United States
                    Environmental Protection
                    Agency
 Air and Energy Engineering
 Research Laboratory
 Research Triangle Park NC 27711
                    Research and Development
 EPA/600/S7-87/018Sept  1987
&EPA          Project Summary
                    Evaluation  of  Utility  Boiler  Radiant
                    Furnace  Residence Time/
                    Temperature Characteristics: Field
                    Tests and  Heat  Transfer Modeling

                    B. M. Cetegen, W. Richter, J. L Reese, J. LaFond, B. A. Folsom, and R. Payne
                      This report describes an investigation
                    of the adequacy of a modeling approach
                    in predicting the thermal environment
                    and flow field of pulverized-coal-fired
                    utility boilers.
                      Two 420 MW. coal-fired boilers were
                    evaluated: a single-wall-fired unit and a
                    tangentially fired unit, representing the
                    two commonest  boilers in the  U.S.
                    Extensive field  measurements were
                    conducted on each  unit to determine
                    detailed temperature,  heat flux, gas
                    species composition, and flow field data
                    for a range of operating conditions.
                      Separate modeling approaches were
                    used to  predict boiler thermal  per-
                    formance and flow  characteristics. A
                    three-dimensional  zone method of
                    analysis was used to predict local and
                    overall heat transfer, temperature pro-
                    files, and fuel burnout. Such predictive
                    tools provide a sophisticated treatment
                    of radiative (Witt transfer, but are de-
                    coupled from the furnace flow field.
                    This input to the heat transfer code was
                    obtained from detailed measurements
                    in reduced scale  isothermal physical
                    flow models of the two boilers.
                      Comparisons between model predic-
                    tions and the detailed field measure-
                    ment data  have demonstrated  the
                    viability of this approach in predicting
                    furnace  performance, and in extrapo-
                    lating limited available data to alternate
                    operating conditions. Overall thermal
                    performance can, in general, be ac-
                    curately predicted; however, the analy-
                    sis has shown that the correct quantita-
                    tive prediction of local temperatures
and other  properties  requires an ac-
curate specification of inhomogeneities
in boiler input conditions, in the boiler
flow field, and in the local distribution
of wall ash deposits.
  This Protect Summary was developed
by EPA's Air and Energy Engineering
Research Laboratory, Research Triangle
Park, NC, to announce key findings of
the research project that Is fully docu-
mented In a separate report of the same
title (see Project Report ordering In-
formation at back).

Introduction
  The injection of dry calcium-based
sorbents into the furnace zone of coal-
fired utility boilers (referred to as the
LIMB process) offers the potential for a
cost effective in-situ SO2 control tech-
nology. However, the effectiveness of the
process is very strongly dependent on the
boiler thermal environment, particularly
the temperature/time relationship experi-
enced by the sorbent particles.  For
example, the ultimate  reactivity of the
sorbent material is very sensitive to the
heating rate and to the peak temperature
levels experienced by the particles. Hence,
optimum sorbent activation requires in-
jection at the appropriate thermal en-
vironment in  the boiler.  On the other
hand, optimum sorbent reactivity must
be accompanied by mixing and dispersion
of the active sorbent  particles in the
furnace gases. Sulfation progresses most
effectively in the temperature window
800-1200°C and requires residence times
on the  order of 1 sec. This provides

-------
further constraints on the effectiveness
of the process. Such considerations  in-
dicate that a detailed knowledge of boiler
thermal environment and the flow field is
necessary for a given boiler application,
both for predicting S02 removal potential,
and in  designing  an effective sorbent
injection system.
  In  order to generate the necessary
boiler information, a purely experimental
approach can be taken, where detailed
in-furnace measurements  of  gas  tem-
peratures and velocities can be made for
each individual coal-fired boiler of interest.
However, such  an approach  requires
many traverses to fully characterize the
complex three-dimensional temperature
and velocity fields typical of most utility
boilers.  Consequently  the  necessary
measurement programs can  be time-
consuming and expensive. An alternative
approach is to use a methodology based
on state-of-the-art analytical or modeling
techniques.  Such  a  methodology,  if
authenticated, could offer a faster, less
expensive, and  more flexible approach
than that provided by purely experimental
methods.
  This report summarizes  an  EPA-
sponsored research program, the primary
objective of which was to investigate the
adequacy of a modeling approach in pre-
dicting boiler thermal  environment and
flow field. The program has included:
  •  Reviewing available furnace thermal
     prediction procedures  and  flow
     models,  and selecting a  suitable
     method  to  evaluate temperatures
     and flow fields in boilers.
  •  Field tests at two coal-fired utility
     boilers  (wall-fired  and tangentially
     fired, representing the  two com-
     monest boilers in the U.S.) to docu-
     ment thermal environment and flow
     characteristics.
  •  Modeling test cases with the selected
     heat transfer/flow model.
  •  Comparing model  predictions with
     field data.
  •  Determining the adequacy and ac-
     curacy of such an approach for future
     predictive evaluations.


 Heat Transfer and Flow
 Modeling Approach
   In this study  separate  modeling  ap-
 proaches have been used to predict boiler
 thermal performance and  flow field. A
 three-dimensional general boiler thermal
 analysis code was used to simulate test
 conditions for which extensive tempera-
 ture, heat flux, and gas species composi-
 tion data were obtained at two boilers.
This heat transfer code is decoupled from
the furnace flow field, but includes  a
sophisticated treatment of the radiative
heat exchange process (mostly respon-
sible for heat exchange in utility boilers)
and  a fully  coupled  solution of heat
balances in a zonal description of  the
furnace enclosure. Since the furnace flow
field is an input to the heat transfer code,
it was obtained from isothermal physical
flow modeling  for the two boilers.  Al-
though there  are purely analytical  ap-
proaches to flow field  prediction, many
uncertainties, assumptions, and difficul-
ties are associated with these approaches,
such  that fully mathematical  solutions
are available only for simple geometries.
Consequently,  reduced  scale physical
modeling is now the preferred approach.
  The heat transfer model, used to predict
the thermal  behavior  of the  two  test
boilers in this program, was EER's three-
dimensional General  Boiler  Thermal
Analysis Code (GBTAC). This is just  one
of several available codes which are based
on  the zone  method  of analysis,  and
which are finding increasing application
in the prediction and design of practical
combustion  systems. Such heat transfer
codes allow predictions of  local  and
overall heat transfer, temperature profiles,
and fuel burnout in boiler combustion
chambers and industrial furnaces depen-
dent on actual furnace geometry, oper-
ating conditions, and fuel and wall deposit
characteristics. All major fuel types  can
be  considered. The EER code can  also
handle heat transfer in  radiant super-
heater sections simultaneously coupled
with  the lower radiant furnace  heat
transfer calculations.
  The furnace heat transfer  model  is
decoupled from momentum  balances.
Thus,  zonal balances of energy,  volatile
matter, char particles,  and O2 are based
on a prescribed mass flux distribution for
calculating convective transport over the
volume zone boundaries. Such decoupling
is typical of zone  models,  and allows
concentrating the computational effort
on an accurate simulation of radiative
heat transfer This has been found neces-
sary  for  reliable thermal  performance
predictions of large boiler furnaces.
   In this present study the distribution of
mass  fluxes through the zone arrange-
ment  of the  heat  transfer model  was
obtained through observations and mea-
surements  in isothermal physical  flow
models.

Test Boilers
   In this study, two coal-fired boilers (a
wall-fired boiler, and a tangentially  fired
boiler) were evaluated. Each boiler was
subjected to an intensive period of mea-
surement,  where detailed in-furnace
probing  was  conducted to determine
thermal conditions, flow field, and overall
performance data as a function of a range
of operating conditions.  Reduced scale
isothermal models were also constructed
of each boiler for the experimental deter-
mination of the flow field, and for the sub-
sequent specification of inputs for heat-
transfer model evaluation.
  The wall-fired boiler selected for this
program was Duck Creek Unit 1, owned
and  operated  by Central Illinois Light
Company (CILCO). This 420 MWB  boiler,
near Canton, IL, was built by Riley Stoker
Corporation. The general arrangement of
the boiler is shown in Figure 1. It is front-
wall-fired, with boiler  capacity  repre-
sentative of medium size wall-fired utility
boilers. The NOX emissions from this unit
were reduced below the limit of 0.7 Ib of
NOX/106 Btu*  by retrofitting the original
burners with Iow-N0x burners.  The fuel
fired at this  unit  is a  high  volatile
bituminous coal which contains approxi-
mately 4% sulfur, and  flue gas  desul-
furization is achieved by a wet scrubber.
  The second  boiler was the 420 MWe
tangentially fired Unit 5 of the Conesville
Generating Station near Conesville, OH,
owned and operated by American Electric
Power Corporation. Figure 2 illustrates
the  overall  arrangement of the  boiler.
This unit is considered representative of
medium size tangentially fired boilers.

Heat Transfer Analysis
  Detailed flow and heat transfer analyses
have been conducted for a wide range of
furnace operating conditions for both the
Duck Creek and Conesville boilers. In
general, the major test cases were chosen
to correspond with those selected for the
field measurement program in  order to
provide  a basis for direct comparison of
measured and predicted results. However,
a number of additional cases were also
established to evaluate the sensitivity of
model predictions to certain model inputs
and assumptions. Tables 1 and 2 list the
test  cases  for the  Duck Creek and
Conesville boilers, respectively. For each
of these cases the modeling procedures
were used to  make  predictions for the
three-dimensional distributions of tem-
perature, velocity, volatile combustion,
unburnt fixed  carbon, 02 concentration,
incident and net heat fluxes,  and wall
surface temperature. These results, com-
pared to measured values, were used to
                                                                                     lb/10»Btu = 043kfl/GJ

-------
          High Temperature
          Pendant Superheater
   Coal Bunkers
                                                             High Temperature
                                                             Reheater
                                                           Primary Superheater
                                                            Primary Reheater


                                                            Economizer
                                                                  Airheater
       Mills
Figure 1     General arrangement of Duck Creek Unit 1 boiler
test the accuracy and  sensitivity of the
model predictions
  Many previous studies for  coal-fired
boiler furnaces using EER's furnace heat
transfer  and  combustion  model  have
shown that the accuracy of the predictions
obtained from this model depends on a
prescription of mass flux distribution and
on  a variety of other model input data.
Since one goal  of the project  was to
establish the amount and type of input
data necessary to establish reliable time/
temperature predictions, the formulation
of additional input data received consider-
able attention in this study. Certain groups
of key input data are difficult to specify.
One such group relates  to the distribution
of ash  deposits  on the furnace heat
transfer surface, and the  specification of
their thermal resistance and emissivity.
A second  important subset of inputs
relates to the distribution of air and fuel
flows to  individual burners (and  hence
local stoichiometries), which is particu-
larly important when overfire or underfire
air is employed for NOX  emission control.
  An example of the importance of cor-
rect specification of the  distribution of
input flows can be found in  the compari-
son between analyses for Cases 1 and 3
for the Duck Creek boiler. Case 1  repre-
sents predictions based on minimal in-
formation of actual boiler operation, while
Case 3 represents a more realistic situa-
tion  where input flows  and stoichio-
metries have been modified based on a
detailed analysis  of field data. Predicted
temperature distributions for these cases
are presented in  Figure 3. The dramatic
effect on the temperature field is clearly
evident, particularly the strong tempera-
ture  stratification caused by the underfire
air jets originating in the lower furnace.
Predicted Case 3 profiles of  local  gas
temperatures and actual field measure-
ments are compared in Figure 4.
  Considering the difficulties inherent to
temperature measurements in full scale
pulvenzed-fuel-fired boiler futnaces  and
the many unavoidable model assumptions
and simplifications, the overall agreement
between measured and predicted profiles
of the gas  temperatures is satisfactory.
Maximum local differences between  cal-
culated  and predicted values occur at
elevations in the upper furnace and very
close to the side walls,  where  steep
temperature gradients are encountered.
Most of the local differences between
predicted and experimental values origi-
 nate from asymmetrical effects of actual
 furnace operation which are not covered
 by the model since symmetry between
 west and east furnace half was presumed.
 The asymmetrical effects  caused by the
 non-uniform secondary air damper set-
 tings (which were changed almost daily)
 are  especially noticeable in  measured
 near-rear-wall temperature  profiles  at
 burner level elevations.
   Predicted and measured CVconcentra-
 tion profiles are in general directly inverse
 to  corresponding  temperature profiles;
 i.e., a low oxygen concentration predicted
 or measured locally is associated with a
 local high temperature value  and vice
 versa. This leads to the conclusion that
 the temperature profiles in the Duck Creek
 furnace are not governed by furnace heat
 transfer  alone  but to  a considerable
 degree by mixing of underfire air with the
 flame zone products, and also by inter-
 burner flow  mixing due to the inhomo-
 geneous firing pattern. Considering that
 the  furnace model used for the current
 anslysis  is primarily a (radiative)  heat
 transfer model, the agreement between
 measured and predicted (^-concentration
 profiles is again reasonable. It is believed
 that a better agreement can only be
 achieved by use of a finer computational
 grid and by more exact modeling of main
 flow development and turbulent exchange
 using expensive fluid dynamics models.
  Similar comparisons between predicted
 and measured results were also obtained
 for the Conesville boiler, where a strong
 impact of the assumed local distribution
 of stoichiometries on flame  zone tem-
 peratures and 02 profiles was observed.
 Figure 5 compares the profiles predicted
 without  and  with the assumption  of
 mhomogeneous air distribution to the
 corners of each burner level. In the case
 of non-uniform  corner  stoichiometries,
 the air flow from the rear wall corners
 was reduced resulting in an increase  of •
 near-rear-wall gas temperatures of up  to
 100 K and more. The effect is particularly
 strong for low load at elevation 18.1 m.
 Compared  to the measured  data, the
 change to  non-homogeneous corner
 stoichiometry improved the predictions  at
 some locations, and at other locations
 the  discrepancies  increased.  However,
 the corner air distribution assumed  in
 Cases 3 and 5 was somewhat arbitrary.
 Due to such operational  uncertainties,
 even the  most sophisticated  computer
 model will probably not yield better agree-
 ment for local profiles than that obtained
 in the current study.
  For the Conesville  boiler  study,  dis-
crepancies of the measured and predicted

-------
         High
         Temperature
         Pendant
         Superheater
                                                                                                 High
                                                                                                 Temperature
                                                                                                 Reheaters
                                                                          '   U  *   JJ'Cr   i L.
                                                                       C'^'t'^t    •&'^.\    'I
                                                                            1  ll.u,U-*t'./IUillU,il llllll, J I
                                                                                                                Air Heater
Figure 2.
               Mills
General arrangement of Conesville Unit 5 boiler

                           4

-------
Table 1.
Case
No.
1
2
3
4
5
6
7
8
9
Table 2.
Case
No.
1
2
3
4
5
6
7
Case Description of Duck Creek Heat Transfer Analysis
Test Mills in
Load Condition Operation
% - -
100 N/A A.B.C
100 1.5 A.B.C
100 1.5 A.B.C
100 1.5 A.B.C.
100 N/A A.B.C
60 Preliminary A.B
60 4 A.C
60 4 A.C
60 4 A.C
Excess
Air
%
18.4
18.4
18.4
18.4
18.4
22,8
20.1
20.1
20.1
UFA
%
25
25
25
25
25
0
0
0
25
Burner Level
Stoichiometry
—
Uniform
Non-Uniform
Non-Uniform
Non-Uniform
Non-Uniform
Uniform
Non-Uniform
Non-Uniform,
Increased
Flame Length
Non-Uniform
Increased
Flame Length
Flow Field
Prescription
—
Initial Corrected
for UFA
Corrected
from Case 1
(Flow Field 1)
Corrected
from Case 2
(Flow Field II)
Corrected from
Case 3
Corrected from
Case 2
Corrected
from Initial Flow.
Ash Hopper Flee ire.
Corrected
from Initial Flow,
Ash Hopper Ftecirc.
Corrected from
Case 7
Corrected from
CaseB
Superheater
Deposits
—
Medium
Medium
Medium
Heavy
Medium
Light
Medium
Heavy
Heavy
Water Cooled
Wing Walls
—
No
No
No
No
Yes
No
No
No
No
Case Description of Conesville Heat Transfer Analysis
Load Test Burner Level
% Condition in Operation
100 N/A B.C.D.E
100 1 B.C.D.E
100 1 B.C.D.E
45 5 C.D
45 5 C.D
45 5 C.D
100 3 B.C.D.E
Excess
Air. %
18.8
21.2
21.2
47.3
47.3
47.3
25.0
OFA
8.3
18.9
18.9
59.4
44.2
59.4
18.9
Burner Level
Stoichiometry
Uniform
Non-Uniform
Non-Uniform
Non-Uniform
Non-Uniform
Non-Uniform
Non-Uniform
Corner
Stoichiometry
Uniform
Uniform
Non-Uniform
Uniform
Non-Uniform
Uniform
Non-Uniform
Flow Field
Prescription
Initial
Upgraded;
Corrected for
Burner Level
Stoichiometry
Corrected
from Case 2
for Corner
Burner
Stoichiometry
Corrected
from Case 2
Corrected
from Case 2
Set-up
from Scratch
Corrected
from Case 3
Furnace Wall
Deposit Pattern
Uniform,
Non-Uniform,
Pattern 1
Non-Uniform,
Pattern II
Non-Uniform,
Pattern 1
Non-Uniform.
Pattern II
Non-Uniform.
Pattern 1
Non-Uniform,
Pattern II

-------
   1550
  = 0.6m
                                               1350
                                                <1350 K'
                                       X = 4.8m
                                                          X=72m
      (a) Case 1: Gas temperatures (K) 100% load, uniform burner level
         load and stoichiometry, 25% UFA.
                                                       1500

                                                       '/7450
   *-
    7400
        7350
'. <7350 /
        '. <7350
     0.6m' -•
       (b) Case 3: Gas temperatures (Kj 100% load, non-uniform burner
          level load and stoichiometry, flow field II, 25% UFA.
Figure 3.   Comparison of predicted temperature distributions for Cases 1 and 3 for the Duck
           Creek boiler.
temperature profiles in the upper furnace
are believed to be mainly due to differ-
ences  in the  velocity profiles. The  in-
fluence of flow field is demonstrated in
Figure 6. This  figure compares tempera-
ture profiles predicted for 45% load at the
radiant superheater inlet plane using two
different flow field  prescriptions: the
standard presumed one (Case 4) and an
upgraded one  based  on field measure-
ments (Case 5). All other model  inputs
were  kept constant. The upgraded flow
field clearly yields a  closer agreement
with measured temperature profiles, and
illustrates the strong influence of this
parameter.
  In addition to the flow field, the thick-
ness and conductivity (thermal resistance)
of ash deposits on boiler walls and other
heat transfer  surfaces,  as input to the
model, can have a large impact on heat
transfer predictions.  Usually these are
specified based on observation of the
distribution of deposit thickness  and
surface texture  in the  boiler, with ap-
propriate  model  parameters  assigned
based  on experience  and standard nu-
merical values. Sensitivity studies based
on  such assumptions have been con-
ducted  and show clearly that, if local
properties (e.g., temperature and heat
flux) are to be correctly predicted, local
variations in deposit thermal resistance
must also be considered. Figure 7 shows
that an increase of the conductivity to
thickness ratio of the ash deposits at the
superheater surface zone layers k = 14
and 15 from  0.160 to 0.400  kW/m2K
resulted in  considerable improvement of
the incident heat flux distribution pre-
dicted in the upper furnace for both load
conditions.
  In addition  to the prediction of local
properties, averaged properties at various
boiler elevations  are  also of interest in
the sorbent injection problem. Of particu-
lar interest is the temperature/time profile
through the sulfation temperature win-
dow, since this is a main parameter that
impacts SOj removal. As shown in Figure
8, the optimum model predictions repre-
sent quite well the measured averaged
temperature distributions  through the
boilers. Such  agreement  indicates also
that overall furnace  performance (e.g.,
mean  exit temperature,  furnace heat
absorption) can be predicted accurately if
model inputs are correctly specified. Cor-
responding mean time/temperature pro-
files predicted for Duck Creek  full  load
Cases 1, 2, and 3 are presented  in Figure
9, and show the impact of the various
flow and stoichiometry distribution  as-
sumptions. In spite of the different model

-------
K
1600
1500
1400
1300
1200
1100
<
o ^
s~ '^\
•
Y=09m
f ...,',,.,
  K
1600

1500

1400

1300

1200

1100
                                                              y = 2.7/n
                   8  10  12  14 16 X.m
                                         K
                                       1600

                                       1500

                                       1400

                                       1300

                                       1200

                                       1100
                                                                                                       Y=35m
0
                  8   10  12  14  16X,m
                                                               10  12  14  16
                                                                           X.m
K
1600 • .'
1500 1
1400 3 /
1300 i
1200
1100
>
0 2
Figure 4.

i 4if • • • • >
r 1 "' Q ^^
j 40m \ •
1 V
1 *
1
\ Y= 46m
\
K
1600
1500
1400
1300
1200
1100
1 	 i 	 . 	 . 	 r- 	 • 	 , 	 , 	 r-i
0 1
1
'S i \
' ' I
1
Y=6.1m
> |


Profiles at 2= 37 9m




4 6 8 10 l'2 14 /'fi X.m 024 6 3 10 12 14 76 X.m
Comparison of measured and predicted gas temperature profiles at 100% load. ( 0 meas Phase 1. O meas. Phase II,
•— • pred. Case 3)
input conditions utilized in the three cases
and in spite of considerably different pro-
files predicted  for furnace hopper and
flame zones, lower furnace exit tempera-
tures (calculated  at platen superheater
inlet plane) differ only by 35 K. The most
realistic model  predictions are probably
achieved  in Case 3. In this  case,  an
average quench rate of 394 K/s is cal-
culated in the upper furnace The other
quench rates predicted are 331 K/s  for
Case 2, and 341 K/s for Case 1.

Conclusion
  Detailed heat transfer analyses for a
wall-fired boiler and a tangentially fired
boiler have been conducted for a range of
furnace  operating  conditions using a
state-of-the-art heat transfer model. Pre-
dictions have been  compared with cor-
responding measurement data obtained
in extensive field trials  on the two test
boilers. In general, the study has shown
that good agreement between predicted
and measured  overall thermal perform-
ance parameters (furnace  exit tempera-
ture, peak radiative  heat flux,  carbon
burnout, etc.) can be achieved if the fuel
and air flows into the furnace, and the
insulating effect  of wall deposits are
properly described.
  Quantitatively correct  predictions of
profiles of local temperatures and other
furnace variables have, however, proven
to be  more difficult to  achieve.  The
analyses have shown that a reasonable
agreement with measured data  can be
obtained only if the inhomogeneities in
the input conditions are  correctly  pre-
scribed, and the flow field incorporated
into the  heat  transfer  model is  a  good
representation  of the flow field in the
actual boiler. This appears to be particu-
larly true when the boiler furnace is
operated under staged combustion con-
ditions,  and strong inhomogeneities in
the distribution of stoichiometry result.
The prediction of local properties is also
affected by the ability to correctly describe
the local distribution and properties of
wall ash deposits,  although this is less
critical than  the flow field specification.
  Due to the sensitivity of absolute model
predictions  to  accurate  prescription of
operating conditions and  model parame-
ters, some model verifications are always
required  especially when the model is
used to support actual design of sorbent
                                          injection systems for existing boilers. The
                                          three methods of model verification are:
                                            • Comparison with design performance
                                              based on manufacturer performance
                                              calculations.
                                            • Comparison with actual overall per-
                                              formance data,  which  requires a
                                              short field test.
                                            • Detailed performance comparison
                                              including comparison with extensive
                                              temperature, concentration profile,
                                              heat  flux,  and  velocity measure-
                                              ments.
                                          The comparison of model predictions with
                                          design performance is only recommended
                                          if no other boiler performance data  are
                                          available. The most desirable and most
                                          efficient approach to model verification is
                                          by using a combination of control room
                                          data and results of an  abbreviated field
                                          test. Detailed in-furnace measurements
                                          are only necessary in exceptional cases,
                                          such as boilers with operational problems.
                                          The high expense of the detailed mea-
                                          surements must be weighed against their
                                          value to model verification.
                                            Some of the weaknesses of the current
                                          model can be eliminated,  and  the  ac-
                                          curacy of absolute performance predic-

-------
                                4/0       x=5?m
                                                                                                             O O
                                                                                       '™fj>
                                                                                       1600^''
                                                                                                    J*''
                                                                                                   <$
                                                                                                                  O
                                                                                                                   \
                                                                                          X= 14.5m
                  8  10  12 Y.m  0  2   4   6  8  10  12 Y.m  0  2   4   6
                                   Gas Tedmperature (K) Profiles at 2 - 26.2m
 14

 12

 10

  8

  6

  4
   X= 1.2m
                                Vo/ % Dry
                                          b  S

                                          Ik
                                         //.
                               --5.1m
                                          K
                                            O
                                                 Vol % Dry
                                                                 8  10  12 Y.m  0   2   4  6   8  10   12  Y.m
                                                                                         Vo/ % Dry
                                                                         X= 106m
                                                                         i   .  n  O
                                                                                            X- 14.5m
          10  12 Y.m  02   4  6   8   10  12 Y.m  0  2   4   6

                    02 Concentration (Vol %. Dry) Profiles at Z - 26 2m
                                                                             8   10 12 Y.m 0   24
                                                                                                          8   10   12 X,m
Figure 5.
Impact of non-uniform corner stoichiometry on gas temperature and O2 profiles predicted for 100 and 45% load (100% load.
O  meas.,  O uncertain data pt., a •—• pred Case 2, uniform corner stoichiometry, a1 •—« pred. Case 3, non-uniform corner
stoichiometry; 45% load:  9  meas., b —• pred. Case 4. uniform corner stoichiometry, i1 	pred Case 5, mon-uniform corner
stoichiometry.
tions can still be increased. A particular
area where improvement can be obtained
for coal-fired boilers is more accurate
characterization of thermal properties of
fly ash  and ash  deposits, including
physical structure,  optical properties,
wave length, and temperature.  This is
also important for the LIMB process since
sorbent injection can affect these pro-
perties. Similarly, the furnace flow pre-
scription could  be replaced by advanced
fluid dynamics calculations which include
prediction of turbulent exchange and the
effects of buoyancy (totally neglected in
the current study). An  increase in ac-
curacy of local temperature profile pre-
dictions of roughly  20-40 K is expected
from such a modification. A fluid dynamics
model  does not  make physical  flow
modeling unnecessary since  physical
models are very  useful  for  validating
numerical flow models under isothermal
conditions.
                                In spite of the limitations of the current
                              model, with respect to absolute accurate
                              performance  predictions, the  present
                              model applications carried  out for two
                              boiler furnaces have undoubtedly shown
                              its usefulness for analysis  of furnace
                              performance and its potential for opti-
                              mization of the LIMB process. The major
                              advantage of the model employed is also
                              its ability to extrapolate limited measure-
                              ment and performance data from baseline
                              conditions to a variety of other operating
                              conditions.

-------
  K

1500


1400


1300
                 X= 1 3m
A
1500
1400


1300
<
1
:~^A^- 1
. *<,''* ^^^C.""*
^ ^*^J
..* 1
-1
X=3.1m \
r' • > > • i
                                                                                                 Profiles at Z = 40 6m
                         8  Y.m
                                                                         8   Y.m
A
7500

7400

7300
*
• • ' •
•
^*^pq---" '.t~"
' I<^ ***
r*

X= 73/T7
f .
02468
Figure 6. Impact of flow dts
1
^
-H
.%-{
"j
-i
•
j
Y.m
tnbut
ff

1500


1400

1300
4
1
-1
1
_. 1

**s '
•% •*
X=727m , 1
^ 1
r f t • • •
                                                    0
                                                                         8
                                                                              Y.m
K
7500
7400
7300
*

\
^f^-r-^"^
^ X= 74.5m **%
I * • • f
                                                                                                    0
                                                                                                               4
                                                                                                                          8   Y.m
            Impact of flow distribution in upper furnace (elevation Z - 40 6m) on temperature proifles predicted for 45% load
            f 9  meas . •—• pred Case 4. original flow field
                        •---• pred Case 6, modified flow field)
Incident Radiative Heat Flux. kW/m2
-« NO Co -k Wi °> ^J
§ 8 § § § § §
1 I I 1 1 1 1
Predictions Conductance/ Thickness
kW/m*/K
7«M/aj», J00% Load 45% Load
K a a' b 6'
70-72 0774 0.057 0.774 0.057
73 0774 0.229 0114 0229
14-16 0.400 0400 0.114 0.400
14-15 SM 0160 0400 0160 0.400 -
Measurements O 9
»^ ^ ,^
i t i i l i l
                                          6       8        10

                                       Furnace Width /X), m
                                                                   72
                                                                            74
Figure 7.
             Impact of deposits in upper furnace on radiative heat fluxes incident on front
             wall at elevation Z = 40 6m

-------
;7oo  -
                                           Q Meas Arithmetic Average

                                           O Meas Arithmetic Average of Few Data

                                           a Pred Arithmetic Average, Case 3

                                           b Pred. Mass-Weighted Average, Case 3
                  10
                                                      \  \
                                  20            30

                                  Vertical Distance (Z), m
                                                           40
                                                                         50
        (a) Duck Creek. 100% Load and Flow Field II
     2000
        1800

        1700

        1600

        1500

        1400
      1200
                         Burner Levels
                          UJU
                       BCDE OF A
                                             Superheater
           . Ash
           Hopper   I
                    nl
                  f/
?00% Load. Condition 1

  a Pred  3-D Model

 a1 Pred  2-D Model

 O  Meas 2nd Test
             \
45% Load, Condition 2
  b Pred 3-D Model

    Pred 2-D Model

    Meas

    Uncertain Data Point
                 •  I
                  10
                           20       30       40

                              Vertical Distance /Z/. m
                                                        50
                                                              60
          (b) Conesville
Figure 8.
         Mean temperature profiles along the furnace predicted for the Duck Creek and
         Conesville boilers and comparison with measurements
                                 10

-------
                         a. Pred. Case 1, Uniform Burner Level Stoichiometry

                         b Pred Case 2, Non-uniform Level Stoichiometry Flow Field I

                         c. Pred. Case 3, Non-uniform Level Stoichiometry Flow Field II
                                                     S'H lnlet
                                                                      Sulfation
                                                                      Window
  §  7/00
           20   15
                       1 0
                                                           2.0
Figure 9.
                 05    0     0.5   10    1.5
                    Mean Residence Time IT), S
Dependence of mean time/temperature prof lies predicted for 100% load on burner
level Stoichiometry and flow field assumption (Duck Creek).
  B. Cetegen.  W. Richter, J. Reese, J. LaFond, B. Folsom,  and R. Payne are
    with Energy and Environmental Research  Corporation, Irvine, CA  92718-
    4190.
  David G. Lachapelle is the EPA Project Officer fsee below).
  The  complete  report  entitled "Evaluation  of Utility Boiler Radiant Furnace
    Residence Time/Temperature Characteristics: Field Tests and Heat Transfer
    Modeling," (Order No.  PB 87-213 112/AS;  Cost: $36.95,  subject to change}
    will be available only from:
           National Technical Information Service
           5285 Port Royal Road
           Springfield,  VA22161
           Telephone: 703-487-4650
  The EPA Project Officer can be contacted at:
           Air and Energy Engineering Research Laboratory
           U S Environmental Protection Agency
           Research Triangle Park, NC 27711
                                                                                  11

-------
                                                 <_>



                                                 •sj
o

00
                                                      01
                                                       c
                                                       v>
                                                       (B
                                                       §
OfVXJC   O
           o
r>  M     a
**s/»om   u^

-------