EPA/600/A-96/019
IMPROVED ALGORITHMS FOR ESTIMATING THE EFFECTS OF POLLUTION
IMPACTS FROM AREA AND OPEN PIT SOURCES
William B. Petersen and Steven G. Perry1
Atmospheric Sciences Modeling Division
Air Resources Laboratory
National Oceanic and Atmospheric Administration
Research Triangle Park, NC 27711
INTRODUCTION
The increasing environmental concern over hazardous emissions from landfills, landfarms,
settling ponds, agricultural fields, surface mines, and surface impoundments has renewed interest
in improving concentration estimates downwind from area sources. Simplistic and numerically
efficient methods (for estimating area source impacts) are the virtual point source approach
(Turner, 1970) as used in the models LONGZ (Bjorklund and Bowers, 1979) and VALLEY (Burt,
1977), and the finite line source approach as used in ISCST (Bowers et al., 1979). The virtual
point source and finite line segment algorithms are most appropriate for estimating concentrations
at downwind distances significantly larger than the side width of the area source (Hwang, 1986
and Weber, 1982). However, these constructs fail for concentration estimates within or near the
area source. Thistle and Londergan (1989) showed that these algorithms behaved in a manner
inconsistent with mathematical and physical principles. They demonstrated that models that use
numerical techniques to approximate the Gaussian point source dispersion function over the area
source are more realistic although computationally intensive. One of the models that they found to
be mathematically and physically consistent was PAL (Petersen and Rumsey, 1987).
Shortly after the Thistle and Londergan study, the Office of Air Quality Planning and
Standards in the U.S. Environmental Protection Agency (EPA) decided to update the area source
algorithm in ISC2 (U.S. EPA, 1992) using the approach contained in PAL. At about the same time
the Office of Research and Development in EPA was involved in a project to improve the
computational efficiency of the area source algorithm in PAL. It is this more efficient version of
the PAL area source algorithm that has been implemented in ISC2 and will be described in this
paper. In addition, the improved area source algorithm is utilized in a new approach to modeling
open-pit sources.
Title II, Part B, Section 234 of the Clean Air Act Amendments of 1990 provide for a
reexamination of the current Environmental Protection Agency's methods for modeling fugitive
particulate (PMio) from open-pit surface coal mines. ISC2 is specifically named as the method
that needs further study.
1 Both authors are on assignment to the National Exposure Research Laboratory, U.S. Environmental
Protection Agency.
-------
To address the requirements of the Act, an open pit algorithm has been developed. Wind
tunnel experimentation (Thompson, 1993, 1994a and Perry et al., 1994) suggests that the
emissions within the pit are not uniformly released from the pit opening. Rather, they show a
tendency to be emitted primarily from an upwind sub-area of the pit opening due to the general
recirculation flow within the pit. The algorithm uses the pit depth and dimensions, height of
emissions above the pit floor, and the wind direction to 1) estimate the fraction of paniculate that
leaves the pit (escape fraction) and 2) to calculate the sub-area dimensions. The new integrated
area-source algorithm is used to model emissions from this sub-area. Initial vertical dispersion
due to turbulence in the pit is also considered. Although the development work for the area source
and open pit algorithms was done with ISC2, both are incorporated in ISC3. In this paper we will
refer to the model incorporating these algorithms as ISC3.
NEW AREA-SOURCE ALGORITHM
Based on the results of an evaluation of area source algorithms performed for EPA
(Thistle and Londergan, 1989), the finite line segment algorithm used in the ISCST model gives
physically unrealistic results for receptors located near the edges and corners of the area. This is
not so much a deficiency in the algorithm as it is a problem in application. The finite line segment
was developed as a simple approximation to the contribution at a receptor from an area source.
Conceptually, it is an improvement over the virtual point source approach without the expense of
numerically approximating the point source function over an area source. It was never the intent
of the developers that the area source algorithm would be applied for nearby receptors. However,
recent concern for environmental impacts from nearby area sources and advancements in high
speed data processing for PC's make it reasonable to revisit the area source problem. This section
provides a brief description of the integration technique used in ISC2.
Wind Direction
Q
•o
s
u
0,0
Receptor
Xmin
Alongwind Direction
Figure 1. xmin and xmax are the distances to the closest and furthest vertices from the area
source, yj and^ arc *« crosswind distances at an arbitrary distance x upwind.
2
-------
The area source algorithms in PAL and ISC3 are functionally the same; the pollutant
impact from an area source is approximated by a number of finite crosswind line sources.
However, the integration techniques are different. To illustrate the integration technique we apply
it for a rectangular area source. Let x be the upwind distance and y the crosswind distance. For a
given source receptor pair, there is a xmin and a x^ (see Figure 1) which coincides with the x of a
vertex. That is, a vertex is represented by (x/.y,) with *„+/ = */, for closure. If we \e\.f(x) be the
dispersion from a point source at a distance x and let y(x) be the crosswind distance to the ih side,
then the integration proceeds as follows: In Figure \,xmin> *„ xb, and x^ are the upwind distances
from the nearest to furthest vertex from a downwind receptor and the four vertices are Xj, A^, Aj,
and Xj. The concentration can be expressed as;
X=
= — Pi = —.
a., a
where;
y "* y
and e(pi) and e(pj) are defined as follows,
Pi j
e(p,) = f -p= exp M).5 p
exp £-0.5
The integrals from -o> top/ and -oo top2 are the cumulative distribution of the standardized normal
distribution and can be determined from tables. Each side of the area source can now be
integrated independently. Figure 2 provides a simplified graphical representation of the
integration by sides. In the actual integration only those portion of the sides that contribute to the
concentration at the receptor are included, that is IX*)/^*) I <4.0. This criteria is a function of
source-receptor geometry, wind direction, and atmospheric stability. f(x) has the following forms:
For stable conditions or unlimited mixing:
/(*) =
exp
+ exp
In unstable or neutral conditions and if cz is greater than 1.6 times the mixing height, L, the
distribution below the mixing height is uniform with height.
/•>
/(*) =
In all other unstable or neutral conditions, that is if az is less than 1.6 time L:
/(*) =
exp
= -00
-Q5(z-H+2NL)2
exp
This infinite series converges rapidly, and evaluation with the integer, N, varying from -4 to +4 is
usually sufficient.
AREA REPRESENTATION ALGORITHM FOR OPEN PIT MINING
The method for modeling particulate impacts from open-pit, surface coal-mine activities is
based on the analysis by Perry et al. (1994) and Thompson (1994a, 1994b) of two wind-tunnel
studies recently performed at the EPA Fluid Modeling Facility in which the impacts of emissions
-------
Wind Direction
Wind Direction
I
5
13
Alongwind Direction Receptor
Xm*
Alongwind Direction
Wind Direction
Alongwind Direction
Alongwind Direction
Figure 2 Graphical representation of the integration by sides. The shaded areas represent a positive contribution. The
horizontal-line pattern represents a negative contribution. The positive and negative contribution cancel everywhere
except the area source.
from a variety of scale model open pits were examined. Perry et al. concluded that, due to
recirculations within the pit, particles that escape show a tendency to be emitted toward the
upwind portion of the pit opening. The pit-mining algorithms determine the escape fraction (for
each particle size group); and establish the location and size of the sub-area from which the
escaped paniculate is modeled with ISC3. It is important to preface that the open-pit algorithms
are designed only for the modeling of impacts outside of the pit caused by emissions within the
pit, that mines are assumed generally rectangular in shape, and that the emissions within the pit are
assumed to be at least somewhat scattered.
-------
ISC3 first estimates the escape fraction as a function of incident wind speed and the
particle size distribution of material released within the pit (Thompson, 1994b). The computation
of escape fraction is particularly important for modeling large particles susceptible to gravitational
settling and less important for PMjQ analyses where the vast majority of material escapes. As
implemented in ISC3, for each of up to 20 particle size categories, an escape fraction (e) is
computed as:
=1.0/1 + -
where
Ur
a
is the settling velocity,
is the approach wind speed at a height of 10m,
is the proportionality constant in the relationship between flux from the pit
and the product of Ur and concentration in the pit (Thompson, 1994a).
v, is already computed in ISC3 (as is done in ISC2) for each size category. Thompson (1994a)
used laboratory measurements of pollutant residence times in a variety of pit shapes and
determined that a single value of a = 0.029 worked well for all pits studied.
The adjusted emission rate (q,) for each particle size category is then computed as:
9i = e i• • f{ • Q
where q is the total emission rate (for all particles) within the pit,/ is the original mass fraction for
the i-th size category, and e, is calculated from the above equation. The adjusted total emission
rate, for all particles escaping the pit, is the sum of the qt. In summary, the amount of particulate
that leaves the open pit mine and is available to impact air quality downwind is a function of the
total emissions within the pit, wind speed, and the particle size distribution of emitted material.
Secondly, ISC3 computes the location and size of a sub-area of the actual pit opening from
which the pit emissions will be modeled. This sub-area source is assumed to be at surface level.
Given an arbitrary rectangular-shaped pit with an arbitrary wind direction as shown in Figure 3,
ISC3 determines the sub-area source in the following way. Based on the wind direction and pit
orientation, the model determines the upwind sides of the pit; in the case of Figure 3, the upwind
sides are 1 and 2. The model then computes the along wind length of the pit (/). / varies between
the lengths of the two sides of the rectangular pit as follows:
/ = L»(l - Q/90) + W.(Q/90)
where 6 is the wind direction relative to the long axis of the pit (therefore 6 varies between 0° and
90°). / is the scaling factor that is used to normalize the depth of the pit.
Wind direction
W
AW
alongwind
length (I)
Figure 3. The effective area for a wind direction of about 45 degrees from the long axis of
the pit. AL and AW (and thus the area) vary with wind direction.
-------
The model user must specify the average height (H) of emissions from the floor of the pit
and the pit volume (V). With these, ISC3 calculates the effective pit depth (de) and the relative pit
depth (dr) as:
de = V/(L.W)
d, = (d.-H)/l
Based on observations and measurements in the wind runnel study, it was clear that the
emissions within the pit are not uniformly released from the pit opening. Rather, the emissions
showed a tendency (based on smoke releases) to be emitted primarily from an upwind sub-area (or
effective area, Ae) of the pit opening and the sub-area size and location depended on the relative
pit depth and the wind direction. The wind tunnel data indicate that if dr £ 0.2, then the effective
area is about 8% of the total opening of the mine (i.e. ^,=0.08). If dr< 0.2, then the fractional area
increases as:
At = (L0-1.7d'r'3)ln
When dr = 0, the effective area is equal to the total area of the mine opening (i.e. Ae=\ .0).
The specific dimensions of this assumed rectangular sub-area are calculated as a function
of 6 such that (see Figure 3):
AW = dJ-^t'W
and,
AL = A(ec^Q)'L
W is defined as the short dimension of the pit and L is the long dimension; AW is the dimension
of the effective area aligned with the short side of the pit and AL is the dimension of the effective
area aligned with the long side of the pit. In the way of a few examples, if Ae = 0.36 at 9 = 45°,
then AW = 0.6-W and AL = 0.6-L; at the extreme of 6 = 0° (wind along the long axis of the pit),
AW = W and AL = Ae-L; at 6 = 90° (wind along the short axis of the pit), AW = /le-W and AL =
L.
The emission rate, qe, for the effective area is such that;
-------
run scale dimensions oi Hi m in tne along now direction and 4DU m in the cross flow direction
(the wind is perpendicular to the long dimension of the pit, i.e. 6 = 90 degrees). The pit is 45 m
deep and the emissions are from the pit floor (H = 0). Therefore the effective pit depth, de , is
simply 45 m and the relative pit depth, dr , is equal to 45/225 = 0.2. With this relative depth, the
effective area for emissions, Ae , equals 0.08 (8 %), and the dimensions of the area are 18 m in the
alongwind direction and 450 m in the crosswind directions. In addition, the pit-induced initial
dispersion is equal to 45 m/ 4.3 = 10.5 m.
MODEL COMPARISONS WITH WIND TUNNEL
z
o
I
111
o
o
o
o
UJ
N
Q:
O
101
•• ISC2-Full Area
'— ISC3- Fractional Area
— Wnd Tunnel
CROSSWIND WIDTH OF PIT
-300 -200 -100 0 100
CROSSWIND DISTANCE (m)
200
300
Figure 4. Horizontal concentration distributions at the downwind edge of a 225m X 450m
pit with effective depth of 45m and source at the pit floor. ISC2 Full Area is the
distribution from the 1SC2 model and ISC2 Partial Area is from ISC3.
The concentration distribution at the downwind edge of the pit for the wind tunnel
measurements and two modeling scenarios are shown in Figure 4. The results labeled FULL-
AREA are those provided by the ISC2 model. The results labeled FRACTIONAL AREA are
those provided by ISC3. For ISC2 where it is assumed that the entire opening of the rectangular
pit acts as a surface level area source with emissions uniform over that area, the modeled
distribution is nearly three times larger than that observed.
In contrast, the ISC3 model is releasing the same total emissions over a much smaller area
on the upwind edge of the actual pit opening (only covering 8% of the pit opening); the plume is
initially dispersion by pit generated turbulence. ISC3 is showing a slight tendency to
overpredicting the distribution particularly near the center of the pit. However, it is clear that the
modified area-source approach is a great improvement over that of ISC2 for impacts at the
downwind edge of the pit where concentrations are expected to be highest. For large downwind
distance (many times the dimensions of the pit), the ISC2 and ISC3 approaches are not expected to
give very different results.
ACKNOWLEDGMENTS
The authors would like to acknowledge Richard A. Strelitz who did the original development work on the
integration technique used for the new area source algorithm and Roger W. Erode who provided review and
refinements to the area source and open pit algorithms and implementation in ISC3.
-------
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.
REFERENCES
Bjorklund J.R. and Bowers J.F., 1979, User's instructions for the Shortz and Longz computer programs
(Volume I), TR-79-131-01. H.E. Cramer Company, Inc., University of Utah Research Park, POB 8049, Salt
Lake City, Utah.
Bowers, J.F., Bjorklund J.R., and Cheney C.S., 1979, User's guide for the Industrial Source Complex (ISC)
dispersion models, Volume I -User Instructions, EPA-450/4-79-030, U.S. Environmental Protection Agency,
Research Triangle Park, North Carolina.
Burt E.W., 1977, Valley model user's guide, EPA-450-2-77-018, U. S. Environmental Protection Agency,
Research Triangle Park, North Carolina.
Hwang S.T., 1986, Methods for estimating onsite ambient air concentrations, ACS Annual Meeting
Anaheim, California.
Perry S.G., Thompson R.S., and Petersen W.B., 1994, Considerations for modeling small-paniculate impacts
from surface coal mining operations based on wind-tunnel simulations, Eight Joint AMS/AWMA
Conference on Applications of Air Pollution Meteorology, Jan. 23-28, 1994, Nashville, Tennessee,
American Meteorological Society, Boston, Massachusetts.
Petersen W.B. and Rumsey E.D., 1987, User's guide for PAL 2.0 - A Gaussian-plume algorithm for point,
area, and line sources, EPA-600-8-87-009, U.S. Environmental Protection Agency, Research Triangle Park,
North Carolina.
Thistle H. and Londergan R.J., 1989, Review and evaluation of area source dispersion algorithms for
emission sources at superfund sited, EPA-450-4-89-020, U.S. Environmental Protection Agency, Research
Triangle Park, North Carolina.
Thompson, R. S., 1993, Wind tunnel simulations of dispersion from surface coal mines. Internal Data
Report, Fluid Modeling Facility, U.S. Environmental Protection Agency, Research Triangle Park, North
Carolina.
Thompson, R. S., 1994a, Residence time of contaminants released in surface coal mines - a wind-runnel
study. Eighth Joint AMS/AWMA Conference on Applications of Air Pollution Meteorology, Jan. 23-28,
1994, Nashville, Tennessee, American Meteorological Society, Boston, Massachusetts.
Thompson, R. S., 1994b, Personal Communication.
Turner D. B., Workbook of atmospheric dispersion estimates, U.S. Environmental Protection Agency,
Research Triangle Park, North Carolina, pp 84.
U.S. Environmental Protection Agency, 1992, User's guide for the Industrial Source Complex (ISC2)
dispersion models, Volume I -User Instructions, EPA-450-4-92-008a, U.S. Environmental Protection
Agency, Research Triangle Park, North Carolina.
Weber E., 1982, Air Pollution Assessment Methodology and Modeling, Plenum Press, New York.
-------
TECHNICAL REPORT DATA
1. REPORT NO.
EPA/600/A-96/019
TITLE AND SUBTITLE
Improved Algorithms for Estimating the Effects of
Pollution Impacts from Area and Open Pit Sources
5.REPORT DATE
6.PERFORMING ORGANIZATION CODE
7. AUTHOR (S)
William B. Petersen and Steven G. Perry
8.PERFORMING ORGANIZATION REPORT NO.
9. PERFORMING ORGANIZATION NAME AND ADDRESS
Atmospheric Sciences Modeling Division
Air Resources Laboratory
National Oceanic and Atmospheric Administration
Research Triangle Park, 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
14. SPONSORING AGENCY CODE
15. SUPPLEMENTARY NOTES
16. ABSTRACT
The increasing environmental concern over hazardous emissions from landfills, landfills, settling ponds, agricultural
fields, surface mines, and surface impoundments has renewed interest in improving concentration estimates downwind
from area sources. Simplistic and numerically efficient methods (for estimating area source impacts) are the virtual point
source approach (Turner, 1970) as used in the models LONGZ (Bjorklund and Bowers, 1979) and VALLEY (Burt,
1977), and the finite line source approach as used in ISCST (Bowers et al., 1979). The virtual point source and finite
line segment algorithms are most appropriate for estimating concentrations at downwind distances significantly larger
than the side width of the area source (Hwang, 1986 and Weber, 1982). However, these constructs fail for
concentration estimates within or near the area source. Thistle and Londergan (1989) showed that these algorithms
behaved in a manner inconsistent with mathematical and physical principles.
Shortly after the Thistle and Londergan study, the Office of Air Quality Planning and Standards in the U.S.
Environmental Protection Agency (EPA) decided to update the area source algorithm in ISC2 (U.S. EPA, 1992) using
the approach contained in PAL. At about the same time the Office of Research and Development in EPA was involved
in a project to improve the computational efficiency of the area source algorithm in PAL. It is more efficient version of
the PAL area source algorithm that has been implemented in ISC2 and will be described in this paper. In addition, the
improved area source algorithm is utilized in a new approach to modeling open-pit sources.
17.
KEY WORDS AND DOCUMENT ANALYSIS
DESCRIPTORS
b. IDENTIFIERS/ OPEN ENDED
TERMS
c.COSATI
18. DISTRIBUTION STATEMENT
RELEASE TO PUBLIC
19. SECURITY CLASS (This Report)
UNCLASSIFIED
21.NO. OF PAGES
20. SECURITY CLASS (Ibis Page)
UNCLASSIFIED
22. PRICE
------- |