DRAFT - Do not Cite or Quote
Consequences Analysis of Using ISC-PRIME Over
the Industrial Source Complex Short Term Model
Staff Report
U.S. Environmental Protection Agency
Office of Air Quality Planning and Standards
Emissions, Monitoring, and Analysis Division
Research Triangle Park, NC 27709
April 1998
-------
DISCLAIMER
This draft report has not been reviewed by The Office of Air Quality Planning and Standards, U.S.
Environmental Protection Agency, and has not been approved for publication. Mention of trade
names or commercial products is not intended to constitute endorsement or recommendation for
use.
11
-------
PREFACE
Before a new algorithm or model is released for use by the air quality modeling
community, there is a need to understand how that algorithm or model will affect concentration
results with respect to measured observations. The incorporation of the Plume Rise Model
Enhancements (PRIME) model into the Industrial Source Complex Short Term (ISCST3) model
to form ISC-PRIME is a major alteration of the way in which building downwash is calculated.
An independent evaluation has been performed in which the concentration output from ISCST3
and ISC-PRIME have been compared with measured observations from three field studies and
one wind tunnel study. Herein is a review of that evaluation, review of a subsequent independent
consequences analysis, and further analysis of data results.
111
-------
Contents
PREFACE iii
FIGURES v
TABLES vii
1. INTRODUCTION 1
2. TECHNICAL DESCRIPTION 1
3. ANALYSIS METHOD 2
4. ANALYSIS RESULTS 4
5. CONCLUSIONS 8
6. REFERENCES 9
APPENDIX A. CONSEQUENCES ANALYSIS SOURCE PARAMETERS A-l
APPENDIX B. TABLE OF CONSEQUENCES ANALYSIS SCENARIOS B-l
APPENDIX C. VARIOUS FIGURES C-l
APPENDIX D. TABLES OF BOWLINE POINT CLOSE-IN CONCENTRATIONS D-l
iv
-------
FIGURES
Figure Page
1. Streamlines Around a Building C-l
2. Depiction of Cavity (Near Wake) and Far Wake Areas C-l
3. Bowline Monitor 1: Quantile-Quantile Plot of All Cases C-2
4. Bowline Monitor 3: Quantile-Quantile Plot of All Cases C-3
5a. Depiction of the AGA Kansas site building and stack layout C-4
5b. Depiction of the AGA Texas site building and stack layout C-5
6. AGA: Residual Plot of ISCST3 and ISC-PRIME Cp/Co Results versus Distance C-6
7. AGA: Residual Plot of ISCST3 and ISC-PRIME Cp/Co Results versus Stability Class . . C-l
8. AGA: Residual Plot of ISCST3 and ISC-PRIME Cp/Co Results versus Wind Speed .... C-8
9. Depiction of the EOCR site building tiers and stack layout C-9
10. EOCR: Residual Plot of ISCST3 and ISC-PRIME Cp/Co Results versus Distance .... C-10
11. EOCR: Residual Plot of ISCST3 and ISC-PRIME Cp/Co Results versus Wind Speed . C-l 1
12. EOCR: Residual Plot of ISCST3 and ISC-PRIME Cp/Co Results versus Stability Class C-12
13. Depiction of the Lee Power Plant site building tiers and stack layout for the wind tunnel
study C-l3
14. Wind Tunnel: Residual Plot of ISCST3 and ISC-PRIME Cp/Co Results versus
Distance (Urban Dispersion) C-14
15. Wind Tunnel: Residual Plot of ISCST3 and ISC-PRIME Cp/Co Results versus
Wind Speed (Urban Dispersion) C-l5
16. Wind Tunnel: Residual Plot of ISCST3 and ISC-PRIME Cp/Co Results versus
Distance (Rural Dispersion) C-l6
v
-------
17. Maximum and Highest of the Second Highest 1-Hour Average Concentrations by
Downwind Distance for a Squat building (34m) with a 35-meter Stack on Corner,
Rural Conditions C-17
18. Maximum and Highest of the Second Highest 1-Hour Average Concentrations by
Downwind Distance for a Tall building (34m) with a 35-meter Stack on Corner,
Rural Conditions C-18
19. Maximum and Highest of the Second Highest 1-Hour Average Concentrations by
Downwind Distance for a Squat building (34m) with a 35-meter Stack 4*Hb
Northeast (NE) of the NE building Corner, Rural Conditions C-19
20. Maximum and Highest of the Second Highest 1-Hour Average Concentrations by
Downwind Distance for a Super Squat building (34m) with a 35-meter Stack on Corner,
Urban Conditions C-20
21. Maximum and Highest of the Second Highest 1-Hour Average Concentrations by
Downwind Distance for a Squat building (34m) with a 35-meter Stack on Corner,
Urban Conditions C-21
22. Maximum and Highest of the Second Highest 1-Hour Average Concentrations by
Downwind Distance for a Tall building (34m) with a 35-meter Stack 4*Hb
Northeast (NE) of the NE building Corner, Urban Conditions C-22
23. Maximum and Highest of the Second Highest 1-Hour Average Concentrations by
Downwind Distance for a Squat building (50m) with a 100-meter Stack on Corner,
Rural Conditions C-23
24. Maximum and Highest of the Second Highest 1-Hour Average Concentrations by
Downwind Distance for a Tall building (50m) with a 100-meter Stack NE of Corner,
Rural Conditions C-24
25. Maximum and Highest of the Second Highest 1-Hour Average Concentrations by
Downwind Distance for a Squat building (50m) with a 100-meter Stack on Corner,
Urban Conditions C-25
26. Maximum and Highest of the Second Highest 1-Hour Average Concentrations by
Downwind Distance for a Tall building (50m) with a 100-meter Stack on Corner,
Urban Conditions C-26
vi
-------
Tables
Table Page
1. Bowline Point Parking Lot Monitoring vs. Calculated Hourly Concentrations D-l
2. Bowline Point Met Tower Monitoring vs. Calculated Hourly Concentrations D-2
vii
-------
1. INTRODUCTION
Deficiencies in the current Building Downwash algorithms are well known. These
deficiencies include: 1) reported over predictions under light wind, stable conditions, 2)
discontinuities in the vertical, alongwind, and crosswind directions, and 3) the assumption that the
source is always collocated with the structure causing down washing.
To address these deficiencies, the Electric Power Research Institute (EPRI) funded
evaluations of a series of field and wind tunnel experiments from which a new downwash
algorithm was derived and called PRIME, Plume Rise Model Enhancements. The PRIME
algorithm was developed, codified, and incorporated into the Industrial Source Complex Short
Term model (ISCST3) by Earth Tech, Inc. The PRIME modified modeled was named ISC-
PRIME. Independent evaluations of ISC-PRIME were conducted and draft documents were
prepared by ENSR, under contract to EPRI. EPRI presented the U.S. EPA with the ISC-PRIME
software, together with the results of independent evaluation and consequences analysis.
EPA has conducted its own Consequences Analysis (CA) of the ISC-PRIME software and
a review of the EPRI independent reports. The analysis consists of verifying the ENSR CA
results, verifying that ISCST3 and ISC-PRIME produce the same results with no building in the
input, an analysis of BPIP as modified for use in preparing input for ISC-PRIME, and an analysis
of the consequences of using ISC-PRIME instead of ISCST3 for building downwash situations.
2. TECHNICAL DESCRIPTION
The current ISCST3 model contains two downwash algorithms, the Huber-Snyder and
Schulman-Scire algorithms. The Huber-Snyder algorithm applies to stacks that are lower than
building height (BH) plus 1.5 times the lesser of building height or projected building width (L).
The Schulman-Scire algorithm supersedes the Huber-Snyder algorithm when stack heights are
less than or equal to BH + 0.5L. Activation of either algorithm is dependent upon the wake
effect height at 2 BH's downwind to be less than or equal to BH + 2L or BH + 1.5L, respectively.
These algorithms have limitations: 1) the stack is assumed to be located in the center of the lee
wall of the structure causing down washing even though it may be upwind, far downwind, or
laterally displaced from the structure, 2) there are discontinuities between the wake and non wake
areas, 3) streamline flow over a structure is not taken into account, 4) plume rise is not adjusted
due to the velocity deficit in the wake or due to vertical wind speed shear, 5) concentrations in the
cavity region are not linked to material capture, and 6) under light wind speed, stable conditions,
concentration estimates are reported to be higher than observed.
To address these issues, the PRIME model was created and incorporates two fundamental
features: enhanced plume dispersion coefficients due to the wake turbulence, and reduced plume
rise caused by descending streamlines and increased entrainment in the wake of a structure. The
PRIME model was integrated into ISCST3 (dated 96113) and named ISC-PRIME (dated 97224).
1
-------
The foundation of the PRIME model is its ability to model the downwind cavity (near
wake) and far wake areas on a three-dimensional scale. The dimensions of the downwashing
structure are used to form an ellipsoidal shape that may consist of a rooftop and downwind cavity,
or a single recirculation cavity (Figure 1). The recirculation cavity is defined as the region
bounded above the roof by the separation streamline formed at the upwind roof edge and bounded
downwind by the reattachment streamline. The lateral sides of the structure form the side of the
ellipsoidal cavity (Figure 2).
Upon the cavity "foundation," streamlines are formed based on the location and maximum
height of the rooftop recirculation cavity, the length of the downwind recirculation cavity and the
building length scale. Upwind streamlines ascent to the point of the maximum rooftop
recirculation cavity height and descend thereafter. There is a gradual decrease in the rate of
streamline descent in the far wake region.
The plume rise component of PRIME is computed using a numerical solution of the mass,
energy, and momentum conservation laws. The model permits arbitrary ambient temperature
stratification, unidirectional wind sheer and an initial plume size. Streamline ascent and decent
effects on plume rise are considered as well as the enhanced dilution due to building induced
turbulence. Dispersion is based on Weil (1996).
3. ANALYSIS METHOD
This CA is divided into several parts. These parts consist of: 1) confirming that with no
building in the input files, both ISCST3 and ISC-PRIME produce the same results, 2) reviewing
the ENSR Independent Evaluation and CA, 3) confirming the modeling results of the ENSR
independent CA, and 4) performing an additional assessment of ISC-PRIME.
In the first part, the input files consisted of a single source (See Appendix A) with four
groups of receptors. The source is a 35-meter height point source with a buoyant plume. There
are four 21 by 21 receptor grid arrays with spacings of 50, 200, 500, and 1000 meters covering an
area from 1 km up to 10 km square and centered on the source. One year of 1964 Pittsburgh,
PA meteorological data was used in all runs. The programs printed the maximum and highest of
the second highest 1-, 3-, 24-hour averaged concentrations for all receptors and the annual
average for all receptors.
In the second part, the ENSR independent evaluation and the CA reports were reviewed.
The ENSR independent evaluation consisted of four data sets:
1. A conventional 1-year monitoring network near the Bowline Point Station, New York
(source type: electric utility; two 600MW units, each with an 86.9-m stack; monitor
coverage consisted of four close-in sites at distances from 251 to 848 meters.)
2
-------
2. A tracer experiment conducted by the American Gas Association (AGA) in Texas and
Kansas (source type: gas compressor station stacks; 63 hours available; tracer sampler
coverage from 50 to 200 meters.)
3. A tracer experiment at the EOCR Test Reactor Building in Idaho (source type:
nonbuoyant releases at 30m, 25m and ground level; 22 release hours.)
4. A wind tunnel study of the Lee Power Plant (source type: 64.8m steam boiler stacks;
each stack 64.8 meters high; numerous cases studied: in neutral conditions, 78
combinations of wind direction, wind speed, and plume buoyancy; in stable conditions, 14
combinations of wind direction and plume buoyancy; tracer sampler coverage at six
distances ranging from 150-900 meters)
In the third part, the ENSR CA was rerun to confirm their results. Five buildings were
used as input. The buildings consisted of a Squat 34m and 50m high building of 60m x 120m
length and width, a Tall 34m and 50 m high building of 30m square base, and a Super Squat 34m
high building of 180m square base (Appendix B). There were two sources 35m and 100m high.
These sources were placed on the northeast corner of each building and then again at a distance of
four building heights to the northeast of each building. Runs were also conducted using no
building at all. 1964 Pittsburgh, PA meteorological data was used for each run. The CA receptor
grids are the same ones used in Part 1 of this analysis.
The fourth part consists of scenarios and runs made to ascertain the characteristics and
consequences of using ISC-PRIME with respect to ISCST3 using the ENSR CA input data sets.
ISCST3 consists of two downwash algorithms, Huber-Snyder, and Schulman-Scire. The ratio of
the height of the stack with respect to the lesser of the building height or projected width will
cause one or the other algorithm to be used in a downwash situation provided that the plume rise
is less than the height of the building plus 2L or 1.5L within two building heights of the source,
respectively. The ENSR CA input buildings, stack heights, and buoyancy provided an adequate
test of these two algorithms. The Cartesian receptor grid networks were replaced with a 20-ring
polar grid network. The binary hourly output files from the EPRI CA runs were reanalyzed for
maximum and highest of the second concentrations, on a ring by ring basis, between ISCST3 and
ISC-PRIME.
Because of the coding of streamlines into PRIME, there are questions about how multi-
tiered buildings should be modeled. Effluents emitted into a streamline can be captured, partially
captured, or not captured by a tier's recirculation cavity. This can have a dramatic effect on
calculated concentrations. If a stack's effluent is captured in the cavity of one tier and not in the
cavity of the highest tier, selection of the proper tier is paramount. BPIP selects the tier with the
highest wake effect height which is not always the correct tier for inclusion into ISC-PRIME.
3
-------
Runs were set up to examine the consequences of using BPIP selected tiers for inclusion
in ISC-PRIME runs. A tall tier was placed on top of a squat building and a stack was initially set
at the far end. Additional runs involved setting the stack to different heights and moving the
stack away from the squat building end. The runs also involve moving the tall tier toward the
stack end of the squat building. The purpose was to see how ISC-PRIME would react to tier
selection and subsequent concentration calculations. The stack emissions rates were set for an
almost neutrally buoyant plume (slight rise). A SCREEN set of meteorological conditions was
used and the wind directions were set to upwind and then downwind directions with a line of
receptors placed along the longitudinal centerline of the squat building out to 5km.
The stack exit velocity and exit temperature were set to produce a near buoyant plume rise
with little momentum. This was done for two reasons. First, this would test the behavior of ISC-
PRIME for a small industrial source venting "fumes" to the outside air from the ambient air inside
the building. What would the concentrations be in the cavity and just outside the cavity?
Second, this type of low plume rise would allow the exploration of plume transport and
dispersion from along various streamlines. The low buoyancy and momentum would allow the
plume to reach "final plume rise" quickly and near the targeted streamline. It would also be used
to gather data to study the effects of tier selection by BPIPPRM.
4. ANALYSIS RESULTS
4.1 Model Runs Without Buildings
In the first part, ISC-PRIME (dated 97224) was run with the same input data set as
ISCST3 (dated 96113) with no building data in the input file. A file comparison showed the only
differences to be in the precision of some of the printed values. There were minor differences in
the 10,000th decimal place and smaller. There were no major differences in any of the runs. Both
models produced the same results.
4.2 Significant Findings of ENSR Independent Evaluation
In the second part, a review of the ENSR independent evaluation's four sets of results was
performed. These sets included the Bowline Point ambient monitoring, the AGA and EOGR
tracer experiment, and Lee Power Plant wind tunnel results. In the following, the most significant
findings of the ENSR evaluation are identified.
Bowline Point Results Analysis
The Bowline Point evaluation results included tables of top 50 monitoring data values
from the plant's Bowline Point, Boat Ramp, Met Tower and Parking Lot monitors. These four
4
-------
monitors are located at distances from 251 to 848 meters from the plant. Three monitors at
greater distances from the plant were not included in the evaluation.
The Bowline Point and Boat Ramp monitors recorded the highest and greater number of
concentration values than the other two close in monitors. Highest observed concentrations were
under stable conditions under moderate (5 to 10 m/s) to high (>11 m/s) wind speeds. Similar yet
conservative concentrations were calculated by both models. However, the highest
concentrations calculated by ISCST3 were produced under moderate wind speeds and more
stable conditions than those observed under ambient conditions where as the highest
concentrations calculated by ISC-PRIME were under high wind speeds and neutral conditions,
the same conditions that also produced the highest observed concentrations.
Quantile-quantile figures of modeled vs. observed concentrations, unpaired in time, show
both models to be conservative (Figures 3 and 4). ISCST3 tends to be more conservative at the
extreme high end of the curve where the results are slightly less conservative with calculated
concentrations becoming nonconservative approximately an order of magnitude below the
maximum concentration. Otherwise, the ISC-PRIME results are generally more conservative
than the ISCST3 values to the point where some of the values are greater than a factor of two.
Except as noted, all top 50 concentrations of both models were conservative and the ISC-PRIME
model was able to produce those concentrations under conditions matching the ambient
conditions producing the highest concentrations.
AGA Results Analysis
This study consisted of two different sites, one in Texas and the other in Kansas. Both
buildings, one in each state, were elongated squat and super squat buildings, respectively (Figures
5a and b). The Texas site had a stack that was offset from the building with a stack height that
was 85% of the building height of 11.37m. The Kansas site had two offset stacks, one 80% of the
building height and the other twice as tall as the building height of 12.19m. A figure of the
receptor layout arcs was provided for the Texas site but not the Kansas site. One interesting
aspect of the study is that the source appears to be upwind of the building. Also, the source is
offset toward a corner where the plume could be effected by the dynamics of wind flow around a
building corner. The arcs are at 50, 100, 150, and 200 meters from the source with the arcs being
at 5L or more from the building where L is the lesser of the building height or projected width.
The observed concentrations were adjusted using a technique that examines the
concentrations at all receptors along an arc and derives an estimate of the Sigma Y and
subsequent maximum concentration value using the program PLMFIT by Irwin (1996). The
resultant estimates were used as input to a series of scatter and residual plots. A residual plot of
the modeled versus observed concentrations versus the four downwind distances showed both
ISCST3 and ISC-PRIME to be conservative and ISCST3 to be more conservative than ISC-
PRIME (Figure 6). At a distance of 200 meters (approximately 17L) downwind of the buildings,
5
-------
it is indeterminate as to whether ISCST3 or ISC-PRIME does a better predictive analysis of
concentrations.
Using the ISCST3 results, two additional residual plots of wind speed and stability class
were generated. The one of the plots shows that there is more scatter in the results as the wind
speeds decrease. In the other plot, for the more stable and unstable stability classes, the scatter is
greater than for the more neutral and less stable of the stable stability classes (Figure 7). Part of
this may be due to the lower number of observations A(2) and E(9) versus B(12), C(35), and
D(34) (Figure 8). Still, there is a definite trend.
Two additional plots were also generated using ISC-PRIME results. The wind speed plot
shows more scatter in the predicted versus observed ratio as the wind speeds decrease. This is
understandable. The other plot showed that even more scatter for stability class E values than
ISCST3 and a more conservative plot for stability class A values (Figures 7 and 8). Except for
stability class E results, there was less scatter in the ISC-PRIME results. This may also help to
explain how stability class A concentrations showed up in the top ISC-PRIME concentrations for
the Bowline Point receptor.
EOCR Results Analysis
The EOCR used a two-tier building with the first tier roof at 7m and the second tier roof
at 25m above plant grade (Figure 9). Both tiers were square with a common corner on the west
side. The shorter tier was super squat and the second tier was tall. There were six rings of
receptors at 50, 100, 400, 800, 1200, and 1600 meters from the plant center. As in the AGA
study, results were adjusted using PLMFIT.
Residual plots of predicted over observed concentrations versus distance showed both
ISCST3 and ISC-PRIME to be conservative overall with almost negligible differences from the
100 meter ring outward (Figure 10). The scatter along the 50 meter ring was less for the ISC-
PRIME runs. There was a slight but notable rise in the bias toward more conservative results for
both models as the downwind distance increased.
The wind speed residual pattern showed that both models produced conservative results
and that ISCST3 was a little more conservative and had a little more data scatter than ISC-
PRIME (Figure 11). The same could be said for the stability class residual plot (Figure 12).
Lee Power Plant Wind Tunnel Analysis
A wind tunnel simulation of the Lee Power Plant was performed with six receptor scale
distances of 50, 100, 400, 800, 1200, and 1600 meters from the source (Figure 13). Short term
sampling averages of 5 minutes were converted to 1-hour averages using the 1/5 power law
(Turner, 1969). Stability classes of rural stable and urban neutral were simulated.
6
-------
Under simulated urban neutral, both models tended to under predict observed
concentrations by less than a factor of eight in almost all cases. ISC-PRIME was generally more
conservative than ISCST3 except close-in at a simulated distance of 150 meters (figure 14).
Under prediction generally occurred at simulated wind speed at or above 20 m/s. The lower the
wind speed, the higher the over prediction and the more conservative the modeled results. In the
0-5.6 m/s category, both models over predicted by a factor of 10 or more (Figure 15).
Under simulated rural stable conditions, the ISCST3 modeled results were conservative by
between one and two orders of magnitude. The ISC-PRIME results were slightly conservative
from 600 meters outward but became quickly non conservative to four orders of magnitude a
distance of 150 downwind (Figure 16).
There is support for this in the Bowline point data. Tables 1 and 2 show that ISC-PRIME
(and ISCST3) tend to under calculate concentrations at the close in Parking Lot and Met Tower
monitor sites.
4.3 Model Runs to Confirm ENSR CA
The ENSR CA runs were duplicated. The EPA duplicated output files were reviewed
against the values reported in the ENSR CA. Only minor discrepancies were found and the
overall conclusions were not affected.
4.4 EPA Runs using ENSR CA Input Data
The same input data sets used in the ENSR CA were used as input to these runs
(Appendix A). The 1964 Pittsburgh, PA meteorological data set was also used. Unless otherwise
noted (eg. "No Bldg"), the 35 m stack results were produced with buildings that were 34 meters
high while the 100m stack results were produced with buildings that were 50 meters high. With
respect to the ISCST3 results, the 35 meter results were produced using the Schulman-Scire
algorithm while the 100 meter results were produced using the Huber-Snyder algorithm.
In the "No Bldg" (no building) cases and using the same input for respective runs, the
maximum and highest of the second highest 1-, 3-, 24-hour and annual average values for all
receptors were the same between respective ISCST3 and ISC-PRIME runs. The ISCST3 and
ISC-PRIME results mirrored each other at all respective 20 downwind distances from 50 meters
to 10 km. This occurred under urban and rural conditions and at stack heights of 35 and 100
meters.
In the Schulman-Scire 35m stack (34m Squat building) cases, the ISCST3 rural
concentrations rise up to maximum peaks above 7500 ug/m3 at or just before 200 meters from the
stacks and then decrease asymtopically whether the stack is on the corner or away from the
building. The ISC-PRIME highest values for stacks located on the corner are less than 30% of
the ISCST3 values (Figure 17). ISCST3, as well as ISC-PRIME, highest of the second highest
7
-------
values are very similar to their respective maximum concentrations in all cases. The ISC-PRIME
maximum value for a Tall building with the stack located on the corner occurred at 50 meters
from the stack and there are indications that the actual maximum may be higher (Figure 18).
When the stack is located to the northeast of the building, the ISC-PRIME maximum values
are 5% to 10% of the respective ISCST3 maximum values (Figure 19).
Under urban conditions, ISCST3 concentrations rise to a maximum peak above 2500
ug/m3 and with their highest of the second highest values appearing to coincide with their
maximum value counterpart. The ISC-PRIME values are 90% of the ISCTST3 values to 25%
greater than their ISCST3 counterparts (Figures 20 and 21). When the stack is moved to the
northeast of the buildings, ISC-PRIME concentrations are about a third of the ISCST3 maximum
concentrations (Figure 22).
In the Huber-Snyder 100m stack (50m building) cases, the ISCST3 rural concentrations
rise up to maximum peaks above 60 ug/m3 after 400 meters from the stacks. The ISC-PRIME
highest values for stacks located on the corner are 20% to 50% greater than their ISCST3
counterparts (Figure 23). ISCST3 and ISC-PRIME highest of the second highest values are close
to their respective maximum concentrations in all cases.
Multimodal peaks appear in many of the ISCST3 and ISC-PRIME curves. This occurs
whether the stack is on the corner or away from the building. These peaks also appeared in
similar plots where no buildings were present in the input.
The ISC-PRIME maximum values for stacks located northeast of the buildings are equal
to 20% greater than their ISCST3 counterparts. However, the highest of the second highest ISC-
PRIME values are equal to 30% less than their ISCST3 counterparts (Figure 24). This is a
significant drop off with respect to possible design concentrations in close to the stack. However,
as distances downwind increase, the ISCST3 and ISC-PRIME maximum and highest of the
second highest values merge or significantly converge by 10 km downwind with the ISC-PRIME
values being at or greater than ISCST3 values. Differences at 2 km are appearent but not overly
significant.
Under urban conditions, ISCST3 concentrations rise to a maximum peak of around 100
ug/m3 and with their highest of the second highest values appearing to be less than their maximum
value counterpart. The ISC-PRIME values are 20 to 50% greater than their ISCST3
counterparts (Figure 25). When the stack is moved to the northeast of the buildings, ISC-
PRIME concentrations are about 20% greater than their ISCST3 counterparts (Figure 26).
5. CONCLUSIONS
When no building data was included in the input files, the ISC-PRIME model was able to
reproduce ISCST3 results for over 1600 receptors, urban and rural settings, two different stack
8
-------
heights and respective values using one year of hourly meteorological data. The incorporation of
the PRIME algorithms into ISCST3 is transparent to users when "no building" input data sets are
used and therefore the PRIME algorithms do not interfere with the no downwash functions of the
ISCST3 model.
The three field study results showed that ISC-PRIME can be more to less conservative
than ISCST3. This is dependent upon the stack-building relationship, downwind distance,
windspeed, stability class and emission factors. Overall and at various distances downwind, ISC-
PRIME tends to be less conservative than ISCST3, but more conservative than observed values.
The EPA analysis using the ENSR CA input data sets shows that the ENSR CA values
can be duplicated and that either model can calculate concentrations that are higher than the other
model's concentrations. Given the same input data, the model output differences are dependent
upon stack location, stack to building height, urban or rural setting, downwind distance. For all
runs, as the downwind distance increases beyond about 1 km, the ISC-PRIME and ISCST3
values converge. After 10 km downwind, the values, in most cases, are practically the same.
The studies so far have dealt with moderate to high stack exit velocities. When low exit
velocities and near ambient temperatures were used to generate slightly buoyant low momentum
plume rises, ISC-PRIME stopped in the middle of execution and reported an exponentiation
error. The problem was reported to EPRI. The revised source and executable code was received
but not in time to revise this part of this Consequences Analysis.
Overall, the ISC-PRIME model calculates conservative results that appear to give better
overall estimates than ISCST3. The design objectives of PRIME appear to have been met.
6. REFERENCES
Huber, A.H. and W.H. Snyder, 1982. Wind Tunnel Investigation of the Effects of a Rectangular
Shaped Building on Dispersion of Effluents from Short Adjacent Stacks. Atm. Env..l6: 2837-
2948.
Irwin, J.S., 1998. Private Communication.
Paine, Robert J., and Lew, Francis, 1997. Results of the Independent Evaluation of ISCST3 and
ISC-PRIME. TR-2460026: Final Report. Prepared for Electric Power Research Institute, Palo
Alto, CA.
Paine, Robert J., and Lew, Francis, 1997. Consequences Analysis for ISC-PRIME. TR-2460026:
Final Report. Prepared for Electric Power Research Institute, Palo Alto, CA.
Schulman, Lloyd, 1998. Private Communication.
9
-------
Schulman, L.L., D.G. Strimaitis, and J.S. Scire, 1998. Development and Evaluation of the
PRIME Plume Rise and Building Downwash Model. Paper No. 4B. 1, presented at the 10th Joint
Conference on the Applications of air Pollution Meteorology, Phoenix, AZ.
U.S. Environmental Protection Agency, 1995a. User's Guide for the Industrial Source Complex
(ISC) Dispersion Models (Revised), Volume 1, 2, and 3. EPA Publication Nos. EPA-454/B-95-
003a-c. Office of Air Quality Planning and Standards, Research Triangle Park, NC.
U.S. Environmental Protection Agency, 1995b. SCREEN3 Model User's Guide. EPA
Publication No. EPA-454/B-95-004. Office of Air Quality Planning and Standards, Research
Triangle Park, NC.
Weil, J.C., 1996. A New Dispersion Model for Stack Sources in Building Wakes, Ninth Joint
Conference on Application of Air Pollution Meteorology with A&WMA, 333-337, American
Meteorological Society, Boston, MA.
10
-------
APPENDIX A
CONSEQUENCES ANALYSIS SOURCE PARAMETERS
The following two sets of point source emissions data were also the data used in the EPRI
Consequences Analysis:
Parameter
Stack 1
Stack2
Stack Height (m)
35.0
100.0
Emission Rate (g/s)
100.0
100.0
Exit Temperature (K) :
432.0
416.0
Exit Velocity (m/s)
11.7
18.8
Stack Diameter (m)
2.4
4.6
The last set of point source emission data were the values that had caused ISC-PRIME to
crash. When raising the exit velocity to 10.1, the program ran just fine. The problem was
identified corrected by the developers.
Parameter
Stack 3
Stack Height (m)
5.0
Emission Rate (g/s)
10.0
Exit Temperature (K) :
280.0
Exit Velocity (m/s)
1.1
Stack Diameter (m)
2.4
A-l
-------
APPENDIX B
TABLE OF CONSEQUENCES ANALYSIS SCENARIOS.
ISC-P RIM E C on sequence Analyst*: Model! ng Scenarios
Case
Bldg. Dimensions
Btdg. Ht. (m)
Dispersion
Stack HL fm)
No Building
N/A
WA
Urban
35
Rural
35
N/A
Urban
100
Rural
100
Squat Build (rig
Stack Adjacent 10 HE
Comor ol Building
60m x120m
34
Urban
35
Rural
35
50
Urban
100
Rural
100
Squat Building
Slack at 4xHb lo NE
Comer of Building
60rn x 1Z0m
34
Urban
35
Rural
35
50
Urban
100
Rural
100
Tall Building
Stack Adjacent to NE
Comer of Building
30m x 30m
34
Urban
35
Rural
35
50
Urban
100
Rural
100
Tall Building
Stack at 4xHb to NE
Comer of Building
30m x 30m
34
Urban
35
Rural
35
50
Urban
100
Rural
1DQ
Super Squat Building
Slack Adjacent 1c NE
Comer of Smtdinci
100m x 1S0m
34
Urban
35
Rural
35
Super 3qilat Building
Slack at 4xHb lo NC
Comer of Building
reomx 180m
34
Urban
35
Rural
35
B-l
-------
APPENDIX C
VARIOUS FIGURES
Figure 1. Streamlines around a building (2). Wind flow is from left to right.
Figure 2. Depiction of ellipsoidal shape of Near Wake (cavity), Far Wake, and Wake Boundary.
C-2
-------
Bawllrte Monitor 1: Quanlile-Quarttile Plot of All Cases
£
10C0 -T=,
acH
O.C01
0.001
o.oi
1DM
Observed Con con [ralfan {iie/m')
Figure 3. Bowline Monitor 1: Quantile-Quantile Plot of All Cases. Note dashed factor of 2 lines, one on either side of the 1:1 ratio line
cutting diagonally across from corner to corner.
C-3
-------
Bowline Monitor 3: Qusrrtll^Quantila Plot of AH Case?
0.001 0,01 0.1 1 10 ¦ 100 1 1DQ0
OLntrrvnH Con contra [Ion fiiflftfl*)
Figure 4. Bowline Monitor 3: Quantile-Quantile Plot of All Cases. Note dashed factor of 2 lines, one on either side of the 1:1 ratio line
cutting diagonally across from corner to corner. Note also that the PRIME curve goes above the 1:1 line by about a factor of 3.
C-4
-------
N
CUHIi^na f A. i
1 SUrt 2
Hl.*9N>n Hi. - 2MJ1U
H.O MCTRHS
Figure 5a. Depiction of locations of the building and stacks used for the BPIP processing for the AGA data base: Kansas site. The
building is 12.19 m high and the stack heights are 9.8 m and 24.4m high.
C-5
-------
s. a uktbrs
Figure 5b. Depiction of locations of the building and stacks used for the BPIP processing for the AGA data base: Texas site. The
building is 11.37 m high and the stack is 9.75 m high.
C-6
-------
AG A; Cp/Co vs. Distance
100
10
D
O
B.
o
^ 1 t
0.1
0,01
50
(22 pts)
100 150
(30 ptsj (20 pte)
ftadtal Distance (m)
200
<18 pta)
Figure 6. Residual plot of predicted to observed concentration ratios versus distance for the AGA data base. The filled in rectangles
represent ISC-PRIME data while the transparent rectangles represent ISCST3 data.
C-7
-------
AGA: Cp/Co vb. Stability
100
3
¦n.
O
0,01
0.CQ1
1
P Pt»)
a
(12 pis)
(35 pis)
Stability Class
4
(34 pte)
5
(9 Ptej
Figure 7. Residual plot of predicted to observed concentration ratios versus stability class for the AGA data base. The filled in
rectangles represent ISC-PRIME data while the transparent rectangles represent ISCST3 data.
C-8
-------
AG A: Cp/Co vs. Wind Speed
0-1,8 2.7-3.7 3.9-5,8 >=6.3
l&pte) (2
-------
Tn ™
pr
\
HtL'uase Hnighla
H 8 UETCR5
Scald-
Figure 9. Depiction of locations of the building tiers and stacks used for the EOCR data base. Tier heights are 7.0 and 25 m high while
stacks along the perimeter are 1 m high and stacks on the tiers are 25 m and 30 m high.
C-10
-------
EOCR: Cp/Co va. Distance
100 —
10 t-rt
O
O
o
1 —
0.01
0,1
so
{43 pts)
100
(54 pta)
400
800
(50 pts) (41 pts)
Radial Distance (mj
120D
(2fl pta)
1600
(29 pts)
Figure 10. Residual pot of predicted to observed concentration ratios versus distance for the EOCR data base. The filled in rectangles
represent ISC-PRIME data while the transparent rectangles represent ISCST3 data.
C-ll
-------
EOCR: Cp/Co vs. Wind Speed
(72 pts} (106 pts) (66 pta)
WIpd Speed (mte)
Figure 11. Residual plot of predicted to observed concentration ratios versus 10-m wind speed for the EOCR data base. The filled in
rectangles represent ISC-PRIME data while the transparent rectangles represent ISCST3 data.
C-12
-------
EOCR: Cp/Co vs. Stability
14 6 6 7
(14 pts] (43 pta) (59 pfcs) (41 pta} (77 pt»)
Stability Class
Figure 12. Residual plot of predicted over observed versus stability class for the EOCR data base. The filled in rectangles represent
ISC-PRIME data while the transparent rectangles represent ISCST3 data.
C-13
-------
N
® 21$ tt
-------
Wind Tunnel: Neutral Stability, Cp/Co vs. Distance (Urban Dispersion)
100
10
o
tj
0.T
0.01
150
(£11 pts)
300
(146 pts)
450
fiOO
750
900
(146 pts} {149 pts} (64 pts)
Racfla] Distance (m)
(64 pts)
Figure 14. Residual plot of predicted over observed concentration ratios versus distance for the Lee Power Plant data base, neutral
(urban) cases only. The filled in rectangles represent ISC-PRIME data while the transparent rectangles represent ISCST3 data.
C-15
-------
Wind Tunnel: Neutral Stability, Cp/Co vs. Wind Speed (Urban Dispersion)
100
0.01
0-5,6
(19 pts)
11.7-12.7
{72 pts)
19.1-26.fl
(134 pfcf)
Wind SfHied (mte)
27.0-33.fl
{306 pts}
>=34.0
(99 pts)
Figure 15. Residual plot of predicted over observed concentration ratios versus 10-meter wind speed for the Lee Power Plant data
base, neutral (urban) cases only. The filled in rectangles represent ISC-PRIME data while the transparent rectangles represent ISCST3
data.
C-16
-------
Wind Tunnel: Stable Cp/Co vs. Distance (Rural Dispersion)
1000
100
10
1
£ 0.1
0.01
0,001
0.0001
0.00001
Figure 16. Residual plot of predicted over observed concentration ratios versus distance for the Lee Power Plant data base, stable
(rural) cases only. The filled in rectangles represent ISC-PRIME data while the transparent rectangles represent ISCST3 data.
; in iTTiTi rif urn ir| turn m hitti pi tirnnik=5==
riT=Y
nr -"fiTT! rn fijn; rr* fnrn rn ;Trn-|T;rrrr-^rrrri::rr: =tt
[tit errunr nr.* urn n? unri Wi tun nrp = irq
TrrnTrrrrrrrrrrrnTTiTTT rrrn rn>7
mrrm
^TfT*
_ f 1 1 _ f
rn ?t?m rn itttj rnpT=r?iTnTrnTf? =ttt7 rn rmrn
¦ nTT!iir^ni;"nifiTrnninTT!fTy;Tnr;TfnTiT;^rnT!nT;rn?i
150
(24 pts]
300
(26 pts)
45D
EDO
{29 pts} (32 pts)
Radial Df®lance (ml
760
(35 pts)
900
(39 pts)
C-17
-------
Max. and H2H Cones by Downwind Distance
Squat Bldg, 35m Stack Height on 34m Corner,
Rural, 1 -Hr Averages
8000.00
6000.00
CO
<
E
D)
3
C
.2 4000.00
•l—i
CT3
•4—1
c
0
o
c
O 2000.00
0.00
0.01
Curve Definitions
- - ISC-PRIME Max
ISCST3 Max
- - ISC-PRIME H2H
ISCST3 H2H
0.10 1.00
Downwind Distance (km)
10.00
Figure 17. Maximum and Highest of the Second Highest 1-hour averaged concentrations versus
downwind distance. ISCST3 max and H2H coincide visually.
C-17
-------
Max. and H2H Cones by Downwind Distance
Tall Bldg, 35m Stack Height on 34m Corner,
Rural, 1-Hr Averages
Downwind Distance (km)
Figure 18. Maximum and Highest of the Second Highest 1-hour averaged concentrations versus
downwind distance. ISCST3 max and H2H coincide visually over much of the ISCST3 curves.
C-18
-------
Max. and H2H Cones by Downwind Distance
Squat Bldg, 35m Stack Height NE of 34m Corner,
Rural, 1-Hr Averages
Downwind Distance (km)
Figure 19. Maximum and Highest of the Second Highest 1-hour averaged concentrations versus
downwind distance. ISCST3 max and H2H coincide visually over much of the ISCST3 curves.
C-19
-------
Max. and H2H Cones by Downwind Distance
34m Super Squat Bldg, 35m Stack Height on Corner,
Urban, 1-Hr Averages
0.01 0.10 1.00 10.00
Downwind Distance (km)
Figure 20. Maximum and Highest of the Second Highest 1-hour averaged concentrations versus
downwind distance.
C-20
-------
Max. and H2H Cones by Downwind Distance
Squat Bldg, 35m Stack Height on 34m Corner,
Urban, 1 -Hr Averages
0.01 0.10 1.00 10.00
Downwind Distance (km)
Figure 21. Maximum and Highest of the Second Highest 1-hour averaged concentrations versus
downwind distance.
C-21
-------
Max. and H2H Cones by Downwind Distance
Tall Bldg, 35m Stack Height NE of 34m Corner,
Urban, 1-Hr Averages
0.01 0.10 1.00 10.00
Downwind Distance (km)
Figure 22. Maximum and Highest of the Second Highest 1-hour averaged concentrations versus
downwind distance.
C-22
-------
Max. and H2H Cones by Downwind Distance
Squat Bldg, 100m Stack Height on 50m Corner,
Rural, 1-Hr Averages
Downwind Distance (km)
Figure 23. Maximum and Highest of the Second Highest 1-hour averaged concentrations versus
downwind distance.
C-23
-------
Max. and H2H Cones by Downwind Distance
Squat Bldg, 100m Stack Height NE of 50m Corner,
Rural, 1-Hr Averages
Downwind Distance (km)
Figure 24. Maximum and Highest of the Second Highest 1-hour averaged concentrations versus
downwind distance.
C-24
-------
Max. and H2H Cones by Downwind Distance
Squat Bldg, 100m Stack Height on 50m Corner,
Urban, 1-Hr Averages
Downwind Distance (km)
Figure 25. Maximum and Highest of the Second Highest 1-hour averaged concentrations versus
downwind distance.
C-25
-------
Max. and H2H Cones by Downwind Distance
Squat Bldg, 100m Stack Height NE of 50m Corner,
Urban, 1-Hr Averages
Downwind Distance (km)
Figure 26. Maximum and Highest of the Second Highest 1-hour averaged concentrations versus
downwind distance.
C-26
-------
APPENDIX D
TABLES OF BOWLINE POINT CLOSE-IN CONCENTRATIONS
!"-»rtI r*g Lot
parking LOC
Parking l^fc
Hank
flDHIffd
f
Come,
Re
1
at, :¦
319
2
i) *
341
J
7-41 t
¦a j m
j* j
1 PI
i u
4
*t,t
]»
5-
74,1
3*1
23
C
3 51
21
t
ti .a
3**
13
1
73.fr
31*
>
13 ,9
7**
21
14
11.3
3 31
11
11 .3
13*
it
H,(
14 B
11
10.4
13 6
13
3*
S*r>
15
3-D
15
31>
it
6t.j
Jit
IT
11
3*1
11
IB
*T.5
31*
L)
19
*7.5
344
*1
10
*1-3
3)T
id
31
J 5 £
33
Jt)
it
3)
151
33
H
fit. 3
HI
H
i(,l
to
14
it
15. 1
151
1*
27
ES. *
)SS
11
ISlI
3 J
li
15.4
141
31
)fl
(it
311
11
41.3
3)1
tl
13
(5.1
ISO
13
*5 .1
144
IS
)4
*4 ,>
11
31
(I.}
D*
14
i«i
*1.4
151
It
)*
*1.1
it
*4.5
-M
31
u*
iO
31
ICi
t*.i
lifi
11
fl
*4.2
SK
1)
12
(4.1
34a
If
41
(} 1
3tS
If
14
fi).4
HI
34
45
(J.4
31)
If
4*
£3.3
34*
4 T
&].]
341
19
(1
43.3
351
IB
4.9
:,3S1
>1
to
IS
'fSCSTJ
Rtnfc IEC1
-PR1HEL
l^cdi£ .
Bay
Re
4 Cone.
LKiy
He
313
13
k. 119.*
*1
13
1L5.3
3)3
1)
3 3)*.}
14 5
14
£5.3
SI
51
3 331.5
311
IT
S3. J
*1T
6
4 3 34.-5
145
1)
li .3
9)
It
5 311.1
14 5
If
13 .1
JlT
T
1 11).4
1*
21
14.&
1*1
It
1 lllr4
14
1C .4
)M
T
1 It?.*
14
li.)
91
IS
t lti?,*
91
ll
11.3
to
7
i« tflu.a
it's.
ll
T.J
124
If
11
T.l
3)4
H
13 PS.4
14 5
n
fi.f
JB4
1
1>. SL..4
it
ij
4.*
11}
33
11 94.4
II
33
f .(
li)
3
15 51.J
II
33
£.5
131
tl
IS 7).1
145
13
t.l
2GL
H
LT 71,i
331
34
*,4
3D 5
14
LI (3.4
IB
31
4.4
51
1
19 55. f
led
34
1.4
51
19
)d 51.3
I?
i.l
331
11
11 5 4.5
3»
It
7. *
14 i
11
».l
321
19
1. 1
IT*
14
31 il>.J
>02
If
304
1)
14 44 *
91
1.3
3)1
*0
31 H4.4
Ml
Lf-
D.l
3 <14
It
3* 14.*
*T
at
n.i
fl
ID
It 44.4
T*
li
D.9
i)i
14
34 41.C
1»
3D
1
1
3) 43.4
M
3D
4.4
UL
JO 3-9. D
ail
11
(J,4
LI 3
*
41 17 r1
3D3
a,4
1*
t
13 nva
»
11
0..1
34 7
S
13 ii-3
14
11
t.i
It*
13
14 15.3
M
t. 1
)lf
4
IS >5.3
an
If
CI, 1
143
3
tt 14.1
n
11
«. )
314
36
)I 13.1
ID*
31
ti.)
314
11
)* 1)4
14
Ifr
ti. 1
44
if
1) IV 3
1(4
0-3
1)1
11
41 )3 .1
u
1*
D.J
n
f
41 )1,4
It
tl
o-a
3)9
1
42 Jt'.t
34
10
5-1
3 43
If
4) |i?,i
33 T
1+
D.l
fl
J 4
It 34.3
ft
1*
0 -I
134
11
li 3?,1
£31
Jfl
D.l
S3
33
tl 31.5.
*3
If.
D.l
13
4) H.t
19
12
D.l
3D
41 3S.}
34
30
D.l
Jli
13
4) as .7
«T
1)
4.1
at
31
5 d ai,i
101
If
Table 1. Bowline Point Parking Lot monitoring and associated ISCST3 and ISC-PRIME
modeling data results for the top 50 1-hour average concentration values each.
D-l
-------
Mot Tenter
Met Twer
Htt Tliner
taift
Cfapa:
Cine.
It-J .4
an.}
2 1*.
117
113 .7
113 ,1
151 .1
I so
141.3
UJ .1
1)1 I
129 .1
115., 4
HI .J
11*. I
ID] .1
:D3 ,*
1D1 .1
M i
)i t
it.(
54
*i .J
91 . 1
it.*
»lr±
¦t,£
H4.1
14.1
H4.1
¦4.4
13 .f
17.1
fd.l
¦ D.l
7 9..1
1*.J
77-J
74 .1
Urfl
71..-t
fit .1
67.*
fiT.l
H ,*
S*r3
fi*/r
E+rt
si.)
Snnk
I3CTST1
Hftnik
hy Hr
94 10
M
91
91
1)
H
ID-i
91
If
H
Si
74
M
11
IIT
311
J*
]1T
15 D
II
39
»1
JJt
fl
li
7t
19
173
in
173
31
iif
•t
Jl-53
71
11 *
t»
!13
1+
ati
ti
151
31)
7*1
74
35
370
31
lit &
Di/ Hr
¦ CfelK.
biy
93 13
1 lii-i
313
5
117 11
3 111.1
313
5
319 31
J 11.1
131
III 11
1 1.1
53
*1 1
5 0.5
111
311 14
4 9,1
114
ill 31
7 D.3
30t
115 11
1 ».l
313
$21 17
t d.l
301
2?D 31
10 fr-l
El
41 1*
11 D.l
313
174 t
ia »,i
111
n
13 0
III
145 11
H t.t
ltif-
14 5 1+
]j 4.4
311
N U
14
1F1
at lb
17
347
ill 12
11 M
313
Ui ^
1? > .&
31*
in U
3Q ».»
E.3
3)1 T-I
21 4.4
2CL
iti 1£
11 a. ft
L9L
145 LT
i ] 4.4
L 91
Lt* Li
;i 1.4
213
337 11
5£ 4.4
113
*1 1*
34 41 f
a*
*1 10
37 41 f
L»i
*11
It 4.*
3-DI
IFl ll
3-1 4 ¦ f
2 b L
n »
30 4.*
1
i*f a
31 f.t
1
US 13
tt 1.1
II
3D3 3
II *¦*
LIS
3D 1 11
14 4.4
ill
1+? Pi
15 f.f
tilt 33
It 4.t
3*1
HI 1
If 4.4
tl?
ill 13
1* 4.4
ti
167 i*
If t.t
at
Ti n
40 4.4
i.**
Mt 15
41 C.a
211
¦) It
43 4.4
ili
t?T IB
4J 4.4
IB?
}DL 3
41 4.4
lit
IT L 5
14 0-4
1*1
9* J1
46 0 .D
±47
T.J6 11
47 0-0
113
It E
It O.o
If!
It* 1
It O.o
S3
ni i
5D 0.0
lfB
Table 2. Bowline Point Met Tower monitoring and associated ISCST3 and ISC-PRIME
modeling data results for the top 50 1-hour average concentration values each.
D-2
------- |