EP A/600/A-96/079
96-TA35.01
Progress in Developing an Open Burn/
Open Detonation Dispersion Model
J.C. Weil
University of Colorado
Boulder, Colorado
B. Templeman
NOAA
Boulder, Colorado
W. Mitchell
U.S. Environmental Protection Agency
Research Triangle Park, North Carolina
Air & Waste Management
ASSOCIATION
For Presentation at the
89th Annual Meeting & Exhibition
Nashville, Tennessee
June 23-28,1996
~
Since 1907
-------
96-TA35.01
INTRODUCTION
Obsolete or unwanted munitions, rocket propellants. and manufacturing wastes require
treatment at Department of Defense (DOD) and Department of Energy (DOE)
facilities. One of the most widely-used treatment methods is open burning (OB) and
open detonation (OD) of the material. Currently, the material destroyed in a single
detonation generally ranges from 100 to 5000 lbs. while the quantity removed by a burn
can be larger and last from minutes to an hour. OB/OD operations are confined to
daytime with atmospheric stability conditions ranging from convective or highly unstable
to near neutral.
OB/OD activities produce air pollutants and require predictions of pollutant concen-
trations to obtain an operating permit. The pollutants include SO2, NOx, particulates,
volatile organic compounds and toxic materials such as metals, seniivolatile organics,
etc.1 Large detonations also generate large quantities of dust/soil that are entrained by
the rising contaminant cloud. OB/OD sources differ from most traditional air pollution
sources in that they have: 1) instantaneous or short-duration releases of buoyant
material rather than continuous releases, and 2) ambient exposure times for the clouds
that can be much less than the typical averaging times ( > 1 hr) of air quality standards.
Atmospheric dispersion models are used to estimate pollutant concentrations given
information on the source and meteorological conditions. However, there is currently 110
recommended EPA dispersion model to address the unique features of OB/OD sources.
The most commonly-used approach is INPUFF,2 a Gaussian puff model, but this has
several limitations as briefly discussed below and elsewhere.3 As a result, a model
development program was initiated under the DOD/DOE Strategic Environmental
Research and Development Program.
This paper briefly summarizes the model development effort which is divided into
''operational" and "research" components. In the following, we give a brief discussion
of the background and overall model design and then describe the operational model
components for instantaneous (OD) and short-duration (OB) sources: the OD model is a
Gaussian puff approach whereas the OB framework consists of integrated-puff and plume
models. The combined OB/OD mode! includes: 1) a continuous treatment of dispersion
as the release condition varies from instantaneous to continuous. 2) cloud and plume
rise obtained from appropriate entrainment models, 3) cloud and plume penetration
of elevated inversions, 4) relative (puff) and total dispersion based on modern scaling
concepts for the planetary boundary layer (PBL), and 5) a capability for the use of
onsite profiles of wind, temperature, and turbulence from a mobile meteorological
platform. The current OB/OD model focuses on the unstable PBL.
2
-------
96-TA35.01
BACKGROUND AND MODEL DESIGN
Background
The development of an OB/OD dispersion model has considered: 1) the limitations of
existing models, 2) current knowledge of turbulence and dispersion in the PBL, and 3) a
mobile meteorological platform under development.
Limitations of existing models. INPUFF has been used to model OB/OD sources
and can handle dispersion from individual puffs or clouds or from a sequence of puffs in
a short-duration release. Although the Gaussian puff approach is suitable for OB/OD
sources, INPUFF has the following limitations: 1) It adopts dispersion parameters
(<7y,tjz) from the Pasquill-Gifford (PG) curves. 2) It includes Briggs' (Eef. 4) plume
rise expressions which apply to continuous releases rather than to instantaneous sources
(puffs, clouds) and does not address thermal penetration of elevated inversions capping
the PBL. 3) It assumes Gaussian velocity statistics for the turbulence, whereas the
vertical velocity statistics in the unstable PBL are positively skewed.-" The skewness
should be included for vertical dispersion.
For OB/OD sources, the PG curves are deficient in that they: 1) are based on dispersion
from a ground-level source and short downwind distances (< 1 km), and 2) are selected
using surface meteorology, which does not account for the PBL's vertical structure. For
large detonations, source buoyancy can carry emissions to several 100 m or the PBL top;
one must then deal with dispersion over the entire PBL.
PBL turbulence. Dispersion in the PBL depends on the turbulence length and
velocity scales which differ for the unstable or convective boundary layer (CBL) and
the stable boundary layer (SBL). For the CBL, the length and velocity scales are the
CBL depth h and the convective velocity scale'k;,; w. = (gwff0h/Qa. where g is the
gravitational acceleration. wQ0 is the turbulent heat flux at the surface, and Qa is the
ambient potential temperature. Typical values of wm and h at midday over land are 1 - 2
m/s and 1 - 2 km. Within the "mixed layer" (O.lh < z < h), the mean wind speed and
turbulence components- longitudinal au. lateral cx^ ~ 0.6tf„.
For the SBL. the turbulence is much weaker with eddy sizes proportional to 2 near the
surface and typically ~ 10s of meters or less in the upper part of the SBL. Models and
observations show that the velocity scale is the friction velocity u. (Ref. 5), which is
typically ~ 0.1 m/s in strong stable stratification.
Knowledge of the PBL turbulence structure has been included in a number of models for
air quality applications.6 One example is AERMOD7 for industrial source complexes.
Mobile meteorological platform. A mobile meteorological platform is being
developed at NOAA's Environmental Technology Laboratory to obtain the PBL
variables necessary for modeling since many DOD facilities are in remote locations.
The platform design includes: 1) a radar wind profiler for obtaining the three wind
3
-------
96-TA35.01
components up to ~ 3 km. 2) a radio acoustic sounding system (EASS) for temperature
measurements. 3) a tnini-SODAR for measuring winds and aw to a height of ~ 200 m,
4) a mini-lidar system for obtaining the PBL depth h, and 5) a portable meteorological
station for measuring near-surface winds, temperature, turbulence, and heat flux. The
dispersion model is being designed for efficient use of these measurements.
Overall Model Design
The overall model design includes: 1) a simple computational or operational framework
for routine problems, and 2) a more detailed or research model for nonroutine problems.
In the operational approach, a Gaussian puff model is adopted for instantaneous sources
and puff, integrated-puff. and plume models for short-duration releases. For the research
framework, a Lagrangian particle and/or puff approach is planned. Both frameworks
will be considered for "onsite" use in a real-time operational mode using data from the
mobile meteorological platform, i.e., for day-to-day decisions on OB/OD operations. The
operational puff and plume models would be used for climatological analyses needed in
risk assessments.
In modeling, the important aspects to address are: 1) all source-related features
including the instantaneous or short-duration nature of the release, buoyancy-induced
rise and dispersion, and cloud or plume penetration of elevated inversions, 2) relative
and absolute dispersion expressions that explicitly include PBL turbulence variables. 3)
meteorological variables including their vertical profiles from the mobile platform, and 4)
a treatment for puff and plume dispersion about complex terrain.
The models discussed in the following sections address points 1 and 2 above and must
be expanded to include points 3 and 4. Further development also will address: 1) a
more complete description of initial source effects (detonation cloud size and height)
and inversion penetration, 2) a more complete PBL turbulence parameterization. 3)
averaging time effects on concentration, 4) the entrained dust source term, and 5}
deposition of gases and particles.
INSTANTANEOUS SOURCES
Dispersion Model
In many OB/OD applications, estimates of peak ground-level concentrations (GLCs)
are required for averaging times ranging from seconds or minutes to an hour or longer.
In making such estimates, we must account for the stochastic or random nature of
turbulence and dispersion in the PBL. That is. we must recognize that the observed
concentration at a receptor is a random variable and should be predicted statistically
through a probability distribution.8,9 The distribution can be parameterized by a
functional form such as a gamma or clipped-normal distribution9 and requires a
minimum of two variables to characterize it—the mean concentration C and the root-
mean-square (rms) concentration fluctuation crc, which is a measure of the width of
the distribution. The peak concentration then can be defined by a specified percentile
value of the cumulative probability distribution, e.g.. the 99.9th percentile level. In the
4
-------
96-TA35.01
following, we discuss approaches for predicting C. ac. cloud rise, and the dispersion
parameters: the functional form of the distribution remains to be selected.
Mean concentration. For instantaneous sources or detonations, a Gaussian puff model
is adopted for predicting the short-term mean concentration (C) field relative to the puff
centroid. The C is the expected or average concentration that would be observed if the
same experiment—same source and meteorological conditions—were repeated a large
number of times.9 The C is given by:
r_ Q
(27r)3/2arIcrry<7r2 P
where Q is the pollutant mass released, U is the mean wind speed, t is the travel time.
he is the effective puff height, and arx, ary. and oTZ are the puff standard deviations
or relative dispersion parameters in the x, y, and 2 directions, respectively. Here, he =
hs + Ah(x) where hs is the source height which is generally zero for OB/OD sources, and
Ah is the cloud rise due to buoyancy; x and y are the distances in the mean wind and
crosswind directions.
The maximum mean concentration Cc at the puff centroid is given by Cc =
Q/[(2n)3^2arxaryarz], where the relative dispersion parameters are generally different,
in the three directions. Iirthe following, we assume arx = ary = arz = aT.
The C field including puff meandering is given by Eq. (1), but with crrx, <7ry, oTZ
replaced by the absolute dispersion parameters—crx,ay,az. A Gaussian distribution
for C is applicable in the SBL where the probability density function (p.d.f.), pw, of
the vertical velocity w is Gaussian. However, for the CBL, a skewed w p.d.f. is more
consistent with laboratory and field data. A skewed p.d.f. is adopted here and is
parameterized by the superposition of two Gaussian distributions.10
The C field due to an ensemble of meandering puffs is derived from pw following the
same approach as applied to continuous plumes.10 The resulting expression for C is
r, Q ( (x-Ut)2 y2 \ ^ Xj ( (z-he-z~)2\
—h j' <2)
where aZJ = oWJxll) and Jj = Wjx/U with j = 1,2. The \3. Wj, and aWJ (j = 1.2)
are the weight, mean velocity, and standard deviation of each Gaussian p.d.f. comprising
pw. Equation (2) applies for short distances such that the plume interaction with the
ground or elevated inversion is weak. The complete expression for C includes multiple
cloud reflections at the ground and PBL top.
The time-averaged concentration can be found from the dose where the partial dose is
defined by xp(x,y,z.t) = /J C(x,y, z,t')dt' and the total dose by xboo — i!){x,y, z.oc).
For clouds with short passage times over a receptor, the average concentration C can be
obtained from C = (^(<2) — where the averaging time Ta = t2 — 11. If the puff
passage time Aarx/U is less than Ta, then C = ipoc/Ta-
~ Ut)~ _ JT_ _ {z - he)
2<7rx 2arv 2°tz
5
-------
96-TA35.01
ftms concentration fluctuation. For elevated sources such as a buoyant cloud, the
rms concentration fluctuation at the ground is largest close to the source and decreases
with increasing distance.9,11 As a first step in predicting crc, we adopt Gifford's12
meandering plume or cloud model which applies to the near-source region: specifically,
it applies when the large-scale eddies in the PBL cause the wandering or meandering
of a small growing plume or cloud. The cloud only occasionally reaches the ground,
but it does so in high concentration due to the small local spread oy; thus, it creates
large "spikes" in the GLC distribution and a large concentration variance.11 In the
following, we model the mean square concentration (c2). where the angle brackets denote
an ensemble average, and then find ac from ac = ((c2) - C2)1//2 (see Ref. 9).
In Gifford's model, the instantaneous concentration distribution in a cloud is assumed
to be a nonrandom, axisymmetric Gaussian distribution about the instantaneous cloud
centroid xc = (xc,yc,zc):
c{x,xc) =
Q
(2tt)3/2ct^
exp
(x - xc)2 (y - yc)2 (z - zc)
21
2 <72 2
-------
96-TA35.01
where aie = {a; + 2a*)1/2.
-------
96-TA35.01
currently underway in a salt-stratified tank at the EPA Fluid Modeling Facility in North
Carolina.
Dispersion parameters. For clouds, ar is dominated by entrainment for short times
with ar = orb = O.I8A/1. where subscript b denotes buoyancy-induced spread. At
intermediate times (t < Ti), the ay may be dominated by ambient turbulence in the
inertial subrange with 0.1/i: e ~ QAwl/h,
,, and Ti ~ 0.7/i/w* (Ref. 10). These approximations coupled with € =
ba^/Ti lead to b = 0.78. To satisfy the long-time oTa limit, we must have a2 = 0.62«i;
di is estimated to be 0.57 from Thomson's two-particle model results. The resulting
parameterization for ora in the CBL is
(Tra 0.36A'3/2 W.X
ir = rrosix w,th x = ~u(9)
where we have assumed t = x/U and A is a dimensionless distance or travel time.
To connect the short-, intermediate-, and long-time relative dispersion regimes in a
continuous manner, we adopt the following parameterization: of - afb + ofa. For clouds
dominated by buoyancy, oTb = 0A2F^4t1^2.
The total or absolute dispersion is necessary to estimate the C for a meandering puff or
plume. The ax and ay in Eq. (2) can be obtained from a parameterization of Taylor's
theory: ax = outj{\ + t/2Tix)1^2 and similarly for ay. The Tix is the Lagrangian time
scale for the u component and can be parameterized by Tlx oc oujh, etc. (e.g., see Ref.
6). For the CBL and the results below, we use Tix — Tiy = 0.7/i/ty. and au = ov —
0.6 w..
Example Results
We have computed the Cc in the buoyant puff and the mean GLC along y = 0 due
to a meandering puff for 0.1 < W < 50 tons. The ay. ax, and oy were calculated as
described in the previous section. In the following, the cloud buoyancy is characterized
by its dimensionless value
in the examples below, we use w, = 2 m/s. h = 1000 m. and U = 5 m/s.
Figure 1 shows calculated values of the peak (Cc) SO2 concentrations in a detonation
cloud. The cloud SO2 mass is estimated from Q = IV • Ej. where Ef (= 2.23 x 10—4;
8
-------
96-TA35.01
Andrulis1) is the SO2 emission factor. At small x. all of the curves have the same slope:
Cc x x-V2 because arb oc x1//2 and the ar is dominated by near the source. For
200 m x ^ 2000 m, some curves exhibit a short region of a nearly constant
with x: this is due to puff trapping in the CBL and a temporary limitation on vertical
dispersion due to the elevated inversion. At large distances (x > 10 km), clouds for all
cases become uniformly mixed in the vertical but continue to spread laterally. Thus, the
Cc tends towards Q/u2a oc Q!x\ the curves for IV = 10-50 tons show that Cc varies
inversely with x for large distances.
Figure 2 shows the mean GLC C along the puff centerline (y = 0) for the same range
of W and Ft values as in Fig. 1. This mean is for an ensemble of meandering puffs and
is obtained from Eq. (2) with reflection terms included. Several interesting features are
found: 1) A non-monotonic variation occurs in the maximum GLC Crn with W and Fx-
2) The variation in Cm for 0.1 < W < 50 tons is only about a factor of 4 even though
the range in Q is a factor of 500; the weak dependence on Q is attributed to the increase
in Ah with Ft- 3) The Cm is of the order of 0.1 /ig/m3. which is the lower bound for Cc
in Fig. 1. 4) The increase in the distance to the maximum concentration with W is due
to the increase in Ah with Ft or W.
For ac > C, the mean GLC along y = 0 due to an ensemble of meandering puffs
probably has little to do with the observed centerline GLC in an individual puff. This is
attributed to the large variability in individual concentration observations when ac > C.
As noted earlier, the computed C in Fig. 2 would be used together with a modeled oc in
a concentration probability distribution to estimate the peak short-term GLC that could
occur downstream of the detonation.
Some preliminary calculations of ac/C at the distance of the maximum GLC were made
for the examples in Fig. 2. For W — 0.1, 1, and 5 tons, the corresponding oc/C was 6. 2.
and 1.8; for W = 10 to 50 tons, the ocjC < 1 and was decreasing rapidly with increasing
distance. The rapid decrease was attributed to the large cloud size (aT) relative to h and
the approach of the concentration distribution to a vertically well-mixed value. The oc
model requires generalization to address the ac at large distances: x > (2 — 3)Uh/w»,
or about 5 - 7.5 km in the current example. This will be pursued in the future.
SHORT-DURATION RELEASES
Dispersion Model
For short-duration releases or burns, our general approach is an integrated puff model in
which the short-term mean concentration relative to the puff centerline is
C
I
tr Qr f{x, 1.1') dt'
0 {^TT^/'^O'Tx&ryGr z
(11a)
/ = exp
(x-U(t-t'))2 y2 (z - he)''
2ffr2x
2<
2 a?z
(11 b)
9
-------
96-TA35.01
where x — (x.y.z). t,' is the puff emission time. t.r is the total release duration. Qr is
the continuous source emission rate. oTX — (7rx(t - t'). and similarly for ary.arz. The
integration in (11a) can be carried out analytically for limiting forms of arx{t - t'). etc.,
but must be done numerically in general.
The integrated puff model also is used for estimating the mean concentration due to a
sequence of meandering puffs by replacing the relative dispersion (crrx, etc in Eq. 11) by
the absolute dispersion—ax, cry, az.
In the following, we calculate the Cc for a short-duration release by numerically
integrating Eq. (11) with orx — ary = aTZ = ar = a\el^2{t - t')3^2/(1 + «2{t - t')jTi).
Similarly, we compute the GLC C for the same release but for the meandering puffs with
ax — ay — 0.6w.(t - t')/(I +0.5(t — t')/Tix)1//2 and Tix = 0.7h/w,. For an infinitely-long
release, the Cc and C values should reduce to those for a continuous plume as shown
below; thus the plume results serve as an upper bound to the concentration values for
the short-duration release.
The mean concentration field relative to a plume centerline is given by
C= expf-4-ii^I). (12)
2irb oryorz \ 2a ,2y 2cr22 /
Here, the plume rise is attributed to buoyancy and is given by A/i = 1.6F^x2^3/U
and its radius is r = 0.4A/i (Ref. 17). Source momentum effects can be included in the
future. As with the puff model, we will assume crry = arz = aT and a;? = afb + afa.
The ora is given by Eq. (11) and the plume <7r6 = r/\/2 = QAbF^2x2^3/U. The Cc =
Qr/(2nUcr2) from Eq. (12); these expressions are expanded to include reflection at z =
0, h.
Example Results
Results are presented for the diinensionless concentration CcUh2/Qr for the plume and
integrated-puff models, with reflection at z = 0,/i included in both. The buoyancy of the
continuous source is characterized by the dimensionless buoyancy flux (Ref. 10):
F. = 77% ' (13)
u wja
where F,b is the so\irce buoyancy flux which is proportional to the source heat flux.
Figure 3 shows the CcUh2 jQr as a function of X for the plume model with F, in the
range 0.001 < F. 0.3. TIig large vciricitiion in tlie dimensionless short r£in.ge ^
1) is due to the buoyancy-induced dispersion or^. As can be seen, CcUh2/Qr decreases
systematically and significantly with an increase in F, due to the increase in arb with
F,. For X > 1, the curves converge to the same limit because at long times the ar is
dominated by ora. which is independent of F„.
Figure 4 presents the CcUh2/Qr for both the plume and integrated-puff models for F, =
0.01 and various values of trm = trw,/h. the dimensionless release duration. The time
10
-------
96-TA35.01
scale h/w. = 500 s for the w, (= 2 m/s) and h (= 1000 ni) used here, so that tT ranges
from 50 s to 500 s or about 1 to 8 min. As can be seen, the plume result (solid curve) is
an upper bound to the integrated-puff model results. The X value corresponding to the
departure of the integrated-puff solution from the plume solution increases as tT, does.
For X > 5. the integrated-puff concentrations can be significantly less than the plume
concentrations; this is due to the finite duration of the source.
Figure 5 shows the dimensionless mean GLC, CUh2/Qr, for the meandering plume
and for a sequence of meandering puffs (finite-duration release). The trends with X
and tr. are similar to those in Fig. 4, but the near-source variation of CUh2/Qr with
X differs from that for Cc due to the different dispersion rates in relative (Fig. 4) and
absolute dispersion (Fig. 5). Note that at X — 0.1, the GLC value is about an order
of magnitude smaller than the Cc (Fig. 4) because the ground is significantly removed
from the elevated plume centroid. However, at X = 10, the concentration values are
quite similar; this occurs because the concentration distribution becomes uniform in the
vertical due to plume trapping and the horizontal spreads (relative and absolute) are
about the same.
ACKNOWLEDGMENTS
This work has been supported by the DOD/DOE Strategic Environmental Research and
Development Program. We are grateful to Seth White for producing the figures.
REFERENCES
1. Andrulis Research Corporation, Development of methodology and technology
for identifying and quantifying emission products from open burning and open
detonation thermal treatment methods. Bangbox test series, Vol. 1, Final Rpt., U.S.
Army Armament. Munitions, and Chemical Command, Rock Island, IL, 1992.
2. W.B. Petersen, "A demonstration of INPUFF with the MATS data base," Atmos.
Environ.. 20: 1341 (1986).
3. J.C. Weil, B. Templeman, R. Banta, and W. Mitchell, "Atmospheric dispersion model
development for open burn/open detonation emissions," in Proceedings AWMA
88th Annual Meeting & Exhibition. 95-MP22A.05, Air and Waste Management
Association, Pittsburgh, PA, 1995.
4. G.A. Briggs. "'Some recent analyses of plume rise observations." in Proceedings of
the Second International Clean Air Congress, H.M. Englund and W.T. Beery, Eds..
Academic Press, NY. 1971, pp 1029-1032.
5. J.C. Wyngaard. ''Structure of the PBL." in Lectures on Air Pollution Modeling, A.
Venkatram and J.C. Wyngaard, Eds.. Amer. Meteor. Soc.. Boston, 1988. pp 9-61.
11
-------
96-TA35.01
6. A. Venkatram and J.C. Wyngaard. Eds.. Lectures on Air Pollution Modeling. Anier.
Meteor. Soc.. Boston. 1988.
7. S.G. Perry, A.J. Cimorelli, R.F. Lee, R.J. Paine, A. Venkatram, J.C. Weil, and R.B.
Wilson, "AERMOD: A dispersion model for industrial source applications." in
Proceedings AWMA 87th Annual Meeting & Exhibition, 94-TA23.04, Air and Waste
Management Association. Pittsburgh. PA. 1994.
8. G.T. Csanady, Turbulent Diffusion in the Environment. Reidel. Dordrect, 1973.
9. J.C. Weil, R.I. Sykes, and A. Venkatram. 'Evaluating air-qualit.v models: review and
outlook." J. Appl. Meteor.. 31: 1121 (1992).
10. J.C. Weil, "'Dispersion in the convective boundary layer," in Lectures on Air
Pollution Modeling, A. Venkatram and J.C. Wyngaard, Eds.. Amer. Meteor. Soc.,
Boston, 1988, pp 167-227.
11. R.I. Sykes, "Concentration fluctuations in dispersing plumes," in Lectures on Air
Pollution Modeling, A. Venkatram and J.C. Wyngaard. Eds., Amer. Meteor. Soc..
Boston, 1988, pp 325-356.
12. F.A. Gifford, "Statistical properties of a fluctuating plume dispersion model." Adv.
Geophys., 6: 117 (1959).
13. R.S. Scorer, Environmental Aerodynamics, Halsted Press, NY. 1978, pp 276-303.
14. J.C. Weil, "Source buoyancy effects in boundary layer diffusion." in Proceedings of
the Workshop on the Parameterization of Mixed Layer Diffusion, R.M. Cionco, Ed.,
Physical Science Laboratory, New Mexico State University. Las Cruces, NM. 1982.
pp 235-246.
15. J.R. Richards, "Experiments on the penetration of an interface by buoyant
thermals," J. Fluid Mech., 11: 369 (1961).
16. D.J. Thomson. "A stochastic model for the motion of particle pairs in isotropic high-
Reynolds-number turbulence, and its application to the problem of concentration
variance," J. Fluid Mech., 210:113 (1990).
17. G.A. Briggs, "Plume rise and buoyancy effects," in Atmospheric Science and Poiver
Production, D. Randerson. Ed.. U.S. Dept. of Energy DOE/TIC-27601, 1984. pp
327-366.
12
-------
96-TA35.01
1000
w (tons):
0.0029 0.1
0.029 1
0.15 5
0.29 10
100
— 0.73
1.46
a>
W-0.1
0.1
100 1000 10000
x (m)
Figure 1. SO2 concentration at detonation cloud centroid as a function of downwind
distance and detonation mass W.
W=50.0
W-0.1
0.001
100
1000
10000
x (m)
Figure 2, Mean ground-level SO2 concentration along puff centerline as a function of
downwind distance and detonation mass; see Fig. 1 for key to lines.
13
-------
96-TA35.01
1000
F.=0.QQ1
0.001
0.003
0.01
0.03
100
w
o
sz
D
C
o
0.3
O
0.1 0.2 0.5 1 2 5 10
X
Figure 3. Dimensionless concentration at plume centroid as a function of dimensionless
downwind distance and dimensionless buoyancy flux F„.
NOTE TO EDITORS
Under the new federal copyright law,
publication rights to this paper are
retained by the author(s).
14
-------
96-TA35.01
000
Plume
Puff. L
F.=0.01
100
0.2
0.5
o
CM
O
0.01
0.1 0.2
0.5
5
10
2
1
X
Figure 4. Dimensionless concentration at plume or puff centroid as a function of
dimensionless downwind distance and dimensionless release duration tTm.
1000
Plume
F.=0.01
100
0.2
0.5
o
eg
_c
ZD
O
0.01
0.5
10
0.1 0.2
5
1
2
X
Figure 5. Dimensionless mean ground-level concentration along plume or puff centerline
as a function of dimensionless downwind distance and dimensionless release duration
15
-------
• ' TECHNICAL REPORT DATA
(Pteaie reed Instruction on the reverie before completing)
"" \
1. REPORT NO. , 2.
EPA/600/A-96/079
3. RECIPI
4. TI.TLE-ANO SUBTITLE \ . , _
, . -Progress in Developing an Open. .
„ ' •' y. Burn/Open. Detonation Dispersion Model -
S. REPORT DATE
6. PERFORMING ORGANIZATION CODE
7. AUTHORISI
Jeff C. Weil and William J. Mitchell
8. PERFORMING ORGANIZATION REPORT NO.
9 PERFORMING organization name and address
NOAA/ETL
325 Broadway
Boulder, CO 80309
10. PROGRAM ELEMENT NO.
NA
11. CONTRACT/GRANT NO.
IAG-DW-13936780
12. SPONSORING AGENCY NAME AND ADDRESS
EPA/NERL
MD-77B
Research Triangle Park, NC 27711
13. TYPE OF REPORT AND PERIOO COVERED
14. SPONSORING AGENCY CODE
IS. SUPPLEMENTARY NOTES
y
16. ABSTRACT /
Open burning (OB) and open detonation (OD) are the
traditional means to dispose of unwanted energetic materials such
as explosives, rocket propellants and manufacturing wastes
containing these materials <;;>This paper presents the results
achieved to date from a multi-agency study focused on improving
our capability to predict accurately the impact which OB/OD
activities have on the health of humans and ecosystems^? This
multi-agency study, which involves the U. S. Environmental
Protection Agency, the National Oceanic -and—Atmospheric
Administration and the Department-of* Defense (DoD), is being
sponsored by DoD's Strategic^fEnvironmental Research and
Development Program (SERDP) .^The primary objectives of the study
are: (1) to develop and validate air pollution dispersion models
which predict how the emissions from OB/OD activities will
disperse in the environment; and (2) to develop a mobile
meteorological measurement system which can characterize the
atmospheric conditions from ground level to 3000 m above ground
level
17. Y ) KEY WORDS AND OOCUMENT ANALYSIS
a. w DESCRIPTORS
b.IDENTIFIERS/OPEN ENDED TERMS
c. COSATi Field/Croup
IS. DISTRIBUTION STATEMENT
1#. SECURITY CLASS (This Report)
21. NO. OF PAGES
20. SECURITY CLASS (This page)
22. PRICE
f PA Farm 2220-1 (R»». 4-77) pkevioui Coition i» oiiOl(tc
------- |