droplets, kg/m3
aa surface energy of sulfuric acid, J/m2
Subscripts
a acid
ad acid mist droplets
af flat acid surface
c Cunningham (correction factor)
cap captured (acid by scrubber spray droplets)
D drag
eq equilibrium
fg flue gas
sa scrubber spray droplets to acid mist droplets
sd scrubber spray droplets
w water vapor
0 initial value at entrance to scrubber
oo free stream, far from droplet surfaces
Xlll
-------
ABSTRACT
The formation of sulfur trioxide (863) during the combustion of sulfur-containing fuels,
particularly coal, can increase significantly following the installation and operation of selective
catalytic reduction (SCR) systems for reduction of nitrogen oxides (NOX). The increased SOs
formation can in turn lead to adverse environmental impacts, including visible near-stack plumes
and increased fine PM emissions, primarily in the form of sulfuric acid (H2SO4) aerosols. The
potential extent of the problem in the electric utility sector is estimated based on the population
of coal-fired utility boilers, the sulfur content of coal burned by each unit, and the likelihood that
units will install SCR and flue gas desulfurization (FGD) systems. Of the 363 large (> 250
MWe) generating plants in the eastern U.S., there is a significant potential that as many as 65
could experience visible H2SO4 aerosol plumes or more serious problems after installation of
SCR and FGD systems, based on the sulfur content of the coal historically used at those plants.
As use of FGD systems increases, it is also likely that utilities will turn to higher sulfur coal,
which can exacerbate this problem. This report describes the mechanisms of 863 and acid
aerosol formation and removal across boiler convection sections, air preheaters, and wet FGD
systems, and presents information from an exploratory study of the absorption of SOs onto coal
fly ash. A model of SOj formation and emissions based on these mechanisms is shown to
accurately predict the stack concentration of 863 for a 1300 MWe pulverized coal-fired boiler,
indicating that the mechanisms described have captured the fundamental behavior of SOs in
utility combustion and flue gas treatment systems. This information can provide the basis for
developing mitigation approaches to reduce the impacts of SO3 formation across SCRs and the
subsequent formation and emission of acid aerosols.
XIV
-------
1. INTRODUCTION
Background
In 2006 the Environmental Protection Agency (EPA) promulgated the "Clean Air Interstate
Rule" (formerly called the "Interstate Air Quality Rule") with the goal of reducing sulfur dioxide
(862) and oxides of nitrogen (NOX) emissions in the Eastern U.S. When fully implemented, the
rule is projected to reduce 862 emissions by over 70% from 2003 levels and reduce NOX
emissions by over 60% from 2003 levels.1 Achieving these goals is expected to involve
widespread use of selective catalytic reduction (SCR) for NOX control and flue gas
desulfurization (FGD) scrubbers for 862 control in coal-fired power plants. In fact, current usage
of these technologies has already led to substantial reductions in emissions of NOX and 862.
However, the use of SCR for NOX control has the potential to enhance the formation of sulfur
trioxide (SOs).2 At elevated concentrations, SOs forms visible plumes and excessive local
concentrations of sulfuric acid (H2SO4) aerosols.
Elevated SOs concentrations are of concern due in part to associated potential adverse health
effects. Particularly in areas that can be affected by sinking plumes that fall to ground level near
the stack, increases in PM smaller than 2.5 um in aerodynamic diameter (PM2.5) caused by high
H2SO4 concentrations are of concern. At typical ambient concentrations, the health impacts of
H2SO4 aerosols are unclear. Some evidence of impaired mucociliary clearance and modest
changes in lung function has been identified,3 but there is limited evidence of significant airway
inflammation or altered bronchial responsiveness as a result of exposure to typical ambient levels
of H2SO4 aerosol.4 In sensitive populations5 or in combination with other pollutants,6'7 exposure
to H2SO4 may lead to more serious problems compared to exposure to H2SO4 alone.
Where plumes intercept the ground, effects associated with the higher concentrations associated
with occupational exposure levels may be more applicable than effects associated with ambient
concentrations. A significant difference is that near-stack plume contact incidents can result in
concentrations much higher than typical ambient levels, but much shorter in duration and less
frequent than occupational exposures. At elevated occupational levels, exposure to 863 can
cause lung irritation, fluid build-up in the lungs (pulmonary edema), third-degree burns to the
skin, and blindness. Long-term, lower level occupational exposure can lead to bronchitis,
emphysema, chronic runny nose, tearing of the eyes, nosebleeds, headaches, nausea, dizziness,
and erosion of the stomach and teeth.8
There have not been any reported instances of the more serious effects associated with non-
occupational exposure to SO3 or H2SO4 aerosol, although physiological responses such as eye,
nose, and throat irritation and breathing difficulty have been reported for episodes involving
plumes containing sulfuric acid aerosols at ground level.9'10 In an incident documented in 2001
involving the American Electric Power Gavin Power Plant, residents downwind of a plant having
a plume that touched the ground complained of irritation of the eyes, nose, and throat; shortness
of breath; and asthma-like symptoms. Measurements of H2SO4 levels in the area were found to
be between 35 and 120 ug/m3.11 At the same time, ground-level SC>2 concentrations were
measured at 20-50 ug/m3, so it is not clear whether the effects were caused by H2SO4, SC>2, or the
combination of the two along with the other pollutants in the plume. In either case, the presence
-------
of increased levels of acid aerosols and 863 is associated with adverse health impacts at
concentrations that have been measured in plumes that have contacted the ground.
In addition to the public health impacts of increased SO3 formation, plant components can be
adversely impacted as well.12'13 Increases in 863 formation and emissions can clearly cause
problems that need to be addressed for several reasons, but as noted above can also be a
consequence of efforts to achieve significant reductions in SC>2 and NOX. This report will discuss
efforts to better understand SO3 formation processes, information that can lead to cost-effective
mitigation approaches and minimization of the adverse impacts of 863.
Project Description
The purpose of this project was to conduct research that would improve EPA's understanding of
the formation mechanisms of SOs and the means by which power plants can minimize the
formation of sulfuric acid emissions. These included characterizing the potential significance of
the problem and assessing interactions between coal type, pollution control system design and
operation, and emissions of 863, H2SO4, NOX and 862.
This research is important to efforts to achieve targeted reductions in emissions of NOX and SC>2
while minimizing adverse consequences of control technology application. Combined use of
SCR for NOX control and FGD for 862 control in coal-fired power plants has the potential to
enhance the formation of SOs. At elevated concentrations, SOs forms visible plumes and
excessive local concentrations of sulfuric acid aerosols. Such indirect effects of SCR use must
be more fully understood in order to meet regulatory goals of reducing ambient levels of SO2,
NOX, and paniculate matter (PM) in a way that does not create new air pollution problems. In
particular, states need information related to SOs emissions and its mitigation for the purpose of
developing State Implementation Plans (SIPs). The tasks performed during this research were
intended to provide that information to EPA and the states.
Earlier work conducted by Southern Research Institute (SRI) for EPA produced an interim report
that described the current state of understanding of SOs formation and mitigation.2 The report
concluded that "the difficulty faced by utility plant operators is that guidelines are not presently
available to define the extent of the problem that may occur at a given site. In the absence of
such guidelines, it is not possible to determine in advance the control strategies that are
technically and economically feasible for a particular plant site." This lack of guidelines is
primarily the result of a lack of sufficient credible data on which to base models for control of
863 emissions. In particular, 863 collection by combustion air preheaters, electrostatic
precipitators and scrubbers were poorly characterized.
The current report describes research conducted by SRI to characterize methods to measure SO3,
develop model plant specifications, estimate acid droplet growth rates, and characterize
absorption and adsorption of SOs by fly ash. The results reported here provide the basis for more
accurate estimates of SOs formation and emissions and identify research needs for further
improvement in understanding SO3 formation and removal across different plant components.
-------
Report Overview
Chapter 2 describes a survey of the potential extent of SOs emissions in the electric utility
industry under current regulations, including the Clean Air Interstate Rule. The chapter includes
a state-by-state tabulation of potential SO3 emissions from individual power plant units and
identified those areas where the greatest uncertainties exist (e.g., SCR conversion rates versus
scrubber removal efficiencies).
Chapter 3 addresses the methods used to measure SO3 in combustion flue gases, and describes
efforts to develop improved measurement methods. In particular, Chapter 3 discusses tests of a
modified sampling probe which permits extraction of a filtered gas sample without passing the
sample through a layer of ash. Results are presented on exploratory tests of this probe for use
with the controlled condensation method. The tests, conducted at SRI's Coal Combustion
Research Facility, demonstrate improvements in 863 measurements using the modified probe.
Additional, preliminary results are presented for tests at a full-scale operating power plant.
Chapter 4 covers the absorption and adsorption of SO3 by coal fly ash for coals with significantly
different ash properties. Absorption and adsorption of 863 and H2SO4 can be a major controlling
factor in the overall process. Chapter 4 reports on tests that passed an SOs laden gas stream over
an ash layer in a temperature and moisture controlled environment. Ashes from various coals
were tested, and analyses are presented on the soluble sulfate levels for the untreated ashes and
the ash specimens after exposure to 863. Test conditions were chosen to be representative of
those found in typical economizer and air preheater sections.
Chapters 5 and 6 describe, respectively, the behavior of SO3 in boiler convection sections and air
preheaters. Chapter 5 provides quantitative results on the formation and removal of 863 across
the convection sections (e.g., convective sections, superheaters, reheat sections, economizer).
Chapter 6 presents results of estimates for SOs removal across four preheaters and compares the
estimates to measured 863 concentrations at the preheater exits. These results can be used to
model 863 formation, removal, and emissions for plants with different configurations and
operating conditions to more effectively estimate impacts of equipment, fuel, and operating
changes on SOs emissions and identify mitigation approaches.
Chapter 7 presents the development of an algorithm to describe the evolution and behavior of
acid mists in wet FGD units, including mist eliminators, and plant stacks. The algorithm is used
to evaluate how changing parameters such as Ft^SC^ concentration, scrubber droplet size, and the
size of droplets exiting the mist eliminator influence the mass fraction of H2SO4 in the stack
aerosol and the scrubber droplets, as well as the acid droplet number concentration and volume
fraction. These results are then used to estimate the impacts of changing scrubber type and are
compared to measurements from a full-scale plant.
The models presented in Chapters 5-7 are in the developmental stages and have had only
minimal validation with measurements. Although they can be used to address performance at
specific plants to identify where SOs is most likely to be formed or removed, there has not been
adequate comparison with measurements to allow significant certainty in the predictions.
-------
Additional measurements are needed along the entire flue gas path at several plants and with
different coals to enable more certain estimates to be made.
-------
2. EVALUATION OF THE POSSIBLE EXTENT OF THE SO3
EMISSIONS PROBLEM IN THE ELECTRIC UTILITY
INDUSTRY
With the promulgation of the Clean Air Interstate Rule (CAIR), the electric utility industry is
beginning to install technologies for control of NOX and SC>2. These newly installed technologies
have the potential to increase emissions of SO3 above current levels. To determine whether such
changes would be likely to impact a significant fraction of plants, and if so, the degree of such
impacts, a simple spreadsheet-based model was developed to evaluate the possible extent of the
issue. The model collected data on basic plant size and design parameters and on coal use
patterns. These data were coupled with reasonable average rates of SO2 conversion to SO3 and
removal of 863 by common pollution control systems to develop an overall picture of what
broad application of SCR and wet FGD scrubbers may mean for SOs emissions. This evaluation
can provide some overview information on the extent to which SOs mitigation measures may be
needed.
However, these estimates are highly uncertain and cannot be used to develop estimates for a
specific plant. The uncertainties stem from the use of "reasonable average" values for SO2
conversion and SO3 removal, and particularly from an inability to predict how plants will work
to meet the CAIR requirements. It is likely that the pattern of coal use will change as some
plants meet the SO2 requirements by switching to lower sulfur coal, while others switch to
higher sulfur coal following installation of FGD systems. Some plants may install either SCR or
wet FGD, while others install both. Use of spray dryer systems or installation of fabric filters
will also reduce a plant's 863 emissions compared to what would be estimated from the
spreadsheet model. Finally, the trading approach used by CAIR will likely result in
combinations of control efficiencies that cannot be incorporated into a simple spreadsheet model.
Each of these factors combine to increase the uncertainties associated with the estimated results.
Even so, the model does provide some order-of-magnitude estimate of the extent to which the
SOs issue may be significant for the electric utility industry.
The evaluation conducted by Southern Research Institute is based on data from the Department
of Energy (DOE) Federal Energy Regulatory Commission (FERC) Annual Steam-Electric Plant
Operation and Design Database for 2002,14 augmented when possible with additional
information obtained from the U. S. EPA15 and an industry database16 to fill gaps in the FERC
database. For each coal-fired power plant surveyed, SOs emissions after anticipated technology
additions for control of NOX by SCR and SO2 by wet scrubbers were predicted. These emissions
are referred to here as "collateral SOs emissions."
To compute the predicted SOs concentrations, SO2 concentrations were calculated from the
reported sulfur content of the coals being burned at each plant. The rate of oxidation of SO2 to
SOs depends on many factors, including the sulfur content of the fuel, the amount of excess air,
and the presence of a catalyst. For this methodology, SOs concentrations produced by the boilers
were estimated using SO3/SO2 ratios developed by Hardman, Stacy and Dismukes for eastern
bituminous, western subbituminous, and lignite coals.17 These ratios are based on measurements
-------
taken during 16 field tests and are shown in Table 2.1 below and are considered to be reasonably
representative of existing information. The coal types used in the calculations were the
predominant fuels for each unit as reported in the database. Subsequent production and removal
of SO3 was calculated for the configuration of downstream controls at each site.
Table 2.1. Ratio of SO3/SO2 Based on 16 Field Tests
Coals Burned
9 Eastern Bituminous Coals
7 Western Subbituminous and Lignite Coals
SO3/SO2 Ratio
Average
0.004
0.0011
Standard Deviation
0.003
0.0005
Table 2.2 summarizes the projected impact of collateral 863 emissions from coal-fired power
plant units in the Mississippi Valley and the Eastern U.S. after anticipated technology additions
for control of NOX and SC>2. The table summarizes the number of units exceeding the specified
concentrations of 863 in their stack emissions broken down by unit generating capacity. The
column labeled "Missing Data" refers to units in the database for which critical information such
as fuel sulfur was missing. The tabulations show the effect of retrofit of SCR and wet scrubbers
on all units. Based on a simplified assumption that the proposed regulations would result in
controls on approximately 75% of the total coal-fired generating capacity, all units larger than
roughly 250 MW would operate with SCR/scrubber combinations. The total capacity for 250
MW and larger plants is estimated here to be 200 GW. The more detailed EPA analysis done for
the Clean Air Interstate Rule (CAIR) estimates 178 GW of capacity will be retrofit with SCR and
slightly higher than that, 187 GW, will be retrofit with scrubbers.18
Table 2.2 has columns for five SOs emissions levels of increasing regulatory or health impacts.
The rows in the table for unspecified unit loads, designated as Unknown, refer to units in the
database for which unit load information was unavailable. Units with scrubbers having stack 863
concentrations of about 4 ppm and higher may have difficulty meeting 20% opacity limits if, as
is becoming more common, the opacity limits are applied to the stack plumes. At stack
concentrations of about 10 ppm and higher, visible "blue" plumes will exist even in the absence
of a scrubber. Plumes from units equipped with scrubbers reach ground level much closer to
stacks than those without scrubbers, given equal stack heights. Because stack gases are less
diluted close to the stack than at greater distances, ground level SOs concentrations near the stack
will be higher for a unit with a scrubber than for the same unit without a scrubber.2 The Gavin
Power Plant and the Cinergy/PSI Gibson Generating Station are two plants that have been faced
with substantial difficulties with local citizens and regulatory agencies because of such ground
level concentrations of SOs.9'19 Potential health effect problems have led to concern at both the
Gavin and Gibson plants where stack concentrations are approximately 13 ppm or higher.
Another potential issue is complying with particulate mass emission rate standards. If the
particulate emission standards were to change such that condensables (such as H2SO4 aerosols)
are included as part of emissions of particulate matter, SOs aerosol concentrations in some stacks
could contribute significantly to the overall particulate loading (1 ppm is equivalent to about 4
mg/dncm).
-------
Table 2.2. Number of units predicted to exceed selected values of SO3 stack concentration under scenario
assumptions. Data cover all coal-fired plants in the Mississippi Valley and Eastern U.S. Note that each column listing
the predicted number of affected plants is a subset of the columns to its right.
Load
Range
(MW)
Total #
of
Units
in
Range
Approx.
Total
Capacity
(MW)
Approx.
%of
Total
Capacity
Included
in Range
SO3>
25 ppm
"Gavin-
like"
cone.
SO3>
20 ppm
SO3>
13 ppm
"Gibson-
like"
cone.
SO3>
10 ppm
"Blue
plume"
w/o
scrubber
SO3>
4 ppm
Opacity
problems
with
scrubber
Missing
Data
1%SCR Conversion
£1,200
900-1,199
700 - 899
400 - 699
250 - 399
<250
Unknown
Totals
Totals >
250 MW
14
13
67
136
133
957
30
1,350
363
18,989
13,378
52,803
74,345
40,690
65,887
266,092
200,205
7%
5%
20%
28%
15%
25%
0
2
0
0
0
23
0
25
2
3
3
6
10
7
62
3
94
32
5
3
12
32
19
163
7
241
78
5
3
15
49
33
266
11
382
116
6
7
42
102
78
590
20
850
255
0
0
0
2
0
17
2
c
o
0.75% SCR Convers
£1,200
900-1,199
700 - 899
400 - 699
250 - 399
<250
Unknown
Totals
Totals >
250 MW
14
13
67
136
133
957
30
1,350
363
18,989
13,378
52,803
74,345
40,690
65,887
266,092
200,205
7%
5%
20%
28%
15%
25%
0
0
0
0
0
8
0
8
0
0
2
1
3
2
26
0
34
8
3
3
10
24
14
124
5
183
59
5
3
12
40
22
197
7
286
89
5
7
40
98
77
564
20
812
247
0
0
0
2
0
17
2
I 0.5% SCR Conversion
£1,200
900-1,199
700 - 899
400 - 699
250 - 399
<250
Unknown
Totals
Totals >
250 MW
14
13
67
136
133
957
30
1,350
363
18,989
13,378
52,803
74,345
40,690
65,887
266,092
200,205
7%
5%
20%
28%
15%
25%
0
0
0
0
0
0
0
0
0
0
0
0
0
0
8
0
8
0
3
3
6
13
9
64
3
101
37
4
3
10
26
15
129
7
194
65
5
5
32
85
69
496
11
703
207
0
0
0
2
0
17
2
In Table 2.2, SOs from SCR conversion was calculated for conversion rates of 0.5%, 0.75%, and
1.0%. This range of scenarios was selected to address the effects of catalyst aging as well as
differences in production rate with new SCR installations. As SCR catalysts age, their NOX
reduction efficiencies decline; however, their SOs oxidation rates remain essentially constant.
Thus, at some point in time additional catalyst volume will need to be installed on each unit in
order to maintain the required NOX performance (the usual approach taken to extend the service
-------
time of installed catalysts). This means that a unit with a 0.5% 862 oxidation catalyst would be
expected to eventually behave like the 0.75% catalyst in the table. Similarly, a 0.75% catalyst
unit would be expected to eventually produce somewhat more SOs than that shown for a 1%
catalyst in the table. Because of the issue of SO3 formation, utilities are strongly pushing for the
absolute minimum 863 conversion, to avoid this problem. So while the additional beds will
result in an increase in conversion, the competing trend is to begin with a catalyst that generates
as little SOs as possible. That pressure will push toward the 0.5% (or even lower) conversion
catalysts.
A summary of the coal sulfur contents in the databases used to generate Table 2.2 are provided in
Table 2.3. It is worth noting that there are reports that several utilities are considering switching
from low sulfur coals to higher sulfur coals after retrofitting units with scrubbers.20"25 Such
actions would result in higher SOs emissions than those shown in Table 2.2 for the same SCR
conversion rates and would offset reductions otherwise achievable by use of lower conversion
rate catalysts.
Table 2.3. Sulfur contents of coals fired in plants in the Mississippi Valley and Eastern U.S.
Load Range
(MW)
£1,200
900-1,199
700 - 899
400 - 699
250-399
200 - 249
<200
Unknown
Totals
Totals >250
Totals > 200
Total
Number in
Range
14
13
67
138
133
74
882
24
1,345
365
439
Total
Capacity
(MW)
18,989
13,378
52,803
74,345
40,690
16,482
65,887
282,574
200,205
216,687
Percentage
of Capacity
Included in
Range
6%
4%
17%
24%
13%
5%
21%
S >3%
3
3
4
23
12
11
82
2
140
45
56
S 2% to
3%
2
0
6
15
6
7
105
5
146
29
36
S 1%to2%
0
0
14
42
42
26
288
4
416
98
124
S<1%
9
10
43
56
73
30
389
11
621
191
221
Missing
Data
0
0
0
2
0
0
18
2
The estimated SOs removal efficiencies provided by pollution abatement systems are
summarized in Table 2.4, with comparison to previously reported reference values ,17'26 Air pre-
heaters were assumed to reduce 863 by 30%. Units with fabric-filter fly ash collectors were
credited with an additional 90% reduction in SOs. Units with cold-side electrostatic precipitators
(ESP) were credited with an additional 25% reduction in SOs. For these calculations no effect
was attributed to either hot-side electrostatic precipitators or wet scrubbers.
Comparison of the columns in Table 2.4 shows the control efficiency values selected are more
conservative than those used in the previous references. Air heaters are commonly estimated to
remove 50% of the SOj reaching them. However, when the above removal percentages were
-------
applied to three plants which have both cold-side precipitators and scrubbers, two plants with
only cold-side precipitators, and two plants with only scrubbers, the predicted values for the SOs
concentrations in the stack gases fell far below the observed values. This led to a reconsideration
of these removal efficiencies to better match measured values. This is shown in Table 2.5.
Table 2.4. Estimated SO3 removal efficiencies (in percent) for control system components
Device
Air Heater
Hot ESP
Cold ESP
Fabric Filter
Wet FGD scrubber
Data Source
HSD(a)
50
0
25
90
0
MH(b)
50
0
50
90
50
Used in This
Report
30
0
25
90
0
(a) Hardman, Stacey, and^Dismukes
(b) Monroe and Harrison
26
Table 2.5. Comparisons of observed and predicted SO3 concentrations
Location
Plant 1
Plant 2
Plants
Plant 4a
Plant 4b
Plant 5a
Plant 5b
Plant 6
Plant 7
Plant 8
Percent
Sulfur in
Coal
2.6
3.75
2.9
0.92
0.92
0.99
0.99
3.35
3.6
0.86
Measured
SO3 (ppm)
14
-35
-13
7.7
9.6
4.0
7.7
2.6
10.7
6.5
SO3 Predicted Here
No SCR
6.3
6.8
0.5% SCR
8.3
11.9
9.2
3.9
3.9
3.2
3.2
5.8
1%SCR
13
18.5
14.3
6.1
6.1
4.9
4.9
6.3
Predictions Based on
MH(a) Removal
Efficiencies
1%SCR
3.4
4.9
3.7
3.1
3.1
2.5
2.5
2.3 (No SCR)
2.4 (No SCR)
2.2
Control System
Cold-side ESP & Scrubber
Cold-side ESP & Scrubber
Cold-side ESP & Scrubber
Cold-side ESP
Cold-side ESP
Cold-side ESP
Cold-side ESP
Scrubber
Scrubber
Cold-side ESP
a. Monroe and Harrison
Review of data obtained at several installations tends to show air heater removal efficiencies in
the range of 20% to 50% with typical values of about 25%. Removal efficiencies across cold-
side precipitators in these same data sets were typically about 25% as well. Similarly, removal
efficiencies as low as 15% to 20% have been measured across some fabric filters and removal
efficiencies across wet scrubbers have been measured as low as 12% to 15%. In the case of air
heaters, the removal efficiencies would be expected to be higher for lower exit gas temperatures,
as illustrated in Figure 2.1. Similarly, removal across precipitators and fabric filters increases
substantially as the gas temperature decreases to and below the acid dew point. Conversely, if the
-------
gas temperature through a precipitator or fabric filter is substantially above the acid dew point,
little removal will take place.
It is clear that the commonly cited estimated removal efficiencies result in under-predicting stack
863 concentrations. In Table 2.4, the value of zero for removal efficiency in scrubbers is not
meant to imply that no removal takes place in them. Rather, it simply means that the removal by
preceding components was over-estimated so much that the results would have been biased
severely low if another 30 to 50 percent reduction were applied. The values in Table 2.2 are
believed to be reasonable estimates of the emissions from units using their current particulate
control systems followed by wet scrubbers.
o
o
o
o
A AO O
O A
OO
250 260 270 280 290 300 310
Exit Temperature, F
320
330
340
350
Figure 2.1. Measured SOs concentrations versus temperature at the exit of an air heater and cold-side electrostatic
precipitator.
Table 2.6 and Figure 2.2 show state-by-state tabulations of estimated SOs emissions from
individual power plant units in the Mississippi Valley and Eastern U.S. Only units with
capacities of 200 MW or more were included, as these are the units that will be most likely
affected by the upcoming regulations. While not specifically related to the pertinent issues of
opacity, ground effects, or local acidic deposition, these data demonstrate the potential for a
significant increase in fine PM emissions relative to the filterable fraction (fly ash). The SOs
emissions estimates were made using the SO3/SO2 ratios developed by Hardman, Stacy, and
Dismukes for eastern bituminous and western subbituminous coals.17 In applying the
correlations it was assumed that the fuels in the 2002 Annual Steam-Electric Plant Operation and
10
-------
Design Database14 are still in use and that the units operate at an average annual load of 75% of
capacity (allowing for some reduced load operation and outages). Table 2.6 and Figure 2.2 show
that of the 17 states included in the state-by-state tabulations, six states (OH, PA, IN, TN, KY,
and FL) account for nearly 70% of the total projected SO3 emissions.
Table 2.6. State-by-state tabulation of estimated SOs emissions from individual power plant units in the Mississippi
Valley and Eastern U.S.
State
AL
AR
CT
DC
DE
FL
GA
IA
IL
IN
KY
LA
MA
MD
ME
Ml
MN
MO
MS
NC
NH
NJ
NY
OH
PA
Rl
SC
TN
VA
VT
Wl
WV
1 % SCR Conversion
SO3
Emissions
(kg/yr)
5.79E+06
1.24E+05
7.19E+05
6.04E+05
8.80E+05
1.24E+07
7.66E+06
1.26E+06
4.90E+06
2.13E+07
1.68E+07
1.29E+06
6.09E+04
6.35E+06
1.02E+05
5.44E+06
1.23E+06
6.71 E+06
1.37E+06
9.02E+06
O.OOE+00
1.02E+06
6.82E+06
4.00E+07
3.09E+07
O.OOE+00
4.86E+06
1.75E+07
5. 11 E+06
O.OOE+00
4.28E+06
5.37E+04
State % of
Total
2.70%
0.06%
0.34%
0.28%
0.41%
5.76%
3.57%
0.59%
2.29%
9.92%
7.85%
0.60%
0.03%
2.96%
0.05%
2.54%
0.57%
3.13%
0.64%
4.21%
0.00%
0.48%
3.18%
18.64%
14.39%
0.00%
2.26%
8.15%
2.38%
0.00%
2.00%
0.03%
0.75 % SCR Conversion
SO3
Emissions
(kg/yr)
4.86E+06
1.05E+05
6.05E+05
5.09E+05
7.41E+05
1.04E+07
6.43E+06
1.00E+06
4.03E+06
1.78E+07
1.42E+07
1.05E+06
5.13E+04
5.35E+06
8.60E+04
4.52E+06
9.74E+05
5.47E+06
1.16E+06
7.60E+06
O.OOE+00
8.56E+05
5.74E+06
3.35E+07
2.60E+07
O.OOE+00
4.09E+06
1.47E+07
4.30E+06
O.OOE+00
3.50E+06
4.52E+04
State % of
Total
2.71%
0.06%
0.34%
0.28%
0.41%
5.78%
3.58%
0.56%
2.25%
9.89%
7.89%
0.58%
0.03%
2.98%
0.05%
2.52%
0.54%
3.04%
0.64%
4.23%
0.00%
0.48%
3.20%
18.67%
14.48%
0.00%
2.28%
8.17%
2.39%
0.00%
1.95%
0.03%
0.5 % SCR Conversion
SOs Emissions
(kg/yr)
3.94E+06
8.50E+04
4.92E+05
4.14E+05
6.03E+05
8.43E+06
5.21 E+06
7.46E+05
3.16E+06
1.43E+07
1.15E+07
8.11E+05
4.17E+04
4.35E+06
6.99E+04
3.60E+06
7.16E+05
4.22E+06
9.40E+05
6.18E+06
O.OOE+00
6.92E+05
4.67E+06
2.71 E+07
2.11E+07
O.OOE+00
3.33E+06
1.19E+07
3.50E+06
O.OOE+00
2.72E+06
3.68E+04
State % of
Total
2.72%
0.06%
0.34%
0.29%
0.42%
5.82%
3.60%
0.52%
2.19%
9.85%
7.96%
0.56%
0.03%
3.00%
0.05%
2.49%
0.49%
2.92%
0.65%
4.27%
0.00%
0.48%
3.22%
18.72%
14.60%
0.00%
2.30%
8.18%
2.41%
0.00%
1.88%
0.03%
Summary
More than half of the units expected to be affected by the new regulations could reasonably
require mitigation of SOs emissions to meet opacity requirements and about a quarter could
require mitigation to alleviate problems related to ground level SO3 concentrations. Evidence in
the trade press suggests that several utilities are considering returning to the use of high sulfur
coals after they install scrubbers. If they do, SOs emissions would be expected to increase to
levels higher than those projected in this report. This study dealt with only coal-fired units. It
should also be noted that similar SO3 emission problems would be experienced at heavy oil-fired
11
-------
units, if NOX control using SCR is applied to them. This analysis suggests that mitigation of
emissions will be needed in some cases to address plume visibility issues, increased acid aerosol
emissions, and PM2.5 emission levels, particularly at plants that install SCR and FGD and use
high sulfur coal. The remainder of this report discusses measurement methods and SO3
formation and removal processes across different plant components. This information can
provide useful guidance for plant operators who are interested in a range of approaches to
avoiding excess collateral SOs emissions.
It should be noted that there is a significant lack of actual measurements of the conversion rate of
SC>2 to SOs in the boiler. This lack makes predictions highly uncertain because of the inability to
validate prediction results against actual measurements. Although the predictions made in the
following chapters are in line with measured results, the scarcity of measurements makes it very
uncertain whether such results can be repeated for the range of units in service. Further
measurements of boiler exit SOs concentrations are key to advancing our understanding of, and
ability to predict, SOs emissions.
SO3 Emissions -1% SCR Conversion
20.00% ^
18.00%
16.00%
14.00%
12.00%
10.00%
8.00%
6.00%
4.00%
2.00%
0.00%
Figure 2.2. SOs emissions for states in the Mississippi Valley and Eastern U.S. emitting > 1% of total projected SOs
emissions for the region (215,000 tonne/year total).
12
-------
3. IMPROVEMENTS IN SO3 MEASUREMENT TECHNOLOGIES
Current SO3/ H2SO4 Measurement and Monitoring Methods
Method 8, the EPA promulgated method for measuring emissions of sulfuric acid mist,27 has a
lower detection limit of about 50 mg/m3, thus it lacks the sensitivity needed for measurements at
electric utility plants. The manual controlled condensation method (CCM) developed by Cheney
and Homolya28 is generally recognized as the most reliable method for measuring S(V H2SO4 at
the levels encountered at power plants.29 In the CCM, a sample gas stream is conveyed through
a heated quartz-lined probe, through a quartz fiber filter and then through a condenser in which
the acid vapor is removed. The probe and filter holder are held at a temperature at or above
550 °F to evaporate condensed H2SO4 and ensure that the SO3/H2SO4 in the sample is entirely in
vapor form before reaching the filter. The condenser is maintained at a temperature above the
moisture dew point but well below the sulfuric acid dew point so that all of the acid is collected
in it. A second filter downstream of the condenser ensures that any aerosolized acid is retained
for analysis with that collected in the condenser itself. After each sample is collected, the
condenser is washed to recover the collected acid. The amount of acid collected is later
quantified in the laboratory by titration. Over the past few years several improvements have been
made to the method as it is applied to coal-fired utility installations.30'31 The changes deal with
minimizing problems arising from the acid vapor being adsorbed on or reacting with ash
collected on the filter upstream of the condenser or the filter medium itself, together with the
problem of ensuring that any sulfuric acid mist in the sampled gas is completely volatilized in the
probe. The first problem is significant upstream of particulate control devices and the second is
especially important downstream of a scrubber where the H2SO4 is, for practical purposes,
entirely in the condensed phase. Both problems must be addressed in automated as well as
manual systems. The CCM as described above was used as a manual reference measurement for
comparison with the results of measurements reported here.
Design and Fabrication of Sample Extraction and Transport System
for SO3/H2SO4
A semi-continuous SO3/H2SO4 monitor was developed by Southern Research Institute under
Cooperative Agreement No. DE-FC26-02NT41593 with the U.S. Department of Energy. A
significant part for that project addressed a major requirement for a field-useable monitor: a
sample extraction system that separates SO3/H2SO4 vapor from in-stack particulate matter (PM)
and transports it efficiently as vapor to the sensor.32
The sample extraction and transport system (SETS) had to extract and maintain or heat the flue
gas to >550 °F, transport the sample stream, separate SO3/H2SO4 from particulate matter, deliver
SO3/H2SO4 to the analyzer, and return excess flue gas and PM to the duct. Heating of the sample
gas is accomplished primarily by the injection of very hot dilution gas (at a dilution ratio of
approximately 1:1) near the probe entrance as well as from contact heating. Flow through the
SETS is driven by an eductor, with the hot sample gas being extracted from the probe through an
annular filter upstream of the eductor. Because of proven performance in similar applications, a
commercially-available probe designed for use in monitoring mercury emissions, the Apogee
Scientific QSIS, was purchased and modified to include the hot dilution gas injection for heating
13
-------
the sample gas as illustrated in Figure 3.1. The QSIS probe and filter element are fabricated from
stainless steel with a proprietary coating applied to all surfaces to minimize reactions with the
sample gas.
The exhausts of some wet FGD scrubbers contain significant quantities of droplets that are too
large (on the order of tens of microns) to evaporate quickly which would lead to their being lost
on internal surfaces of the sampling nozzle and transport line. Overtime, such droplets deposited
on surfaces would dry, leaving a residue that would accumulate and interfere with normal
operation of the system. Therefore, when monitoring downstream of wet FGD scrubbers the
SETS should incorporate an inertial collector at the probe inlet to remove these droplets. In the
past this approach has been found necessary for sample extraction in wet streams. In particular,
SRI developed a series of such procedures for the California Air Resources Board for making
particle size measurements in wet process streams that have been successfully utilized for over
10 years.33 Fortunately, the condensed acid tends to reside in droplets having diameters of a few
micrometers and smaller.
Connect to Compressed Air
Eductor
Heated Air
Apogee Control Unit
O
To Blow
Back Venturi
Pressure Lines
TC
Heated Line to
Monitor
Venturi AP
Figure 3.1 Schematic diagram of the Apogee QSIS probe modified for use with the SOs monitor.
The QSIS probe extracts a flue gas sample at a flow rate of approximately 10 to 12 acfm through
a 3/4-inch diameter sample probe. The extracted gas then passes axially along the length of a l/2-
inch ID cylindrical porous filter element. A slipstream of sample gas is pulled radially through
the filter element at a flow rate of a few liters per minute and directed to the sample collection
and measurement system. Due to the high axial velocities in the sample probe and core of the
14
-------
cylindrical filter element, particles entrained in the gas stream are prevented from depositing on
the inner walls of the filter. However, fine particles are capable of being trapped in the filter
element over time. As a result, the filter element must be periodically removed for cleaning by
backwashing. As in the original SETS design, flow through the Apogee system is driven by an
eductor, with the analyzer extracting about 2 slpm of hot sample gas upstream of the eductor.
The excess gas not used for the actual measurement is returned to the host duct.
Prior to proceeding with the purchase and modification of a QSIS probe for use with the monitor,
a brief series of tests was performed to check for any obvious problems with the use of the probe
for SO3/H2SO4 sampling. These tests were carried out in SRI's Coal Combustion Research
Facility (CCRF) in conjunction with an ongoing mercury measurement program which utilized
two such probes, one sampling at a point in the CCRF at which the gas temperature was about
550 °F and the second down stream of the first, following two heat exchangers, at which the gas
temperature was about 325 °F. A heated hose like that to be used between the probe and
condenser for the SOs monitor was used at each location to convey a sample stream to a
conventional controlled-condensation condenser setup. Data obtained with this setup were then
taken for comparison with data obtained with conventional controlled condensation setups that
were being operated to obtain information for the mercury emissions program. The results of
these tests are shown in Figure 3.2. The agreement was excellent at the 550 °F location and was
reasonably good at the 325 °F location although the results with the QSIS probe were slightly
higher than those from the conventional probe at the 325 °F location, perhaps due to a bias
introduced by particulate matter on the filter of the CCM sampler. (The differences in
concentrations between the two locations results from losses in the heat exchangers.) These
results were deemed satisfactory enough to proceed with the purchase and modification of a
QSIS probe for use with the SO3/H2SO4 monitoring system.
Tests conducted as part of the current project compared measurements made utilizing the
modified probe with those obtained using a conventional CCM setup. These tests were
conducted at two conditions: (1) with a flue gas stream in which the H2SO4 present would be
entirely in the vapor phase and (2) a gas stream in which the majority of the H2SO4 present
would be in the condensed phase. Condensation of the H2SO4 in the flue gas for the second test
condition was induced by means of a water spray in the duct upstream of the sampling location.
An Illinois Basin coal with a 3.5 % sulfur content was fired in the CCRF for these tests.
Sampling was done at a location about 30 feet downstream of the 325 °F location of Figure 3.2.
As is the case for air preheaters on full-scale utility boilers, the air preheaters of the CCRF
remove significant fractions of the SCV H2SO4 formed in the furnace. The air preheaters were
operated to provide less cooling than that used in normal CCRF operations in order to obtain
higher H2SO4 concentrations and to provide assurance that the H2SO4 was entirely in the vapor
phase for the first test condition. A nominal temperature of 400 °F was selected as the air
preheater exit temperature based on the amount of cooling expected from the water spray to be
used to induce condensation by cooling for the second test condition. A target temperature of 270
°F was selected for the second test condition. The saturation vapor pressure of H2SO4 at 270 °F
was low enough that the majority of the H2SO4 vapor could be expected to condense.
15
-------
CM
O
25
20
15
ro
©
Q.
Q.
8 10
* Conventional CCM at 325F Location
• Apogee Probe at 325F Location
• Conventional CCM at 550F Location
A Apogee Probe at 550F Location
8:00
10:00
12:00
14:00
16:00
18:00
Figure 3.2. Comparison of SO3 concentrations measured by conventional controlled-condensation methods (CCM)
with those measured using Apogee QSIS probes and a heated hose to deliver samples to a controlled-
condensation condenser.
Three Apogee probe operating conditions were used during the tests without the water spray with
a flue gas temperature of 393 °F. During this phase of the testing the probe was operated without
dilution and with dilution of flue gas with filtered hot air; first with the dilution air at 394 °F and
then with the dilution air at 1113 °F. The mixed gas temperature was 559 °F at the higher
dilution air temperature. The dilution ratio used in these tests was 1:1 on a mass basis. The
dilution air temperatures cited here are those measured at the exit of the dilution air heater. The
actual temperature of the dilution air as it enters the probe is somewhat lower. No change was
noted in the dilution-corrected concentrations under these conditions as illustrated by the first
block of data points in Figure 3.3. The temperatures of the dilution air and the mixed sample and
dilution gas are given on the figure for each test condition. The initial rise in the SOs/ H2SO4
concentration seen at the beginning of the test was the result of losses in the air preheater that
diminished as an ash coating built up on the heat exchanger surfaces which were initially clean.
The amount of cooling provided by the water spray was less than expected at the sampling
location, perhaps because of incomplete evaporation in the transit time between the point of
injection and the sampling location. Therefore the air preheater exit gas temperature was adjusted
16
-------
70
60
50
Q.
Q.
O
c
0)
u
O
O
fO
O
w
40
30
20
10
• Apogee Corrected to 3% O2
• Conventional CCM Corrected to 3%O2
Baseline: Tgas = 393F
w/o Hot Dilution
^ b
•
•
•
• •
•
•
•
Baseline: Tgas = 393F
with Hot Dilution
Dilution Air Temp. Dil.AirTei
394F 1113F 1079F
_•" • •
H •• •
. '
. ••
•
Mixed Temp. Mixed Ter
394F 562F 559F
/
Reduce Tgas to ~ 360 F
to obtain target Temp.
with spray
Spray on: Tgas =
273F with Hot
^ Dilution fc
ip.
Dilution Air Temp.
1276F 1290F86S
"" ••
• • •
•
% *•
ip.
•
•
•
Mixed Temp.
559F 559F 43<
Spray on
w/o Hot
^ Dilution ^
F
•
• •
i.'
Time
Figure 3.3. Comparisons of SO3 concentrations measured by the conventional controlled-condensation method
(CCM) with those measured using the modified Apogee QSIS probe incorporating the hot gas dilution
approach to evaporating condensed
downward somewhat to about 360 °F in order to obtain an acceptably low gas temperature at the
test location. This resulted in a noticeable decrease in the total SCV H2SO4 in the duct.
The dilution air temperature had to be raised to about 1280 °F when the water spray was
employed in order to obtain the target 560 °F temperature for the mixed gas. The Apogee probe
was also operated with a nominal 870 °F dilution gas temperature, resulting in a mixed gas
temperature of 434 °F. At the latter condition a very slight drop in the measured SCV H2SO4
concentration was noted as compared to that measured with the 560 °F mixed temperature.
Finally, the Apogee probe was operated without the hot dilution gas at which point the measured
concentration fell to a value close to the saturation concentration at the flue gas temperature,
reflecting the loss of the condensed phase SCV H2SO4 in the filter. The large drop in the value
measured with the QSIS probe without the hot dilution air while the spray remained on at the end
of the test series resulted from condensed acid no longer being evaporated in the probe.
It appears to have taken about five hours after the water spray was turned on for the conventional
CCM probe to reach steady-state conditions and provide reliable results. Evidently it took that
long for the internal surfaces of the CCM probe to get hot enough to evaporate the condensed
phase acid. Again, the lower steady-state values measured with the conventional CCM as
compared to those obtained with the Apogee probe through most of this test series are believed
17
-------
to have resulted from adsorption onto paniculate matter which collects on the CCM filter. The
QSIS probe as adapted for SOs measurements is shown in Figure 3.4.
Following the tests conducted at the SRI pilot-scale Coal Combustion Research Facility, a short
series of tests were conducted on a stack downstream of a scrubber on a large coal-fired utility
boiler. Results of measurements using a standard controlled condensation probe and filter setup
with results using the inertial-separation hot-gas-dilution probe are given in Table 3.1. All values
have been corrected to 3% oxygen. The intended value of mixed gas temperature in the dilution
probe, 550 °F to 600 °F, was never obtained during these tests. Too much of the transport tubes
and mixing zone were in the flue where they were cooled beyond the capability of the dilution
gas heater. A mixed gas temperature of 445 °F was as great a value as could be achieved on Test
1 . At that temperature the dilution probe results approached those of the conventional probe but
were clearly low when the mixed gas temperature was only 419 °F. During Test 2, additional
insulation was added to the dilution probe, but the target mixed gas temperature still could not be
reached. A simple solution to the problem has been found: simply add a spool piece and probe
extension so that the hot gas transport tubes and mixing zone can be located outside the flue. Had
the mixed gas been as hot as intended, one could be reasonably certain that all of the condensed
acid in the flue was evaporated and thus would have been measured. However, complete
evaporation may or may not have been achieved during the tests when the temperature was 445
°F to 490 °F, so it is uncertain as to whether all of the acid present in the sample was available
for measurement downstream of the filter.
Table 3.1 Comparison of measurements made using a conventional CCM probe with those made with the hot-gas
dilution probe downstream of a full-scale utility scrubber.
Standard Controlled Condensation Probe
Test Run
1
2
SO3, ppm @ 3% O2
Average
19.3
18.0
Range
18.7-20.7
17.6-18.6
Hot-c
Mixed Gas
Temperature, °F
419
445
480 - 490
as Dilution Probe
SO3, ppm @ 3% O2
Average
14.1
18.4
17.0
Range
12.7-15.5
18.4-18.4
16.2-17.9
Recommended future work includes:
1. Performing tests utilizing the hot gas dilution system to check for artifact SOs formation by
oxidation of 862.
2. Performing tests downstream of a scrubber to verify performance when the predominant
form of SOs is condensed-phase H2SO4 in a low temperature gas stream.
18
-------
Waste Exhaust
Mam Flow
Shut-off
Dilution Gas
Flow Meter
Sample Inlet
^^
Mounting Flange
Dilution Gas
Heater
Total Flow
Venturi Meter
Ventun Taps
Dilution Gas
Mixer
Sample Take-offs
Filter Housing
Figure 3.4. Modified Apogee QSIS sampling probe. Upper: probe with insulating jacket in place. Lower: Probe with jacket removed to show principal components.
(The waste exhaust normally points opposite the direction shown.)
-------
4. EXPLORATORY STUDY OF SO3 ADSORPTION BY
COAL FLY ASH
Introduction
Removal of 863 or H2SO4 from the flue gas can take place through a number of
mechanisms: by uptake by fly ash, by condensation on the heat exchanger surfaces in air
preheaters, by reaction with reagents injected for control of vapor phase SOs/H^SO^ or
by condensation and subsequent collection in the form of particles. Significant removal
can result from uptake by fly ash but the mechanisms are not fully understood and at
present cannot be well predicted. This study was undertaken to provide data which might
be used to aid in the development of a predictive model for uptake of SO3/H2SO4 by fly
ash from coal-fired utility boilers.
Approach
Nine sets of ash samples from a wide range of coal types were selected from an inventory
of ashes that had previously been characterized during studies related to resistivity
modification by SOs injection. These ashes had been collected either by EPA Method
1734 directly from flue gas streams or as hopper samples from dry particulate control
devices (either electrostatic precipitators or fabric filters). All had been previously
exposed to flue gas streams having unknown concentrations of 863 prior to collection.
The ashes selected included ashes from high and low sulfur eastern bituminous coals, low
and moderate sulfur western sub-bituminous coals, a Powder River Basin coal, and a
North Dakota lignite.
The ashes were exposed to synthetic flue gas atmospheres in an apparatus normally used
for studies related to fly ash resistivity and ash resistivity modification by moisture and
sulfur trioxide.35 The apparatus consisted of a continuous-flow generator which produces
a synthetic flue gas stream which was passed through a bell jar containing the samples in
a temperature controlled oven. The synthetic flue gas was humidified, and heated to a
temperature above that at which SOs and moisture combine to form H2SO4, after which
controlled amounts of 863 were added to obtain the desired 863 concentration. Previous
experience in resistivity work with this apparatus had indicated that over a 72 to 96 hour
exposure time SOs would permeate and apparently approach equilibrium in a 1 mm thick
ash layer.36 Fresh ash samples were spread as 1 mm thick layers in the bottoms of 12
petrie dishes and placed in the bell jar for each exposure condition. The layout of the
samples in the bell jar is illustrated in Figure 4.1. The arrows shown in Figure 4.1 show
the positions of the SOs inlet and outlet gas streams. Duplicate samples of three of the
nine ashes were included in each test to provide information on the reproducibility of the
results. Experiments were performed at four temperatures (350 °F, 375 °F, 450 °F, and
850 °F) and two 863 concentrations (20 ppmv and 70 ppmv). In each test, prior to adding
SOs to the gas stream, the samples were cycled up to 850 °F and held at that temperature
overnight. In addition to the exposure tests, one baseline run was made in which the
samples were subjected to the 850 °F temperature cycle but not exposed to 863.
Following the exposure to 863, the samples were removed from the bell jar as rapidly as
20
-------
possible and the covers were placed on the petrie dishes after which they were
individually extracted for measurement of soluble sulfates.
A typical set of SO3 concentrations at the inlet and outlet of the bell jar are shown over
the course of a test in Figure 4.2. As can be seen, the uptake of 863 by the ash was
initially fairly high, but declined fairly rapidly over the course of about two days, after
which it continued to decline but at a much slower rate, indicting that the limiting
capacity of the samples was never reached.
Results
The exposure conditions and results of the measurements are provided in Table 4.1. The
percentage of the total entering the bell jar that was taken up by the ash is provided in the
table together with the SOs mass balance for each set of exposures. The mass balances
indicate that the recovery in the analyses of SOs taken up by the ash was essentially
complete. The results are shown plotted versus the base:acid ratios of the ashes in Figure
4.3. The curves in Figure 4.3 are present only for purposes of tying the results for each
test condition together and have no physical or chemical significance. Given the strong
correlation between SOs uptake and the base:acid ratios of the ashes, it appears that
chemi-sorption rather than physical adsorption dominates the process and that
temperature plays a strong role as well. Based on rather crude extrapolations of the
concentrations of SOs at the outlet of the exposure chamber, it is estimated that the
limiting capacities of the ashes are perhaps twice the maximum uptakes found in these
experiments with the uptake rates being concentration and temperature dependent.
Following the general approach used in developing predictive equations for ash
resistivity,36 ash compositions based on atomic percentages of cations were calculated
from the weight percentages as oxides shown in Table 4.2. The resulting atomic
percentages are given in Table 4.3. Multiple regression was then used to obtain fitting
coefficients for use as a predictive tool for SOs uptake by ash based on the ash chemistry.
Results of the regression are provided in Table 4.4. A plot of values predicted by the
regression equation versus those measured in this study is shown in Figure 4.4.
21
-------
Table 4.1. Exposure conditions and results
Ash ID
301
9896-1-68
303
143
304
304 dup
9896-1-69
129
129 dup
305
305 dup
9896-1-57
Experiment
Duration, hours
Oven Temperature, °C
Oven Temperature, °F
Nominal SOs Cone., ppm
Percent Capture by Ash
Mass Balance, %
Ash Type
Low SW SB (a)
E Bit (b)
Low S W SB
Low S W SB
Low S E Bit (c)
WSB(d)
ND Lignite (e)
Hi S E Bit (f)
PRB(g)
Baseline
12
454
850
0
Exposure
96
177
351
20
21.5
96.7
Exposure
74.75
190
374
20
26.7
96.1
Exposure
74.72
232
450
20
26.3
94.4
Exposure
93.4
232
450
70
19.2
100.9
Exposure
96.5
454
850
70
25.8
99.5
Post Exposure Soluble Sulfate, %
0.44
1.58
0.88
0.88
0.10
1.51
4.04
0.78
2.49
2.16
4.05
2.82
2.76
0.86
0.83
4.16
6.55
6.15
1.59
1.68
4.69
1.95
3.67
2.55
3.46
0.95
0.91
3.98
6.19
6.01
1.59
1.71
4.54
2.46
4.51
2.94
3.36
1.32
1.24
4.54
6.56
6.47
1.99
2.15
4.99
3.62
7.31
4.73
6.10
1.78
1.81
8.39
14.36
12.91
2.72
2.70
10.22
3.21
8.82
11.43
10.25
1.19
1.00
10.71
14.27
13.45
2.87
3.21
13.92
(a) Low sulfur western subbituminous coal
(b) Eastern bituminous coal
(c) Low sulfur eastern bituminous coal
(d) Western subbituminous coal
(e) North Dakota lignite
(f) High sulfur eastern bituminous coal
(g) Power River Basin coal
-------
Table 4.2. Properties of ashes used in adsorption study
Ash ID.
301
9896-1-68
303
143
304
9896-1-69
129
305
9896-1-57
Coal Type
LowSWSB(b)
E Bit (c)
Low S W SB
Low S W SB
Low S E Bit (d)
WSB(e)
ND Lignite (f)
HiSEBit(g)
PRB(h)
Base/Acid
Ratio(a)
0.149
0.354
0.462
0.535
0.092
0.698
1.558
0.242
0.901
Ash mineral analyses, wt%
Li2O
0.03
0.04
0.05
0.01
0.04
0.01
0.02
0.05
0.02
Na2O
0.51
1.20
0.34
1.13
0.19
4.00
1.58
0.34
1.70
K2O
1.7
2.40
0.42
0.70
2.70
1.90
0.20
3.10
0.52
MgO
1.3
1.80
6.30
4.00
0.85
5.70
8.90
1.10
4.40
CaO
4.4
7.4
19.5
22.7
0.56
19.9
32.2
2.20
27.9
Fe203
5.0
13.1
4.3
4.8
4.1
9.0
12.6
12.5
9.9
AI203
25.8
25.5
24.1
21.6
32.2
14.4
12.3
27.1
19.3
SiO2
59
45.6
41.2
38.8
56.4
42.7
22.6
50.5
27.4
TiO2
1.70^
2.10
1.50
1.90
2.30
0.92
0.70
1.80
2.60
205
0.31
0.32
0.31
1.40
0.15
0.27
0.30
0.33
1.10
S03
0.35
1.60
0.94
1.70
0.18
1.60
7.90
0.57
3.10
(a) Base/Acid ratio = (Na2O+K2O+MgO+CaO+Fe2O3)/(AI2O3+SiO2+TiO2)
(b) Low sulfur western subbituminous coal
(c) Eastern bituminous coal
(d) Low sulfur eastern bituminous coal
(e) Western subbituminous coal
(f) North Dakota lignite
(g) High sulfur eastern bituminous coal
(h) Power River Basin coal
-------
Table 4.3. Atomic concentrations of cations for ashes used in adsorption study
Ash ID
301
9896-1-68
303
143
304
9896-1-69
129
305
9896-1-57
Li
0.047
0.066
0.075
0.015
0.065
0.015
0.030
0.085
0.031
Na
0.385
0.952
0.245
0.830
0.149
2.839
1.132
0.278
1.288
K
0.844
1.253
0.200
0.338
1.395
0.887
0.094
1.667
0.259
Mg
1.126
1.639
5.219
3.370
0.766
4.643
7.318
1.031
3.826
Ca
2.739
4.843
11.612
13.745
0.363
11.651
19.032
1.483
17.437
Fe
0.875
2.409
0.719
0.816
0.746
1.480
2.092
2.366
1.738
Al
7.068
7.343
6.315
5.755
9.178
3.710
3.199
8.036
5.308
Si
22.626
18.383
15.112
14.471
22.505
15.398
8.228
20.962
10.548
Ti
0.490
0.637
0.414
0.533
0.690
0.250
0.192
0.562
0.753
P
0.044
0.048
0.042
0.194
0.022
0.036
0.041
0.051
0.158
S
0.076
0.367
0.196
0.361
0.041
0.328
1.635
0.135
0.679
-------
Table 4.4. Results of multiple regression to obtain predictive equation for uptake of SO3
Regression Statistics
Multiple R
R Square
Adjusted R Square
Standard Error
Observations
0.973909
0.948498
0.942555
0.081164
59
Analysis of Variance (ANOVA)
Regression
Residual
Total
Intercept
log(SOs cone., mg/am3)
log(T, R)
Log(Li+Na)(a)
Log(K)
Log(Mg+Ca)(b)
Log(Fe)
df
6
52
58
Coefficients
-2.99121
0.513845
0.785485
0.270711
-0.1177
0.292563
0.288818
SS
6.308791
0.342558
6.651349
Standard Error
0.512215
0.062655
0.185629
0.077058
0.05934
0.081426
0.056989
MS
1.051465
0.006588
tStat
-5.83974
8.20119
4.231475
3.513064
-1.98347
3.592989
5.068003
F
159.6114
P-value
3.44E-07
6.13E-11
9.45E-05
0.000926
0.052606
0.000725
5.43E-06
Significance F
1.1E-31
Lower 95%
-4.01904
0.388119
0.412993
0.116082
-0.23677
0.12917
0.174462
Upper 95%
-1.96337
0.639571
1.157977
0.42534
0.001375
0.455956
0.403174
(a) Li + Na = Sum of atomic concentrations of Li and Na
(b) Mg + Ca = Sum of atomic concentrations of Mg and Ca
-------
ON
Figure 4.1. Layout of ash samples in the exposure chamber as seen from above. Simulated flue gas enters as shown by the arrow at the left of the figure and
exits as shown by the arrow at the top of the figure. The upper number is the sample position number. The center number is the ID of the ash exposed
in that position during an initial check for reproducibility and positional dependence and the third value is the SOs uptake by the ash in that position in
this initial check.
-------
60
50
"3) 40
(0
i 30
c
0
o
c
O
o
o" 20
10
Inlet Concentration
Outlet Concentration
0.00 0.50
1.00
3.00 3.50
1.50 2.00 2.50
Time, days
Figure 4.2. Typical concentration versus time measurements at the inlet and outlet of the SOs exposure chamber.
4.00
-------
18
16
14
12
» Baseline SO4
• Total SO4, post 350F 20 ppm
• Total SO4, post 375F 20 ppm
• Total SO4, post 450F 20 ppm
• Total SO4, post 450F 70ppm
A Total SO4, Post 850F 70ppm
0.0
0.2
0.4
0.6
1.2
1.4
1.6
0.8 1.0
Base:Acid Ratio
Figure 4.3. Total soluble sulfate found in ash samples after exposure to SOs at the conditions indicated versus ash base:acid ratios.
1.8
-------
16
14
=
o
'55
0)
Di
o
O
V)
0)
£1
_3
O
V)
10
Line^f perfect agreement
6 8 10
Measured % Soluble SO4
12
14
16
Figure 4.4 Plot of SO4 on ash as predicted by the regression equation and the values measured in this study.
-------
5. FORMATION OF SULFUR TRIOXIDE IN THE CONVECTION
SECTION OF COAL-FIRED ELECTRIC UTILITY BOILERS
Formation of SO3 in the Convection Section
Southern Research Institute, working in cooperation with the U.S. Environmental Protection
Agency, Electric Power Research Institute, and Tennessee Valley Authority, is developing a
model relating SOs formation in the convection sections of pulverized coal-fired boilers to coal
properties, convection section design, and operating conditions. The model is intended to provide
electric utilities with guidance on coal specification, sootblowing schedule, and excess air
adjustment to minimize the contribution of convection section processes to formation of sulfuric
acid, subject to the other constraints on fuel selection and system operation.
The processes incorporated in the model are: catalytic oxidation of SCh to 863 by oxide scale on
convection section tubes, catalytic oxidation of SC>2 by ash deposits on the tubes, catalytic
oxidation of SC>2 by suspended fly ash particles, and adsorption of SOs and H2SO4 by fly ash.
The model is a refinement and significant extension of one developed by Peter Walsh and
coworkers at Pennsylvania State University, working with the Consolidated Edison Co. of New
York, Electric Power Research Institute, Empire State Electric Energy Research Corp., and
Florida Power & Light Co. on particulate emissions from residual oil-fired boilers, and with the
New York State Electric and Gas Corp. on the blue plume at Hickling Station in Corning, NY.
That work is documented in two reports and an article.37"39
The basis for the calculation of catalytic oxidation of SO2 to SOs by tube scale, ash deposits, and
fly ash is a chemical mechanism in which SC>2 is oxidized by iron oxide. The oxide, which is
reduced in the process, is returned to its higher oxidation state by reaction with oxygen. This
system of reactions, coupled with diffusion of SOs from its point of formation on the iron-
containing surfaces of tubes, deposits, or suspended fly ash to the free stream, reproduces the
complex dependence of 863 formation on temperature, excess oxygen, coal sulfur content, heat
exchanger surface area, extent of fouling of the tubes, and boiler load. The calculation includes
flue gas, tube, and deposit surface temperatures; reaction rates, including approach to
equilibrium; contact times; and transport of SOs. There is no diffusion limitation on the supply
of the reactants, 862 and 62, because they are present in great excess. The calculation is done
for each row of convective section tubes from the furnace exit through the economizer.
Measurements at an oil-fired boiler showed that a situation that leads to excessive SOs formation
was the growth of fouling deposits on the usually relatively cool surfaces of the primary
superheater toward the back end of the convection section, where tube spacing is relatively close
and the tube surface to gas volume ratio is high. The growth of deposits in this region eventually
leads to deposit surface temperatures in the most active temperature range, and the combination
of these temperatures with high surface to volume ratio generates high levels of 863.
A related problem is illustrated in Figures 5.1, 5.2, and 5.3, for the case of the stoker-fired boiler
at Hickling Station.38 The top panel in Figure 5.1 shows the surface temperature distribution on
clean superheater tubes without fouling deposits on either the steam or fire sides. Sulfur trioxide
30
-------
formation is most rapid in the range of catalyst temperatures from 900 to 1400 °F. The catalyst,
in the case of clean tubes, is iron oxides in tube scale, nearly 100% iron oxides on the SA213
Til and T22 tubing. Sulfur trioxide formation over clean tubes, shown in the bottom panel of
Figure 5.1, is most rapid near the entrance to the superheater where the temperature is at or
above 900 °F. At the lower temperatures further back, 863 formation is negligible.
The case of tubes heavily fouled on the fire side is shown in Figure 5.2. Surface temperatures
through almost the entire superheater are in the most active range. The catalyst is now the
relatively low concentration of iron oxides (~ 0.13 wt%) in the ash deposits. The combined
effect of higher temperature and lower iron oxide concentration is a modest increase in SOs
leaving the superheater, from about 25 to 30 ppmv.
The case of tubes moderately fouled on the steam side but clean on the fire side is shown in
Figure 5.3. Temperatures through much of the superheater are in the most active range and most
of the surface exposed to flue gas is catalytic tube scale. This combination results in a marked
increase in SO3 leaving the superheater, to approximately 75 ppmv. These examples illustrate
some of the complexity of the 863 formation process and the usefulness of a mechanistic model
in providing rational explanation for observations in the field.
Comparisons of model predictions in the Hickling report38 with measured values at various
locations through the plant are shown in Figure 5.4. As can be seen, the model as it existed at the
time appeared to provide plausible results.
Mechanisms of SO3 Formation
Sulfur trioxide is formed by oxidation of sulfur dioxide according to the overall reaction,
SO2 + 1/2 O2-> SO3 (5.1)
Two distinct reaction mechanisms are possible under conditions in boilers: homogeneous (gas)
and heterogeneous (solid-catalyzed). In the flame and postflame regions, above about 1200 K
(1700 °F), the steady-state concentration of SO3 is approximately described by formation and
destruction reactions of SO2 and 863 with oxygen atoms:40
SO2 + O + M->SO3 + M (5.2)
SQ3 + O--> SO2 + O2 (5.3)
"M" is any molecule, such as N2, CO2, H2O, O2, etc., that collides with SO2 and the O atom at
about the same time and carries away energy, stabilizing the product. All of the possible
collision partners are not equally efficient, a refinement that a detailed treatment might take into
consideration. Discussions of this homogeneous reaction system are given by Merryman and
Levy,40 Cullis and Mulcahy,41 and Smith, Wang, Tseregounis, and Westbrook.42
31
-------
LU
O.
Q.
CO
o
CO
1300
1100
900
700
500
80
60
40
20
0
l l 1 l
SURFACE
TEMPERATURE
SO,
0
8 10 12 14
TUBE NO.
Figure 5.1. Surface temperature and SO3 formation in the superheater under the assumption of negligible
fouling and tube scale exposed to flue gas.
38
32
-------
u?
o
QL
2
1-
0.
Q.
"5
CO
o
UUU
1100
900
700
500
80
60
40
20
0
(
— • J i i i i i
^~l~^-^_
SURFACE
~ TEMPERATURE
i i i i i i
i i i j i i
- S03
- —
ff** *****"
JL 1 i i I 1
3 2 4 6 8 10 12 1-
TUBE NO.
Figure 5.2. Surface temperature and SOs formation in the superheater under the assumption of heavy fouling on the
fire side, with ash deposits exposed to flue gas.
38
33
-------
0
R
Q_
LU
H
?
CL
Q.
.-
Q
W
I JUU
1100
900
700
500
80
60
40
20
0
c
] I 1 1 I 1
^^^-n_ -
SURFACE L-T_
~ TEMPERATURE
i i t i i i
i i i i i i
„—•—•—•• — » —
—••'***"*"" '
,x**
*
y* ~
X
1 1 1 I 1 1
) 2 4 6 8 10 12 1-
TUBE NO.
Figure 5.3. Surface temperature and SOs formation in the superheater under the assumption of moderate fouling on
the steam side, with tube scale exposed to flue gas.
38
34
-------
SUPERHEATER
OUT
E
Q.
Q.
in
O
03
80
60
40
20
FURNACE
OUT
GENERATION
BANK OUT
ECONOMIZER
IN OUT
ESP ID FAN
IN OUT
0
4 6
TIME (seconds)
8
10
, 38-
Figure 5.4. Comparison of the calculated SO3 formation to measurements for the Hickling Station modeling. The
model calculations were done under the assumptions of moderate fouling and a mixture of tube scale
and ash deposits exposed to flue gas.
In a region where conditions are not changing too rapidly, the system of reactions (5.2) and (5.3)
approaches a steady state in which the ratio of SOs to SC>2 is constant, independent of the oxygen
atom concentration.40 At the furnace exit, far from the flame zone, where the oxygen atom
concentration might reasonably be assumed to approach equilibrium at the local temperature, a
useful approximation, in the absence of a direct measurement, may be to assume that 862 and
SOs are also in equilibrium at that temperature. This would typically give an SOs volume
fraction on the order of 10 to 20 ppmv in combustion products from fuel containing 1 to 2 wt %
sulfur at temperatures in the vicinity of 1560 K (2350 °F). Measurements in a pilot-scale
combustion tunnel firing residual fuel oil43 support the assumption that 863 volume fractions in a
boiler furnace and at the furnace exit are of roughly this magnitude.
The equilibrium distribution of the sulfur oxides shifts toward 863 as temperature decreases.
Therefore, as combustion products cool, near the furnace exit and further downstream,
-------
equilibrium favors additional conversion of 862 to SOs, but the rate of the homogeneous
reactions decreases with decreasing temperature, closing off this route for SOs formation. The
rate of SO2 oxidation may still be appreciable, however, in the presence of a heterogeneous
catalyst. Solid-catalyzed reactions are thought to be the source of the high levels of SO3
sometimes observed at the economizer outlet.
The best-known catalyst for SC>2 oxidation is vanadium pentoxide, "V^Os, used to generate SOs in
the manufacture of sulfuric acid. V2O5 is a component of selective catalytic NOX reduction
(SCR) catalysts and the ash constituent typically making the greatest contribution to catalytic
SOs formation in high-sulfur residual oil firing. Because of the importance of catalytic oxidation
of SC>2 in the manufacture of a major industrial chemical, this reaction has been the subject of
numerous investigations.44
A comparison of the catalytic activities of iron and vanadium oxides for SC>2 oxidation, for the
specific purpose of understanding SOs formation in oil-fired boilers, was made in laboratory
experiments performed by Wickert.45 His results demonstrated two important features relevant
to boilers: (1) iron oxide was an active catalyst, only slightly less active than vanadium
pentoxide, and with peak activity shifted to slightly higher temperatures than that for vanadium
oxide, and (2) during a fixed residence time in contact with iron pentoxide, conversion of SC>2 to
SOs was greatest over the range of temperatures from roughly 760 to 1030 K (900 to 1400 °F).
The evident activity of iron oxide and the abundance of iron oxide in coal ash and tube scale
suggest that it is likely to be the principal catalyst for SOs formation in coal-fired boilers. Iron
oxides are present in tube scale, in ash deposits on tubes, and in suspended fly ash. Significant
conversion is observed only over a limited range of temperature because at higher temperatures
formation of SOs is limited by the equilibrium distribution of sulfur oxides and at lower
temperatures by the rate of the solid-catalyzed reaction.
Models for both homogeneous formation of SOs and heterogeneous formation catalyzed by ash
deposits in oil-fired boilers were developed by Squires.46 Squires treated the temperature
dependence of conversion in the presence of catalyst by incorporating, in the rate coefficient for
the heterogeneous reaction, a temperature dependence derived from the measurements of
Wickert.45 This approach has the disadvantage that calculated SO2 oxidation could proceed to
unrealistically high conversion in a system having long gas-catalyst contact time, regardless of
the equilibrium constraint.
Development of the model described below also began with a study of SOs formation in the
convection sections of residual-oil-fired boilers.37'39'47 The reactions were described using rate
coefficients having Arrhenius temperature dependence, with conversion limited by the
equilibrium between sulfur oxides. The oil-fired case did not require that a contribution from
suspended ash be considered, because the ash concentration and its surface area per unit of gas
volume are small. The model was applied to a coal-fired stoker boiler by Walsh, DeJohn,
Bower, and Rahimi.38 These authors included an approximate treatment of the contribution to
SOs formation from catalysis by fly ash, which was small, but significant. In pulverized coal
firing, the fly ash concentration and contribution to SOs formation are expected to be larger than
in a stoker-fired boiler, therefore a more rigorous treatment of the contribution to SOs formation
36
-------
by catalytic reaction on suspended fly ash has been implemented in the present version of the
model.
Model for Catalytic SO3 Formation in the Convection Section
The kinetics of heterogeneous oxidation of SC>2 were reviewed by Urbanek and Trela.44 The
essential features of the system are oxidation of SC>2 by a metal oxide, accompanied by reduction
of the metal to a lower oxidation state having little or no catalytic activity. The active higher
oxide is regenerated by reaction with oxygen. The cycle can be represented as follows:
SO2 + FeOx — > SO3 + FeOx-i (5.4)
FeOx.i + 1/2 O2 — > FeOx (5.5)
Equations 5.4 and 5.5 are the overall reactions, not elementary steps, and ki and k2 are global
rate coefficients. The reactions are assumed to be first order with respect to the gaseous species
and the iron oxides. The rate of formation of 863 by reaction 5.4 is then (please see the list of
nomenclature for definitions of the symbols):
rSo,=kJFe0xSCsXs02 (5.6)
When the catalyst composition is steady, the mass fraction of the higher oxide, FeOx, can be
written in terms of the total mass fraction of iron oxides, Ype2o3, from an ash analysis, or any
other measure of total iron content, and the relative rates of formation and destruction of the
higher and lower oxides:
rso.
V
X,., .
(5.7)
The factor in brackets is introduced to account for the approach of the rate to zero as the 863
concentration adjacent to the catalyst surface approaches its equilibrium value. At steady state,
the rate of SOs formation at the deposit surface equals its rate of transport from the surface to the
free stream:
ShD,0
fso = s^-s Cbl (Xso s - Xso ) (5.8)
3 d *'s 3
Combining Eqs. 5.7 and 5.8 to eliminate the unknown concentration of SO3 at the deposit
surface:
rso =kSC (Xso e -Xso) (5.9)
37
-------
in which the effective rate coefficient is:
* = „ „ 1 r n (5.10)
1
ShDs03Cb
cgx
i C,l
S0,,eq
^0,
1 1
""I X S02 ""2 X02
d Cg
Under conditions of interest in boilers the extent of conversion of 862 to SO3 is small, so the
changes in 862 and 62 concentrations accompanying SO3 formation may be neglected. The rate
of SO3 production, Eq. 5.9, is equated to the rate of change in molar flux of SO3 with position in
the convection section. Integration of the resulting expression, with the boundary condition X$o3
= Xso3,o att= t0, gives an expression for the change in SO3 mole fraction with residence time in a
row of tubes in which deposit composition and surface temperature are uniform:
Inspection of the rate expression, Eq. 5.9, and effective rate coefficient, Eq. 5.10, shows that
there are four regimes of SO3 formation: (1) equilibrium; (2) control by diffusion through the
concentration boundary layer adjacent to the deposit (near equilibrium at the surface); (3) control
by O2; and (4) control by SO2.
The temperature dependence of SO3 at equilibrium in the products of combustion of 1.41 wt %
sulfur coal with 5 vol % excess oxygen is shown in Figure 5.5a and the temperature dependence
of the effective rate coefficient and its components is shown in Figure 5.5b, as functions of the
reciprocal of absolute temperature. The total volume fraction of sulfur oxides (SC>2 + SO3) is
1170 ppmv. The equilibrium volume fraction of SO3 (Figure 5.5a) is the ultimate value reached
if the mixture were allowed to stand at the specified temperature, with or without catalyst, for a
long period of time. Referring to the Fahrenheit temperature scale at the top in Figure 5.5a, we
see that at equilibrium at 2200 °F (1478 K), the gas would contain 10 ppmv SO3; at 1800 °F
(1255 K) it would contain 40 ppmv SO3; at 1400 °F (1033 K) it would contain 300 ppmv SO3;
and at 1000 °F (811 K) it would contain 1000 ppmv SO3.
In the flow of combustion products through a convection section, in which the volume fraction of
SO3 is rising from a low value but unable to keep up with the increase in equilibrium
38
-------
TEMPERATURE (°F)
2200 1800 1400 1000
600
EFFECTIVE
RATE
COEFFICIENT
,6
.8 1.0 1.2 1.4 1.6
1000/TEMPERATURE (K)
Figure 5.5. Equilibrium and kinetic constraints on SO3 formation, (a) Equilibrium mole fraction of SO3 in the
products of combustion of Lee coal (wt %, as received: moisture, 11.62; ash, 17.51; volatile matter,
22.92; fixed carbon, 47.95; sulfur, 1.41; and 10,508 Btu/lb HHV, as received), (b) Rate coefficients for:
diffusion of SO3 from the surface of a tube or deposit to the free stream; oxidation of SO2 by reaction
with metal oxide, Reaction (5.4); reoxidation of reduced metal oxide to the active state, Reaction (5.5);
and the overall reaction, when these three processes all occur simultaneously.
concentration associated with the decrease in temperature, the equilibrium concentration at the
local temperature of the point of formation (the catalyst surface) will not be exceeded. The curve
shown in Figure 5.5a therefore represents the maximum possible SOs that could be found in flue
gas adjacent to a catalytic surface at the specified temperature.
39
-------
Walsh, Mormile, and Piper39 proposed the rate coefficients shown in Figure 5.5b to simulate
formation in a residual-oil-fired boiler. The line labeled "diffusion" is the mass transfer
coefficient for SO3 formed at the surface of a tube or ash deposit to the free stream. The dashed
and dot-dash curves are effective rate coefficients for reactions (5.4) and (5.5). They are termed
"effective" because the concentration of catalytic metal oxide on the surface and the distribution
of the oxides between the higher and lower oxidation states are also incorporated in the
coefficients. The solid curve is the overall effective rate coefficient governing the formation of
863. The overall rate coefficient is never greater than the rate coefficient for the slowest process
at a given temperature, which shifts from SC>2 oxidation at low temperatures, to catalyst (metal
oxide) regeneration at intermediate temperatures, to mass transfer at the highest temperatures.
At high temperatures equilibrium also limits the conversion of SO2 to SO3. A connection
between diffusion control and equilibrium at the catalyst surface is expected because, when both
reactants (SC>2 and 02) are present in excess, a limitation by mass transfer implies that the
chemical reaction at the deposit surface is much faster than diffusion and, therefore, that the
concentration of the product, SO3, at the surface approaches its equilibrium value.
Combination of the upper limit imposed by equilibrium (Figure 5.5a) with the limit imposed by
the rate coefficient for SOs formation and the residence time of gas in contact with catalyst,
results in the characteristic behavior observed by Wickert,45 i.e. conversion of SO2 to SO3 is
greatest over a narrow range of temperatures. At lower temperature, the rates of the reactions are
too slow and at higher temperatures the equilibrium concentration (the maximum possible
concentration in the situation under consideration) of SOs is low.
The choice of active surface area is a question where little guidance is available. Several
possible assumptions were examined by Shareef, Homolya, and Mormile.48 In the present work,
the surface area of the convective tubes was used as the basis for the estimate of catalyst surface
area, and the mass fraction of vanadium oxides in deposits provided an estimate of the fraction of
the surface which was active. Measurements of the vanadium content of the outer surface of
deposits from the boiler showed little variation in vanadium content through the convective
49
section.
Simultaneous SO3 Formation on Ash Deposits and Suspended Particles
The equations describing SOs formation catalyzed by iron oxide in suspended fly ash particles
and by iron oxide in stationary deposits or tube scale are the same, Eq. (5.11), but with their
respective external surface areas per unit of gas volume, S, Sherwood numbers, Sh, characteristic
lengths, d (tube and mean particle diameters), mass fractions of iron oxides, YFe2o3, and
temperatures, T (flue gas temperature for fly ash and surface temperature for deposit or clean
tube). The total rate of SOs formation, including the reaction on tube surface or ash deposits,
subscript "t," and the reaction on suspended fly ash, subscript "p," is the sum of the contributions
from the two parallel processes:
^ = k. St Cg (Xso^t - XSO) ) + kpSp Cg (Xso^p - XS03) (5.12)
40
-------
A derivation parallel to that for a single catalytic surface, above, gives the following expression
for the mole fraction of SOs in flue gas at the outlet from the space surrounding a row of tubes,
in the presence of suspended fly ash, when the mole fraction at the inlet to the space is Xso3,o-
ktSt+kpSp
(5.13)
' SOi.eq.p „
P P'
-p" p
exp[-(ktSt+kSp)(t-t0)]
The first term on the right-hand side, which also appears as the first term inside the large
brackets, is an average equilibrium SOs mole fraction, weighted by the relative rates of formation
of SOs over deposit or tube surface and fly ash.
Calculations
Design data and operating conditions for the boiler of a large coal-fired electric generating unit
were provided by the utility to which it belongs. The model described above was used to
calculate 863 formation in the convection section between the furnace exit and the inlet to the
economizer at full load. This region included, in order, beginning at the furnace exit, the
secondary superheater, reheat superheater, primary superheater, and horizontal reheat
superheater, a total of 140 rows of tubes. For the present, preliminary calculations the flue gas
temperature change from the furnace exit to the economizer inlet was divided into 140 equal
increments, corresponding to a temperature change of approximately 5 K (9 °F) across each row
of tubes. The resulting distribution of gas temperatures through the region under consideration is
shown in Figure 5.6. The temperature drop shown in the figure for the economizer was
estimated simply by dividing the overall temperature change and residence time into four equal
increments.
The difference between steam inlet and outlet temperatures for each tube bank was divided into a
number of increments equal to the number of rows of tubes in the bank to estimate the steam
temperature in each row. The general direction of steam flow in each bank was taken into
consideration, where an overall direction could be identified, but finer details of the flow were
not considered, for example in some sections of the secondary and reheat superheaters, where the
steam flow enters through tubes in the middle of a bank, then goes to upstream and downstream
tubes in the same bank. The actual volumetric flow rate of flue gas was calculated at each tube
row and used, with the tube diameter, to determine the residence time in the empty space
associated with that row, the Reynolds, Nusselt, and Sherwood numbers for the tubes, and the
tube surface area per unit of gas volume. The tubes were assumed to be covered with ash to the
extent that the heat transfer resistance from the deposits plus tube metal plus steam film was
equal to the external film resistance in the flue gas. The deposit surface temperature was
assumed to be uniform around the perimeter, along the height or length, and over the width of a
given row of tubes.
41
-------
A mean particle size of 10 micrometers was assumed for the fly ash. Its concentration in the flue
gas was calculated from the coal analysis and local flue gas volumetric flow rate, assuming that
20% of the ash feed was retained in the boiler. Loss of ash by deposition and fall-out in the
convection section was not considered. The fly ash temperature was set equal to the flue gas
temperature at every point. The formation of 863, catalyzed by fly ash, in the cavities within and
between convective tube banks was included, as well as in the space surrounding the tubes.
The calculation was begun at the furnace exit, with the equilibrium SOs in flue gas containing
SC>2 and 62 corresponding to the coal analysis and typical excess air, at the furnace exit gas
temperature. The calculated volume fraction of SOs at this point is 12 ppm. The increment in
the volume fraction of SOs was calculated for each of the 140 rows of tubes using Eq. (5.13) and
local values for the geometry, temperatures, and flow conditions, with the results shown in
Figure 5.7, as the volume fraction of 863 versus flue gas residence time. The rate coefficients,
ki and k2, for the iron-oxide-catalyzed reactions (5.4) and (5.5), respectively, were given the
same values used by Walsh, DeJohn, Bower, and Rahimi38 to explain their measurements in
Boiler No. 1 at Hickling Station. The values assigned to the parameters used in the calculations
are listed in Table 5.1.
Table 5.1 Conditions for the SO3 model run whose results are shown in Figures 5.6
and 5.7.
Parameter
Barometric pressure, inch Hg
Excess air
Stoichiometric air/fuel mass ratio, kg/kg
Coal feed rate, kg/s
Mass fraction sulfur in coal, kg/kg
Mass fraction ash in coal, kg/kg
Mass fraction iron oxides in ash, SOs free
Mass fraction iron oxides in tube scale
Average mass fraction iron oxides presented to flow over tubes
Ratio of heat transfer coefficients, (steam+deposits)/gas
Pressure at furnace exit, inch water column
Furnace exit gas temperature, °F
Average molecular weight of flue gas, wet, kg/kmol
External surface area-weighted mean particle size, m
Fly ash apparent density, kg/m3
Weight fraction of ash feed to furnace bottom
Weight fraction of ash feed to economizer hopper
Volume fraction O2 in furnace exit gas, wet
Catalyst deactivation rate, activation energy, J/kmol
Catalyst deactivation rate, preexponential factor
Catalyst regeneration rate, activation energy, J/kmol
Catalyst regeneration rate, preexponential factor
Value
29.50
0.24
8.570
145.56
0.03
0.08
0.2
1.0
0.2000
1
0
2350
29.1
0.000010
2500
0.2
0.05
0.038
2.40E+08
4.794461 E+14
5.00E+07
0.959806
42
-------
2500
2000
£
3 1500
"ro
o>
Q.
E
0)
1000
re
O
500
0
.
X
X.
\
"*^
Secondary
Superheater
-L -L
\x
X
Reheat
Super-
heater
j.
4 Gas temperature
+ Deposit temperature
Horizontal
Reheat *
^i
» • • • ^
^N
* + + + + •%,*,
Primary
Superheater
\
\
+-
+
•
•
Economizer
0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0 1.1
Time, seconds
Figure 5.6. Modeled temperature profiles through a 1300 MW pulverized coal-fired utility boiler.
1.2 1.3 1.4 1.5
-------
100
90
80
Q.
Q.
c"
.0
"•5
(0
0)
u
c
o
o
co
o
70
60
Superheater
0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0 1.1 1.2 1.3
Time, seconds
1.4 1.5
Figure 5.7. 50 3 concentration profile through a 1300 MW pulverized coal-fired utility boiler as predicted by the model.
-------
Results and Discussion
The surface temperature of the deposit on the first row of secondary superheater tubes, at the
furnace exit, was estimated to be 1130 K (1574 °F), above the high end of the temperature range in
which catalytic 863 formation is thought to be most rapid. Nonetheless, the formation rate is
significant. The discontinuities in the SOs profile in the figure are cavities between tube bundles.
The small positive slopes of the line segments in these regions are indicative of the contributions to
863 formation by suspended fly ash. Its contribution is small throughout the convective section
because the concentration and external surface area of fly ash are low.
At the entrance to the outlet (third) section of the secondary superheater, the deposit surface
temperature is 1068 K (1463 °F), still above the most favorable range for SO3 formation, but the
side-to-side spacing of the tubes has been reduced by half and the number of tube assemblies
doubled, compared to that in the first two secondary superheater sections, so the surface area per
unit of gas volume is doubled, with a corresponding increase in the rate of SOs formation.
At the entrance to the reheat superheater the deposit surface temperature is 1036 K (1405 °F) and
the tube density is the same as at the exit from the secondary superheater. The rate of 863
formation in this region remains high. In the second section of the reheat superheater, however, the
rate is still high but beginning to decline. The deposit surface temperature at the exit from this
section is 868 K (1103 °F), near the temperature at which the highest conversion was observed by
Wickert45 in the presence of iron oxide. Evidently, the temperature dependence of the activity
obtained by fitting the Hickling Station measurements is not the same as that of the iron oxide
samples examined by Wickert.
In the short section at the entrance to the primary superheater (the entrance with respect to the flue
gas flow, it is actually the outlet with respect to the steam flow) the rate of SOs formation again
shows a marked increase, due to another reduction in side-to-side tube spacing and near doubling of
the surface area of ash deposits per unit of gas volume. From the slopes of the line segments
between the following widely spaced (in the flow direction) tubes, one can see the contribution to
SOs formation from the heterogeneous reaction on suspended fly ash. In the last two sections of the
primary superheater, the rate of SOs formation remains high.
The deposit surface temperature at the entrance to the horizontal reheat superheater is 794 K (970
°F). At this point the rate of SOs formation shows a large decline, due at least in part to a factor of
two reduction in the side-to-side tube spacing. Also, surface temperature is evidently approaching
the lower end of the most active range, because the rate of 863 formation leaving this section has
practically dropped to zero. The surface temperature is now 727 K (849 °F), in agreement with the
lower end of the active range of temperatures observed by Wickert.45
Absorption of SO3 and H2SO4 by Fly Ash
Walsh, DeJohn, Bower, and Rahimi38 inferred a rapid decline in SOs at temperatures below about
700 K (800 °F) in the flue of Boiler No. 1 at Hickling Station. This is approximately the
temperature at which reaction with water vapor begins to convert SOs to H2SO4. A preliminary
treatment of this process was incorporated in the present calculation by considering the rate of
diffusion of H2SO4 to suspended fly ash, in combination with the capacity of ash for SOs/H^SC^
uptake described in another section of this report. For this calculation the temperature drop and
45
-------
residence time in the economizer were divided into four equal increments, as mentioned briefly
above. The results of these calculations, also shown in Figure 5.7, agree favorably with the
measured value at the economizer exit. The comparison of the predicted and measured values at the
economizer outlet using reasonable values for the controlling parameters in this instance is
encouraging. However, substantially more testing is needed for validation of the model before it can
be relied upon as a useful tool for predicting SOs emissions from coal-fired boilers.
46
-------
6. DEVELOPMENT OF AN ESTIMATOR FOR REMOVAL OF
SO3/H2SO4 ACROSS AIR PREHEATERS
Condensation of H2SO4 on the surfaces of combustion air preheaters can remove significant
amounts of the incoming SO3/H2SO4, depending on the incoming concentrations and the air
preheater exit temperatures. Although a detailed model of the processes related to the behavior of
H2SO4 in air preheaters exists, an easily used, reasonably accurate means of estimating losses across
rotary air preheaters would be useful. Such a method is described here.
The surface temperatures of the heat exchange elements in the air preheater are cooler than the
incoming flue gas. As the elements of the air preheater rotate the surfaces are heated by the
incoming flue gas, resulting in a gradient in the outlet flue gas outlet temperatures. For the purposes
of this estimator it is assumed that the effective surface temperature for H2SO4 condensation at any
point in the rotation has some constant offset from the exit gas temperature at that point. The acid
dewpoint at any point on the exit face is then taken to be equal to the flue gas temperature at that
point less this offset and the H2SO4 concentration is taken to be the saturation concentration at that
temperature. Typical ranges of gas temperatures across the exit faces of air heaters from the cold
end to the hot end are 50 °F to 100 °F.
Table 6.1 shows measured average concentrations at the exits of four rotary air heaters together
with values predicted as described above using two offset temperatures. Figure 6.1 shows plots of
estimated losses versus average exit temperature for a range of air heater inlet concentrations using
a value of 35 °F for the temperature offset between the local exit gas temperature and the local exit
saturation temperature. Curves are shown for two ranges of gas temperature variation across the exit
faces of the preheaters: 50 °F and 100 °F. Figure 6.2 shows plots of the estimated average H2SO4
exit concentrations for the same conditions used in generating Figure 6.1. The flue gas moisture
concentration will have some affect on the exit H^SC^ concentrations as illustrated in Figure 3.
Figures 6.4 and 6.5 show curves similar to those in Figures 6.1 and 6.2 resulting from the use of a
30 °F offset temperature rather than the 35 °F offset. Figures 6.6 through 6.9 show plots of
measured air heater exit H2SO4 concentrations together with concentrations predicted by the model
using a 30 °F offset temperature.
Table 6.1. Measured and predicted average exit SO3 concentrations for four rotary air heaters.
Site
Rotary No. 1
Rotary No. 2
Rotary No. 3
Rotary No. 4
Measured, ppm
Inlet
3.1
50
42
50
Outlet
0.9
23
18
24
Predicted, ppm
30 degree offset
1.7
30
17
30
35 degree offset
1.4
26
13
26
47
-------
280
290
300 310 320
Average AH Exit Temperature, F
330
—•—120/50
-O" 120/100
—•—80/50
- O- 80/100
50/50
- 50/100
40/50
' 40/100
20/50
--0-- 20/100
10/50
•o- 10/100
340
350
Figure 6.1. Estimated SOs/h^SC^ losses across combustion air preheaters versus average air preheater exit temperature for a temperature offset of 35 °F. The
first value of each pair in the legend is the preheater inlet SO3/H2SO4 concentration in ppm and the second value of the pair is the spread in exit gas
temperature between the cold side and the hot side of the preheater exit.
-------
120/50
120/100
80/50
'80/100
A—so/50
50/100
40/50
'40/100
20/50
280
290
300 310 320
Average AH Exit Temperature, F
330
340
350
Figure 6.2. Estimated air preheater exit SOs/h^SC^ concentration versus average air preheater exit temperature for a temperature offset of 35 °F. The first value
of each pair in the legend is the preheater inlet SO3/H2SO4 concentration in ppm and the second value of the pair is the spread in exit gas temperature
between the cold side and the hot side of the preheater exit.
-------
250
270
290
310
330
350
Average Outlet Temperture, F
370
390
Figure 6.3. Effect of flue gas moisture content on the estimated air preheater exit SO3/H2SO4 concentration for a temperature offset of 35 °F. Values shown are
calculated for an inlet concentration of 50 ppm with an air preheater exit temperature spread of 100 °F.
-------
(A
(A
O
120/50
O" 120/100
80/50
'80/100
A-- 50/100
40/50
40/100
20/50
--0--20/100
10/50
10/100
55 40%
280
290
300 310 320
Average AH Exit Temperature, F
330
340
350
Figure 6.4. Estimated SO3/H2SO4 losses across combustion air preheaters versus average air preheater exit temperature for a temperature offset of 30 °F. The
first value of each pair in the legend is the preheater inlet SOs/h^SC^ concentration in ppm and the second value of the pair is the spread in exit gas
temperature between the cold side and the hot side of the preheater exit.
-------
120/50
0-- 120/100
80/50
'80/100
A— so/50
- -50/100
40/50
40/100
20/50
0-- 20/100
280
290
300 310 320
Average AH Exit Temperature, F
330
340
350
Figure 6.5. Estimated air preheater exit SOs/h^SC^ concentration versus average air preheater exit temperature for a temperature offset of 30 °F. The first value
of each pair in the legend is the preheater inlet SO3/H2SO4 concentration in ppm and the second value of the pair is the spread in exit gas temperature
between the cold side and the hot side of the preheater exit.
-------
(Jl
UJ
250
255
260
265 270 275
Gas Temperature, F
280
285
290
Figure 6.6. Comparison of predicted and measured SOs concentrations at the exit of rotary air heater No. 1 for a temperature offset of 30 °F. The measured inlet
concentration was 3.1 ppm. The measured average was 0.9 ppm while the predicted average was 1.7 ppm.
-------
290
300
310
320
330
340
350
Temperature, F
Figure 6.7. Comparison of predicted and measured SOs concentrations at the exit of rotary air heater No. 2 for a temperature offset of 30 °F. The inlet
concentration was 50 ppm. The measured average was 23 ppm while the predicted average was 30 ppm.
-------
E
Q.
Q.
c"
O
^5
(0
+J
a)
o
c
o
O
«
O
(O
280
290
300
310
320
330
340
Temperature, F
Figure 6.8. Comparison of predicted and measured SOs concentrations at the exit of rotary air heater No. 3 for a temperature offset of 33 °F. The inlet
concentration was 42 ppm. The measured average was 18 ppm while the predicted average was 17 ppm.
-------
_
ON
E
Q.
Q.
c"
O
^5
s
+J
a)
o
c
o
O
«
O
(O
290
300
310
320 330
Temperature, F
340
350
360
Figure 6.9. Comparison of predicted and measured SO3 concentrations at the exit of rotary air heater No. 4 for a temperature offset of 30 °F. The inlet
concentration was 50 ppm. The measured average was 24 ppm while the predicted average was 30 ppm.
-------
7. EVOLUTION OF ACID MIST IN WET FLUE GAS
DESULFURIZATION UNITS AND STACKS
In the initial EPA report on work conducted by SRI on SOs emissions, Farthing and
coworkers assessed the effects of various pollution control technologies on sulfuric acid
emissions and stack plume opacity.2 Among the types of equipment and processes
evaluated were ESPs, SCR, and wet FGD. Marked differences were noted in the
behavior of buoyant, higher velocity plumes formed by hot stack gas in the absence of
wet FGD and plumes formed by denser cold gas leaving a stack downstream from wet
FGD equipment.
In the evaluation of anticipated acid emissions and light scattering coefficients of plumes,
it was found that very few data are available on the behavior of acid mist in wet FGD. A
calculation was presented by Farthing et al. showing the expected growth of acid droplets
by coagulation following nucleation and condensation of the acid on cooling at the
entrance to a scrubber.2 The calculation assumed a fixed acid composition, neglecting
the possibility for growth of the acid droplets by absorption of water vapor from the high
humidity flue gas in the scrubber and stack.
To improve upon the accuracy of calculated acid mist properties, a more detailed analysis
of acid mist evolution during wet FGD was conducted, considering coagulation of the
acid droplets with each other, absorption of water vapor by the droplets, and coagulation
of the acid droplets with spray droplets in the scrubber and with droplets carried over
from the scrubber to the stack.50 The principal findings of that work are summarized
below.
Under the assumptions and conditions investigated, the principal role of a wet SC>2
scrubber, from the point of view of sulfuric acid mist, is to provide conditions for
nucleation of acid droplets and a long residence time for their growth by coagulation and
absorption of water vapor. Downstream from a wet FGD unit, the principal role of the
stack, again from the point of view of sulfuric acid, is to provide additional residence
time for growth of the acid mist by coagulation. Under all of the conditions examined,
the mean size of acid droplets at the stack exit was predicted to have grown, through the
combined effects of coagulation and absorption of water vapor, to sizes within the range
able to scatter visible light.
According to the calculations and within the ranges of the parameters investigated, the
only factor having significant influence on the volume fraction of acid mist leaving the
stack was the original H^SC^ content of the flue gas. When the initial H^SC^
concentration was increased, the ultimate acid droplet volume fraction increased more
rapidly than the H2SO4 concentration itself. Doubling the H2SO4 at the scrubber inlet
from 10 to 20 ppmv increased the acid droplet volume fraction at the stack exit by a
factor of 2.5, from 5.6 to 14 ppmv. The increased absorption of water vapor in the
presence of higher acid concentration was due to the decrease in vapor pressure at the
surfaces of the larger droplets that were formed, due to the Kelvin effect. The decrease in
vapor pressure increased the rate of absorption of water vapor by the acid droplets.
57
-------
In spite of the fact that the mechanism for the increase in volume fraction of acid mist in
the scrubber is vaporization of water from scrubber droplets and absorption of the vapor
by the acid, variation of the size of scrubber spray droplets over the range from 0.7 to 2
mm had hardly any influence on the properties of the acid. Transfer of water from the
scrubber droplets to acid droplets appears to be regulated more by the dependence of the
vapor pressure of the acid droplets on their H2SO4 content than on the specific surface
area, mass transfer coefficient, or concentration of the scrubber spray.
The principal effect of decreasing the scrubber spray droplet size was to increase capture
of acid by spray droplets in the scrubber, although the fractions of acid captured were not
large enough to have significant influence on acid mist droplet size and volume fraction.
The fraction of the original H2SO4 retained in the scrubber increased from 0.33% to 7.6%
on decreasing the spray droplet size from 2 to 0.7 mm, due to increased mass transfer to
the smaller droplets and to the increase in spray droplet concentration as the terminal
velocity of the spray droplets approached the gas velocity. Spray droplets smaller than
0.7 mm were outside the range over which the model is applicable, because closer
approach of the droplet terminal velocity to the gas velocity led to unrealistically high
droplet concentrations characteristic of the "fluidized" regime.51 It was noted that most
of the acid capture was predicted to occur at very early times, when the acid nuclei are
still small, so the calculated capture may be influenced by the assumption that acid
nucleates immediately to form H2S(VH2C) on entering the scrubber.
According to the model, the source of water vapor that could be absorbed by acid mist in
the stack is scrubber droplets carried over from the mist eliminator, not the water vapor
present in the flue gas leaving the scrubber. This is because the acid at this stage is dilute
(~ 1 wt% H2SC>4), so the vapor pressure of water adjacent to acid mist droplets is not far
from the vapor pressure over pure water. A continuous supply of water vapor is therefore
needed if there is to be any significant uptake of water by the acid. Under the conditions
investigated, an increase of 10 to 15% in acid mist volume fraction by absorption of
water vapor in the stack was typical. Cooling of the stack gas, not taken into
consideration in the calculation, is a possible mechanism by which water vapor could be
maintained close to saturation in the presence of its absorption by acid mist. According
to the model and calculations, the most important process occurring in the stack is the
increase in droplet size due to coagulation. The acid droplet size was estimated to
increase by 50% during 15 seconds of residence time in the stack.
The model exhibits some unexpected and counter-intuitive behavior, such as (1) a
decrease in H2SC>4 concentration in acid droplets at the stack exit on increasing the H2SC>4
content of the flue gas, (2) insensitivity of acid volume fraction to scrubber spray droplet
size, and (3) a decrease (though it is slight) in acid mist volume fraction on increasing the
size of scrubber droplets passing the mist eliminator. The potential of such calculations
to assist in the interpretation of observations of sulfuric acid behavior in the field would
appear to justify further development of the model. To make the model a more powerful
and useful simulation tool, the following refinements were proposed:
• Evaluate the effects of scrubber and stack temperature.
58
-------
• Compare the results of the calculations with measurements of acid
concentrations at the entrance and exit from wet FGD units.
• Using the scattering efficiency for visible light versus particle size, determine
relative opacities for the acid mist at the stack exit for comparison of the
effects of scrubber and stack conditions on plume visibility.
• Incorporate a droplet size distribution for the scrubber spray.
• Locate experimental data to improve confidence in the assignment of the
value for the size of scrubber droplets passing the mist eliminator.
• Examine and refine, if necessary, the description of nucleation to provide a
better model for droplet growth at short times and improve the simulation of
capture of acid mist in the scrubber.
• Allow for cooling of flue gas as a driver for transfer of water vapor to acid
mist in the stack.
Background
In a recent EPA report, Farthing et al. assessed the effects of various pollution control
technologies on sulfuric acid emissions and stack plume opacity.2 Among the types of
equipment and processes evaluated were electrostatic precipitators (ESP), selective
catalytic reduction (SCR), and wet flue gas desulfurization (FGD). Marked differences
were noted in the behavior of buoyant, higher velocity plumes formed by hot stack gas in
the absence of wet FGD and plumes formed by denser cold gas leaving a stack
downstream from wet FGD equipment.
In the evaluation of anticipated acid emissions and light scattering coefficients of plumes,
it was found that very few data are available on the behavior of acid mist in wet FGD. A
calculation was presented by Farthing and coworkers, showing the expected growth of
acid droplets by coagulation following nucleation and condensation of the acid on
cooling at the entrance to a scrubber.2 The calculation assumed a fixed acid composition,
neglecting the possibility for growth of the acid droplets by absorption of water vapor
from the high humidity flue gas in the scrubber and stack.
The present report describes calculations of acid mist evolution during wet FGD,
considering coagulation of the acid droplets with each other, absorption of water vapor by
the droplets, and coagulation of the acid droplets with spray droplets in the scrubber and
with droplets carried over from the scrubber in the stack.
The type of SC>2 scrubber under consideration is a counterflow design, in which flue gas
travels upward and scrubber spray droplets travel downward through the gas. The
general outline of the model for the evolution of acid droplets in the scrubber and stack is
as follows. Flue gas is cooled immediately to 333 K (140 °F) on entering the scrubber,
causing nucleation of sulfuric acid. The acid nuclei travel upward through the scrubber at
the average gas velocity, growing by coagulation with each other and by absorption of
water vapor from the flue gas. Some acid droplets are removed from the flow by
coagulation with scrubber spray droplets. The acid and scrubber spray droplets are
59
-------
characterized using single, volume mean sizes. Temperature, gas velocity, and droplet
velocity are considered to be uniform throughout the scrubber and stack.
At the outlet from the scrubber, the mist eliminator is assumed to remove all but 5 ppmv
of the scrubber liquid flow. The droplets carried over are assumed to have a mean size of
10 um under the conditions considered as the base case. Absorption of water vapor by
acid mist droplets, coagulation of acid droplets with themselves, and coagulation of acid
droplets with the remaining scrubber droplets continue in the stack.
In order to perform the calculations, we require the density and surface energy of sulfuric
acid and the vapor pressure of water over sulfuric acid as functions of acid composition.
The correlations for these properties, derived from data in the literature, are given in
Appendix A.
Immediately after their nucleation, acid droplets grow very quickly by coagulation and
condensation. Because growth at early times is so rapid, an exact calculation of the
initial size of the acid nuclei was not thought to be necessary. A useful approximation is
to take the initial size of the nuclei to be the diameter of single molecules of H2SO4-H2O.
This size was estimated by taking the diameter of a sphere having the volume occupied
by the molecule, determined from the density of 84.5 wt % sulfuric acid, given by the
correlation in Figure Al. Under this approximation, the initial number density of acid
nuclei is simply the number concentration of sulfuric acid molecules in the flue gas. The
initial diameter and values for the other constants used in the calculations are given in
Table 7.1.
Development of Mist Behavior Equations
Coagulation of Acid Mist Droplets
The coagulation of acid droplets is described approximately by the rate expression for the
decrease in number concentration of a monodisperse aerosol (Friedlander, p. 193):52
dN
ad
where Nad is the number concentration of acid mist droplets and kad is the rate coefficient
for their coagulation. The rate coefficient is given, for uniform-size droplets, by
(Seinfeld, p. 395;53 Friedlander, p. 19252):
(7.2)
in which dad is the diameter of the acid mist droplets and Dad is their diffusion coefficient.
The factor, /?, in Eq. (7.3) describes the transition of droplet diffusion behavior from the
continuum, through the transition, to the free molecule regime (Seinfeld, p. 395):53
1 + Knn(] af-j
/? = - ^^L - (7.3)
60
-------
where Knad>ad is the Knudsen number for diffusion of acid droplets toward each other,
equal to the ratio of the mean free path of the droplets, Xad, to the droplet radius, dad/2
(Seinfeld, p. 395):53
Knad,ad = ^ (7.4)
dad
The mean free path of acid mist droplets in flue gas was evaluated using the form of the
relationship between the mean speed of particles, their diffusion coefficient, and their
mean free path required for consistency with the correction factor specified in Eq. (7.3),
[Seinfeld (pp. 326, 335-338, 395)]:53
Dad = —r=cad^ad C7-5)
in which cad is the mean speed of the droplets:
<- = M"! (7-6)
The derivation by Seinfeld53 (pp. 395-398) is for different-sized particles, for which the
2 21/2 / —
mean velocity is defined as cy^ = (c\ + c^ ) . This introduces a factor of V2 in Eq.
(7.5), when the particles are identical, and changes the equation from its more familiar
form, D = l/2 cX. The Knudsen number is then (Seinfeld, p. 395):53
a
Knad,ad = - -: - (I •')
caddad
The diffusion coefficient for the droplets is given by the Stokes-Einstein relation with
53
Cunningham correction (Seinfeld, p. 324):
k TC
(7.8)
and the Cunningham correction factor, Cc, is (Seinfeld, p. 317) :53
Cc =l + {^/g;^[1.257 + (0.4e"^)]} (7.9)
where Knfgtad is the Knudsen number for acid droplets in flue gas, the ratio of the mean
free path of the average flue gas molecule to the droplet radius (Seinfeld, p. 326):53
61
-------
u
(7.10)
ad
The mean free path is determined from its relationship to the molecular speed and
diffusion coefficient:
The molecular diffusion coefficient, Dw/g is estimated by treating the flue gas as a binary
mixture of water vapor with the other combustion products. The mean speed of flue gas
molecules is:
Cfn =
(7.2)
and the Knudsen number is then:
Absorption of Water Vapor from Flue Gas by Acid Mist Droplets
All the while that they are coagulating, the acid mist droplets may also absorb water
vapor from the flue gas. Simultaneous evaporation of water from scrubber liquor
droplets maintains the flue gas at high relative humidity. The process can be represented
as follows:
H2O(scrubber spray droplets) = H2O(vapor) (R7.1)
H2O(vapor) = H2O(acid mist droplets). (R7.2)
The rate of change in concentration of scrubber liquor droplets, Csd (kg/m3 gas), due to
evaporation of water from the droplets, is given by the expression describing diffusion of
water vapor from the surface of a droplet, through the concentration boundary layer
surrounding the droplet, and into the free stream:
dcsd _ ShsdDw,fg
,sd ~Lw,voJ I7-14)
dt d
sd
in which Shsd is the Sherwood number for mass transfer between scrubber droplets and
flue gas; Dwfg is the molecular diffusion coefficient in a pseudo-binary mixture of water
vapor and flue gas, ds(j is the mean scrubber liquor droplet size; Ssd is the specific external
surface area (m2/kg) of scrubber liquor droplets; and Cw>sd and CW;00 are the concentrations
62
-------
(kg/m3 gas) of water vapor at the surface of scrubber liquor droplets and in the free
stream, respectively. The Sherwood number was determined using the correlation of
Frossling54 and Ranz and Marshall55 (please see the Nomenclature Section for definitions
of symbols not mentioned here in the text):
Shsd = 2 + 0.6Rel/2Scl/3 (7.15)
The rate of change in acid mist droplet concentration, Cad (kg/m3 gas), due to absorption
of water vapor, is similarly described by the rate of diffusion of water vapor from the free
stream, through the concentration boundary layer surrounding an acid mist droplet, to the
surface of the droplet:
,
dt dad
where the symbols have definitions analogous to those in Eq. (7.14), but now refer to
properties of the acid mist rather than scrubber spray droplets. During evolution of the
acid mist, its droplets grow from the free molecule, through the transition, to the
continuum regime. To account for mass transfer at all three stages, the Sherwood number
was based on the interpolation relation of Dahneke,56 given by Seinfeld (pp. 33S-336):53
1 + Knw nfj
= 2 - ^ - (7.17)
in which Knw>ad is the Knudsen number for water vapor molecules diffusing to acid mist
droplets in flue gas, the ratio of the mean free path of H2O to the droplet radius:
ad
The mean free path of water vapor molecules in flue gas was evaluated using the same
general relationship between mean speed, diffusion coefficient, and mean free path as
before (Seinfeld, p. 338):53
in which Cn2o is the mean speed of water vapor molecules:
'"
=—
07.20)
^ '
The Knudsen number is then:
63
-------
(7.21)
Assuming the system is close to its steady state throughout the scrubber, we can equate
the net rate of change in water vapor concentration, represented by the sum of Eqs. (7.14)
and (7.16), to zero and solve for the pseudo-steady concentration of water vapor in the
free stream, Cw,m.
Shu,
w,ad
1+ ^
Shsd
, *~> sd^ sd^w,sd
-% (7.22)
1 + dad
Sh
'sd
-j Ot/ Ot/
dsd
This expression is substituted for the free stream water vapor concentration in Eq. (7.16),
which then becomes:
dJ7T- 1 C'"'""C""°" i <723>
i 6/tY 6/tY i oCY oCY
rfaJ "«/
Replacing the effective overall rate coefficient in Eq. (7.23) by the symbol, ksa, and
changing from concentrations to partial pressures of water vapor, we have:
dC ,1 W
^^L = ksa(Pw Sd-Pwad)— (7.24)
dt sa^ w'sd w'ad) RT ^ }
Because the curvature of the surface of the small acid droplets is so high, especially at
early times, the equilibrium pressure of water vapor at the surfaces of the droplets is
significantly higher than that over a flat liquid surface. The effect of curvature on the
vapor pressure was evaluated using the relation presented by Equation 7 in Yue:57
dadPaRT
1 , Ya dpa 3 Ya daa
Pa dYa 2 0-fl dYa
(7.25)
The density, pa, and surface energy, cra, of sulfuric acid, and the equilibrium vapor
pressure of water over a flat sulfuric acid surface, Pw,afi are given by the correlations in
Figures Al, A2, and A3, respectively, of Appendix A. The derivatives of the density and
surface energy with respect to composition were also evaluated using their respective
64
-------
correlations. The vapor pressure of H2SO4 over aqueous sulfuric acid is negligible at the
temperatures of interest here.
Coagulation of Acid Mist Droplets with Scrubber Liquor Droplets
The last process to be considered is the coagulation of acid droplets with the much larger
scrubber spray droplets, treated as a diffusion process analogous to the vapor transport
processes discussed above:
n
(7-
..
dt d
sd
Coagulation of acid droplets with scrubber droplets changes the number density of the
acid mist, as well as its mass concentration. The number density can be estimated by
dividing the mass concentration by the average droplet mass, Nad = Cadlmad, therefore,
from Eq. (7.26) we have, approximately, for uniform-size acid mist droplets:
dNad ShsdDad c r \T n TT\
—7— = -- -, - ^sd^sd^ad (7-27)
dt dsd
in which the Sherwood number for the scrubber liquor droplets, Shsd, is given by Eq.
(7.15) and the diffusion coefficient for the acid mist droplets, Dad, by Eq. (7.8).
Calculation Procedure
The mass concentration of acid mist is increased due to absorption of water vapor by the
droplets, but decreased due to loss of acid mist droplets to scrubber liquor. Adding the
contributions from these two processes, according to Eqs. (7.24) and (7.26):
i (r,
,, ~ ksa Vw,s ~ w.a ~ -- ~,
dt RT dsd
This equation was integrated approximately over small increments of time, At, as follows:
Cad (t + At) = Cad (0 + kfl
Cad=Cado at t = 0
Coagulation of acid droplets with themselves does not change their mass concentration.
The number density of acid droplets, on the other hand, is decreased both by coagulation
of the droplets with themselves and by their coagulation with scrubber spray droplets.
Adding the contributions from these two processes, according to Eqs. (7.1) and (7.27):
65
-------
ShsdDad ? r AT n ic\\
3 - SsdCsdNad (7-30)
This equation was integrated approximately over small increments of time, At, as follows:
= Nad(t)-
2
kad (t)Nad (0 + SsdCsdNad (0
(7.3 la)
2
(7.3 Ib)
Following the addition of an increment in time and calculation of new mass and number
concentrations using Eqs. (7.29a) and (7. 3 la), the new average acid droplet mass is found
from:
(7.32)
The cumulative fraction of the original acid captured by scrubber spray droplets is
found from:
fca (t + &) = fca (0 + saSsdCsdCad (t)--At (7.33a)
fca = 0 at t = 0 (7.33b)
The new acid composition, resulting from its dilution by absorption of water vapor, is
given by:
(7.34a)
Ya =0.845 at t = 0 (7.34b)
where Ya = 0.845 is the mass fraction of H2SO4 in the first acid nuclei formed, assumed to
have composition,
Evolution of Acid Mist in the Stack
The same processes of acid mist coagulation, absorption of water by acid droplets,
evaporation of water from scrubber droplets, and coagulation of acid mist with scrubber
droplets continue in the stack, downstream from the wet scrubber and mist eliminator.
However, the characteristics of the system are changed in the following ways: (1) there
are marked decreases in scrubber droplet mass concentration and droplet size across the
mist eliminator, and (2) the remaining scrubber droplets have approximately the same
velocity as the gas and therefore have Sherwood numbers for mass transfer equal
66
-------
approximately to 2. The effects of lower Sherwood number and smaller concentration on
mass transfer are offset somewhat by the smaller droplet size so, depending upon the
conditions in a particular situation, the net change in rates of evaporation of water vapor
from scrubber droplets carried over and coagulation of acid droplets with the scrubber
droplets may be less than might be expected.
In contrast to the situation in the scrubber, where acid mist droplets coagulating with
scrubber spray droplets are removed from the flue gas, acid captured by carried-over
scrubber droplets in the stack is emitted as sulfate, though in a different particle size
range from acid mist.
Results and Discussion
The results of the calculations of acid mist properties as functions of time through a wet
SC>2 scrubber and stack are shown in Figures 7.1 to 7.5. The results for a base case are
presented in Figure 7.1, followed by figures showing the effects of changing individual
parameters characterizing the H2SO4 concentration entering the scrubber, the scrubber
spray quality, and mist eliminator performance. The parameters and their values were:
H2SO4 concentration, 10 and 20 ppmv; scrubber spray droplet size, 0.7, 1, and 2 mm; and
size of droplets leaving the mist eliminator, 10 and 20 um. Droplet size was chosen as
the indicator of mist eliminator performance because its influence is stronger than that of
collection efficiency. The flue gas residence time in the scrubber was fixed at 5 seconds,
followed by up to 15 seconds of residence time in the stack. Identical scales are used on
the ordinates of the figures, for ease of comparison of acid properties under the various
sets of conditions.
Base Case
The following conditions were assigned for the base case: H2SO4 volume fraction in flue
gas entering the scrubber, 10 ppmv; scrubber spray droplet size, 1 mm; and size of
droplets passing the mist eliminator, 10 um. The values of the other parameters used in
the calculations are given in Table 7.1. The evolution of acid mist properties in the
scrubber and stack under the base case conditions is shown in Figure 7.1. The individual
panels in the figure are, from the top: (a) the acid mist droplet size; (b) mass fraction of
H2SC>4 in the acid mist droplets; (c) the number concentration of acid mist droplets; (d)
the volume fraction of acid mist (m3 droplets/m3 flue gas); and (e) the cumulative fraction
of the original acid captured by coagulation of acid mist with scrubber spray droplets.
Both coagulation and absorption of water vapor contribute to the growth of acid mist in
the scrubber, shown at the left in Figure 7. la. The progress of coagulation is best seen in
the decline in number concentration with time, shown on a logarithmic scale in Figure
7. Ic. Growth by coagulation is most rapid at early times, when the number concentration
is high and the droplets are small. Absorption of water vapor has no direct effect on
number concentration, though it influences the rate of coagulation through its effects on
the volume fraction of acid mist and its droplet size.
The progress of absorption of water vapor is best seen in Figure 7.Id, showing the
increase in volume fraction of acid mist with time. Absorption of water vapor by the
droplets is most rapid at early times, when the acid concentration in the droplets is high
and the equilibrium vapor pressure of water at the surfaces of the droplets is low, so there
67
-------
B
0
E
03
b
c
o
tj
CO
10
(A
OS
0.8
0.6
0.4
0.2
0
0.04
0.03
0.02
0.01
0
Scrubber
1x10r
W
0
Q
E
Q.
Q.
C
O
ts
(0
c
o
UL
1x10
1x1014
20
15
10
5
0
0.08
0.06
0.04
Stack
a. Acid Mist Droplet Size
b. Mass Fraction H SO in Droplets -
c. Acid Droplet Number Concentration
d. Acid Droplet Volume Fraction -
e. Fraction H SO to Scrubber Droplets .
10
15
20
Time (seconds)
Figure 7.1. Properties of acid mist as a function of time in the SC>2 scrubber and stack—base case:
H2SO4 entering scrubber, 10 ppmv
mean spray droplet size, 1 mm
mean droplet size passing the mist eliminator, 10 urn
68
-------
is a large difference in water vapor concentration between the free stream and acid
droplet surfaces. The importance of this can be seen by examination of Figure 7.1b,
showing the decline in acid concentration in the acid mist droplets with time due to
dilution by absorbed water vapor. The initial condensation nuclei are assumed to contain
84.5 wt % H2SC>4. During only 5 seconds of residence time in the scrubber, this
concentration is reduced by absorption of water vapor to 0.71 wt%. Corresponding to
this change in composition, the vapor pressure of water adjacent to the acid droplets rises
from approximately 0 to 19954.8 Pa, compared with 19955.5 Pa at the surface of the
scrubber spray droplets, which are the source of water vapor. At an acid concentration of
0.71 wt % there is less than 1 Pa of partial pressure difference driving the transfer of
water from scrubber droplets to acid mist so, by the time they reach the scrubber exit, the
absorption of water vapor by acid droplets has become relatively slow. Coagulation of
acid mist droplets with each other has no direct effect on the volume fraction of acid mist
(m3 acid droplets/m3 flue gas), though it influences the rate of absorption of water vapor
through its effects on droplet size and the consequent effects on water vapor pressure at
the droplet surfaces (Kelvin effect) and mass transfer coefficient.
The evolution of the ability of the scrubber spray to capture acid mist is shown in Figure
7.1e. According to the model, only 2% of the original sulfuric acid is retained in the
scrubber. Most of the capture occurs at very early times, when the sizes of the acid
condensation nuclei are still small and have large diffusion coefficients and high rates of
mass transfer to the much larger scrubber droplets. Because the properties of the acid
mist at short times are evidently so important to this process, the result may be sensitive
to the assumption that all of the H2SO4 nucleates as H2SO4-H2O instantaneously on
entering the scrubber.
Evolution of acid mist in the stack, shown in the right-hand three-quarters of the panels in
Figure 7.1, is dominated by coagulation. Because the mist eliminator removes most of
the scrubber droplets and greatly reduces the source of water vapor, the free stream water
vapor partial pressure drops to only 0.002 Pa above the partial pressure of water vapor
adjacent to the dilute acid droplet surfaces, shutting off the transfer of water from
scrubber droplets to acid. There is little increase in the volume fraction of acid mist in
the stack, as shown in Figure 7. Id, and little further dilution of the acid in the mist, as
shown in Figure 7.1b. Coagulation, however, continues in the stack, as shown by the
factor of three decrease in number concentration (Figure 7.1c) and 50% increase in
droplet size (Figure 7. la) from stack entrance to exit. Coagulation of acid mist with the
low concentration of scrubber droplets passing the mist eliminator is not significant in the
stack (Figure 7.1e). At the stack exit, under the base case conditions, the acid mist
droplets are estimated to have a mean size of 0.5 um, are expected to contain 0.6 wt %
H2SO4, and occupy 6 ppm by volume of the stack gas.
Effect of H2SO4 Concentration
Increasing the initial H2SC>4 from 10 to 20 ppmv changes the properties of the acid mist
as shown in Figure 7.2. The number concentration of acid droplets at the scrubber and
stack exits is only slightly increased (Figure 7.2c), by 16 and 12%, respectively, because
higher initial number concentration increases their coagulation rate. On the other hand,
the volume fraction of acid mist at the scrubber outlet is increased from 5 ppm (base
case) to 13 ppm, a factor significantly larger than the doubling in initial H2SC>4
69
-------
Table 7.1. Values of the parameters used in the calculations for the base case.
Property
Temperature in scrubber
Average molecular weight of flue gas
Flue gas velocity in scrubber
Scrubber spray droplet downward
velocity, relative to the scrubber
Liquid-to-gas ratio in scrubber
Drag coefficient for the scrubber
droplets
Scrubber liquor spray droplet diameter
Density of scrubber spray droplets
Initial mass fraction of h^SC^ in
H2SO4-H2O, the condensation nuclei
Density of sulfuric acid
Surface energy of sulfuric acid
Equilibrium vapor pressure of steam
over a flat sulfuric acid surface
Symbol
T
Wfg
Ufg
Usd
LG
CD
dsd
Psd
Ya0
Pa
oa
Pw.af
Value
333.15 K (140 °F)
30
3.7m/s(12ft/s)
1.5m/s(5ft/s)
0.02 m3/actual m3
(150 gal/1 000 acf)
0.54
1 mm
1225kg/m3
Ww2so4/WH2so4 f/2o = 0.845
Figure A1
Figure A2
Figure A3
Source
typical
typical
typical
typical
typical
a
b
15 wt% solids
c
d
c
Initial diameter of acid condensation
nuclei
Viscosity of flue gas
Molecular diffusion coefficient in
water vapor - flue gas mixture
Fraction of scrubber liquor flow rate
passing the mist eliminator
Mean size of scrubber liquor droplets
passing the mist eliminator
Flue gas residence time in scrubber
0.597 nm
1.458x10'° T
71+110.4
Dw
1.87xlO~lu T
5x10'e
10 urn
5s
kg/(m-s)
nrf/S
h
estimate
typical
a. Michalski, 2000.51
b. Adjusted to give a terminal velocity for the scrubber spray droplets of 5.2 m/s (17 ft/s), equal to
usd, using the drag coefficient of Michalski, 2000.
51
c. Perry et al., 1984, pp. 3-65, 3-66, and 3-237.58
d. Morgan and Davis, 196959; Myhre et al., 1998.60
e. Diameter of a sphere having the volume occupied by one molecule of h^SCU-hbO liquid at 333.15 K.
f. Approximated using the viscosity of nitrogen from Hilsenrath et al., 1955.61
g. Approximated using the binary diffusion coefficient for water vapor-nitrogen from Marrero and Mason,
1972.62
h. 300 MWunit, liquid-to-gas ratio of 150 gal/1000 acf, emission limit of 0.03 lb/106 Btu, slurry is 15 wt%
solids, and 1/4 of particulate matter in stack is scrubber solids.
70
-------
E
"
E
03
b
c
g
ts
2
LL
CO
(/}
03
Scrubber
1X101
if)
c
0
Q
1x101
1x10
14
Q.
a.
c
o
"•5
03
C
O
ts
03
LL
.0
O
20
15
10
5
0
0.08
0.06
0.04
0.02
Stack
a. Acid Mist Droplet Size
b. Mass Fraction KSO in Droplets -
24
c. Acid Droplet Number Concentration
d. Acid Droplet Volume Fraction
e. Fraction H2SO4 to Scrubber Droplets .
10
15
20
Time (seconds)
Figure 7.2. Properties of acid mist as a function of time in the SC>2 scrubber and stack—increased H2SO4
case:
H2SC>4 entering scrubber, 20 ppmv
mean spray droplet size, 1 mm
mean droplet size passing the mist eliminator, 10 urn
71
-------
concentration. The increase in volume fraction, due to increased absorption of water
vapor, is consistent with increased dilution of the acid, as shown by comparison of
Figures 7.1b and 7.2b. The mass fraction of H2SO4 in the acid mist at the scrubber outlet
is now only 0.55 wt %, compared with 0.71 wt % at the same location under the base
case conditions. The higher absorption of water vapor in the present case is driven by the
lower vapor pressure of water at surfaces having larger radii of curvature, characteristic
of the larger acid droplets (Kelvin effect57). The combination of increased amounts of
condensing species and small change in droplet number concentration result in a
significant increase in mean acid mist droplet size at the scrubber outlet, from 0.33 um in
the base case to 0.44 um on doubling the H2SO4 concentration. The fraction of the acid
captured in the scrubber (Figure 7.2e) decreases in inverse proportion to the increase in
initial H2SO4 concentration, from 2% in the base case (Figure 7. le) to only 1%.
As in the base case, there is little absorption of water by acid mist in the stack, because
the mist eliminator removed of most of the scrubber droplets, which were the source of
water vapor in the scrubber. Neither the droplet composition (Figure 7.2b) nor the
droplet volume fraction (Figure 7.2d) changes significantly in the stack. Coagulation of
the acid mist, however, continues in the stack, increasing the mean droplet size from 0.44
um at the scrubber outlet to 0.65 um at the stack exit, a fractional increase similar to that
observed in the base case. Little coagulation of acid mist with scrubber droplets passing
the mist eliminator is expected (Figure 7.2e).
Scrubber Droplet Size
The effects of increasing the mean scrubber spray droplet size from 1 mm to 2 mm and
decreasing it to 0.7 mm are shown in Figures 7.3 and 7.4, respectively. Compared with
the base case, Figure 7.1, there are no large changes in acid mist droplet size,
composition, number concentration, or volume fraction in either the scrubber or the stack.
The properties of acid mist leaving the stack are practically the same in all three cases.
According to the model, the only significant change in the behavior of the system is an
increase in the capture of H^SC^ in the scrubber as the spray droplet size decreases,
shown by comparison of Figures 7.1e, 7.3e, and 7.4e. The fraction of the original H^SC^t
captured by 2-mm spray droplets was 0.33%, the fraction captured by 1-mm droplets was
2%, and the fraction captured by 0.7-mm droplets was 7.6%. A corresponding slight
reduction in the volume fraction of acid mist leaving the stack can also be seen (Figures
7.1d,7.2d, and 7.3d).
Further reduction in the size of the scrubber spray droplets below 0.7 mm is outside the
range of application of the model. The terminal fall velocity of 0.5 mm droplets is 3.7
m/s, equal to the upward gas velocity in the scrubber. As the spray droplet size
approaches 0.5 mm, the calculated concentration of droplets increases to physically
unrealistic levels. In practice this corresponds to the "fluidized" regime,51 not described
by the present model. The calculated volume fractions of droplets in the scrubber were 2,
5, and 11% for the spray droplet sizes of 2, 1, and 0.7 mm, respectively. The increase in
acid capture with decreasing size is due to the combined effects of increasing specific
surface area and increasing concentration of the spray droplets.
We note that, according to the calculation, capture of acid by scrubber spray droplets
occurs primarily at early times when the acid nuclei are still very small and have high
72
-------
2 scrubber and stack—increased scrubber
droplet size case:
H2SC>4 entering scrubber, 10 ppmv
mean spray droplet size, 2 mm
mean droplet size passing the mist eliminator, 10 urn
73
-------
0
05
Q
c
o
t>
05
05
0.8
0.6
0.4
0.2
0
0.04
0.03
0.02
0.01
Scrubber
1x10'
20
15
10
5
0
2 0.08
c
•B 0.06
o
2 0.04
•§ 0.02
E
Q.
Q.
C
g
"o
OJ
§
Stack
a. Acid Mist Droplet Size
b. Mass Fraction H SO in Droplets -
c. Acid Droplet Number Concentration
d. Acid Droplet Volume Fraction -
e. Fraction H SO to Scrubber Droplets
10
15
20
Time (seconds)
Figure 7.4. Properties of acid mist as a function of time in the SC>2 scrubber and stack—decreased scrubber
droplet size case:
H2SC>4 entering scrubber, 10 ppmv
mean spray droplet size, 0.7 mm
mean droplet size passing the mist eliminator, 10 urn
74
-------
diffusion coefficients. This is shown by the rapid rise in the fraction of H2SO4 captured
near time = 0 in Figure 7.4e. The present estimates of acid capture in the scrubber may
be influenced by the assumption that nucleation of acid begins with instantaneous
formation of H2SO4-H2O at the inlet. A more detailed description of the nucleation of
acid droplets may be needed to achieve high accuracy in the calculation of acid removal
by the scrubber. The presence of a distribution of droplet sizes in the scrubber spray also
needs to be considered, including a mechanism for removal of droplets having terminal
velocities near the gas velocity, so their concentrations are properly bounded. Possible
mechanisms for removal of droplets near the critical size are coalescence of the droplets
and a distribution of gas velocities in the scrubber.
Performance of the Mist Eliminator
The last case to be considered is the effect of spoiling the performance of the mist
eliminator. We chose to simulate this by increasing the mean size of scrubber droplets
passing the mist eliminator, rather than decrease the collection efficiency, because the
base case already assumes that one-half of the particulate matter emissions limit is
contributed by scrubber solids (Table 7.1, footnote h), so there is little room left for
adjustment of the efficiency. The effect of increasing the mean size of scrubber droplets
in the stack from 10 to 20 um is shown in Figure 7.5. Interestingly, the most visible
effect, compared with the base case, is a small decrease in the acid mist volume fraction
at the stack exit with increasing droplet size. This is due to the decrease in rate of supply
of water vapor to the free stream from the larger size scrubber droplets, limiting the
transfer of water from the scrubber droplets to the acid mist. The acid concentration in
the mist droplets at the exit from the stack is slightly higher, and the droplet size slightly
smaller, than in the base case. With respect to acid mist, and under the conditions
investigated, the function of scrubber droplets not collected by the mist eliminator is to
provide a small supply of water vapor to increase the volume fraction of acid mist in the
stack. An increase of 10 to 15% in acid mist volume fraction by absorption of water
vapor in the stack was typical. The concentration of scrubber droplets is too low, and
both acid mist and scrubber droplets are too large, for the scrubber droplets to be
effective as scavengers of acid mist in the stack. Larger increases in the volume fraction
of acid mist by absorption of water vapor or capture of acid mist by scrubber droplets in
the stack might be possible under some conditions, for example in the presence of low
efficiency of mist removal at the scrubber exit or unusually small size of droplets passing
the mist eliminator.
Comparisons of Predicted and Measured Acid Droplet Size
Distributions
Over the past several years Southern Research has developed a technique for analyzing
the material collected on the stages of cascade impactors to separately ascertain the size
distributions of the primary sources of the material collected in the various size fractions:
fly ash, scrubber solids, and condensed H2SC>4. Examples of these breakdowns by
constituent are shown in Figures 7.6 through 7.8. As can be seen, the bulk of the
condensed H2SC>4 is found in particles having diameters of a few tenths of a micrometer.
Comparisons of the volume mean diameter of the condensed acid from the model to
extrapolations from the measured values of the diameters of the acid are shown in Figure
7.9. The extrapolations were made by assuming constant number concentrations with the
75
-------
E
CD
•5
co
Q
c
.o
"o
co
LL
(0
CO
CO
0.8
0.6
0.4
0.2
0
0.04
0.03
0.02
0.01
0
Scrubber
IxlO1
V)
o
O
1x10
1x10
,14
CL
Q.
C
O
TS
2
u.
5
c
o
'•s
CO
20
15
10
5
0
0.08
0.06
0.04
0.02
0
Stack
a. Acid Mist Droplet Size
b. Mass Fraction KSO in Droplets -
24
c. Acid Droplet Number Concentration
d. Acid Droplet Volume Fraction -
e. Fraction H SO to Scrubber Droplets .
10
15
20
Time (seconds)
Figure 7.5. Properties of acid mist as a function of time in the SC>2 scrubber and stack—increased size of
droplets passing the mist eliminator case:
H2SC>4 entering scrubber, 10 ppmv
mean spray droplet size, 1 mm
mean droplet size passing the mist eliminator, 20 urn
76
-------
diameters being proportional to the cube root of the 863 concentration. As can be seen,
the predictions compare favorably with values based on measurement. The validity of the
predictions are further strengthened when predicted opacities in the plume from a 1300
MW coal-fired using a wet scrubber for 862 control are compared with those measured
by a certified "smoke reader" as shown in Figure 7.10. The predictions and
measurements in the latter case were made for the point downwind of the stack at which
the condensed water fog had evaporated.
Conclusions and Recommendations
Under the assumptions and conditions investigated, the principal role of a wet SC>2
scrubber, from the point of view of sulfuric acid mist, is to provide conditions for
nucleation of acid droplets and a long residence time for their growth by coagulation and
absorption of water vapor. Downstream from a wet FGD unit, the principal role of the
stack, again from the point of view of sulfuric acid, is to provide additional residence
time for growth of the acid mist by coagulation. Under all of the conditions examined,
the mean size of acid droplets at the stack exit is predicted to have grown, through the
combined effects of coagulation and absorption of water vapor, to sizes within the range
able to scatter visible light.
According to the calculations and within the ranges of the parameters investigated, the
only factor having significant influence on the volume fraction of acid mist leaving the
stack was the original H2SO4 content of the flue gas. When the initial H2SO4
concentration was increased, the ultimate acid droplet volume fraction increased more
rapidly than the H2SO4 concentration itself. Doubling the H2SO4 at the scrubber inlet
from 10 to 20 ppmv increased the acid volume fraction at the stack exit by a factor of 2.5,
from 5.6 to 14 ppmv. The increased absorption of water vapor in the presence of higher
acid concentration was due to the decrease in vapor pressure at the surfaces of the larger
droplets that were formed, due to the Kelvin effect. The decrease in vapor pressure
increased the rate of absorption of water vapor by the acid droplets.
In spite of the fact that the mechanism for the increase in volume fraction of acid mist in
the scrubber is vaporization of water from scrubber droplets and absorption of the vapor
by the acid, variation of the size of scrubber spray droplets over the range from 0.7 to 2
mm had hardly any influence on the properties of the acid. Transfer of water from the
scrubber droplets to acid droplets appears to be regulated more by the dependence of the
vapor pressure of the acid droplets on their H2SO4 content than on the specific surface
area, mass transfer coefficient, or concentration of the scrubber spray.
The principal effect of decreasing the scrubber spray droplet size was to increase capture
of acid by spray droplets in the scrubber, although the fractions of acid captured were not
large enough to have significant influence on acid mist droplet size and volume fraction.
The fraction of the original H2SO4 retained in the scrubber increased from 0.33% to 7.6%
on decreasing the spray droplet size from 2 to 0.7 mm, due to increased mass transfer to
the smaller droplets and to the increase in spray droplet concentration as the terminal
velocity of the spray droplets approached the gas velocity. Spray droplets smaller than
0.7 mm were outside the range over which the model is applicable, because closer
approach of the droplet terminal velocity to the gas velocity led to unrealistically high
droplet concentrations characteristic of the "fluidized" regime.51 It was noted that most
77
-------
of the acid capture was predicted to occur at very early times, when the acid nuclei are
still small, so the calculated capture may be influenced by the assumption that acid
nucleates immediately to form H^SCVH^O on entering the scrubber.
According to the model, the source of water vapor that could be absorbed by acid mist in
the stack is scrubber droplets carried over from the mist eliminator, not the water vapor
present in the flue gas leaving the scrubber. This is because the acid at this stage is dilute
(~ 1 wt % H2SO4), so the vapor pressure of water adjacent to acid mist droplets is not far
from the vapor pressure over pure water. A continuous supply of water vapor is therefore
needed if there is to be any significant uptake of water by the acid. Under the conditions
investigated, an increase of 10 to 15% in acid mist volume fraction by absorption of
water vapor in the stack was typical. Cooling of the stack gas, not taken into
consideration in the present calculation, is a possible mechanism by which water vapor
could be maintained close to saturation in the presence of its absorption by acid mist.
According to the present model and calculations, the most important process occurring in
the stack is the increase in droplet size due to coagulation. The acid droplet size was
estimated to increase by 50% during 15 seconds of residence time in the stack.
The model exhibits some unexpected and counter-intuitive behavior, such as the decrease
in H2SO4 concentration in acid droplets at the stack exit on increasing the H2SO4 content
of the flue gas (Figures 7.1b and 7.2b), the insensitivity of acid volume fraction to
scrubber spray droplet size (Figures 7.3d and 7.4d), and the decrease (though it is slight)
in acid mist volume fraction on increasing the size of scrubber droplets passing the mist
eliminator (Figures 7. Id and 7.5d). The potential of such calculations to assist in the
interpretation of observations of sulfuric acid behavior in the field would appear to justify
some further development of the model. To make the calculation a more powerful and
useful simulation tool, the following refinements are proposed:
• Evaluate the effects of scrubber and stack temperature.
• Compare the results of the calculations with measurements of acid
concentrations at the entrance and exit from wet FGD units.
• Using the scattering efficiency for visible light versus particle size, determine
relative opacities for the acid mist at the stack exit for comparison of the
effects of scrubber and stack conditions on plume visibility.
• Incorporate a droplet size distribution for the scrubber spray.
• Locate experimental data to improve confidence in the assignment of the
value for the size of scrubber droplets passing the mist eliminator.
• Examine and refine, if necessary, the description of nucleation to provide a
better model for droplet growth at short times and improve the simulation of
capture of acid mist in the scrubber.
• Allow for cooling of flue gas as a driver for transfer of water vapor to acid
mist in the stack.
78
-------
o
c
T3
0.06
0.05
0.04
0 0.03
Q
G
O
0.02
0.01
0
~~ Solids, less Carbon
• ' H2SO4 + 5H2O
~~ Carbon
Total Volume
0.01
0.1
10
Diameter,
Figure 7.6. Measured size distributions of acid mist and other particulate components downstream of scrubber module "A."
Concentrations are shown on the basis of particle volume. The H2SO4 concentration in the flue gas was 2.6 ppmv.
-------
oo
o
0.06
0.05
0.01
Acid Alone
'Solids Alone
Sum Acid Alone + Solids Alone
0.1
10
Diameter,
Figure 7.7. Measured size distributions of acid mist and other particulate components in the stack downstream of a scrubber system.
Concentrations are shown on the basis of particle volume. The H2SO4 concentration in the flue gas was 2.3 ppmv.
-------
bU
50
E/m
40
O
C
^^
^J)
E
IM/dLogD
> c.
> c
<-> £U
10
n
I
I
!•
4
I
J
^*\
{' \\
\
\
3
\
\
I
\
^
*
t\
\ "
i
r
• •
p
•4
r~
i m
*
•<
i
" ^ •"!
^^2-7-06 H2SO4+H2O
- +- 1-04H2SO4+H2O
"!•
.,
k4
^
^1-1-i
:-t-
*.
k-
i— -i
0.1
10
100
Diameter,
Figure 7.8. Measured size distributions of acid mist in the stack as shown in Figure 7.7 after replacement of the scrubber system with one of a different type.
Concentrations here are shown on the basis of particle mass.
-------
07
./
Of
JO
E
_» n e
3. U-O
H
L.
fi ^
u/ U.*t
.2
Q A 0
«- °-3
.S*
Q.
O A O -
t 0"t
Q
04
. 1
•
/
'
(
jr
;
'
r-
,^
--
^^
'
*'
'.
^
i
•
— -
— '
-*"
'
_^ -
--*
i
-4— Extrapolated from
measured values
• Predicted values
0 5 10 15 20 25 30
H2S04, ppm
Figure 7.9. Comparison of droplet diameters extrapolated from measurements with those predicted by the model. The
extrapolations were made by assuming constant number concentrations with the diameters being proportional to the
cube root of the SO3 concentration.
-------
100
90
80
70
'Predicted Plume at dia. = 70m, Acid only
Predicted Plume at dia = 70 m w Part. & NO2
Reported by VEE Smoke-reader
10
15
H2SO4, ppm
20
25
30
Figure 7.10. Comparison of predicted plume opacities versus H2SO4 concentration with those measured by a certified "smoke reader" for a 1300
MW unit with a pollution control system consisting of an SCR followed by a cold-side ESP and an SO2 scrubber.
-------
8. REFERENCES
1. U.S. Environmental Protection Agency, "Rule To Reduce Interstate Transport of Fine
Particulate Matter and Ozone (Clean Air Interstate Rule)," Federal Register, Vol.
70, pp. 25162-25405, 2005.
2. Farthing, W.E., P.M. Walsh, J.P. Gooch, J.D. McCain, W.S. Hinton and R.F. Heaphy,
Identification of (and Responses to) Potential Effects of SCR and Wet Scrubbers
on Submicron Particulate Emissions and Plume Characteristics., Office of
Research and Development, EPA-600/R-04-107, Research Triangle Park, NC,
August, 2004.
3. Leikauf, G., D.B. Yeates, K.A. Wales, D. Spektor, R.E. Albert and M. Lippmann,
"Effects of sulfuric acid aerosol on respiratory mechanics and mucociliary particle
clearance in healthy nonsmoking adults," AIHA Journal, Vol. 42, pp. 273 - 282,
1981.
4. Wyzga, R.E. and LJ. Folinsbee, "Health effects of acid aerosols," Water, Air, & Soil
Pollution, Vol. 85, pp. 177 - 188, 1995.
5. Folinsbee, L.J., "Human health effects of exposure to airborne acid," Environmental
Health Perspectives, Vol. 79, pp. 195-199, 1989.
6. Schlesinger, R.B., J.T. Zelikoff, L.C. Chen and P.L. Kinney, "Assessment of
toxicologic interactions resulting from acute inhalation exposure to sulfuric acid
and ozone mixtures," Toxicology and Applied Pharmacology, Vol. 115, pp. 183-
190, 1992.
7. Jakab, G.J., R.W. Clarke, D.R. Hemenway, M.V. Longphre, S.R. Kleeberger and R.
Frank, "Inhalation of acid coated carbon black particles impairs alveolar
macrophage phagocytosis," ToxicologicalLetters, Vol. 88, pp. 243-248, 1996.
8. New Jersey Department of Health and Senior Services, Sulfur Trioxide Hazardous
Substance Fact Sheet, http://www.state.nj.us/health/eoh/rtkweb/1767.pdf
accessed February 2005, 2005.
9. Downing, B., "Power Plant Releases Clouds Of Sulfuric Acid," Akron Beacon Journal,
Akron, OH, p. 1, Sect. A, August 2, 2001.
10. U.S. Environmental Protection Agency Region 5, "Request for the ATSDR to
evaluate air monitoring data and assess any adverse human health effects on the
residents of Cheshire, Gallia County, Ohio," to Senior Regional Representative,
Agency for Toxic Substances and Disease R., Chicago, IL, July 24, 2001.
11. Agency for Toxic Substances and Disease Registry, Health Consultation: Gavin
Power Plant, Cheshire, Gallia County, Ohio,
http://www.atsdr.cdc.gov/HAC/PHA/gavinpower/gpp_p 1 .html accessed July 10,
2006.
12. Zetlmeisl, M.J., K.J. McCarthy and D.F. Laurence, "Investigation of some factors
affecting nonflame generation and suppression of SOs," Industrial Engineering
and Chemistry Product Research and Development, Vol. 22, pp. 710-716, 1983.
13. Srivastava, R.K., C.A. Miller, C. Erickson and R. Jambhekar, "Emissions of sulfur
tri oxide from coal-fired power plants," Journal of the Air & Waste Management
Association, Vol. 54, pp. 750-762, 2004.
84
-------
14. U.S. Department of Energy, Annual Steam-Electric Plant Operation and Design
Data., Energy Information Administration, Form EIA-767, Washington, DC,
2002.
15. U.S. Environmental Protection Agency, 2002 National Emissions Inventory Data &
Documentation,
ftp://ftp.epa.gov/EmisInventory/2002finalnei/point sector data/national/, August
7, 2006.
16. The Mcllvaine Company, Utility Environmental Upgrade Tracking System, The
Mcllvaine Company, http://www.mcilvainecompany.com/, 2006.
17. Hardman, R., R. Stacey and E. Dismukes, "Estimating sulfuric acid aerosol emissions
from coal-fired power plants," presented at DOE-FETC Conference on
Formation, Distribution, Impact, and Fate of Sulfur Trioxide in Utility Flue Gas
Streams, Pittsburgh, PA, March, 1998.
18. U.S. Environmental Protection Agency, Analysis in Support of the Clean Air
Interstate Rule Using the Integrated Planning Model, Office of Air and Radiation,
Clean Air Markets Division, Washington, DC, May 28, 2004.
19. "Madigan, Kaid comment on lawsuit filed Monday against Cinergy/PSI," Daily
Republican Register, Mt. Carmel, IL, August 10, 2004.
20. Cassell, B., "Murray to expand Galatia Coal operation in Illinois into high-sulfur
steam," SNL Energy Coal Report, Charlottesville, VA, January 24, 2005.
21. Cassell, B., "Alliance Resource Partners works on major high-sulfur mine project,"
SNL Energy Coal Report, Charlottesville, VA, January 24, 2005.
22. Barber, W., "Cinergy emissions control spending grows to $465 million in 2005,"
SNL Energy Coal Report, Charlottesville, VA, January 24, 2005.
23. Cassell, B., "Numerous scrubbers to cut SO2 emissions, boost high-sulfur coal," SNL
Energy Coal Report, Charlottesville, VA, January 31, 2005, 2005.
24. Cassell, B., "LG&E, KU to remake coal market with new coal unit, scrubbers," SNL
Energy Coal Report, Charlottesville, VA, January 31, 2005.
25. Cassell, B., "PPL boosts high-sulfur coal with Brunner, Montour scrubbers," SNL
Energy Coal Report, Charlottesville, VA, February 28, 2005.
26. Monroe, L. and K. Harrison, An Updated Method for Estimating Total Sulfuric Acid
Emissions from Stationary Power Plants, Southern Company Services, Research
& Environmental Affairs, Internal Report, Birmingham, AL, March, 2003.
27. U.S. Environmental Protection Agency, "Method 8 - Determination of Sulfuric Acid
and Sulfur Dioxide Emissions from Stationary Sources," Code of Federal
Regulations, Title 40 Part 60 (Appendix A-4), 2004.
28. Cheney, J. and J. Homolya, Characterization of Combustion Source Sulfate
Emissions with a Selective Condensation Sampling System, Office of Air Quality
Planning and Standards, EPA-600/9-78-020a, Research Triangle Park, NC, 1978.
29. Maddalone, R. and N. Garner, Process Measurement Procedures - H2SO4 Emissions,
Office of Air Quality Planning and Standards, EPA-600/7-79-156, Research
Triangle Park, NC, 1979.
30. Blythe, G., C. Galloway, C. Richardson and R. Rhudy, "Improvements to the
controlled condensation measurement method for sulfuric acid," presented at
EPRI-DOE-EPA Combined Utility Air Pollution Control Symposium: The Mega
Symposium, Atlanta, GA, August 16-20, 1999.
85
-------
31. Blythe, G., C. Galloway and R. Rhudy, "Flue gas sulfuric acid measurement method
improvements," presented at AWMA EPRI-DOE-EPA Combined Utility Air
Pollution Control Symposium: The Mega Symposium, Chicago, IL, August 20-
23,2001.
32. McCain, J.D., R.S. Saltzman and B. Carney, "Development of a low maintenance,
field ruggedized 863 CEM," presented at Air Quality V, Arlington, VA,
September 19-21, 2005.
33. McCain, J.D., Development of Particle Size Test Methods for Sampling High
Temperature and High Moisture Source Effluents., Southern Research Institute,
California Air Resources Board Contract Al32-084, Birmingham, AL, June,
1994.
34. U.S. Environmental Protection Agency, "Method 17 - Determination of Particulate
Matter Emissions from Stationary Sources," Code of Federal Regulations, Title
40 Part 60 (Appendix A-6), pp. 426-431, 2004.
35. Bickelhaupt, R.E., Measurement of Fly Ash Resistivity Using Simulated Flue Gas
Environments, Industrial Energy Research Laboratory, EPA-600/7-78-035,
Research Triangle Park, NC, March, 1978.
36. Bickelhaupt, R.E., A Technique for Predicting Fly Ash Resistivity, U.S.
Environmental Protection Agency, Industrial Energy Research Laboratory, EPA-
600/7-79-204, Research Triangle Park, NC, August, 1979.
37. Walsh, P.M., Properties of Particles Formed during Combustion of Residual Fuel
Oils, Empire State Electric Energy Research Corp., Research Report EP90-15,
New York, NY, October, 1996.
38. Walsh, P.M., T.E. DeJohn, E.T. Bower and M. Rahimi, Measurements of Sulfur
Trioxide and Sulfate in the Products of Coal Combustion in a Traveling Grate
Stoker, Pennsylvania State University, University Park, PA, June, 1994.
39. Walsh, P.M., D.J. Mormile and B.F. Piper, "Sulfur trioxide formation in the presence
of residual oil ash deposits in a electric utility boiler," American Chemical
Society, Division of Fuel Chemistry, Vol. 38, pp. 294-300, 1993.
40. Merryman, E.L. and A. Levy, "Sulfur trioxide flame chemistry - H^S and COS
flames," Thirteenth Symposium (International) on Combustion, The Combustion
Institute, Pittsburgh, PA, pp. 427-436, 1971.
41. Cullis, C.F. and M.F.R. Mulcahy, "The kinetics of combustion of gaseous sulphur
compounds," Combustion and Flame, Vol. 18, pp. 225-292, 1972.
42. Smith, O.I., S.-N. Wang, S. Tseregounis and C.K. Westbrook, "The sulfur catalyzed
recombination of atomic oxygen in a CO/O2/Ar flame," Combustion Science and
Technology, Vol. 30, pp. 241-271, 1983.
43. Walsh, P.M., W.F. Farmayan, T. Kolb and J.M. Beer, "Sulfur trioxide formation in
high sulfur residual oil flames," American Chemical Society, Division of Fuel
Chemistry, Vol. 31, pp. 112-121, 1986.
44. Urbanek, A. and M. Trela, "Catalytic oxidation of sulfur dioxide," Catalysis Reviews,
Science and Engineering, Vol. 21, pp. 73-133, 1980.
45. Wickert, K., The Oil-Firing Yearbook, Kopf, Stuttgart, 1963.
46. Squires, R.T., "The kinetics of SOs formation in oil-fired boilers," Journal of the
Institute of Energy, Vol. 55, pp. 41-46, 1982.
86
-------
47. Walsh, P.M., A.N. Sayre and J.M. Beer, Reduction of Particulate Emissions from
High Asphaltene Heavy Fuel Oil Flames, Empire State Electric Energy Research
Corp., Research Report EP85-21, Albany, NY, November, 1988.
48. Shareef, G.S., J. Homolya and DJ. Mormile, "SOs formation over fuel oil fly ash
deposits," presented at the 1990 862 Control Symposium, New Orleans, LA, May
8-11, 1990.
49. Piper, B.F. and T. Kokoska, Opacity study - data to support SOs formation analysis,
KVB, Inc., KVB 81-21620-1447, prepared for Consolidated Edison Company of
New York, 1983.
50. Walsh, P.M., Formation of Acid Mist and Evolution of its Properties in the Wet SO2
Scrubber and Stack of a Coal-Fired Electric Utility Boiler', University of Alabama
at Birmingham, Report prepared for Southern Research Institute, Birmingham,
AL, April 5, 2005.
51. Michalski, J.A., "Development of droplet size distribution in FGD spray towers,"
Atomization and Sprays, Vol. 10, pp. 105, 2000.
52. Friedlander, S.K., Smoke, Dust, and Haze: Fundamentals of Aerosol Dynamics,
Oxford University Press, New York, 2000.
53. Seinfeld, J.H., Atmospheric Chemistry and Physics of Air Pollution, John Wiley &
Sons, New York, NY, 1986.
54. Frossling, N., "Uber die Verdunstung fallender Tropfen (The evaporating of falling
drops)," Gerlands Beitrage zur Geophysik, Vol. 52, pp. 170-216, 1938.
55. Ranz, W.E. and W.R. Marshall, Jr., "Evaporation from drops, Parts I and II,"
Chemical Engineering Progress, Vol. 48, pp. 141-146, 173-180, 1952.
56. Dahneke, B., "Simple kinetic theory of Brownian diffusion in vapors and aerosols," in
Theory of Dispersed Multiphase Flow, Mayer, R.E. (ed.), Academic Press, New
York, NY, 1983.
57. Yue, G.K., "A quick method for estimating the equilibrium size and composition of
aqueous sulfuric acid droplets," Journal of Aerosol Science, Vol. 10, pp. 75-86,
1979.
58. Perry, R.H., D.W. Green and J.O. Maloney (eds.), Perry's Chemical Engineers'
Handbook, McGraw-Hill, New York, NY, 1984.
59. Morgan, J.L.R. and C.E. Davis, "The properties of mixed liquids. I. Sulfuric acid-
water mixtures," Journal of the American Chemical Society, Vol. 38, pp. 555-568,
1916.
60. Myhre, C.E.L., C.J. Nielsen and O.W. Saastad, "Density and surface tension of
aqueous H2SO4 at low temperature," Journal of Chemical Engineering Data,
Vol. 43, pp. 617-622, 1998.
61. Hilsenrath, J., C.W. Beckett, W.S. Benedict, L. Fano, H.J. Hoge, J.F. Masi, R.L.
Nuttall, Y.S. Touloukian and H.W. Woolley, Tables of Thermal Properties of
Gases Comprising Tables of Thermodynamic and Transport Properties of Air,
Argon, Carbon Dioxide, Carbon Monoxide, Hydrogen, Nitrogen, Oxygen, and
Steam, U.S. Department of Commerce, National Bureau of Standards, NBS
Circular 564, Gaithersburg, MD, 1955.
62. Marrero, T.R. and E.A. Mason, "Gaseous diffusion coefficients," Journal of Physical
and Chemical Reference Data, Vol. 1, pp. 3-118, 1972.
87
-------
APPENDIX A. CORRELATIONS FOR SULFURIC ACID
DENSITY, SULFURIC ACID SURFACE ENERGY, AND
THE VAPOR PRESSURE OF WATER OVER SULFURIC
ACID AT 333.15 K (140 °F).
-------
APPENDIX B. COMPUTER CODE FOR THE
CALCULATION OF ACID MIST PROPERTIES IN THE
SCRUBBER AND STACK.
------- |