EPA/600/A-95/114
11.10	SENSITIVITY OF THE INDUSTRIAL SOURCE COMPLEX MODEL
TO INPUT PARAMETERS
Donna B. Schwede"*
Atmospheric Sciences Modeling Division
Air Resources Laboratory
National Oceanic and Atmospheric Administration
Research Triangle Park, NC 27711
James O, Paumier
Pacific Environmental Services, Inc.
Research Triangle Park, NC 27709
1. INTRODUCTION
In recognition of the need for a state-of-the-science
model for estimating pollutant concentrations, as well as
dry and wet deposition of these pollutants, the U.S.
Environmental Protection Agency (USEPA) released a
new version of the Industrial Source Complex-Short Term
model (ISCST2) (USEPA, 1992). This new version,
ISCST3, integrates the algorithms for modeling simple
terrain found in ISCST2 and the algorithms found in the
COMPLEX I model, a USEPA screening-level model for
complex terrain applications. In addition, the model
includes a newly developed algorithm for modeling dry
deposition of particulates, algorithms for modeling wet
deposition, and a new algorithm for modeling area
sources. The model is described in an updated user's
guide (USEPA, 1995).
In this paper we examine the sensitivity of predicted
concentrations, dry deposition fluxes and wet deposition
fluxes to input parameters related to deposition of
particles. We consider the effects of dry and wet plume
depletion, the shape of the particle size distribution, the
resolution of the particle size distribution, the particle
density, scavenging coefficients, and the use of gridded
terrain data. The results reported here should be
considered preliminary until the analysis can be repeated
using alternate data sets.
2.	TEST PARAMETERS
A test data set was created for use in the sensitivity
analysis. A single stack, emitting particulate matter, with
a height of 100 m was used for all of the tests except the
terrain grid tests. A shorter stack was used for those tests
to insure high pollutant impact on the terrain. A source-
centered, gridded-polar receptor network was used for all
the sensitivity tests. Thirty-six radials, from 10° to 360°
every 10°, were used, with receptors defined along each
radial at distances from 0.1 to 20 km. The receptors were
defined on flat terrain for most of the sensitivity tests,
except the terrain grid tests. Gridded terrain elevation
data, described in section 3.6, were used for the terrain
sensitivity tests. One year of hourly meteorological data
was used for all tests.
3.	RESULTS
In this section we describe the results of the
individual sensitivity tests and speculate as to the reasons
for these results. We chose to examine the highest 25
values of concentration, dry deposition flux, and wet
deposition flux unpaired in space or time since these are
of concern in many regulatory applications. All cases
except those described in section 3.1 were run including
the effects of plume depletion due to dry and wet
deposition.
* Corresponding author address: Donna B. Schwede, U.S. EPA, MD-80, Research Triangle Park, NC, 27711.
+ On assignment to the National Exposure Research Laboratory, U.S. Environmental Protection Agency.

-------
TABLE I
Diameter(pm)
0.32
0.55
0.7
0.9
1.55
3.4
5.56
8.37
12.66 17.61
Mass Frac.
0.13
0.04
0.07
0.05
0.13
0.06
0.06
0.18
0.23 0.05
Scav. Coeff.*
1.4
0.88
0.6
0.56
1.32
2.44
3.41
4.74
7.05 10.22
Diameter(jjm)
0.32
1.38
4.61
8.37
15.53




Mass Frac.
0.13
0.29
0.12
0.18
0.28




Scav. Coeff."
1.4
1.13
2.99
4.74
8.82




Diameter(pm)
1.26
6.78
15.53






Mass Frac.
0.42
0.3
0.28






Scav. Coeff.*
0.98
3.98
8.82






' - the scavenging coefficients should be multiplied by 10
3.1	Depletion
Depletion accounts for the mass lost from the plume
due to deposition. To test the effects of plume depletion
on the resulting maximum concentration and deposition
fluxes, ISCST3 was run using an example particle size
distribution with dry depletion only, wet depletion only,
and no depletion. Maximum concentrations were lowered
due to the dry depletion and were unaffected by the use of
the wet depletion option. Since wet deposition fluxes
were generally high for the test cases used, it makes sense
that maximum concentrations would occur when the wet
deposition flux and, therefore, the wet depletion is
minimal. Figure 1 shows the effect of dry and wet
depletion on the maximum dry deposition fluxes. Both dry
and wet depletion have an effect due to the reduced mass,
however the dry depletion appears to have more of an
effect than wet depletion because the receptors with
maximum dry deposition flux do not necessarily
correspond to the receptors with maximum wet deposition
flux. The use of dry depletion has little effect on the
maximum predicted wet deposition flux since the highest
wet deposition fluxes generally occur close to the source,
where ground level concentrations and dry deposition
fluxes are lower.
3.2	Mass Fraction
The deposition velocity calculated by the dry
deposition algorithm and the scavenging coefficient
specified by the user for use in the wet deposition
algorithm are a function of particle diameter, so the
distribution of the mass at a particular particle size is an
important input to the model. We selected three particle
size distributions (Set 1, Set 2, Set 3) for use in this test
that are typical of different control strategies that might be
used with a municipal waste combustor. These
distributions are plotted in Figure 2. The same scavenging
coefficients were used for each distribution. Figure 3
shows the effect of using these different distributions on
the maximum predicted dry deposition flux. We see that
there is a greater difference in the dry deposition flux
between using set 1 versus set 2 compared to set 2 versus
set 3. This is likely due to the peak in the set 1
distribution at 10 jam which would result in high
deposition, whereas there is less of a difference in the
shape of distributions 2 and 3. As expected, the maximum
concentrations are lower for the sets with higher dry
deposition. The wet deposition flux is less sensitive to the
distribution used than is the dry deposition flux. This is
because the relative differences between scavenging
coefficients for different size categories is less than the
differences between dry deposition velocities.
3.3 Particulate size resolution
The resolution (number) of particle size categories
was varied from 10 to five to three, with a corresponding
variation in the mass fraction and scavenging coefficients.
The 10 size categories correspond to set 1 in the mass
fraction tests described above. The 10 categories were
combined to form five categories. Diameters from
combined categories were used to calculate a mass mean
diameter for the new category. Scavenging coefficients
were calculated using a formula described in section 3.4
below. The five categories were similarly combined to
form the three categories. The size categories and
conresponding mass fractions and scavenging coefficients
used are shown in Table 1.

-------
TABLE2
Diameter(iam)
0.32
0.55
0.7
0.9
1.55
3.4
5.56
8.37
12.66
17.61
Base case
1.4
0.88
0.6
0.56
1.32
2.44
3.41
4.74
7.05
10.22
Base case + o
0.85
0.54
0.37
0.34
0.8
0.28
0.18
0.12
0.11
0.27
Base case - o
2.31
1.46
0.99
0.92
2.17
4.02
5.63
7.82
11.63
16.85
The results show that changing the number of
categories has little effect on concentration. Figure 4
shows the comparison of maximum dry deposition fluxes
for the three sets of size categories. Coarser resolution
results in an increase in the predicted dry deposition flux.
As a result of combining categories, the distribution
becomes skewed toward larger particles, which likely
causes the estimated dry deposition flux to increase. The
trend for wet deposition is not as clear. The maximum wet
deposition flux is greatest for the case with 5 categories,
while the wet deposition estimates for the case with 3
categories fall between those of the 10 category case and
the 5 category case. This may have occurred because the
distribution for the case with 3 categories is heavily
weighted at a diameter of about 1 |im which corresponds
to a very low scavenging coefficient.
3.4	Particle density
Density was varied to simulate different particulate
materials. Four densities were modeled: 1.0, 1.5,2.0 and
5.0 g cm'3. The first three densities are more
representative of combustion materials from a facility that
burns traditional fossil fuels, while the fourth density is
more representative of metals that may be emitted as a
result of burning waste-derived fuels. The results for dry
deposition are shown in Figure 5. As expected, an
increase in particle density results in an increase in the
maximum dry deposition flux. The increased density
likely promotes gravitational settling, a component of dry
deposition. Due to the effects of plume depletion, the
increased deposition results in decreased maximum
concentrations. Density variations showed no effect on the
maximum predicted wet deposition flux since maximum
wet deposition usually occurs close to the source where
ground level concentration and dry deposition are
minimal.
3.5	Scavenging coefficients
Three sets of scavenging coefficients were modeled.
The first set of coefficients is based on a particle size
distribution used in previous sections. The following
formula, developed by Crouch (1993) based on the work
of Jindal and Heinhold (1991), was used to obtain the
scavenging coefficient, a, for each particle size category
for liquid precipitation scavenging:
In a = 	-	 + B + Cz(z + z.)2 + p
(z + Z,,)2 + X
where A= 0.0523, B = -8.569, C = 0.214, z0 = 0.0786, z,
= 1.288, X - 0.0419, and z = log,„ (d), where d is the
particle diameter in microns, n is a random variable with
a normal distribution with a mean of 0 and variance, a2, of
0.242 (i.e., N(0,o2)).
The coefficients then are obtained from ew<0. The second
and third sets were obtained from the first using ln(a) ± o,
where o is the standard deviation and equal to the square
root of o2 above. The resulting scavenging coefficients for
liquid precipitation are given in Table 2.
The effect on the maximum wet deposition fluxes is
shown in Figure 6. As expected, increasing the
scavenging coefficients increases the predicted wet
deposition flux. Varying the coefficients has no effect on
the maximum concentrations and dry deposition fluxes
since the maximum values for wet deposition occur close
to the source where concentration and dry deposition will
be minimal.
3.6 Effect of terrain grid
In this series of tests, the effect of the use of a terrain
grid was examined. The terrain grid is used only in the
calculation of the plume depletion due to dry deposition
where an integration of the material deposited along the
plume path is performed. A 10 km by 10 km Cartesian
gridded terrain network, with terrain elevation specified
every 100 m in the north-south and east-west directions,
was defined for ISCST3. An elevation was assigned to
each receptor in the polar grid. The ISCST3 model was
run both with and without this terrain grid. In the absence
of a terrain grid, the model represents the effect of terrain
by linearly interpolating between the source elevation and
receptor elevation.

-------
The use of a terrain grid hrs no effect on the
maximum predicted wet deposition. It does, however
affect the concentration and dry deposition. Maximum
concentration estimates decreased by up to 4% and the
maximum dry deposition flux decreased by as much as 7%
when the terrain grid was used. These results likely
depend on the terrain being considered.
4. CONCLUSIONS
We have reported the results of initial sensitivity
testing of the ISCST3 model to input parameters
controlling deposition. While these results are not
comprehensive, they provide useful information to model
users in selecting model inputs. For the model
sensitivities explored, the predicted maximum
concentrations and deposition fluxes responded in a
manner that is supported by the technical basis of the
model. Further tests should be done with other test
conditions to affirm our results.
PLUME DEPLETION
0006	0.00?	0.008	0000	0010	0,011
DRY DEPOSITION WITHOUT DEPLETION (g m*)
Figure 1. Comparison of the predicted maximum dry
deposition flux for ISCST3 model run where depletion
was not considered with cases where dry or wet depletion
was considered.
5. REFERENCES
Crouch, Edmund, 1993: Personal Communication.
Jindal, M. and D. Heinhold, 1991: Development of
Particulate Scavenging Coefficients to Model Wet
Deposition from Industrial Combustion Sources. Air and
Waste Management Association, Presented at the 84th
Annual Meeting, Vancouver, B.C., June 16-21,1991.
USEPA, 1995: User's Guide for the Industrial Source
Complex (ISC3) Dispersion Models Volume II -
Description of Model Algorithms (Draft). EPA-454/B-
95-0Q3b, Office of Air Quality Planning and Standards,
U.S. Environmental Protection Agency, Research Triangle
Park, NC.
PARTICLE SIZE DISTRIBUTIONS
100	» » « • • I
PARTICLE DIAMETER (wm)
Figure 2. Plot of the three particle size distributions used
in the sensitivity analysis.
USEPA, 1992: User's Guide for the Industrial Source
Complex (ISC2) Dispersion Models Volume II -
Description of Model Algorithms. EPA-450/4-92-008b,
Office of Air Quality Planning and Standards, U.S.
Environmental Protection Agency, Research Triangle
Park, NC.
DISCLAIMER
This paper has been reviewed in accordance with the U.S.
Environmental Protection Agency's peer and
administrative review policies and approved for
presentation and publication. Mention of trade names or
commercial products does not constitute endorsement or
recommendation for use.

-------
MASS FRACTION
PARTICLE DENSITY
0.006	0.004	0.007	0.000
DRY DEPOSITION FOR SET t (g m*)
Figure 3. Comparison of maximum dry deposition flux
predicted by ISCST3 for the three different distributions
of the mass fraction.
6
S «
LIJ
Q
O
5 1.1


• 1 P-iJacr*'
•—e.lOgcm1
•••- 9 • 10 9 Off1
A <
'c Cy-ia'-s
0.004	0.005	0.006	0.007	0.008
DRY DEPOSmON WITH DENSITY = 1.0 g cm-3 (g m'2)
Figure 5. Comparison of maximum dry deposition flux
predicted by ISCST3 for 4 different particle densities.
SCAVENGING COEFFICIENT
1.08
1.07
1.0S
1.0S
1.04
1.03
1.02
1.01
1.00
0.90
PARTICLE SIZE RESOLUTION
0.004	0.005	0.006	0.007	0.006
DRY DEPOSITION WfTH 10 CATEGORIES (g m*2)
Figure 4. Comparison of the maximum dry deposition
flux predicted by ISCST3 for the original 10 category
particle size distribution and the 5 and 3 category
distributions.
O 14
UJ
<0
2
a
s
o
2 t.o
O
P
5

		a	

WET DEPOSITION FOR BASE CASE (g m*)
Figure 6. Comparison of the maximum wet deposition
flux predicted by ISCST3 for the base case scavenging
coefficients and variations (±o) of the base case.

-------
TECHNICAL REPORT DATA
1. REPORT NO,
EPA/600/A-95/11 4
2 .
TITLE AND SUBTITLE
Sensitivity of the Industrial Source Complex Model
to Input Parameters
5.REPORT DATE
6.PERFORMING ORGANIZATION CODE
7. AUTHOR (S)
SCHWEDE, D.B.1, J.O.Paumier2
.PERFORMING ORGANIZATION REPORT NO.
9. PERFORMING ORGANIZATION NAME AND ADDRESS
1Same as Block 12
2Pacifie Environmental Services, Inc.
RTP, NC 27711
10.PROGRAM ELEMENT NO.
11. CONTRACT/GRANT NO.
12. SPONSORING AGENCY NAME AND ADDRESS
NATIONAL EXPOSURE RESEARCH LABORATORY
OFFICE OF RESEARCH AND DEVELOPMENT
U.S. ENVIRONMENTAL PROTECTION AGENCY
RESEARCH TRIANGLE PARK, NC 27711
13.TYPE OF REPORT AND PERIOD COVERED
Conference paper, FY-95
14. SPONSORING AGENCY CODE
EPA/600/9
15. SUPPLEMENTARY NOTES
16. ABSTRACT
In recognition of the need for a statc-of-thc-science model for estimating concentrations and dry and wet deposition of air pollutants, the U.S.
Environmental Protection Agency has developed an improved modeling technique. The newest version of the Industrial Source Complex model (ISC3)
integrates dry deposition, wet deposition, and complex terrain algorithms into the Industrial Source Complex - Short Term (ISCST2) model as well as a
new area source algorithm and algorithms for modeling open-pit source types. The dry deposition algorithm couples a deposition velocity calculated from
a resistance model with a modified source depletion algorithm. The wet deposition algorithm is a simple scavenging coefficient approach. Complex
terrain algorithms from the COMPLEX I model were used. The new area source algorithm is a numerical integration approach. The method for modeling
open-pit sources was derived from a fluid modeling study. The additional capabilities provided in ISC3 make the model an important tool for modeling
for ambient air quality standards and for performing multi-pathway risk assessments that require concentration and deposition estimates as their starting
point.
To better understand these new algorithms, particularly the deposition algorithms, we examined the sensitivity of the model results to various
input parameters. The predicted dry deposition flux depends on the specification of the particle size distribution since this affects the calculated deposition
velocity. We varied both the shape of the size distribution and the resolution of the distribution. The dry deposition algorithm also includes a depletion
algorithm which integrates the mass-loss along the plume path. The terrain elevations along the plume path can be interpolated between the source and
receptor elevations or can be specified explicitly using a terrain grid. We examined the effect of the use of the terrain grid as well as the resolution of the
grid on the calculated depletion. The wet deposition flux depends on the scavenging coefficient which is specified as a function of particle size for a unit
precipitation rate. For a given particle size distribution, wc varied the scavenging coefficients from a base value to one standard deviation higher and
lower than that value. Three point sources were run using one year of National Weather Service meteorology. The parameters listed above were varied
independently and the concentration and deposition values were compared for I-h, 3-h, 24-h and annual averaging times. Of the parameters tested, the
modeled results are most sensitive to the specification of the scavenging coefficient, followed by the shape of the particle size distribution and the
resolution of the particle size distribution, the sensitivity to terrain specification depended highly on the receptors examined.
KEY WORDS AND DOCUMENT ANALYSIS
DESCRIPTORS
b.IDENTIFIERS/ OPEN ENDED TERMS
cCOSATI
18, DISTRIBUTION STATEMENT
RELEASE TO PUBLIC
19 SECURITY CLASS (This Report)
UNCLASSIFIED
21.NO. OF PAGES
20, SECURITY CLASS (IMt Page)
UNCLASSIFIED
22. PRICE

-------