EPA 903/9-75-019
DECEMBER, 1976
WIND TUNNEL MODELING STUDY OF
THE DISPERSION OF SULFUR DIOXIDE IN
SOUTHERN ALLEGHENY COUNTY, PENNSYLVANIA
\] -^ f... .'. -
U.S. Environmental Protection Agency
Region ffl
Philadelphia, F EPA Report Collection
Information Resource Center
US EPA Region 3
Philadelphia, PA 19107
-------
-------
I
I
EPA 903/9-75-019
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
WIND TUNNEL MODELING STUDY OF THE
DISPERSION OF SULFUR DIOXIDE IN SOUTHERN
ALLEGHENY COUNTY, PENNSYLVANIA
Prepared by
G.R. Ludwig and G.T. Skinner
Li c fir' r:r> '-';-! P,-;fe'i?
»\ ij>. -;c-r,iv,at!&n Resourca
I!'.! C:;>:t!.ut Street
Prepared for: h, jl1;;!^, PA 1SI07
U.S. Environmental Protection Agency
Region HI
Philadelphia, Pennsylvania 19106
December 1976
Calspan Corporation
P.O. Box 235
Buffalo, New York 14221
-------
I
I
I
I
I
I
I
I
I
I
I
f
I
I
I
I
I
i
I
ACKNOWLEDGMENT
The authors wish to thank Dr. Peter Finkelstein, EPA Region III
Meteorologist, who was EPA Project Officer in this work. His understanding of
the problems posed by this program and his guidance and help in their solution
were invaluable. We are also indebted to Mr. Harry Geary of H.E. Cramer
Company Inc. and to Mr. Ron J. Cheleboski, Director of the Allegheny County
Bureau of Air Pollution Control for their timely assistance in providing in-
formation required for the design of both the physical model and the emission
inventory model. Finally, we wish to thank Dr. William H. Snyder of the EPA
Environment Sciences Research Laboratory for his suggestions on the prepara-
tion of this report.
111
-------
I
i
i
i
I
I
i
i
i
i
I
i
i
i
i
i
i
i
i
ABSTRACT
This report presents the results of a wind tunnel model study to
determine the ground-level SO,, concentrations in the Clairton Area of Allegheny
County, produced by emissions from stationary sources within the area. The
study was designed to provide data under flow conditions which correspond
roughly to those which prevailed during an air pollution episode at Liberty
Burough, Clairton in January 1973. The inventory of emission sources used in
the tests corresponded to the estimated 1973 full-scale inventory for 49
sources which were located within the confines of the modeled area.
The test program included flow visualization studies and quantitative
measurements of ground-level concentrations of a tracer gas from which full-
scale SO,, concentrations could be calculated. The concentration level measure-
ments were performed for four wind directions. At each wind direction, three
wind speeds were studied with neutrally stable flow, and for one of these a
temperature inversion condition was also tested.
The report presents details of the model scaling laws, the test
facilities, the model, test procedures, and the experimental results. The
results include measurements of the velocity and temperature profiles at
several locations above the model as well as ground-level concentrations. The
latter are presented in terms of full-scale SO,, concentrations.
-------
I
I
I
I
I
I
I
I
I
I
I
1
I
I
I
I
I
I
I
TABLE OF CONTENTS
Section Page
I INTRODUCTION 1
II SCALING CRITERIA 3
A. Modeling of Atmospheric Winds 4
B. Modeling of Stack Emissions 6
1. Exit Momentum Scaling 7
2. Buoyancy Scaling 9
3. Pollutant Concentration Scaling 12
C. Sampling Time 14
D. Summary 16
III TEST FACILITIES 17
A. The Atmospheric Simulation Facility 17
B. Pollutant-Concentration Measuring System 18
C. Auxiliary Equipment 22
IV WIND TUNNEL MODEL 24
A. Model Scale 24
B. Model Construction 28
C. Emission Inventory and Source Locations 29
D. Stack Emission Model 30
V DESCRIPTION OF TEST PROCEDURES 37
VI EXPERIMENTAL RESULTS AND DISCUSSION 41
A. Model Atmospheric Wind Conditions 41
1. Approaching Flow 41
2. Flow Over the Model 42
B. Flow Visualization 46
C. Ground Level Concentrations 48
1. Preliminary Near Field Test 48
2. S02 Concentrations 50
3. Summary 60
VII
-------
TABLE OF CONTENTS (Cont'd)
Section Page
VII CONCLUDING REMARKS 62
APPENDIX A - Minimum Permissible Ground Roughness
Reynolds Number 65
APPENDIX B - Table of Ground Level S02 Concentrations
Calculated from Model Test Data 67
REFERENCES 127
Vlll
-------
I
I
I
I
I
I
1
I
I
I
I
I
I
I
I
I
I
I
I
I. INTRODUCTION
This report presents the results of a model study to determine the
dispersion of stack emissions over a strip of land located near the Monongahela
River in the Clairton area Southeast of Pittsburgh. The objective of the study
was to determine ground level concentrations of the stack effluents from steel
mills and power plants which are located in the area. Both qualitative flow
visualization studies and quantitative concentration level measurements were
made.
The model study was performed in the Calspan Atmospheric Simulation
Facility (ASF). This specialized wind tunnel, which is described in the text,
was designed for the specific purpose of modeling the wind in the lower atmos-
phere. During the course of the current program, the ASF was modified to in-
clude the capability of generating an elevated inversion layer over the model.
The test program included studies under neutrally stable flow conditions and
studies in the presence of the elevated inversion. The inventory of stack
emissions used in the model tests corresponded to the estimated 1973 full-
scale emission inventory for forty-nine of the stacks which fell within the
confines of the modeled area.
Tests to determine ground level SO- concentrations were performed
for three wind velocities in neutral flow and one wind velocity with an
elevated inversion, for each of four wind directions. In addition, the model
emission inventory was operated in separate sub-groupings of sources in order
to optimize the sensitivity of the concentration measurement system and at the
same time allow identification of the separate contributions from various groups
of stacks. In all, a total of forty tests were performed to determine ground
level concentrations of SO,, under the various wind and stack combinations.
-------
Each test provided quantitative concentrations of S02 at twenty
ground level sampling points. In all of the tests, two of the sampling point
locations were held fixed at locations corresponding to full-scale continuous
monitoring stations at Glassport and Liberty Borough- The remaining eighteen
sampling points were located on the basis of flow visualization studies to
provide a satisfactory distribution of samples for determining overall ground
level concentrations of SO . The inclusion of the two full-scale monitoring
stations in the sample array for all tests allows an assessment of the
effectiveness of these monitoring stations as pollution alert monitors.
In the sections that follow, the scaling criteria for the model tests
are presented first. This is followed by descriptions of the test facilities,
the wind tunnel model, and the test procedures. Next, the experimental re-
sults are presented in three steps, First, the flow conditions over the model
are described and then the results of flow visualization studies are discussed
briefly. Finally, the measured ground level concentrations are presented as a
series of maps which show full-scale SO,, concentrations in ppm calculated from
the model data and the scaling laws. Some concluding remarks on the program
are presented in the final section. Two appendices are attached at the end of
the report. In Appendix A, a discussion of the minimum permissible Reynolds
number based on ground roughness height is presented. The ground level con-
centration data are listed in tabular form in Appendix B.
-------
I
I
I
I
I
I
I
I
I
I
I
I
i
I
I
I
I
I
I
II. SCALING CRITERIA
In attempting small-scale modeling of flows in the atmospheric
boundary layer, care must be taken to ensure that all important features of
the full-scale situation are represented in the model. Broadly speaking,
these include the ambient wind environment, including both the mean and tur-
bulent characteristics, as well as the local terrain. In stack emission
studies such as presented here, one must also model the relevant features of
the exhaust gases, namely; exit momentum, buoyancy and pollutant concentra-
tion. The dynamics of such flows involve inertial, viscous and buoyancy
forces, as well as turbulent transport. The scaling criteria presented here
are mathematical statements of the requirement that each of these forces be
present in the same relative degree in the model as in the full-scale situation.
This section is divided into four parts. The scaling requirements
for modeling the atmospheric wind are discussed in II-A. The additional
scaling requirements for modeling of stack emissions are presented in II-B.
This is followed by a discussion of sampling time for the model measurements
in II-C. Finally, a summary of the scaling equations is presented in II-D.
The following notation is used.
C = pollutant volume concentration (He in model, SO^ in prototype)
(dimensionless)
Cs = volume concentration of total stack gas (dimensionless)
9
a = acceleration of gravity (meters/sec )
Ji = characteristic length (meters)
,f> = pressure (pascal = newtons/meter )
Q = mass flux of S0~ from prototype stack (kilograms/sec)
t = time (sec)
T = temperature (°K)
u-^ = reference velocity (meters/sec)
u.^ = model reference wind velocity (meters/sec) measured
1.22 meters (4 feet) above model river
f. = full-scale reference wind velocity (meters/sec) corresponding
to u-ry,re.f (i.e., full-scale wind velocity at 2926 meters
above river)
-------
u.s = stack-exit velocity (meters/sec)
a* = friction velocity (meters/sec); u# = CC
2
\l = volume (meters )
% = strearawise coordinate (meters)
it - horizontal cross-stream coordinate (meters)
z = vertical coordinate (meters)
Z0 = characteristic ground roughness length (meters)
3
A = S0? (tons per year)/He (cm /min) for any stack
2
£> = kinematic viscosity (meters /sec)
^ = density (kilograms/meter )
p* = density of S0? at ambient temperature
2
f = Reynolds shear stress at the wall (pascal = newtons/meter )
4> = volumetric flux of He from model stack (meters /sec)
( )^ = in ambient air
C ), = model
(. ")^ - prototype
( )s = at stack exit
A. MODELING OF ATMOSPHERIC WINDS
The primary constraint on modeling of atmospheric flows comes from
the need to ensure that the tunnel boundary layer, formed over the rough floor,
conforms to the "fully-rough wall" flow criterion. When this criterion is
satisfied, the mean-velocity profile i? logarithmic, namely
u
= const, x 2t>q ( ) (1)
U^ Z.g /
where a is the mean velocity at a height z , and the characteristic roughness
length of the flow, z0 , is related numerically to some average dimension of
the roughness elements on the floor.
Many have studied the general problem of simulating atmospheric winds
in a laboratory facility. The conclusions reached in these studies are that
the ratio of a typical length scale JL to the roughness length of the wind
profile 2 should be matched between the model and prototype (full-scale)
approaching flows:
-------
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
(2)
and that the approaching model flow should be "fully rough", i.e., there
should be no laminar sublayer in the tunnel boundary layer. Usually this
latter criterion is satisfied when a Reynolds number based on zn and the
friction velocity u^ is more than about 3:
<** Z.
(3)
Of the two conditions, (2) and (3), it is more important to satisfy condition
(3). In addition to these criteria, it is also necessary to make certain that
the turbulence spectra of the tunnel flow are suitably scaled reproductions of
the atmospheric flow. When these conditions are met, the wind environment in
the tunnel flow is a proper representation of the atmosphere, for neutrally
stable conditions.
The problem of actually generating the required flow in a laboratory
facility is one that has received a great deal of attention in recent years.
A wide variety of approaches is available for the development of the proper
flow; these involve the use of various types of roughness elements, fences,
spires, and jets transverse to the flow. At Calspan, the approach that has
been used is that of a matched fence/rough floor combination (see Section III).
With this technique, the appropriate logarithmic mean velocity profile as well
as a turbulence spectrum representative of that in the neutral atmosphere is
generated.
In addition to the above scaling criteria for modeling of the neutrally
stable atmosphere, another relationship is required when thermally stratified
flows are to be modeled. A simple derivation of the scaling between model and
prototype in thermally stratified flows can be obtained as follows, with the
understanding that all full-scale temperatures are measured relative to the
potential temperature (i.e., 7= T '- Pz where P is the adiabatic lapse rate
and T' is the absolute temperature).
-------
The vertical velocity attained by unit mass of gas in a given interval
of time, as a result of buoyancy, must scale with the horizontal velocity.
Since the vertical acceleration is given by
^*" />«. ^ '*
the velocity acquired in time interval At is Q ^ At .
' 'a,
Taking the ratio of this to the reference velocity, we obtain
Atm
Using the time-scaling relation = -- -j^- we obtain
~AT^ ~ 7^ ~J^ i*^~
This result is equivalent to scaling based on Richardson number in a form
which may be called a bulk Richardson number, B0 = uI T* . Use of ths
parameter Bg for a scaling law has been discussed in Reference 1. For
j
practical purposes, the ratio -^ in Equation (4) is essentially unity. Thus,
Equations (2), (3) and (4) constitute the scaling criteria for the
flow approaching the model. Proper simulation of stack'emissions requires
that additional scaling criteria be satisfied. These are discussed below.
B. MODELING OF STACK EMISSIONS
There are three aspects to be considered regarding the scaling laws
for stack emissions. These are:
(1) Exit momentum scaling
(2) Buoyancy scaling
(3) Pollutant concentration scaling
-------
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
Since the current literature contains various opinions as to the
proper form of the appropriate scaling laws (see for example, Reference 5), it
is relevant to discuss these laws in greater detail. In the following dis-
cussion, it is a basic assumption that the model is scaled geometrically and
that the distribution of eddies and eddy sizes is correctly modeled to scale
(in a statistical sense) in the ASF. Then, since all properties of the flow
are determined by the turbulent mixing, all other properties of the flow such
as mean-velocity profile, Reynolds stress distribution, spectral energy dis-
tributions, and (in the case of stratified flow) deviations from the potential
temperature will be correctly scaled. That this is so in the ASF has been
2-4
demonstrated by extensive measurements in pilot studies, which formed the
basis for design of the ASF, and also during calibration of the ASF itself.
1 . Exit Momentum Scaling
Near the exit of a stack, the velocity and direction of motion
of any small volume of the plume depends on the ratio of the horizontal momentum
given to it by that portion of it which came from the ambient wind, to the
vertical momentum contributed by the stack gas. This is true in the early
part of the mixing process, before buoyancy has had time to contribute signi-
ficant vertical momentum. Since any model motion must correspond to a possible
full-scale motion, the above momentum ratio must be the same in model and full-
scale. The ambient air in the small volume may have come from different levels
with, therefore, different velocities. However, these can all be expressed in
terms of the momentum per unit volume at some reference height where the
velocity is a.^ , since the fractions of gas that come from all levels, and
from the stack, must be the same in the model and in full-scale. Thus, the
momentum scaling near the stack exit requires that
This result can equally well be derived by considering the
average direction and momentum at any cross -section of the plume. The surface
enclosing all the mixing region is denoted by A. The cross-section B is
bounded by the surface A. The stack exit surface is C.
-------
The flux of horizontal
momentum, $fl , entering the plume
through A must flow out through B.
Likewise, the flux of vertical momentum,
(j) , entering via the stack must also
flow out through B. The ratio $c / (j>fl
must be the same in the model and in
full scale.
v-
A
Writing u(z) for the velocity at any height z ,
n
Also, u. (z) can be functionally related to the reference velocity, u.
as follows
Thus,
For correspondence between model and full-scale, the surface integral must
be expressible in the form
r r
/ / (f(z) n)\f(z)\dfi = const. I3-
J Ja
where J! is a characteristic length.
The flux of vertical momentum is simply
Ps
Thus, the momentum ratio becomes
-------
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
Hence, for correspondence between the model, ( ) w, > an-d the full-scale
prototype, C ) ^ , we must have
Am *L ^ = P*+ "5% ^
- -2 n2 2- D*
Pa.U-a.r~ * Pa. ^a^b * +
where we have assumed /oaryi - p ** P
'a.
ft JL "
Since -r^- = -^ the relation becomes, as
A
P* U-*^ Pa. ^a-p
2. Buoyancy Scaling
As the plume from a stack diffuses downstream, its motion must
become buoyancy dominated. The velocity and direction of motion of any small
volume, AV , is determined by the ratio
Stream momentum in AV
Buoyancy momentum in A V
In a volume AV of the plume, a fraction C is stack gas and a
fraction (1-C ) is entrained air. The entrained air may have originated at
many different heights, but its velocity will be some function of position
and time and may be written as a function of the reference velocity u.^ , Thus,
Stream Momentum in A I/ = AV (1-C ) /oa u.a /, ( X , u , z , t
)
The buoyancy momentum in the volume AV is calculated as the
impulse delivered by buoyancy to the fraction C in its travel to the location
(x, y, z) at time t. If we consider A V small enough, then all of the fraction
C may be considered as having traversed the same route. The time taken to
traverse this route may be written as a space and time function of the charac-
teristic time scale, J./U.^ .
-------
Thus, the buoyancy force on A V = A V Cg g ( fls - p^ ) and it
has acted for a time = £ (x, y, z, t) J-/u-A producing the
?
Buoyancy Momentum in AV = Al/C5 Q (ps - p^ f (Z,u,z,t)-~
Q *C 0 0-
Thus, the ratio
_ 2 /,
Stream Momentum in A V _ _ 1-CS A, a^ f, (Xt y,z,t)
Buoyancy Momentum in A I/ £s fyOs ~ p^) aji -f.^2, c/,z , t,>
Since the model motion must be a possible full-scale prototype motion, this
ratio must be the same in the model and prototype at corresponding points in
space and time. Furthermore, the functions f and -f2 must have the same
numerical values in the model and prototype at corresponding points in space
and time because they are geometrical functions of the scaled motion. Thus,
the scaling relationship becomes
'- Csn Pa. "a ' ~ C^p Pa. "2+
Generally, where buoyancy dominates C << 1 and the relation can be written
- X r r_S-rn
Csn ' ~^T ^ (7)
To find a relation for the ratio of the stack-gas concentration
in the model to that in the prototype, we may simply observe that stack-gas
concentration must be proportional to stack exit velocity and, far downstream,
inversely proportional to stream velocity. Thus
More formally, we may again consider a cross section, B, of the plume - say,
a vertical cross section. The flux of stack gas through B must equal the flux
emanating from the stack. That is
e.
where the bar denotes time averaging.
10
-------
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
Again, writing C and u. as functions of some reference concentration and
velocity
= usfis
Invoking the necessary correspondence between model and full scale, and noting
Z vr1 ^ Sy" ^
that // f, dB is proportional to J. , since
and
n
H
we have
S-jb
as before.
Finally, combining equations (7) and (8), the following equation is obtained,
P
7 -
In practice, this equation is usually combined with the momentum scaling
equation (6) to give
a
2.
'fie
5W , _
UO)
leading finally to
JL
(lOa)
This equation determines the model stack density for any choice of the
parameter
1. =
-
Ordinarily, it is not necessary to scale the volumetric (or mass)
flux at the stack exit, and no further restriction need be imposed on the
scaling of the dynamics of the plume.
Equation (9) can be shown to be in agreement with the buoyant plume rise re-
lation recommended by Briggs (Reference 6, equations 4.19c and 4.32) and with
the relation for touchdown point of negatively buoyant plumes advanced in
Reference 7 (equation 6.3). Moreover, the maximum plume rise of negatively
buoyant plumes (Reference 7, equation 6.2) scales in accordance with
equations (6) and (9) above.
11
-------
Equation (10) indicates that the wind velocity over the model
must be reduced in proportion to the square root of the geometric scale ratio
between the model and full-scale. In many situations, it is necessary to
model relatively large areas and the geometric scale ratio must be chosen so
that the area will fit in the wind tunnel. This may require values of J.^/Jl^
which are very small, say on the order of 1/2000. Then, it becomes desirable
to increase the values of the other terms in Equation (10) so that the re-
quired model wind velocities do not become so small that Equation (3) cannot
be satisfied. This is done using highly buoyant gas mixtures for the model
stack effluents. It can readily be seen from Equation (10) that a decrease in
psrr< will increase u.^^ for given values of the other variables. In other
words, the buoyancy and inertial forces must not only be in the proper ratio,
but must be large enough in absolute terms to keep the influence of molecular
viscosity negligible, and satisfy Equation (3). This results in an envelope
or "window" of experimental conditions within which one must operate to simu-
late the full-scale flow properly.
3. Pollutant Concentration Scaling
The inventory for full scale stack emissions usually gives
pollutant fluxes in mass/unit time. In the model we deal with He fluxes in
volume/unit time. For any stack let the full-scale mass flux of SO be Q and
the model volume flux of He be ^ . The density of SO at ambient temperature
A very simple derivation of the concentration scaling law can be
obtained as follows. At some instant mark a cross section of the plume, A.
Let a short time, t, pass and observe the new position of the cross section, B.
A volume, V, is contained between A and B. Since the model motion must be a
possible full-scale motion, we can loo< at this same picture on both scales
and say the following.
12
-------
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
The transient times between A and B are not the same for the two
scales. They are related as follows.
a.
(12)
where UL^ is the reference velocity, and J- is a characteristic length. The
volumes, V, also differ
3
J
p
1 3
' w>
(13)
The average concentration of SO- in the prototype volume is given by
p"
(14)
Note that, since we are primarily concerned with the far field where the plume
;k
is near ambient temperature, we use the SO density, p , at ambient
temperature. The average concentration of He in the model volume is given by
V
(15)
Equations (12) through (15) can be used to obtain the ratio c. /cyn . Since the
initial cross section, A, was completely arbitrary, this ratio must also hold
for all values of C^/C^ at corresponding points in model and prototype. Thus,
a,
(16)
13
-------
Equation (16) is written in a form consistent with the relaxation
of the volumetric (or mass) flux modeling law, as mentioned following Equation
(11), and thus may seem unfamiliar at first.
C. SAMPLING TIME
A final discussion, regarding the comparison of model results with
full scale, relates to the well-known fact that in full-scale the averaging
time has a distinct effect on the measured concentrations. This is not the
case in model tests in the ASF. The model results correspond to short-time
averaged full-scale measurements, taken over not more than 10 or 15 minutes in
most cases. Briefly, what is involved here is the following. The frequency
spectrum of wind gusts in full-scale always shows a null, or near null, in the
Q
range 1 to 3 cycles per hour. Thus, it is theoretically correct to separate
the spectrum into two parts at a frequency in that range, and deal with phenomena
associated with each part separately. In the ASF the high-frequency portion
related to the ground-induced turbulence is fully simulated. The low-frequency
portion related to meandering of the wind, diurnal fluctuations, passage of
weather systems, annual changes, and so on, must be considered separately. In
particular, a correction ' ' for meandering of the wind can be applied if
necessary, to compare with longer term averaging. Thus, since the effective
full-scale averaging time is independent of model averaging times, one can
choose the model averaging time to provide data which are repeatable to within
a specified accuracy. The model averaging times required to obtain a given
accuracy can be estimated from statistical considerations as described in the
following paragraphs. However, as noted above, the data so obtained will cor-
respond to full-scale data measured over not more than 10 to 15 minutes.
For a statistically stationary process one can form an average of any
quantity by taking N independent samples, adding their values and dividing by N.
If one were to do this many times, one would obtain a distribution of average
values having some standard deviation from the true mean. The ratio of this
-1/2
standard deviation to the true mean value is approximately N , in most cases.
This ratio may be regarded as a typical fractional error in a quantity measured
by averaging N independent samples. Thus, to keep this error within 10% of the
mean requires about 100 samples; to keep it within 1% requires about 10,003 samples.
14
-------
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
We have stressed that the samples must be independent. That means that
that the system (the air flow around the model in the ASF) must "forget" what
it was doing in the time span between samples -- an independent sample can be
obtained once the correlation with the last value has essentially vanished.
Now we do not actually sample at long intervals, as indicated above.
We generally deal with continuous data or with digitized data taken at a high
sampling rate. However, we are still essentially bound by the same rules, and
to estimate the required averaging time we can proceed along the following
lines. The model is immersed in a
boundary layer of thickness, cT ,
typically about 1 meter. The velocity,
Uref , near the top of the boundary
layer may be anything from roughly 1 to
25 meters per second. The biggest
eddies in the turbulent flow essentially
span the boundary layer, so that we are
not assured of an independent turbulence
picture until the boundary layer has moved a distance of about 6" . We can say
that most of the boundary layer moves at a velocity close to U-ref , so we can
take "independent" samples at a rate U-re!r/6' per second. We can now construct
an equation which relates the sampling time, t, required to obtain a given
fractional error, a~ , to the tunnel reference velocity, U-ref . From the
above discussion,
7
rf. '
-------
It should be noted that, in the case of turbulence measurements, high
frequency components require the same averaging time as discussed above because
they are the products of the breakdown of the large (low frequency) eddies.
Therefore, they are subject to the same statistical considerations.
As indicated in the derivation, Equation (17) is only approximate.
In practice, it has usually been found that somewhat shorter averaging times
provide the required accuracy. To establish a suitable value at the start of
any program, a few averages are generally checked as a function of integration
period.
D. SUMMARY
The basic scaling criteria used in this investigation can be
summarized as follows.
Ground Roughness Reynolds Number - Equation (3)
Temperature Inversion - Equation (5)
Stack Exit Momentum - Equation (6)
Stack Emission Buoyancy - Equation (10)
Pollutant Concentration - Equation (16)
Approximate Sampling Time Required - Equation (17)
In addition, Equation (2) which relates a typical length scale to the
characteristic roughness length of the ground is satisfied approximately in
the ASF in that (i) geometric scaling for the model is used, and (ii) the
approach flow to the model is developed over a considerable length of random
roughness elements whose mean height is scaled to the full-scale terrain up-
wind of the modeled area.
16
-------
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
III. TEST FACILITIES
A. THE ATMOSPHERIC SIMULATION FACILITY
The Calspan Atmospheric Simulation Facility (ASF), is designed main-
ly for studying atmospheric flow phenomena. This wind tunnel differs from
the conventional aeronautical wind tunnel in two important respects, namely,
the wind shear and the degree of turbulence. Every effort is made in a con-
ventional aeronautical wind tunnel to assure a smooth, uniform flow, free
from turbulent gusts. In contrast to this, a wind tunnel for simulating the
lower atmospheric flow requires a relatively thick turbulent boundary layer
within which the mean and turbulent properties are similar to those in the
atmosphere.
In order to simulate these effects properly, a wind tunnel must be
constructed in a very unconventional way. The particular method developed
at Calspan for this purpose ~ is to use a fence, protruding from the floor
of the tunnel, followed by a length of floor that is covered with roughness
elements. This combination assures both the desired shear, and the associ-
ated turbulent gust spectrum as well. Figure 1 shows an exterior view of the
facility. The rough floor, consisting of wooden blocks in this case, can be
seen upstream of the model in Figure 2. The fence, which is a solid aluminum
plate, protruding from the floor at the beginning of the flow development
region, is also visible in this figure.
The facility is 36 meters long. The test flow is developed generally
over a length of about 15 m downwind of the intake, leaving approximately
9 m available as a test section. The tunnel test section is 2.4 m wide by
approximately 2.1 m high. The tunnel ceiling is adjustable to allow the axial
pressure gradient to be set near zero. The turbulent shear layer occupies
roughly the lower meter of the test section. A variable-pitch fan powered
by a two-speed motor pulls air through the tunnel at speeds from less than
0.5 meter/second to 25 meters/second.
17
-------
Two turntables are incorporated into the floor of the ASF. Both of
them have a diameter of 2.24 meters. These turntables can be located at
various axial locations. In the current program, the center of the turntable
used was located 15.9 meters from the beginning of the flow development re-
gion.
For the present series of tests a liquid-nitrogen (LN ) delivery
system was installed above the ASF. To produce temperature-inversion con-
ditions, LN was poured from four sintered-metal vapor-phase separators
through holes two abreast in the ceiling of the ASF near the beginning of
the flow development region. Each pair of LN streams fell onto a metal
plate covered with wooden blocks similar to those of the rough ground. The
LN? scattered fairly uniformly over the two plates. As the LN~ evaporates,
it both cools the air and adds cold N» to the flow.
B. POLLUTANT-CONCENTRATION MEASURING SYSTEM
The most common application of the; ASF to pollution dispersion is
related to the emission from multiple sources (stacks) in an industrial area.
These emissions are generally hot and have reasonably high exit velocities.
As discussed in Section II, it is necessary to simulate the proper stack
emission buoyancy, exit momentum, and pollution content. This is done by
using gas mixtures to simulate the emission from each stack. The mixtures
are made up of nitrogen, helium, and hydrogen. The helium in the gas mix-
tures is the pollutant simulant and the mass flow of helium is made propor-
tional to the mass flow of pollutant (SO in this program). The nitrogen and
hydrogen components in the gas mixtures are adjusted to provide the correct
buoyancy and exit momentum for the stack emissions.
In the current study, forty-nine stacks were simulated at a scale
ratio of 1:2400. The gas flow rates required to make the emission gas mix-
tures were very small, so it was not possible to use conventional flow maters
or metering valves. Instead, the component mixtures were generated by metering
gases through capillary tubing trimmed to the required flow rate with an
18
-------
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
internal length of piano wire. The complete gas generating system consisted
4
of three regulated gas supplies, H9, He and N at 5.9 x 10 pascal gauge (8.5
psig), and a network of calibrated capillaries to connect each stack, or
group of stacks, to the gas supplies. Stacks were grouped together wherever
several stacks had the same, or nearly the same emissions. A single set of
two or three calibrated capillaries could then serve to regulate the flows
for the group. In these cases the distribution to the group was through suf-
ficient lengths of tubing to ensure that each stack received an equal share.
Pressure differentials generated by the boundary-layer flow over the model
are so small, at the low speeds used in these studies, as to be completely
negligible.
The emission-gas generating system was used in conjunction with a
multiple channel sampling system to determine concentrations of pollutant
(helium) at various points downstream from the stacks. The concentration
sampling system is illustrated in Figure 3 where it is shown connected to a
helium leak detector which is used for quantitative analysis. Briefly, the
sampling system consists of a ring of 24 chambers (the numbered cylinders
on the left of Figure 3), which are initially pumped down to a hard vacuum,
into which the samples are drawn through 3.7 meter long capillaries. Three
of these capillaries were taken to calibration gases, viz.
(1) dry N_ containing about 1 ppm He
(2) 50 ppm He in Air
(3) 975 or 1078 ppm He in Air (according to gas analysis)
The reference mixtures and their analyses were supplied by the Matheson Company.
The other capillaries were exposed to the mixtures drawn from the 20 sampling
points on the model and one upstream reference to determine background level.
Each capillary is connected to the top of a sample collection chamber
through a solenoid valve electrically driven so that all 24 solenoids can be
opened or closed simultaneously. The bottom of each collection chamber is
open to a vacuum plenum which is held at roughly 10 torr by a diffusion
19
-------
pump backed by a large mechanical pump. A single plate valve is used to seal
off all 24 chambers from the vacuum plenum at the start of sample collection
in the chambers.
The method of collecting the samples is as follows. The collection
chambers are pumped down to a hard vacuum (^10 torr) with the capillary
end of the chambers closed by the solenoid valves. Then with the conditions
for a test established (stacks operating and capillaries exposed to the proper
calibration gases and flows to be sampled), the capillaries are flushed for
15 seconds by opening the solenoid valves. This is sufficient time to draw
legitimate samples into the full lengths of the capillaries. The solenoid
valves are then closed and the chambers are pumped down to a hard vacuum.
This takes about 15 seconds. During this time the capillaries return to
atmospheric pressure but they now contain legitimate samples. Once the hard
vacuum is attained in the chambers, the large plate valve at the bottom of
the chambers is closed to seal off all 24 chambers from the vacuum plenum.
Finally the solenoid valves are reopened for 300 seconds to allow samples
into the chambers. The solenoid valA'es are then closed to seal the collection
chambers which now contain the collected samples at a final pressure of 1 or
2 torr.
At the end of sampling, each chamber is analyzed for helium concen-
tration by connecting it, in turn, to the measuring system through an elec-
trically driven scanning valve. The measurement is made on a helium leak
detector in which the pressure is regulated by the fixed geometrical (area)
relationship between an inlet pinhole at the scanning valve and the outlet
restriction of a butterfly valve which is part of the leak-detector. Since
each sample chamber is at the same pressure, the leak detector provides a
direct reading of the concentration level.
The system has been trimmed so that, when all channels are exposed
to the same source, the readouts lie within ±5% of the mean, down to concen-
trations of about 5 ppm helium. Trimming was done by adjusting the capillary
flows. The capillaries are 3.7 meters long and have a nominal internal dia-
meter of .051 cm. They were threaded with lengths of .041 cm diameter piano
20
-------
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
wire. The length of each wire was adjusted to effect the trimming of the flow
-4
rate. Thus the leak detector always sees the same pressure, about 10 torr,
and its readings depend only on the concentration of helium. Calibration
mixtures in the 3 calibration channels allow direct standardization on each
scan. The response time is governed by the time to flush the volume between
the pinhole and the butterfly, which is a few seconds. About 10 seconds
averaging was used to obtain repeatable results. An automatic electronic
control, activated when the sample channel is switched, waits for a selected
time to allow for equilibration, then integrates the output for whatever
number of seconds is selected. A complete scan of the 24 channels took about
10 minutes. At the lowest concentrations some time is required to get com-
plete "clean-up" of helium in the system if it has just been used at much
higher concentrations. The overall performance can be relied upon to within
±10% down to 5 ppm. With special care, even greater precision can be achieved.
Since the system uses a buoyant constituent (helium) as the pollution
simulant gas, large percentages can be used and the equivalent sensitivity to
SCL at full scale is much greater than the actual sensitivity to He at model
scale. Another advantage of using He as the simulant is that the background
concentration in normal air is low, generally about 5-1/4 ppm. The resolu-
tion of the equipment, defined as that concentration which produces a reading
equal to the maximum random fluctuation, is at least 1/2 ppm, but the system
must be kept extremely clean, and adequate time must be allowed for the
system to equilibrate at each reading, if one intends to utilize it to its
full capability. In this study the ratio of full scale ppm SO to model ppm
-4 -3
He ranged from 6 x 10 to 3 x 10 . It was our objective to maintain a
resolution of .01 ppm SCL or better. This required a He resolution in the
model measurements ranging from 17 ppm when simulating the Clairton and Irvin
complexes to 3 ppm when simulating the Glaus plant, Mitchell and Elrama
simulations being intermediate. Thus, in general, the system capability was
not fully taxed so that moderately fast sampling was possible for most
measurements while maintaining a full-scale resolution of considerably better
than .01 ppm SCL. The system resolution finally attained in this program is
discussed in Section IV-D.
21
-------
C. AUXILIARY EQUIPMENT
One of the problems encountered in modeling of stack emissions in the
ASF is the lack of reliable instruments to measure the very low velocities re-
quired by the scaling laws. Such measurements are usually made with hot-wire
anemometers. However, the calibration of these instruments is dependent on
the temperature as well as the velocity of the flow. When the flow temperature
varies with position (in the flow), for example, when modeling an inversion,
hot-wire techniques for determining flow velocity become difficult and time
consuming. Thus, it was decided to take a different approach to velocity
measurements in the presence of a mocel temperature inversion. A heat-pulse
anemometer, developed at Calspan, was used for this purpose. This instrument
operates as follows.
A hot-wire, similar to a conventional hot-wire anemometer, is raised
in temperature step-wise 100 times per second. It is maintained at a constant
elevated temperature for 0.5 millisec. It can be shown theoretically that
if the heat transport is mainly by convection rather than conduction (large
Peclet number), then the temperature jump at any point downstream is affected
symmetrically by the diffusion process, and the mid-point of the jump occurs
at a time determined by fluid velocity and the distance from the heated wire
to the point. That is, the mid-point, or point of inflection, is independent
of the thermal properties of the fluid.
The Calspan heat-pulse anemometer consists of two fine wires at
right-angles to each other, and to the mean flow velocity. The upstream wire
is pulsed electrically and the downstream wire acts as the temperature sensor.
A signal-processing circuit determines each transit time as described above,
rejecting any unsatisfactory measurements, and integrates over 1,000 samples
(i.e., for approximately 10 seconds) to produce a voltage proportional to the
average velocity. Lacking a longer-term integration period we have generally
summed about 5 or 6 readings to obtain a more useful average.
Calibration of the heat-pulse anemometer was accomplished by placing
it in a uniform flow and comparing its output with the flow velocity. The
flow velocity was determined from the measured time of passage of a smoke
22
-------
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
puff between two streamwise locations in the flow. The useful range of the
velocity calibration depends on the distance between the pulsed wire and
the sensor wire in the heat pulse anemometer. The particular instrument used
in this program had a useful calibrated range between approximately 0.15 m/sec
and 0.5 m/sec.
In addition to the use of the heat-pulse anemometer for measuring
flow velocities in the modeled inversion layer, temperature profiles in the
layer were measured with a rake of thermistors. The thermistor rake is ap-
parent on the lower right side of Figure 2. This rake is mounted on a tra-
versing mechanism which can be positioned anywhere in the ASF test section.
The rake has an attachment point on the bottom for the heat pulse anemometer.
The combined rake system was used to obtain the required temperature and ve-
locity profiles.
Finally, the smoke puff technique was used to measure reference
velocities in the approximately uniform flow 4 feet (1.22 meters) above the
model river level.
23
-------
IV. WIND TUNNEL MODEL
In general, it is desired that the model be designed to be as ver-
satile as practicable. However, the scaling and mechanical construction
problems associated with building a model to simulate a large full scale
area introduce certain limitations which are discussed in the following para-
graphs .
The Clairton Model SO Inventory supplied to Calspan indicated that
there are 49 emission sources of varying degrees of importance along the
Monongahela River. These sources are generally distributed along a strip of
land about five kilometers wide by twenty kilometers long with its long axis
running approximately parallel to the N.E. direction.
As will be shown, it was not possible to select a model scale suffi-
ciently small to allow this entire area to be modeled within the confines
of the turntable of the ASF. However, it is desirable to incorporate as much
of the area as possible on the turntable, so that, when the upstream details
are not sufficiently unique to require specific modeling, the general rough
ground can be continued up to the turntable which can then simply be rotated
to change wind direction. Additional sources and terrain can be added up-
stream for those wind directions where they are required. The location of
the center of the turntable and the extent of detailed modeling for each wind
direction is given later in this section. The following paragraphs (Section
IV-A) are concerned with the selection of the smallest permissible scale ratio
between the model and the prototype. The model construction is presented in
IV-B and the source inventory and their locations are given in IV-C. Finally,
the model for the stack emissions is discussed in IV-D.
A. MODEL SCALE
The fully rough flow criterion presented in Eq. (3) of Section II
provides a means of selecting the smallest possible model scale. This cri-
terion states that the Reynolds number, *, °- , based on the friction
24
-------
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
velocity, IL^. and the characteristic roughness length, T.O , of the ground
should be large enough to ensure that the flow will be fully rough. The
roughness Reynolds number can be restated in terms of the actual physical
height, & , of the roughness elements or the height, -fas , of an equivalent
roughness consisting of closely packed sand grains. Reference 13, Equation 20.37
provides the following breakdown of the flow regimes between hydraulically
smooth and fully rough flow,
hydraulically smooth
transition 5 < - < 70
7^
^*-k$
completely rough - > 70
In general, the height of actual roughness elements can be related to an
equivalent sand grain roughness height. For the irregular type of ground
roughness in the Clairton area, it was estimated that the equivalent sand
(18)
grain roughness would be about the same as the actual physical height,
^m . That is, in the model, ^~~ **
viscosity, i) , it remains to evaluate
That is, in the model, ^~~ ** ~ ~ Knowing ~k^ and the kinematic
Our experience in calibrating the flow in the ASF has indicated that
the ratio , where u^ , is measured in the free-stream near the ceiling,
e
+
lies in the range - ~ 0.05. Meteorological wind velocity data are typically
Uv>1ref
measured at a height of 10 meters above ground at a location corresponding to
a local maximum ground elevation. The ratio between u-^^^ and the velocity,
um , at a height in the model corresponding to the full scale measuring
station is approximately 2.5. That is u^ & 2.5 u-^->,0 . Hence, u* x, 0.125 u-m
and the Reynolds number of the roughness elements in the model becomes
-5 2
where we have taken -y = 1.46 x 10 meters /second, am is in meters per
to x
second and -fe^ is in meters.
25
-------
For geometric scaling between the model and prototype ground roughness,
-& J. a -A~,
jT^22- = -~ , where -k^ is the height of the full scale roughness and -j^ is
model to prototype scale ratio. For neutrally stable flow, the model velocity
u-m is related to full scale conditions by Equation (10). That is
P* - fl*~ *r» &* +
^f V Pa.' PPfr *+ V/°s^
Preliminary calculations for model stack emissions which would provide proper
simulation of the sources listed in the Clairton Model S0? Inventory for 1973
indicated that we could use a density defect ratio, V"s? ~o> °^ approxi-
mately 1.8. Thus,
7.8
<^10 / I*
~k U-m
Substituting the expressions for -^- and - ^'° in Equation (19) ,
we have
-^5- » &.{,xio* VTT^1 ( ) a*, -k_^ (20)
Equation (20) expresses the model ground roughness Reynolds number in terms of
the model to prototype scale ratio, the full scale meteorological wind velocity
in meters/second, and the full scale terrain roughness height in meters. In-
spection of the Clairton area elevation contours on the 1:24,000 U.S. Geo]ogical
Survey Maps for the Glassport and McKeesport P.A. quadrangles indicated that a
value of 91 meters (300 feet) is reasonable for -k^ . Using this value, Equation
(20) becomes finally
1,05 x 10
^
« /^w,v- r2i)
Equation (21) was used to construct the following table which shows
the effect of various scale ratios and full scale winds on the model ground
roughness Reynolds number.
26
-------
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
TABLE I
Model Ground Roughness Reynolds Number As a Function
Of Scale Ratio And Full Scale Wind Speed
Full Scale
Wind Speed
10 Meters Above Ground
u-pio m/sec
2
4
6
8
10
Jw, /
JLf
af-ks
1}
51
101
152
202
253
7200
U-n /sec
rtlt
0.19
0.39
0.48
0.77
0.97
J i
*+
U-i.~ks
jJ
18
36
54
71
89
24oo
Urcf m/SSC
0.14
0.27
0.41
0.55
0.68
Jm r
Jt-r> 3(000
a* &s
i)
10
19
29
39
49
a m/sec
""rtf
0.11
0.22
0.34
0.45
0.56
The wind tunnel free stream velocity, u^ , has been included in Table I to
illustrate how slowly the tunnel must be run in order to obtain proper simu-
lation.
Comparison of the values of -5 given in Table I with the criteria
JLm f
given by the inequalities (18) shows that for -. = -j^^- , the flow is fully
rough over almost the complete wind speed range, for ~- = / , the flow is
&v* 1
in the transition regime, and for
"f3 3 lo OO
, the higher wind velocities lie
~- -
in the fully rough regime and the lower wind velocities lie in the transition
regime.
There is some question as to how far into the transition region one
can go before the model conditions depart significantly from the full-scale
conditions. In Appendix A, this matter is discussed in some detail, and it is
shown that good modeling probably continues down to a Reynolds number in the
neighborhood of 40. Referring again to Table I, it can be seen that, on this
basis, at a scale ratio of 1 in 2400 testing could be carried out to equivalent
full-scale velocities down to less than 6 m/sec. Therefore, this scale ratio
was selected for the model.
27
-------
B. MODEL CONSTRUCTION
A wooden model was designed and constructed at Calspan based on in-
formation supplied by EPA, Region III, Philadelphia, Pennsylvania, H.E. Cramer
Company, Inc., and the Allegheny County Health Department. The terrain models
were built by photo-enlarging U.S. Geological Survey 1:24,000 scale maps.
Three maps were required,
Glassport, Pennsylvania, SW/4 Pittsburgh 15" Quadrangle
McKeesport, Pennsylvania, SE/4 Pittsburgh 15" Quadrangle
Monongahela, Pennsylvania, NW/4 Brownsville 15" Quadrangle
Prints, enlarged to a scale of 1 in 2400, were glued to sheets of
plywood which were then cut along contour lines to produce the model as seen
in Figure 2, which shows the assembly for the 225° (approximately s-w) wind
study. Where the slope of the ground was gradual, 0.63 cm (1/4 inch) sheets
of plywood were used corresponding to contour intervals of 15.2 meters (50
feet). Where the ground was steep, 1.27 cm (1/2 inch) sheets of plywood were
used corresponding to contour intervals of 30.5 meters (100 feet). The edges
of the plywood contour sheets were not smoothed, in order to assist in main-
taining the aerodynamically rough wall condition required in the atmospheric
boundary layer simulation.
A portion of the terrain model, containing the Clairton complex, was
separately mounted on one of the ASF turntables. This circle represented a
full-scale diameter of 5.36 kilometers centered at 4462000 m N and 595800 m E
on the grid. The amount of terrain modeled upstream and downstream of the
turntable varied with wind direction according to the locations of the emission
sources and the regions to be studied for S09 concentrations. Table II
Zj
indicates the extent of precise terrain modeling for each wind direction.
Figure 4 shows a sketch of the modeled area overlaid on the geological survey
maps. Beyond the modeled areas wooden blocks, visible in Figure 2, were dis-
tributed to maintain the characteristic roughness length generated in the
approaching flow. The latter regions are indicated by dashed lines in
Figure 4.
28
-------
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
TABLE II
Terrain Modeled in ASF
Wind
Direction
225°
180°
331°
145°
Distance Modeled*
Upstream
14.6 km
5.8
5.2
6.4
Downstream
5.8
7.3
6.4
5.2
Distances measured from center of turntable.
Wind angles are refereneced to the northerly grid direction of the 1000 meter
Universal Transverse Mercator Grid (UTM), Zone 17. This grid was drawn on the
maps used to produce the S0? concentration maps. Its northerly direction has
been called "REFERENCE NORTH".
Models of the structures and stacks at Mitchell, Elrama, Clairton and
Irvin were added to the terrain model. These were made as accurately as possible
from the information supplied. The exit diameters of the stacks were held as
close as possible to scale size, and a system to supply the stacks with the
correct gas mixtures at the correct flow rates was installed below the tunnel.
The gas generating system for the stacks has been described in Section III-B.
C.
EMISSION INVENTORY AND SOURCE LOCATIONS
A total of 49 stacks were modeled in this study. These stacks com-
prised the major sources of S09 emissions from four locations along the
Monongahela River. These are,
(1) Mitchell Power Plant (4 stacks)
(2) Elrama Power Plant (4 stacks)
(3) Clairton Coke Works (33 stacks)
(4) Irvin Works (8 stacks)
29
-------
The locations of these sources are shown on the terrain map of Figure 4. The
full-scale inventory of emissions from these sources is given in Table III.
In Figure 4, source locations are indicated by plus signs with numbers adjacent
to them. The numbers correspond to the source number listed in Table III. As
indicated in Table III, the numbering system for the sources is not consecutive.
The numbers correspond to
vided the inventory data.
14
The numbers correspond to those used by H.E. Cramer Company, Inc. who pro-
As illustrated in Figure 4, the Mitchell and Elrama Power Plants and
the Irvin Works did not fall within the confines of the turntable in the ASF.
Thus, these sources were modeled only for wind directions where they might
contribute to the SO concentrations in the Clairton area which was modeled on
the ASF turntable. The sources modeled for each of the tested wind directions
will be discussed shortly. In all cases, the Clairton Coke Works was modeled.
This area contained the majority of the stacks. The locations of the individual
stacks in the Coke Works are shown in Figure 5.
D. STACK EMISSION MODEL
As noted above, the 49 stacks listed in Table III were modeled in
this study. The Calspan simulation procedure is to use light gases, H and
He, for buoyancy, and to make the flux of He proportional to the S02 output.
The balance of each mixture is made up with N_. Each sample taken in an experi-
ment is analyzed for He content and the result converted into S02 concentration.
The flow rates for the various components of the gas mixture for each stack
were calculated in accordance with the scaling laws discussed in Section [I.
Since the model contained a very large number of stacks (for modeling
purposes) with different emission characteristics, the emission scaling cal-
culations become an optimization problem to achieve useful measurement sensi-
tivities. This problem was attacked by setting up a computer program to calcu-
late the H , He and N2 fluxes required to simulate each stack, for various
values of aavyi/da^ and of Q /p* <$* (see Section II; Equations (10) and (161).
It is best to keep the velocity ratio the same for all tests for two reasons;
first, it makes for better consistency in the approaching boundary-layer flow,
and, secondly, it minimizes the number of calibrated gas flows that has to be
provided for the series of tests.
30
-------
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
TABLE III
*
1973 Full-Scale S02 Emission Inventory Simulated in Model Tests
**
Source
1 Clairton Underfire #1
2 " " #2
3 " " #3
7 " " #7
8 " " #8
9 " " #9
10 " " #10
11 " " #11
12 " " #12
13 " " #13
14 " " #14
15 " " #15
16 " " #16
17 " " #17
18 " " #18
19 " " #19
20 " " #20
21 " " #21
22 " " #22
23 " " #12A
24 Clairton B § W #1
25 Clairton CE #2
26 Clairton Benzene Boiler
27 it " ii
28 Clairton Blast Furnance
30 Clairton Claus Plant
31 Irvin 3 § 4
so2
Emission
(T/yr)
578
i t
1 1
ii
ti
ii
M
I I
It
II
II
II
It
II
II
II
H
II
II
II
3,730
1,175
588
It
303
5,074
824
Stack
Height
(m)
69
ii
ii
65
ii
n
69
TI
1 1
1 T
M
If
61
ii
76
H
M
II
I f
69
50
n
53
1 T
60
58
53
Temperature
(°K)
700
n
H
II
II
It
II
II
1 1
It
II
tl
11
It
It
II
II
It
1 1
II
455
n
616
it
716
561
646
Volume
Flux
(m-Vsec)
37.27
H
n
35.87
ti
n
37.27
ii
H
37.74
n
n
32.13
tl
32.30
58.43
n
n
it
35.87
92.57
72.33
36.55
n
180.58
18.03
54.55
Stack
Radius
(m)
1.22
n
it
1.27
n
it
1.22
it
n
1.31
ii
n
M
it
1.46
2.14
H
i r
i i
1.52
1.37
1.06
0.99
n
1.88
0.61
1.79
31
-------
TABLE III (Cont'd)
**
Source
32 Irvin 5 £ 6
33 Irvin 7
35 Elrama
36
37
38 "
39 Mitchell
40 "
41 "
42 "
43 Irvin Reheat
44 " "
45 " "
46 " "
47 " "
48 Clairton Reheat
49 " "
50
51
52 " "
53 "
54 " "
S02
Emission
(T/yr)
1,232
937
12,079
I?
13,920
20,935
27,142
6,769
I!
T 1
365
ii
1 1
ii
ii
131
1 1
1 1
ii
ii
ii
ii
Stack
Height
On)
75
27
65
M
t !
89
82
70
it
ii
21
I T
11
M
t?
31
1 1
ii
ii
26
1 1
ii
Temperature
C°K)
633
483
416
430
ti
416
403
467
M
n
700
n
1 1
n
n
811
n
1 1
n
i r
n
ii
Volume
Flux
(m^/sec)
79.62
33.40
198.95
n
229.45
299.14
534.81
223.64
ii
H
89.57
n
n
H
M
27.50
1 1
H
1 1
n
1 1
H
Stack
Radius
(m)
1.60
0.92
2.15
it
H
2.30
3.05
2.15
M
M
1.23
n
H
n
n
0.65
1 1
I I
1 1
M
If
M
Data supplied by H.E. Cramer Company, Inc.
t
Source numbering system is not consecutive. It corresponds to system us.ed in
Reference 14. With this system, numbers 3, 4, 5, 6, 29, and 34 are omitted.
32
-------
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
Using the computer program, it was established that it would not be
possible to simulate the complete inventory at one time and maintain reasonable
sensitivity. This situation arises because the source inventory (Table III)
contains stacks where the highest outputs of SO- in general occur with relatively
low temperatures, and vice-versa. Hence, the program was broken down into
four parts, each with one group of stacks fully simulated. The separate parts
will be described shortly.
The computer program quickly showed when a particular choice of
parameters was unsuitable, by indicating negative fluxes of one gas component.
A suitable choice of the parameter f^ (Equation 11), used in conjunction with
Equation (10) to calculate model stack density, was
([ = 1.8 (22)
This value was used throughout. The stack exit velocities, and hence volume
fluxes, were calculated from Equation (6).
The amount of He used in each fully-simulated stack was set by the
choice of a parameter defined as
x Flux of SO? (Tons/year in Full Scale) f ,
Flux of He (cm-Vmin.)l J
This can be converted to the quantity Q* /p* used in Equation (16). The
concentration ratio between full-scale S0_ and model He becomes
-~ - 3.01 x 10~6 A (24)
CyYl
where Cp = -p-pm SO- in full scale
He measured in model tests.
Equation (24) holds for the complete model study. The parameter X
was varied for the four different groups of stacks. The four groups of stacks
for which different values of X were used are as follows,
33
-------
(1) Mitchell
(2) Elrama
(3) Clairton (excluding Glaus) and Irvin
(4) Glaus (Glaus Plant is source number 30 in Table III and Figure 5)
When testing for dispersion of SO,, from Mitchell, Elrama, or Glaus,
interaction effects between the plumes from these sources and the plumes from
the Clairton Works were simulated even though the Clairton Works were not
under study for S0» dispersion. This was done by equipping the Clairton com-
plex, including the Claus plant, with two gas-supply arrangements, one in
which He was included to simulate the SO emissions, and one in which He was
excluded but the correct exit momenta and buoyancies were maintained so that
dynamic coupling between plumes would be retained when the SO emissions were
not being simulated.
The division of the emission modeling into four groups of stacks re-
quired that separate tests be performed to obtain concentration measurements
for the complete inventory of stacks. Thus, for a given wind condition (di-
rection, velocity and stability) each stack group was operated separately and
ground level helium concentrations were measured downstream. The separate
results were then converted to full scale S0? concentrations and added to
obtain the S0~ concentrations for the complete inventory. Table IV lists the
various combinations of stack groups used for each wind direction. The second
column in Table IV lists the stack group under test for helium concentration
levels and the third column lists the stack group which was operated without
helium to allow for plume interaction effects. The fourth column lists the:
parameter A (Equation (23)) used for each test group and the fifth column
shows the scaling factor between the measured He concentration levels and the
full-scale S0~ concentration levels.
The last column in Table IV provides an estimate of the resolution of
the gas sampling system at low levels in terms of full scale SO,., concentration.
The values in this column are based on calibrations of the model gas sampling
system when a true concentration of 10 ppm He is inserted into the sampling
system. The spread in the calibration data was approximately -S ppm He under
34
------- OCR error (C:\Conversion\JobRoot\00000AP6\tiff\2000WEEG.tif): Unspecified error
-------
these conditions. At high concentration levels, the calibrations indicated an
accuracy better than -10 percent. Typical calibrations for the various
channels of the sampling system are given in Figure 6. As noted in Section
III-B, our objective in this study was to maintain a resolution of -0.01 ppm
in full-scale SO concentration levels. As indicated in column 6 of Table IV,
this objective was attained for all stack groups except the Glaus Plant where
it is estimated at -0.015 ppm S0?. The Glaus Plant (Source number 30 in
Table III) is a particularly difficult modeling situation because it has high
SO,, emissions combined with moderately low temperature and very low volume
flux. The low temperature and volume flux limits the amount of He that can be
used if correct scaling is to be preserved.
All of the stack groups listed in Table IV were tested at three
reference wind velocities. In addition, the tests were repeated at one of
the reference wind velocities with the LN~ generated temperature inversion.
This made a total of forty tests to determine concentration levels. Thus,
the procedure of separating the complete emission inventory into separate
groups for testing required an extensive amount of concentration testing.
However, one advantage of the procedure adopted is that it allows assessment
of the individual contributions from the various stack groups to the overall
ground level concentrations of S0-
36
-------
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
V. DESCRIPTION OF THE TEST PROCEDURES
The appropriate portions of the terrain model were installed in the
ASF and wooden-block rough-floor sections were used to fill in the upstream
and downstream portions of the tunnel. In Figure 2, which is a view looking
upwind with the configuration of the 225° (sw) wind series of tests, most of
the details can be seen. Starting in the far distance, the 30 cm-high trip
fence is clearly visible. This is followed by a predominantly white region
which is the wooden-block rough floor. The slightly darker bands of unpainted
blocks are at the locations where liquid nitrogen (LN?) is poured onto the
tunnel floor to create an inversion. These sections consist of metal plates
with a pattern of blocks laid out to encourage uniform distribution of the
LN2 which is delivered to the roof of the tunnel by a distribution system ter-
minating in four sintered-metal phase separators which pour smooth streams of
LN~ through holes in the tunnel ceiling and down onto the metal plates.
The terrain model is the black contoured area, with the Elrama, and
Clairton complexes plainly visible. The Mitchell Power Plant is beneath the
right hand of the man in the background. Intruding from the right is the
thermistor rake mounted on the traverse mechanism, with an attachment point
for a heat-pulse anemometer. These instruments were used to obtain inversion
temperature profiles and velocity profiles.
The tunnel speed was set at a reference height of 4 feet (1.22 meters)
above the model river. This corresponds to 2,926 meters in full-scale. A
smoke-puff generator was installed at this height and the time of passage
between two axial locations measured with a stop-watch. The three reference
velocities used, were
o.^ = 0.22, 0.41, and 0.60 meters/sec in the model tests, or
u~p = 8, 15, and 22 meters/sec expressed in terms of full-scale values.
The full-scale values are accurate to approximately *1 m/sec. Reference velocities
at other heights can be obtained through the velocity profiles, which are dis-
cussed later.
37
-------
The model inventory of effluents was set up by adjusting the flow
rates through suitable stainless steel capillary tubes with lengths of piano
wire inserted in them. Each capillary was calibrated by measuring the volume
of water displaced from a graduated glass cylinder in a measured time interval.
The back pressure on the capillary was at most, a few centimeters of water,
whereas the supply pressure was 598 cm H^O above ambient (8.5 psig). Thus,
the calibration technique was accurate to better than 1% in most cases. Only
for some of the very small flows, where it became difficult to read the grada-
ated cylinder with sufficient accuracy, did the calibration accuracy deteriorate.
However, these small flows contribute little to the overall result. Each
capillary was adjusted by trimming the inserted wire until the flow rate was
within 10% of the required rate in all but the lowest flow cases. The Helium
flow rates, which have a very direct bearing on the results, were set up as far
as possible to considerably better than ±10% accuracy.
Smoke visualization was used to assist in locating appropriate sampling
points for the first two series of tests. The smoke was generated by inserting
test tubes containing a small amount of Titanium Tetrachloride just below the
tunnel floor and passing the stack gas supplied through these test tubes.
The gas sampling system, which is described in Section III-B, was
connected via long plastic tubes to the sampling points on the model. The
method utilized through this program to bring the sample gases to the capil-
laries in the sampling system was as follows. A multiport plenum was held at
a pressure of 12.7 cm H-0 below ambient by a simple jet pump driven by
compressed N~. Small diameter plastic tubes were strung from the sampling
points on the model to the ports of the jet-pump plenum. At each port where a
plastic tube was connected, a specially adapted Tee fitting was used which held
the end of a sampling capillary very close to, and coaxial with, the tube to
which the plastic line was attached. The slight gap between them was thus
exposed to the jet-pump plenum suction in such a way that the capillary opening
was fully immersed in the flow of its particular sample mixture.
38
-------
I
I
I
I
I
I
I
I
I
I
I
I
I
1
I
I
I
I
I
The lengths of plastic tubing required to reach between the model
sampling locations and the gas sampling system were quite long in some cases.
This made it important to know the maximum total time delay between the sampling
location on the model and appearance of the sample in its holding chamber in
the system. This total delay was determined by testing to be less than 30
seconds with the plastic tubing contributing less than 18 seconds to the total.
Thus, throughout the program, the stacks and the jet pump were turned on 45
seconds before measurements commenced to provide a generous margin of safety
in flushing time.
As discussed in Section II-C, the averaging time for making measure-
ments in the ASF is not directly related to the averaging time in full-scale.
The latter has an effect on the result arising from meandering of the wind which
is not duplicated in the model tests. The choice of averaging time in the
model study is simply a matter of repeatability, so that to some accuracy the
results can be expected to be consistent. Equation (17) of Section II-C was
used to estimate the averaging time required for 10 percent accuracy in these
tests. For the reference velocities used, the results are
Full Scale reference velocity, u-1bref = 8, 15, 22 m/sec
Model reference velocity, um = 0.22, 0.41, 0.60 m/sec
Model sampling time for 10% accuracy, t5 = 454, 244, 167 sec
As noted in Section II-C, Equation (17) is only approximate. Usually in
practice, these estimates are conservative. It was found that the gas
sampling system could be made to work well with sampling times up to 300
seconds. Thus, a 300 second averaging time was adopted for all of the
measurements.
Finally, a comment should be made regarding the upstream reference
sample, which critically affects the data at very low concentration levels.
During the course of a day's operation the background helium level in the
laboratory rises slightly. In these tests, the amount of helium used was
quite small and the laboratory is a very large building. Nevertheless, the
resolution required was such that the background level of helium had to be
39
-------
known accurately. It was realized early in the program that the IN- used to
create the inversion contained less helium than normal air. Since the
gas in contact with the model was predominantly N,-, which had evaporated from
the LN» poured on the floor upstream, the background He concentration was
lower than in the ambient air.
Thus, during the neutral stability tests the background He concentra-
tion was simply that of the ambient air in the laboratory and the position of
the upstream reference sample was immaterial. When the temperature-inversion
system was operating, however, it was necessary to use a ground-level samp]e
upstream in order to obtain a true background reading. It did not appear
that the ambient air penetrated significantly to ground level as the flow
passed over the model during the inversion runs. Except for the first few
neutral stability tests, a ground-level upstream reference was used throughout.
40
-------
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
VI. EXPERIMENTAL RESULTS AND DISCUSSION
The experimental results are presented in three parts. The flow
conditions over the model are presented in VI-A. Flow visualization studies
are discussed in VI-B. Finally, the ground level concentration measurements
are presented in VI-C.
A. MODEL ATMOSPHERIC WIND CONDITIONS
1. Approaching Flow
The problem of selecting a model scale for the extensive area
required for this study has been discussed in Section IV-A. In that section,
a scale of 1:2400 was selected on the basis of the estimated ground roughness
Reynolds number attainable. With the scale selected, it is possible to obtain
a more accurate evaluation of this Reynolds number in the flow development
region upstream of the model.
The approaching flow was developed over a random distribution of
wooden blocks with a mean height of 3.81 cm, equivalent to 91.4 meters in full
scale. This random roughness height provided a good match for the mean height
of the model terrain. Moreover, the flow over the random roughness elements had
been calibrated through extensive measurements with hot-wire anemometers. The
calibration measurements provided the following results for the approaching flow;
Ground Roughness Length, Z0 = 0.12 cm
a*
Ratio of Friction Velocity to Reference Velocity, = 0.052
Thus, the ground roughness Reynolds number is
^*»
7 * rn^ it
, . . ^- S). . . ^~ k-V> .-nr
T) T)
= 4.27 u-r»nf where i) = 1.46 x 10"5 meters2/sec
The model ground roughness Reynolds number at the velocities
selected for the tests become:
41
-------
Full-Scale reference velocity, ^-^r&f. = 8, 15, 22 m/sec
Model reference velocity, ^,~iref = 0.22, 0.41, 0.60 m/sec
Model ground roughness Reynolds No., LL*'»Z°'» = Q.94, 1.75, 2.56
These Reynolds numbers are shown in Figure 7 on a graph which illustrates the
effect of lowering the Reynolds number below the recommended value of approxi-
mately 3 (Equation (3) in Section II) . Such effects have been discussed in
Reference 13; Figure 7 has been prepared from the experimental data given
there in a fashion similar to that discussed in Appendix A. The dashed lines
in Figure 7 enclose the scatter of experimental data presented in Reference 13.
Figure 7 shows that as the Reynolds number is lowered from a
value of about 3, the effective roughness length increases at first. The
Reynolds numbers for the three velocities used in the tests (indicated by
arrows), fall at the beginning of the increase of effective roughness length.
Only the highest velocity, a^f = 2^ m/sec, is within the range for completely
rough flow. The intermediate velocity ( ^-p^. = 15 m/sec) provided a Reynolds
number where the effective roughness length has just begun to increase. The
mean deviation from completely rough conditions is about 6 percent in terms of
effective roughness length. This deviation is within the accuracy of determining
roughness lengths from measurements and is thus considered to be acceptable.
The lowest reference velocity used can be seen to be definitely in the transi-
tion Reynolds number range, with an effective roughness length about 20 percent
greater than that for fully rough flow. Thus, the lowest test velocity was
probably too low to provide the proper model flow.
2. Flow Over the Model
Velocity and temperature profiles in the flow over the model
were measured at four locations for the configuration used with a 225 degree
wind direction. This configuration had the most extensive amount of model
terrain and the largest number of stacks (see Figure 4). The heat-pulse
anemometer and thermistor rake were used to make the velocity and temperature
measurements. The measurement locations were as follows:
42
-------
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
(1) Above the Mitchell Power Plant
(2) Above the Elrama Power Plant
(3) Above the approximate middle of the Clairton Coke Works
(4) Above the Liberty Borough School Monitoring Station
The mean velocity profiles at each location are shown in
Figures 8 through 11 and the temperature profiles in the LN~ generated model
inversion are shown in Figures 12 through 15. All measurements were made with
a model reference velocity, u~^ = 0.41 m/sec corresponding to u^ = 15m/
sec in full scale. The data are presented in terms of equivalent full-scale
values. The velocity profiles are estimated to be accurate to within better
than 10 percent of the true value. The accuracy with which the temperature
profiles were determined corresponds to about -1/2°C in terms of full scale.
The latter accuracy was governed by the ability to average turbulence generated
temperature fluctuations displayed on the chart recordings of the thermistor
outputs.
The velocity profiles in Figures 8 through 11 are presented for
neutral stability and for the model inversion. The data for the Mitchell
Power Plant shows an additional velocity profile (dashed line labeled High-
Speed Calibration Profile in Figure 8). This profile was measured over the
random wooden block ground during tunnel calibration. Conventional hot-wire
anemometry was used to obtain this profile. The reference velocity was much
higher than those used herein, u.^ = 13 m/sec. providing a ground roughness
Reynolds number, ^^ « 55, well above the transition range shown in
T)
Figure 7. This same ground was used to develop the flow approaching the model.
For the 225 degree wind direction, it ended just upstream of the Mitchell Power
Plant (see Figure 4) so the neutral stability profile at Mitchell should agree
reasonably well with the calibration profile. As can be seen in Figure 8, the
agreement between the low-speed ( w.yr,ref = 0.41 m/sec) neutral stability profile
and the high-speed ( a^ ref = 13 m/sec) calibration profile is remarkably good.
These profiles can, therefore, be assumed to hold for all except the lowest
reference speed used in the tests. The heat-pulse anemometer used to obtain
the low-speed velocity data did not read below about 0.15 m/sec. Hence, we
could not check the velocity profile corresponding to a full-scale reference
velocity of 8 m/sec which is only 0.21 m/sec in the model.
43
-------
Inspection of the neutral stability velocity profiles in
Figures 8 through 11 shows that the presence of the Monongahela River valley
causes a distortion in the velocity profiles. The degree of distortion can
be seen by comparison of the neutral stability profiles for Elrama, Clairton,
and Liberty Borough with that for Mitchell. The Mitchell profile has been
added to Figures 9, 10 and 11 to facilitate the comparison. The Elrama pro-
file shows a small amount of distortion close to the river valley. The small
differences at high altitudes are within the probable measurement accuracy.
At Clairton, the velocity profile is distorted below about 300 meters. The
distortion is most pronounced at the Liberty Borough School which is in the
wake of the steep bluffs on the North-East side of the river near the Clairton
Coke Works (see Figures 4 and 5). This wake apparently reaches an altitude
of about 1300 meters at the Liberty Borough School. The profile for Liberty
Borough is shown as a dotted line below 400 meters because the heat-pulse
anemometer readings are questionable at the apparent low velocities encountered
in this region.
The velocity profiles measured at the various locations above
the model in the presence of the LN_ generated temperature inversion are
shown as solid points in Figures 8 through 11. The velocities measured with
the temperature inversion are, in general, considerably higher than those
measured for neutral stability and the shapes of the profiles are different.
for the the two cases. The higher velocities associated with the temperature
inversion will have an effect on the ground level concentration data presented
later, since the velocities at stack (and plume) heights are affected. Thus,
the inversion SO,, data measured with a full scale reference velocity, u-p =
15 m/sec, will tend to compare with the neutral atmosphere data taken with
u.^ = 22 m/sec.
The temperature profiles in Figures 12 through 15 show that the inversion
over the model began at Mitchell approximately as a ground level inversion with
a gradient of 3.2°C per 100 meters. Then, since the model itself was not
cooled, the inversion became elevated with an unstable layer beneath it close
to the ground. The elevated inversion persisted as far as the Liberty Borough
44
-------
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
School with a maximum temperature gradient of approximately 1.9°C per 100
meters.at a full-scale heights between 300 and 500 meters above the river
level. In all cases, the heights where the maximum temperature gradients were
observed corresponded closely to the heights where the maximum velocity
gradients were observed.
One of the problems encountered in interpreting model test data,
or for that matter full scale data, is the selection of a location for measuring
the mean wind velocity. In full scale these are usually measured at meteoro-
logical stations which may be remote from the area of interest. Moreover, the
anemometers are located at low altitudes, typically 10 meters above local
ground level. The measured wind velocities can be influenced by the local
terrain in a fashion similar to that apparent in the neutral stability profile
for the Liberty Borough School (Figure 11). In the ASF, the possibility of
local terrain influences on the reference wind velocity is avoided by selecting
a measuring location well above the terrain, in this case at an effective full
scale height of 2926 meters above the river bed. However, this reference
velocity is still required to have a known relationship with some full-scale
meteorological station. In the current program, the continuous monitoring
station at Liberty Borough School provides a satisfactory reference location
since it is located within the confines of the modeled area. The neutral
stability velocity profile at this location (Figure 11) indicates that the
velocity near ground level is approximately 0.4 times the reference velocity
at 2926 meters. Thus, for example, a full scale reference velocity of 15
meters per second is equivalent to approximately 6 meters per second measured
at the Liberty Borough School. Similarly, with the temperature inversion, the
above factor becomes 0.6 and a 15 m/sec reference velocity corresponds to
approximately 9 m/sec at Liberty Borough.
Velocity profiles were measure only for the 225 degree wind
direction- However, the data measured at Elrama and Clairton (Figures 9 and
10) suggest that the velocities near ground level are relatively insensitive
to the local terrain. Thus, wind direction should not greatly influence the
relationship between the reference velocity and the velocity at the Liberty
45
-------
Borough School. Therefore, the relations discussed in the previous paragraph
will be used for all four of the wind directions studied. Both the reference
wind velocity and the approximate velocity at Liberty Borough School will be
noted in the presentation of the ground level concentration data.
B. FLOW VISUALIZATION
As mentioned in Section V, smoke flow visualization was used to
assist in locating appropriate sampling points for the first two series of
tests with wind directions of 225 degrees and 180 degrees. Some photographs
of these tests are shown in Figures 16 and 17 for the 225 degree wind direction.
The Clairton Coke Works, with many of the stacks carrying TiCl. to produce
smoke, is shown in Figure 16(a) under neutral stability conditions and in
Figure 16(b) under inversion conditions. The main stack of the Mitchell
power plant is shown operating under neutral stability conditions in Figure
17(a), and the main stack of the Elrama power plant under inversion conditions
in Figure 17(b).
At the model scale of 1:2400, it was very difficult to generate the
smoke emissions for any length of time because deposits of TiO tended to
accumulate rapidly on humid days and either completely block the small stack:;
or change the effective stack geometry. Thus, a great deal of time was spent
in cleaning out deposits before each test. An attempt to minimize the for-
mation of Ti02 deposits was made by using a concentric tube design for the
largest stacks on the Mitchell and Elrama power plants. The emission gas
mixture was divided and TiCl . vapor was introduced only to that portion of the
gas that went through the central tube. The dry emission gas in the outer tube
shielded the TiCl. vapor from atmospheric moisture until it had exited from
the stack. This arrangement worked fairly well as evidenced in Figure 17.
However, it could not be used on smaller stacks because of their size.
The small size of the model stacks and the low flow rates required
for proper scaling of the buoyant plumes generated a laminar flow within the
stacks. In full scale these stack flows are generally turbulent. In the
46
-------
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
immediate vicinity of the stack exit, plume dispersion with turbulent emissions
is more rapid than it is with laminar emissions. The small plume dispersion
with the laminar model emissions is evident in Figures ,17(a) and (b) very close
to the stack exits. Farther from the stack exit, plume dispersion becomes
dominated by atmospheric turbulence. The latter region is also evident in
Figure 17. In general, the plume rise of an initially laminar plume will
exceed that of an initially turbulent plume because the turbulent plume mixes
more rapidly with the ambient air. However, the initial plume rise observed
in these tests with laminar stack emissions was quite small before atmospheric
turbulence began to dominate the mixing process. For example, at the Clairton
Coke Works (Figure 16) and at the Mitchell powerplant (Figure 17 (a)) atmospheric
turbulence began to dominate the mixing process before the plumes had completely
crossed the model river valley. Thus, except for the ground level regions
close to the stacks, the quantitative concentration level measurements should
scale properly if the model sampling time is long enough to compensate for
the lack of initial turbulent diffusion near the stack exits. This point is
discussed further in Section VI-C.
It was shown in Section VI-A that the minimum wind speed used in this
study (a,., f = 8 m/sec) was probably too low to provide the proper model flow
for the atmospheric wind. The flow visualization studies tended to confirm
this in a qualitative way. The smoke plumes from the stacks showed an inter-
mittent behaviour at the lowest test velocity. At times, the plumes rose to
substantial heights and at other times the plumes acted approximately as they
did at the higher test velocities. This behaviour was particularly noticeable
at the Mitchell power plant where the approaching flow has the least distance
for development downstream of the fence. Thus, it is concluded, on the basis
of the Reynolds number calculations in Section VI-A and the smoke visualiza-
tion observations, that the lowest test speed did not provide proper modeling
of the atmsopheric wind. The ground level concentrations which were obtained
at this lowest test velocity are not included in the presentation of the ex-
perimental results. However, these data are listed in the tables in Appendix B
for those who are interested.
47
-------
C. GROUND LEVEL CONCENTRATIONS
1. Preliminary Near Field Test
In the discussion of the flow visualization tests, it was noted
that the model flow leaving the stacks was laminar rather than turbulent.
Dispersion of these laminar plumes is not as rapid as that of full-scale
turbulent plumes. The flow visualization showed that plume dispersion be-
came dominated by atmospheric turbulence a short distance from the stack exits.
Thus, the model plume concentration levels should provide representative fuLl-
scale values at moderate to large distances from the stacks. However, the
lack of initial model plume dispersion will affect the ground level concen-
tration measurements close to the stacks, even for distances where atmospheric
turbulence has begun to dominate the mixing process. In this region, a long
time sample of the concentration at a point will provide approximately correct
values but shorter time samples will deviate from the correct value because
the lack of initial turbulent diffusion accentuates the timewise intermittency
in the local concentration levels. The estimates in Section V for the sampling
time required in the model tests do not allow for this intermittency. Hence,
for sampling points very close to a source, we might expect that the maximum
error in the measurements may exceed the 10 percent value estimated in
Section V for a sampling time of 300 seconds.
The possible error in the measurements close to a source can be
evaluated by comparing the results of separate test runs taken under identical
operating conditions. Table V shows the results obtained on two successive
days with the model of the Mitchell power station operating with a neutrally
stable atmosphere. The model reference velocity ^^^f was 0.60 m/sec
(equivalent to 22 m/sec in full scale) and the wind direction was 225 degrees
for these tests.
The location of the sampling points listed in Column 1 of Table V can
be seen in Figures 18, 19 and 20. (These figures will be discussed shortly..)
Approximate full-scale distances downwind of Mitchell are listed in Column 2.
48
-------
m
CQ
c/i
0
H
rt
f-l
rt
&
0)
co
PS
H
T3
0
fn
3
in
rt
CD
S
">
PS
O
H
4-J
PS O
rt 0
i-H !/)
cx -^
PS S
f-i O
0 'H CM
5 4-> CM
O O
OH 0 ||
r-H -H
-H Q
0
^ TJ
O PS
4-> -H
0
!-H
PH
4->
PS
CD
O
C
o
u
LO -H
CN O
in CM o
PS r-l
o > 0
4-> fn
H 0 0
-d j2 o
PS p.
O t/)
U O
S 0
bO 4-> 4H
PS < 0
H at
CD
rt rt 0
fn to ,-)
0 4-> rt
_P< 3 O
o 0 co
2 i
O -H
c
O
oj
a.
S <
O
u
s
0
rH
IX, 0
bO
PS rt
O rH
H 0
4-1 >
rt 5
H
>
0
Q
PS
0 E
H 0
+-J $-1
rt ix,
f-t
P t)
PS 0 rt
04-14-1
o rt rt
PS ^H Q
0 3
CJ O 4->
^H U1
CM rt 0
0 U H
CO
0 E 0
1 (1 V- 1
r 1 }_1^ TJ
rt PH O
O v ' S
CO
1 r-H
> 1 0
i 1 >
3 0
m J
TJ
PS
H
0 3
4-) pS CD
rt 3 U
eon
,_j / v ^-«
1-1 1 1 ^J
X O
O 0 CO
f-l 0
Pli C 4-4
PH rt O
< 4->
W
H
Q
bO
PS H
H 4-J 0
-H PS ^
PH-H 6
e o 3
rt (X 2
co
0\°
CN
0
co
PH
PH
CN]
t
0
2
jl
+^
t/J
(11
vy
H
^H
O
1 *
1/1
0
H
/ \
0
i~H
rt
o
OT
i
i
r-H
3
4H
J
:-,
S
CM
O
+J
00
, (
(/!
too
H
IX,
0
0
co
O Ttf- K> vd [~^
** * .-K-X-X**
r-~ CM en oo -*o
r-H ^f CN
TtLOLO'^tOOOCNCN^-tOCNi'^-^O
OOLO^OOrsiOr-lOOOOO
ooooooooooooo
ooooooooooooo
CTl^^Dt~O^OK)LOt~O'*v£)'^-OOt^
or^t-o^Or-HOit-or--fOOr-Ho
^tO-^t^t-OCMr-HOOOOOO
ooooooooooooo
r~-i>OLnr-Hr itocDt^-^ocMcoi-ooo
l-.OOrM^tCNLOCTlLOvO'^'OtNi 1
LOr-HLO^t-OCNi-HOOOOOO
ooooooooooooo
l^OOOOOOOLOLnLO(MCN^)-OOr-H^t
OOOOr-HrHr-HK}^3-\£>^OOOO
r-H
r-HC\IK)^)-LnvDI^OOCTli
i I
O
H
X!
4-J
49
-------
Sampling point number 1 is at river level directly downwind of Mitchell. The
next three points are near the top of the escarpment with sampling point
number 2 close to one edge of the plume. The remaining points are at in-
creasing distances downwind. Columns 3 and 4 in Table V list the ground level
concentrations in terms of ppm SCL in full scale and Column 5 shows the devia-
tion in these data from the average value. The deviations are presented in
terms of ppm SCL and also in terms of percent for those values which exceed
a resolution of 0.01 ppm SO .
As noted previously, our objective in these tests was to maintain
a resolution of approximately ±0.01 ppm S02 or an accuracy of -10 percent
(whichever was larger). While the data in Table V show that it is possible.
to find results which exceed these limits close to a source, they also show
very good consistency a short distance downwind. Only sampling point numbsrs
1, 2 and 8 exceed the above limits and the deviation in sampling points 1 and
2 is expected. Sampling point number 8, which is further downstream, shows a
deviation of 0.012 ppm from the average. This is a negligible amount greater
than the desired resolution. Thus, Table V indicates that except for sampling
locations very close to a source, the desired accuracy was attained. The
fact that a lesser accuracy may be attained at the first one or two sampling
points closest to a source should be borne in mind when using the data in
this report.
2. SO? Concentrations
The ground level concentration data are presented in tabular
form in Appendix B. These data were used to prepare isopleths of the full-
scale ground level S02 concentrations. As explained in Section IV-D, the
tests were performed with various combinations of stacks operating at different
times. (The various stack groups which were studied have been listed in
Table IV.) Thus, it is possible to identify the individual contributions from
each stack group to the overall ground level concentrations. This has been
done for some of the stack groups. Selected individual results and the over-
all results corresponding to operation of all stacks are presented in
Figures 18 through 39. Table VI is an index to these figures.
50
-------
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
oa
o
H
H->
nj
rH
P
O
R
O
(1)
O
rH
bO
R
H
4->
0)
rH
a,
(D
rH
bO
H
P-,
O
-P
X
-H
rH
H 3 rH
Oj 03 I 1
U U off
r-T
r-H -N O rH nj U
H rH rH
S PJ U uB-
W 0
3 R
nJ O
i i i i
U <
R
O t/)
M 3
M rt
ni U
U uir
oj
U4
i i
r 1
(D
U
si
e
(U -H
r-H 2£ X^
Oj 4->
O Q) -i-H O
Ul o O ^
UH M-l
0}
OS
0
T3 H-> bO
ROD
H (D T3
S rH v '
*rH
Q
(1) rH
^H <0
3 -f~l
&o g
H 3
ti-. 2:
XXX
XXX
X
XXX XXX XXX
XXX
XXX
't? 'R" R 't? 'P? 'R' 'R"
o o o o o o o
H -H -H -H -H -rH -H
w 10 in t/) in to tn
rH rH rH rH rH rH rH
(D ?> ^ r-*
R R R R R R R
i 11 ii ii i i i i ii i
S / V J ^ / V 1 V J \ / \ /
LOCMLO LOOJLO LO
-------
The data in Figures 18 through 39 are presented in terras of
full-scale ground level concentrations of SCL. On each figure, sampling point
locations are denoted by plus signs with an adjacent flag. The circled
number in the flag indicates the number assigned to the sampling point and the
uncircled number is the full-scale concentration of S02 in ppm. The full-
scale SO,, concentrations were calculated from the measured helium concentra-
tions in the manner explained in Section IV-D. With the 225 degree wind
direction (Figures 18 through 29) a total of 41 sampling point locations were
used with the various stack groups. Each of these locations was assigned a
permanent number. However, only 20 locations could be accommodated by the gas
sampling system at one time. The 20 locations selected for use with each
stack group in Figures 18 through 29 are indicated by the flags which contain
S0~ concentration data.
Isopleths of S02 concentration levels have been drawn as solid
lines through the data in the figures. The SO concentration levels for the
isopleths are indicated at breaks in the solid lines. For the majority of
Figures 18 through 39, the sampling point locations used with the various
source groups were coincident, so the overall S0~ concentrations could be
obtained by adding the contributions of SO,, from each source group. However,
for the 225 degree wind direction, many of the sampling point locations used
with the Mitchell and Elrama source groups were not coincident with each
other or with those used with the Clairton arid Glaus source groups. Thus, to
obtain the total S02 concentration levels from all of these source groups
(shown in Figures 27, 28 and 29), isopleths were drawn through the available
data for the Mitchell and the Elrama source groups, as in Figures 18 through
23, and the isopleths were then combined with the Clairton and Glaus data
(Figures 24 through 26).
Although the program reported herein was intended primarily as a
data gathering study, there are some features in the data which are worthy of
specific comment. However, before proceeding to these comments, it is appro-
priate to point out what trends one might expect in the data.
*
The source locations for Mitchell, Elrama, Glaus and Irvin are indicated in
these figures when appropriate. The locations of the remaining sources
within the Clairton Works are shown in Figure 5.
52
-------
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
In general, we expect to find that the effect of increasing the
wind speed will be to increase the immediate downwind S02 concentrations be-
cause the plumes are increasingly held down. Farther downwind, increasing
wind speed should decrease the far field SCL concentrations because of
stretching out of the plume gases in the wind direction. The presence of an
inversion will generally tend to hold the plumes down and increase the SO,,
concentrations both near the source and in the far field for a given wind
velocity. These general effects may be modified considerably in the present
case by the very rough terrain in the area of interest and the presence of the
Monongahela River Valley with its steep bluffs. The valley and bluffs may
tend to steer the local wind and to promote plume downwash near the sources.
Such effects can be seen in some of the SO- concentration data which are dis-
cussed in the following paragraphs under sub-headings which identify the sources
simulated and the wind direction.
225 Degree Wind; Mitchell Simulated (Figures 18, 19 and 20)
The classic effects with changing wind speed are evident in
Figures 18 and 19 for neutrally stable flow. Increasing the wind speed
increases the local SO,, concentrations and decreases the far field concen-
trations. There is some evidence at the higher wind speed (Figure 19) of
local steering of the flow by the river valley. The isopleths in the near
field appear to be deflected to the left of the prevailing wind direction.
Moreover, the SO,, concentration level at sampling point number 1, which is at
river level, is comparable in magnitude to the maximum values observed near
the top of the bluff (sampling point numbers 3 and 4). This suggests that
there is a considerable downwash effect which is caused by the presence of
the relatively steep bluffs near the Mitchell Power Plant. The model bluffs
near Mitchell can be seen in Figure 17(a). At the lower wind speed (Figure 18)
neither the flow steering or downwash effects are as apparent.
The effect of the inversion (Figure 19) is difficult to separate
from the effect of changes in wind speed. As discussed in Section VI A.2, the
velocity profiles measured with the inversion indicated that the velocity at
53
-------
stack and plume height was increased when the inversion was generated. It was
suggested in that discussion that the inversion data should be compared with
the neutral atmosphere data taken at a reference velocity u,^, , - 22 m/sec.
If this is done, then the measured velocities at plume height with and without
the inversion are approximately equal. For example, the velocities at the
Liberty Burough School become 8.8 m/sec in the neutral atmosphere (Figure 19)
and 9 m/sec with the inversion (Figure 20).
A comparison of the data in Figures 19 and 20 shows that the In-
version had only a small effect on the near field concentration levels.
Steering of the flow by the river valley is not evident in the inversion data
but there is still a large downwash into the river valley at sampling point
number 1. Moreover, the concentration observed on the bluffs immediately
downwind of Mitchell are comparable in both cases. In the far field, the in-
version increased the observed concentration levels by a factor of approxi-
mately two in the Clairton area.
In general, the far field data for the Mitchell Power Plant show
a very slight tendency towards deflection of the isopleths around the bend :in
the river at the Clairton Coke Works. However, the number of sampling points
in this region was limited, so the slight steering of the plume indicated by
the isopleths may not be real.
225 Degree Wind; Elrama Simulated (Figures 21, 22 and 23)
For neutrally stable flow (Figures 21 and 22), the effects of
changing wind speed on the ground level S0? concentrations are not as con-
sistent as they were with Mitchell. In the near field, the ground level con-
centrations increase as expected with increasing wind speed. However, the
far field data remain at levels as high as or higher than those observed at
the lower wind speed. In addition, the higher speed data (Figure 22) show
greater lateral diffusion to the left after the bend in the river at Clairton.
Apparently, the higher wind velocity holds the plume down enough that the
bluffs near Clairton can deflect some of it to the left.
54
-------
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
The most apparent effect of the inversion (Figure 23) is to
increase the SCL concentration levels observed in the far field. In the near
field, the plume does not appear to reach ground as close to Elrama as it did
for neutral flow at either wind velocity. The presence of the river bend has
no apparent effect in this case.
225 Degree Wind; Clairton (Including Glaus) Simulated
(Figures 24, 24 and 26)
The ground level SCL concentrations for this case arise as a
result of a highly complex combination of multiple stacks in the presence of
steep bluffs perpendicular to the wind direction. The model terrain near the
sources can be seen in Figure 16 and the distribution of the 33 sources is
shown in Figure 5. In the model emission inventory (Table III) the single
largest contribution of SCL comes from the Glaus Plant (source number 30).
The location of this source is identified separately in the SO concentration
figures. Two other sources have higher than average SO,, emissions. These
are source numbers 24 and 25 in Figure 5. In Figures 24 through 29 these
two sources are located near the river bank directly below the circled
number 30. Sampling point numbers 24, 26, 30, 32, 33 and 40 are at river
level. The remainder are at considerably higher elevations which are listed
in Appendix B.
Inspection of Figures 24, 25 and 26 shows that there is con-
siderable downwash of the plumes in all these cases. The S0_ values at river
level just downwind of the Glaus Plant (sampling point number 26) are comparable
to, or larger than, those observed on the bluffs (sampling point numbers 25,
27 and 28). The highest values occur at the lowest wind speed, which is to
be expected since downwash occurred at all of the tested wind velocities. Even
the data at the lowest wind velocity ( ^pre{ = 8 m/sec), which are not pre-
sented in the figures because the ground roughness Reynolds number was too low,
indicated a very large downwash just downstream of the Glaus Plant.
55
-------
The near field concentration isopleths in Figures 24, 25 and 26
show evidence of the contribution from source numbers 24 and 25 as well as
from the Glaus Plant. This is most noticeable in Figures 25 and 26 where
locally high concentration levels can be seen near sampling point number 31,
In the far field, the S00 concentrations decrease with increasing
wind speed in the usual fashion. The inversion (Figure 26) had an unusual
effect in this case. The far field concentration levels decrease in comparison
with either of the neutrally stable flow tests and the plume trajectory is
deflected slightly to the left. The low concentration levels with the inver-
sion are probably the result of increased wind velocities downwind of the river.
Velocity profiles above the Liberty Borough School (Figure 11) show a rapid
increase in velocity with height between about 450 and 550 meters. This is
the same region where the temperature gradient is a maximum (Figure 15). If
portions of the plume from the Clairton area are deflected upwards by the
river bluff to heights above 400 meters, then these portions will be above
the elevated inversion layer and will be swept away rapidly by the high wind
velocity.
225 Degree Wind; Mitchell, Elrama, Clairton (Including Glaus)
Simulated (Figures 27, 28 and 29)
In these figures, the far field SO,, concentration isopleths from
the Mitchell and Elrama plumes have been added to the isopleths from the
Clairton Coke Works. The wind direction for this series of tests was such
that the plumes from Mitchell and Elrama were almost directly in line with
each other. The composite plume from these two sources contributes almost all
of the SCL to the isopleths to the right of sampling point number 41.
For the particular wind direction used in this test series, "he
Glassport monitor was completely bypassed by the plumes from the various sources
and the Liberty Borough monitor just picked up the left edge of the composite
plume. However, a small decrease in wind angle would place the Liberty
Borough monitor within the plumes from Clairton and Mitchell and a slightly
larger decrease would bring in the Elrama plume. If it is assumed that the
56
-------
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
isopleths do not change shape rapidly with small changes in wind angle, then
one can rotate the isopleths from each source group about the source to obtain
the S0? levels at Liberty Borough for slightly different wind directions.
For example, if this procedure is used for a neutral wind at a velocity of
6 m/sec at the Liberty Borough monitor (Figures 18, 21 and 24) and the Glaus
Plant is used as the center of rotation for the data in Figure 24, then one
obtains SO- concentration levels at Liberty Borough of approximately 0.08 to
0.09 ppm S02 for wind angles between 200 and 218 degrees.
The above result compares quite favorably with full-scale data
measured at Liberty Borough on 18 January 1973. During hours 13 and 14 (EST)
on that date, the wind speed was relatively steady between 6.2 and 6.7 m/sec,
the wind angle was close to 220 degrees, and the atmospheric stability was
neutral (see Tables 6-5 and 6-7 of Reference 14). The full scale SCL concen-
trations measured at Liberty Borough under these conditions were between
0.115 and 0.13 ppm S02-
180 Degree Wind; Clairton (Including Glaus) Simulated
(Figures 30, 31 and 32)
The results shown in these figures are similar to those obtained
with this source group at a wind direction of 225 degrees. There is evidence
of plume downwash at sampling point numbers 6 and 8 which are at river level .
The effects of contributions from the Glaus Plant and from source numbers 24
and 25 are apparent in the kidney shaped isopleths. In this case also, the
wind direction was such as to place the Glassport and Liberty Borough moni-
toring stations on the edges of the plume so that neither monitor location
received much S02. Of course, a small change in wind direction should place
one or the other of the monitoring stations within the plume. For example,
rotating the isopleths in Figure 30 about the Glaus Plant indicates that
about 0.07 ppm S0? would be measured at Liberty Borough for a wind angle of
200 degrees. This agrees well with the value 0.08 ppm, obtained by rotating
the isopleths in Figure 24 to the same wind angle.
57
-------
There is some question that SCL levels at the Glassport monitoring
station can be estimated easily for wind angles other than those tested. This
monitor is near river level and the SO- levels may be influenced by local
flows around the river bluffs. The data obtained for a wind direction of
145 degrees, which are presented below, show evidence of such influence.
180 Degree Wind; Glaus Plant Simulated (Figure 33)
This figure is included to show the contribution of the Glaus
Plant to the ground level S0? concentrations from the Clairton Coke Works.
The results can be compared to those in Figure 30, which shows the SO,, iso-
pleths for all of the Clairton Coke Works, including the Glaus Plant.
145 Degree Wind; Clairton (Including Glaus) Simulated
(Figures 34, 35 and 36)
The wind direction used in this test series was chosen because
it placed the Glassport monitoring station directly downwind of the majority
of sources in the Clairton Coke Works. Thus, if the plumes were not much
affected by the presence of the river valley and bluffs, the Glassport monitor
should be centered approximately in the composite plume. We understand that
this particular monitoring station has consistently provided S0« concentration
levels which are lower than expected for wind directions near 145 degrees. The
model test data in Figures 34, 35 and 36 provide an illustration of how the
low readings arise.
In all three cases, shown in Figures 34, 35 and 36, the ground
level isopleths deflect around the right side of the Glassport monitor. Tiis
monitor is close to the river level, so one might speculate that the composite
plume is lifted over the monitor. However, sampling point numbers 12 and 19
are also near river level, upwind and downwind of the Glassport monitor.
These sampling points detected significant quantities of SO-. Moreover, tne
upwind river-level sampling point (number 12), detected the highest SO-
levels measured anywhere in the area. Thus, it appears that the composite
plume reaches ground well before the Glassport monitor, but that the plume is
58
-------
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
locally deflected to the right by flow disturbances from bluffs on the sides
of the river valley. The direction in which the local flow is deflected is
rather surprising since the river valley angles slightly to the left near the
Glassport monitor.
331 Degree Wind; Claifton (Including Glaus) and Irvin Simulated
(Figures 37, 38 and 39)
The final wind direction studied in this test program was chosen
to place the Irvin Works directly upwind of the Clairton Coke Works. With
this configuration, the Glassport monitor lies approximately midway between
the source groups.
For neutral flow (Figures 37 and 38), wind speed has a large
effect on the ground level data. At the lowest velocity (Figure 37), the
plumes apparently rise quite high and the SO,, levels are low everywhere
except directly downwind of the Glaus Plant, at sampling point number 16. At
this sampling point, the separate tests with the Glaus Plant simulated by
itself showed that the total SCL concentration level (0.08 ppm) is composed of
0.06 ppm from the Glaus Plant and 0.02 ppm from all the remaining sources.
The Glassport monitor shows no detectable SCL at this low wind speed. For the
higher wind speed (Figure 38), the SCL levels increase everywhere near the
river valley. The plume from Irvin appears to reach ground in the vicinity of
the Glassport monitor and relatively high SCL levels are detected there. Con-
centration levels of SCL downwind of the Clairton Coke Works also increase
significantly at the higher wind speed. At sampling point number 16, the SCL
contribution from the Glaus Plant decreased to 0.04 ppm while that from the
remaining sources increased to 0.14 ppm.
With the inversion (Figure 39), the SCL levels downstream of
Irvin are low and comparable in magnitude to those observed in neutral flow
at the lower wind speed (Figure 37). However, the isopleths are deflected to
the left, along the river valley, during the inversion. Downstream of the
Clairton Coke Works, the isopleths are again deflected to the left by the
59
-------
inversion. Sampling point numbers 13, 15 and 18, which are to the left of tne
Clairton Coke Works, show significant increases in SCL level with the inver-
sion. All of the SCL at these sampling points came from sources other than the
Glaus Plant. At sampling point number 16, the relative contributions from
the Glaus Plant and from the remaining sources are 0.04 and 0.07 ppm SCL
respectively.
3. Summary
In the preceding paragraphs, the model test results have been
discussed with a view toward identifying unusual dispersion patterns caused
by the presence of the Monongahela River Valley with its steep bluffs. Ground-
level concentration data were presented in terms of full-scale S0? concentra-
tion levels for four wind directions, two wind speeds in neutrally stable
flow, and one wind speed with an inversion.
The results indicate that the river valley and bluffs do have an
effect on the dispersion patterns. The largest effects were observed on the
plumes from the Clairton Coke Works. For South and South-West wind directions,
the plumes from the Coke Works appear to be deflected down into the river
valley at the base of the bluffs. Concentration levels comparable to or
higher than those observed near the top of the bluffs were obtained at the
base. For a wind direction of 145 degrees, the isopleths of the ground level
SO,-, show an unusual deflection around the Glassport monitoring station,
missing this monitor almost completely. In the absence of local flow distur-
bances from the bluffs, the Glassport monitor would lie near the center of
the plumes for this wind direction.
The effect of the inversion used in these model tests was in-
consistent. The inversion appeared to hold down the plumes from Elrama and
Mitchell, increasing the far-field S09 levels over those observed with neutral
^
flow for comparable wind velocities at the Liberty Borough School. This is
the expected result. However, with the Clairton Coke Works as the source of
emissions, the far field SCu levels sometimes decreased with the inversion
60
-------
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
(225° wind direction), sometimes remained almost unchanged (180° wind direc-
tion) and sometimes the whole isopleth pattern changed shape (331° wind di-
rection) . The inconsistent effects of the inversion on the plumes from the
Clairton Coke Works are probably a result of flow disturbances caused by the
steep bluffs.
For all except one test, the model results showed very low S07
concentration levels at the locations of the Liberty Borough and Glassport
monitoring stations. In the one exception, the Glassport monitor detected
relatively high concentration levels downwind of the Irvin Works (331° wind
direction). The low observed levels at Liberty Borough appear to be a result
of the choice of wind directions for the test program. Other wind directions
would probably have placed this monitor within the plumes. However, the same
general comment does not appear to apply to the Glassport monitor. This
monitor was almost completely bypassed by the composite plume from the Clairton
Coke Works when it should have been situated in the middle of this plume (145°
wind direction). Thus, the Glassport monitor appears to be located in an area
where local flow disturbances from the bluffs are critical in determining the
detected SCL levels. We understand this result is consistent with full-scale
observations where the Glassport monitor has been detecting lower SCL levels
than expected under similar circumstances.
Finally, an indirect comparison between the model test results
and full-scale measurements of SO,, concentration levels at the Liberty Borough
monitor has been made for one wind condition. Full-scale hourly concentrations
between 0.115 and 0.13 ppm SO,, were measured on 18 January 1973, during a
relatively steady wind with a velocity between 6.2 and 6.7 m/sec, a direction
close to 220 degrees, and conditions of neutral stability. The model test
results obtained with neutral flow at a wind direction of 225 degrees were
used to estimate S02 levels at the Liberty Borough monitor for wind directions
near 220 degrees and a velocity of 6 m/sec at Liberty Borough. The estimates
based on the model test results indicated approximately constant SO, concen-
tration levels of 0.08 to 0.09 ppm for wind directions between 200 and 218
degrees. This compares favorably with the full-scale values noted above.
61
-------
VII. CONCLUDING REMARKS
The model study reported here was a formidable undertaking in many
respects. The large number of stacks to be modeled (49), each with its own
emission characteristics, made the design and fabrication of the stack gas
generating system highly complex. In addition, the extent of model terrain
required a model-to-prototype scale ratio (1:2400) which was much smaller than
that usually used. This, in turn, required careful consideration of the model
scaling laws and led to some compromises to the final design.
Because of the small scale of the model, it was not possible to obtain
turbulent flow within the model stacks.' The plumes leaving the stacks were
initially laminar. As such, the model plumes may have risen higher very close
to the stacks than the corresponding turbulent plumes from the full-scale
stacks. However, smoke visualization of the model plumes showed that plume
dispersion rapidly became dominated by atmospheric turbulence a short distance
from the stacks. Moreover, ground level concentration measurements showed
evidence of considerable downwash close to the stacks in most tests, pre-
sumably because of the very rough terrain and the river valley. This suggefits
that the initially laminar flow at the stack exits did not have a large effect
on the results.
In addition to the presence of laminar flow in the model stacks, the
small scale of the model and the large extent of terrain modeled limited the
range of wind speeds and directions which could be tested during this pro-
gram. The necessity for extensive modeling of the terrain upstream and down-
stream of the turntable in the ASF made it impractical to perform tests for
more than a few wind angles, since each wind angle required its own model
terrain. The small scale of the model limited the wind speeds which could be
simulated properly to values above about 6 m/sec measured in full-scale at
the Liberty Borough School.
62
-------
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
One series of tests was performed for a smaller wind speed, but the
ground roughness Reynolds number corresponding to this speed was too small for
proper flow simulation. The minimum wind speed limit would have been even
higher if the full-scale terrain in the area had not been very rough, with
hills and valleys about 90 meters in depth.
One of the aims of this study was to investigate the effects of an
inversion on ground level SO- concentrations. To do this required the develop-
ment of a system for generating an inversion in the ASF. A liquid nitrogen
distribution system for generating an inversion was developed and tested as a
part of this program. The results are reported in the text.
The purpose of this model testing program was to provide data on the
ground level concentrations of S0~ generated by the various sources in the
area. The main body of the results is presented in terms of full-scale ground
level SO,, concentrations for the various test conditions.
The discussion of the results is oriented towards identifying situa-
tions where the local terrain appears to modify normal plume dispersion
patterns. In general, it was found that the presence of the river valley with
its steep bluffs did have local effects on the plume dispersion. The concen-
tration data indicated the presence of considerable downwash near the face of
the bluffs in the area near the Clairton Coke Works and to a lesser extent
near the Mitchell Power Plant. In addition, there is evidence that the river
valley and bluffs can deflect the plumes to a certain extent. The most unusual
plume deflection observed, occurred for a wind direction which was chosen to
place the Glassport monitoring station directly downwind and in the center of
the composite plume from sources in the Clairton Coke Works. The test results
showed that the plume was deflected around the Glassport station, missing it
almost completely.
The effect of the model inversion on the ground level concentrations
can be described as inconsistent. The inversion had the expected effect on
the plumes from the Mitchell and Elraraa Power Plants, increasing the measured
63
-------
concentrations in the far field. However, the effect of the inversion on the
plumes from the Clairton Coke Works appeared to depend on the wind direction,
sometimes increasing ground level concentrations and sometimes decreasing them.
These effects are discussed in the main body of the text.
In conclusion, it is clear from the data presented that the terrain
does influence dispersion of the plumes. It is believed such influence would
be impossible to predict without model tests or extensive full-scale
measurements.
64
-------
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
APPENDIX A
MINIMUM PERMISSIBLE GROUND ROUGHNESS REYNOLDS NUMBER
In many cases of modeling atmospheric flows in a wind tunnel, it is
not practical to satisfy completely the fully rough flow boundary conditions
for the flow over the model under all wind conditions. It is then of interest
to estimate how small the roughness Reynolds number can be before significant
deviations occur from the ideal flow. One approach to this problem is pre-
sented briefly in the following paragraphs.
The mean velocity profile near the ground can be expressed as (Ref.
13, Eq. 20-32)
-- +- 5
(A-l)
= 5-75 JLoa ( } + B
\ -k
where & = a constant
-A-s = the height of closely packed sand grain roughness
u~ - mean velocity at height 2
U-x - friction velocity
Z = vertical height above ground
Equation (A-l) can be rewritten in terms of a characteristic roughness length
?-0 as follows
^L = 5.75 loa -*-
m* ? Z0 (A- 2)
where we have used the relation
£ V (£*)
a
For fully rough flow S> = 8.5, making ~ = 30.1
65
-------
4
To plot -2- as a function of the roughness Reynolds number, we can use
:2
for B from Equation (A-3) . The result is shown in Figure 40.
* T
Figure 20:21 of Reference 13, which is a plot of 8 vs * , substituting
a
It can be seen that -=- increases gradually in the transition zone
as ^ 5 is decreased below a value of 70. For ^-^ = 40, the mean value
of -^- is about fifteen percent above its value in the completely rough
^o
regime. In practice, it is difficult to determine values of Za over a givsn
roughness to much greater accuracy than plus or minus ten percent, even with
very careful measurements and an independent determination of u.^ , for example
from the Reynold s stress profile. In many field evaluations of zg , both
(MX and za are estimated from measured mean velocity profiles. That is, ti*
is not determined independently. In th'ese cases, it is unlikely that calcu-
lated values of Z0 approach the accuracy mentioned above. Thus, it appears
consistent with the general accuracy of the modeling to allow testing down to
model roughness Reynolds number of forty. This is equivalent to a value of
about 1.3 if the Reynolds number is based on Z0 rather than ~&s .
66
-------
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
APPENDIX B
TABLE OF GROUND LEVEL S02 CONCENTRATIONS
CALCULATED FROM MODEL TEST DATA
Ground level S0~ concentrations calculated from the model test results
are listed in the following tables. In each table, Column 1 identifies the
sampling point by an assigned number. The locations of the sampling points
are shown in the figures identified in brackets below the Column 1 title.
Column 2 shows the local ground elevation above river level for each sampling
point. The remaining columns list the SO,, concentration levels obtained with
the various source groups operating. In columns identified as totals in the
title, SO,, levels were obtained by adding contributions from the appropriate
individual source groups. Apparent negative S02 concentration levels listed
in the tables were all smaller than the resolution of the concentration
measuring equipment. These small negative values were taken as zero when
adding contributions to obtain total SO- concentration levels.
With the 225 degree wind direction, the twenty sampling channels
available in the testing equipment were not sufficient to provide a good
distribution of sampling point locations for all four source groups. Accord-
ingly, a total of 41 sampling point locations were used in these tests, but
only 20 of these were used with any one source group. Thus, it was necessary
to interpolate the measured data to obtain the totals for all 41 sampling
locations. This was done by drawing isopleths through the available data for
each source group to estimate the SO levels at the missing locations. Data
estimated in this way are shown in brackets in the columns for the individual
source groups. All of the data calculated from direct measurements are given
to three decimal places. Data estimated by interpolation from the isopleths
and also total concentrations obtained by addition are listed to two decimal
places.
67
-------
All of the data measured during the course of the test program are
listed in the tables, including data taken at the lowest wind speed which was
too low for proper modeling. The latter are listed for the sake of completa-
ness and are identified in the title of the table as taken in the absence of
proper simulation.
68
-------
Full-Scale Ground Level Concentrations of SC>2 Calculated from Model Test Data
(Data listed on this page were measured at a wind speed too low for proper modeling)
Wind Direction: 225°
Wind Velocity: Full Scale - 8 m/sec @ 2926 meters above river level
Model - 0.22 m/sec @ 1.22 meters above river level
Sampling
Point No.
(See Figs
18-29)
/
.3
J3
y
.<
(*
i
£
°l
lo
li
13
IA
/V
1 £L
/6
11
IJ>
/9
30
<£!
.3.2
c23
£
6".<
-e?
0
Hie
0
/*$
f/(c
fs
Ss '
S6~
.tt
O
SS
l/(*
rt Sampl(
Borough
SO? coi
the ava:
U desigi
thus wi:
S02 Concentration Levels (ppm)
Mitchel 1
. c $(#
CC9
.c \y.
.OCf
.609
L^tf
-_(./*?
.CbX
t>S 7
( }
. /L- ')
.OLC.
( }
rt^
./C'l
( )
( )
( }
( )
( ^
.C90(
\
C )
,&/o
( )
ns in br;
sured da-
ling poii
tribute 1
Liaus
Plant
//
U
L(
U
U
U
U
U
U
U
.U ..
(-(
U
U
U
.a_. ..
u
a
U '
u
n
L(
f.<
.06/
.C/f
. 6 '&
. 3V6J
, it's
-.r>d/
.oo ^
c
. 00^
, oo/
.O/b
.{)O&
c>
ckets art
:a.
ts are u]
o the tol
uiairton
Excluding
Glaus
U
U
U
U
14
a .
U
u
u
u .._.
u _
u .._..
u
u
u
u
u __
u
u
_.... t(
u
n
.001 .
. 06 (
./A~6>
./,?£k
, G5'/
,061
0
.OSS'
. 06 £
, C?76
.o(*t,
. 00 f_
.033
<0.3?
,0^/
-_±O£3__->
,^/a .
estimate
iwinH of
al SO,, 1^
"
Total
All
Sources
.£>(e
.fjl
,6J
.r, I
f,/
c^
.C3
,i,1
.Ck
/'/
,03
1*
.11
,11
,/'J
.^1
0
.d(*
. 0 /
,/.*
d from i
VIP sourc(
vel.
;opleths'
; and
T3<;pd nn
.
1
L_ _t -
r
--
69
-------
Full-Scale Ground Level Concentrations of S0? Calculated from Model Test Data
Wind Direction: 225°
Wind Velocity:
Full Scale - 15 m/sec @ 2926 meters above river level,
6 m/sec @ Liberty Borough School
Model - 0.41 m/sec @ 1.22 meters above river level,
0.16 m/sec @ Liberty Borough School
t
Sampling
Point No
(See Figs
18-29)
/
3
^
V
f
(s
1
J
1
/o
1!
13
/3
/V
/o'
1(6
n
n
/9
__£$_
Cj
r> __
/.T/
//6
.^.r
jr
5" 6"
r,r
0
cf>^
l/(*
rt Sample
Borough
SO., Cone
baSed or
U desigr
thus wi
SO Concentration Levels (ppm)
Mitchell
fFig 18.
J3s
,W7
, 00 j
J9l
.06,0
,w
.0-?f
f.os\
03 1
. o/s
(. O i~ }
Lu.j\
0(*9
(w\
( o/^
(,c^\
(,0(, .
( ' .03'
. A
035
,+jyo
. 3/3
./(.I
.OOd
^
-.oo/
,oo/
.oo/
- .ao*/
0
.//
.O/
,o?
,07
C
,04
.03
.DC
,03
.0 /
from is
he snurc
:yel, .
Total
All
Sources
O. 13
a oy
0 .£ (r
G /O
0 01
/9
0(.
_ .csr
,of
,'JO
03
£l
,/A"
./V
, IX
.04
,o(*
j^
.id
-------
Full-Scale Ground Level Concentrations of S02 Calculated from Model Test Data
Wind Direction: 225°
Wind Velocity: Full Scale - 22 m/sec @ 2926 meters above river level,
8.8 m/sc @ Liberty Borough School
Model - 0.60 m/sec @ 1.22 meters above river level,
0.24 m/sec @ Liberty Borough School
Sampling
Point No.
(See Figs
18-29)
/
0?
3
^
I g
' 6
7
J'
?
/6 . ._
//
;<#
13
/V
/ s
/'
.J/3
. 196
0^
.OW
(.\
.no -? n
,^/7
,0/f
( C3 )
Lc/
f.o/.
(o\
(n\
_loj .
0
.Of) 3
(,ca\
(,C,3 }
r>
.po/
(,oi^
(n}
lot}
,0/3
(0}
r
Sampler
entratior
the avad
ates samj
1 not COT
Elrama
fFig 221
(J
u
U
U
(J
u
u
u
u
,3*7
0
,31, 9
,,^3
.A-?9
VY?
.(.C(*
,G^/J
.1/3.
,o?/
. OC.-3
,C?3
, C'V/
.nJJ-
, c.y.y
,f07
Lof\
LoT}
(.cA
<£./'.
to]
(o\
(0}
, OO~(o
(.03)
(.03)
(.01)
.po.-r
__(,o-
L'
U
u
u
. u
o
u
u
u
- U
u ..
u
(j
,.., u
- . no*
.CO/
.r:DP
./.rt/
.001
&
. 00 /
, no?
. noa
- . nos
. .,W
iOo/
.009
O
....Q_ _|
ket.s arp
sured dal
ts are uri
n thft. t"1
Clairton
Excluding
Claus
U
u
u
u
u
O
J
(J
u
O
u
u.
o
u
u
u
u
u
u .
u
. -M
u
,nc,3
,??i- 7
.£>?«/
.Of/
.01J
.m#
. 00(
tot
.0*
.0 3
0
.01.
f T-nm i <;
he sonrr.
ivel.
Total
All
Sources
fFig 281
.41
.01
....vv
,vv
01
. J?/
.0?0
.03
.07
.(,0
r- ^
r_^9
J5"
/
-------
Full-Scale Ground Level Concentrations of SO- Calculated from Model Test Data
Wind Direction: 225° (with inversion)
Wind Velocity:
Full Scale - 15 m/sec A 2926 meters above river level,
9 m/sec (i Liberty Borough School
Model - 0.41 m/sec @ 1.22 meters above river level,
0.25 m/sec @ Liberty Borough School
Sampling
Point No ,
(See Figs
18-29)
/
,-3
^
V
o
A,
7
?
9
/O
//
13.
/3
/y
A-r
/6
'1
If
/9
-JO
o2/
X??
-23
d?y
<3S
£l*
>£^L
}
(.01 .
,QO^
(.0^.
(.01.
(.03}
/9jc2
C)
t
Sampler
entratio
the ava
ates sam
1 not co
Elrama
(Ficr 231
U
U
U
U
U
U
U
U
U
.6 ft
-.fJOl
.3 Iff
C(ff
.us-
.^'9
,06/
,/(,')
./37
.c$y
c
.o(*e
.._dca
.Hbs
.0/9
.COS'
re, )
(o )
(o)
U>\
(0)
(6}
O
o
(.03)
(,0/.)
to)
0
( ,oy )
(o)
(, 01 )
.nao
(0)
s in bra
lable me
ling poi
tribute
Claus
Plant
U
(J
U .__
U
i)
U
0
U
u
-U -- -
u
u
U ,
(.
u
LL.
U
u ,.,
u
. -LL
. .U
u ....
- . oo /
0
. rtOA
_^2£L_
,/6£> .
.A^y
.009
,oo/ __
.00 / .
c) ...
.00 f
,oo/
, O^3
.oo9
fj
-*aoi
,079
,M
o
,o^/
o
,0£
,0$
,0*
o
,03
.o/
o
,03
from is
he sourc
vel. ....
Total
All
Sources
fFig 291
,y^
13
-_-y<*
.30
.O£
.^
,0(0
.07
. Of
./.-$-
.(J7
.,33
./y
,^v
./a
,04
,/9
,/9
-J.L-
-^0^__
IA
,/v5
.//
07
,09
./I
.if
,/3
,/o
,
,O<3-
,os~
.^L^l___
,oy
,o£>
,03
pl eth-S
and,
-------
Full-Scale Ground Level Concentrations of S02 Calculated from Model Test Data
(Data listed on this page were measured at a wind speed too low for proper modeling.)
Wind Direction- 180°
Wind Velocity: Full Scale - 8 m/sec @ 2926 meters above river level
Model - 0.22 m/sec @ 1.22 meters above river level
1
Sampling
Point No.
{See Figs
30-331
1
o
3
V
.V
6
S)
₯
y
/c
//
/^
/J *
/V
A5 **
/6
77
//
/ q
-yO
" Glassp
* » . . ;- r
Libert
Ground
Elevation
\bove
River
fmetersl
O
l/(f
fs
IKo
f.5'
^)
^c'
/)
.f^"
^vT
/?v
f'tf
o
,5^5
.f ^
/3y
//(e
C'
6
y/6
1
)rt Sampl
r Borough
S02 Concentration Levels
Cppm")
L'laus
Plant
(-)Q^-
fr
.OotC)
, r>&/
073.
.001
0
^
O
,009
0
0
< oo/
c
c
. 00 -2
-.CO/
,oo/
-. oo/
,oo/
*"Y*
Sampler
Jlairtori
Excluding
Zlaus
, 00(a
ft
,OO 9
,00 /
./)/£:
. Jl
0
.003
o/V
,003-
.M/
, OO'/
Total
Clairton
& Glaus
O /
n
r, d
.09
,/>£--
,0(o ._
o
. 03.
,O3
o
O
A
ft
/)
£>
ft
\
I
_ . - -
73
-------
Full-Scale Ground Level Concentrations of SO- Calculated from Model Test Data
Wind Direction: 180°
Wind Velocity: Full Scale - 15 m/sec @ 2926 meters above river level,
6 m/sec @ Liberty BoroughSchool
Model - 0.41 m/sec @ 1.22 meters above river level,
0.16 m/sec @ Liberty Borough School
Sampling
Point No.
tSee Figs
sn-33)
/
,-5
^
V
^
(o
'I
J'
9
A;
//
/47
/3 *
/-/
/3T**
/to
/7
//
19 .
0?£
* Glassp
*»_., "
Libert
Ground
Elevatio
Above
Riv^r
(meters")
(0
//6>
85
//6
/r
c-
vs
0
^
fS'
/&/
-6'y
/)
.
.M&
,a>;j
, 6fl3
. n*>a
,009
.{,03
.003
.w
, CO/
,637
.r,ot
DO 3
,80*/
,noz
;r
Sampler
^lairton
Excluding
Claus
,n,33
,oo/
,r*3.y
,009
,&')
-------
I
I
Full-Scale Ground Level Concentrations of S0? Calculated from Model Test Data
Wind Direction: 180°
Wind Velocity:
Full Scale - 22 m/sec @ 2926 meters above river level,
8.8 m/sec @ Liberty Borough School
Model - 0.60 m/sec @ 1.22 meters above river level,
0.24 m/sec @ Liberty Borough School
Sampling
Point No.
(See Figs
30-33.) .._
/
-3
3
4
S~
In
1
s
3
/O
1!
J3
13 *
/V
/f**
i(a
/"?
_//
/Q
,36
* Glassp
Libert
Ground
Elevatio*
Above
River
Cmeters")
O
J/(a
fs
//(,->
fS
0
JS
r,
f.r
is ...
/.W
5'5
Cj
,
//6
irt Sampl
^ RO3?OLlffh
S02 Concentration Levels
fppm)
Claus
Plant
.F>O<3
-.00,3
- .(0(0^
-. f)f)3
03;?
,f#£
. n,~3.<
.r>&3
- , OOJ
03 /
.009
.CO/
.vox
-.no?
- ()O£
. f.. u
,n/(}
.r&£
.Mb.
.flQJ.
^r
.Samp! PT*
Clairton
Excluding
Claus
.61?
OOI
<&?-/
.007
. 0,f6
.0&j>
,A&O
./At,
- Of)/
,r>,v(s
.CJ.f
.cio
.not
o
,£'03
,Avvr
fjZ')
00-5
.fj/0
,6Qf
Total
Clairtnr
FT riaii<;
(Fig 3.1)
,f)^
D
.n.z
.n/
, nf
./^
,a3
,t>
-------
Full-Scale Ground Level Concentrations of SO^ Calculated from Model Test Data
Wind Direction: 180° (with inversion)
Wind Velocity: Full Scale - 15 m/sec @ 2926 meters above river level,
9 m/sec § Liberty Borough School
Model - 0.41 m/sec @ 1.22 meters above river level,
0.25 m/sec @ Libert)' Borough School
Sampling
Point No.
(See Fig
30-331
/
.Q
3
</
S
(»
']
$
9
JO
//
j<3
/3 *
/V
/s**
/6>
in
//
/X
0
.<6"
X&
/3V
/LL
o
o
IHt
rt Sampl
Borough
S02 Concentration Levels
(ppm)
Claus
Plant
,nn/2
.M/
06 /
0
.Of}
.OS")
,G/0
,f>0&
.6C3
033
,nn^
o
,C02
.M/
. 0/6'
.0/3
,cn/
,001
,003
,0fi#
r
Sampler
lairton
xcludins
laus
.046,
,003
,C*:O
.0,9 A
,O7 /
O7&
.Off
.A/ 7
,C&$
,r>3f
,0.49
,090
.M3
,00.^
,0/3
. 0,30.
,C3//
.(")/6
Total
Clairtor
S Claus
Fie 321
.os-
0 _
.OS
.03
,/0
,/3
.09
, 4*2 .
.01 .
.07
,0
-------
I
I
Full-Scale Ground Level Concentration of S02 Calculated from Model Test Data
(Data listed on this page were measured at a wind speed too low for proper modeling )
Wind Direction: 145°
Wind Velocity: Full Scale - 8 m/sec @ 2926 meters above river level
Model - 0.22 m/sec @ 1.22 meters above river level
Sampling
Pni nt Nn
-\-Se5 Fig
34-36)
/
c?
.?
V
9
/o
// **
13
/3
/-'/
y^'*
Jt>
/"?
~ //
'"/I
.-30
Glasspc
**, . v --
Liberty
Ground
P.I f»va-H m
ADOV6
River
(meters)
4"-5"
^,5-
v5^T
5
rO
.-r^
7/6-
^-T
(0
6"5~
rt Sample
Porpugh.
S02 Concentration Levels
(ppm)
Llaus
Plant
.fipf
no^i
,003
,DC^
,OA(c
.003
.MV
,/)09
,oc/
,ncfjO£
,cfo
,Mt
.6/6-
O
lOax
,O/f
.Of?/
,031
.flM
r
Sampler
Clairton
pxcludinj
Claus
.OO/
G
0
.06 f
,00^
.001
,00^
.0/0
0
,nc.f
£
.oil
<%y,
.CM
6
,M/
,063
,no9
,003
, OCX
Total
riairt.ni
fi Clans
,n/
0
&
.T;/
>0f
o
,01
,02
n
Gl
0
,0? . .
0- .
,03
o
0
,0 /
,fj3
,03
L6(.
,
'
_
..
i }
r
77
-------
Full-Scale Ground Level Concentrations of S07 Calculated from Model Test Data
Wind Direction: 145°
Wind Velocity: Full Scale - 15 m/sec @ 2926 meters above river level,
6 m/sec (d Liberty Borough School
Model - 0.41 m/sec @ 1.22 meters above river level,
0.16 m/sec @ Liberty Borough School
"
Sampling
Point No
(See Fig
S4-361
/
J?
.3
4
<5
6
^
/
?
/6
'" // * *
/^
/J
/V "'
/5' *
/6
n
i#
/9
-3O
' Glassp(
**T' . ~ 7~
Libert^
Ground
:levatioi
Above
Rivpr
(rafttprs)
^To"
U2 Concentration Levels
(ppm)
Glaus
Plant
,0(jJ
£>
<£>(}/
.00 /
J)0/
.Mf
,COI
, £GS
C
fi/1
0
.6&/
- . 063
.OA9
- .(Y\l
./)f>l
.OO/
fi
.M(e
.0/1
r
Sampler
Clairton
Excludinj
Claus
, r>r>/
o
,oo/
,do./
,063
,66,3
,£03-
,033
0
.MO
0
.30
,OGf
,Mt
,f
, oo-y
.0-6S
,03$
.,
Total
Clairtoi
S Claus
fFig 34^1
fl
0 ...
0
/)
O
Q
.01
,Cd_ ___
O
.££. _
^
,c-P3.
61
,/^
6.. __
£l_
0
^
.M .
,W
L .
i
1-
78
-------
I
I
I
Full-Scale Ground Level Concentrations of S0? Calculated from Model Test Data
Wind Direction: 145°
Wind Velocity:
Full Scale - 22 m/sec @ 2926 meters above river level,
8.8 m/sec @ Liberty Borough School
Model - 0.60 m/sec @ 1.22 meters above river level,
0.24 m/sec @ Liberty Borough Sc
Sampling
Point NO
(.see t-'ig
34-361
1
£
s"}'
V
$~
^
^
j1
9
/£
//**
«
/3
M
IS*
/i
/I
//
n
3 0 _^
Glasspc
Libert>
Ground
Elevatioi
Above
Ri V^T
(met <=-!-<;)
v£3~
5"^
0-5-
0
n
#2 Concentration Levels
fppml
Claus
Plant
.OC'V
M /
./9V
,0/3,
,0.15
,00 7
r
Sampler
Clairton
Fxrliirlinj
fllaiis
, o/c>
,s>a/
.MX
.MI
,0^
.ne>z
, M-J
,r,36
,va/
, C}/(-.
, OM
JAC
,003
,A^
,001
iCoa
, 00
,0/
,0/
,ai
./)/
,n/
.oV-_ .
0
,0,3
.rt/
.,30
n
,os-... ..
.of
,o/
, o-+
.07
,00
r .0 y.
iool
79
-------
Full-Scale Ground Level Concentrations of SO Calculated from Model Test Data
Wind Direction: 145° (with inversion)
Wind Velocity:
Full Scale - 15 m/sec @ 2926 meters above river level,
9 m/sec (4 Liberty Borough School
Model - 0.41 m/sec @ 1.22 meters above river level,
0.25 m/sec @ Liberty Borough School
amp ling
oint No.
(see Fig
34-361
/
£
3
4
,-f
(a
1
j
9
/c-
//* *
;-2
/3
/y
Af *
/b
/-?
JS
19
,3O
* <-!
Cil a^<;p
**T 1
Libert]
Ground
Elevation
Above
River
(meters!
£S
5*
SS
0
0
/
.CAI
,M)/
.OGb
.Of)/
.067
c
,091
0
OOI
,OC/
,0£)''
.fl/3
,OC'£i
'T
Samp! pr
lairton
xcluding
laus
. C)l/
.oca
,061
,c>//
o
, ^ /
n
,aw
,oo/
,C)OJ
o
. C-5L
,003
,0/d
0
,oo<;
o
,OSf
. o/x
,U)J
Total
Clairton
ft C1a,,<=
Fig 361
.0'
0
0
,n/
o
ft
o
,03
6
.Cl
0
JS>
o
,c*/_ ..
o
.o/
0
.r>(o
,03
'01
____
-,-
i
-
80
-------
I
I
I
Full-Scale Ground Level Concentrations of S02 Calculated from Model Test Data
(Data listed on this page were measured at a wind speed too low for proper modeling.)
Wind Direction: 331°
Wind Velocity: Full Scale - 8 m/sec @ 292b meters above river level
Model - 0.22 m/sec @ 1.22 meters above river level
Sampling
Point No
(See Fig
37-3Q)
/. _.
0?
3
y
6 *
b
1
j
4.- ._.
SO**
u // - -
A5
A3
H .
is-
!(
n
us
/y
"~~ ^r>
Glassp
**Libert
Note: U
t
Ground
F. ] pvat i or
Above
River
("meters!
cT5-
0
//(e
-T6-
A
L--
/.J
o?V
c
_/A^T. .
=?5-
9V
C
0
_^k__.
_1A-__H
G
ML i
yrt_SamiiL
r Borough.
designat
lus will
S02 Concentration Levels
u-^^ tojm^
TTaus
Plant
U
U
_kL___.
U
U
(_i
Y~ ^
U
U
U .
1 '
^
,__-_,__
(_l
(J
, £Od
_,C/2._.
,t'^J?
,CC(
~,06u>
_^l£J£__n
i"
;r
Sampl ar
;s samplii
tot contr
LlairTon
5 Irvin
jJiiLludiry
Glaus
f' 1
- .L.CI
ig points
.bute to
"Total
Glair ton
C 1 aus__ _£_
Irvin_. .
0
__u2/__
O
0
.01
0!
t.)
Cl
,01
6
c '
, Cl
_^L_
c
j2I_
,A-5
C; -
1 ,^;/
,(/
/"} -
are upwi
:he total
id of sou
S02 leve
:ce and,
-§,
_
1
_
-
81
-------
Concentrations of S02 Calculated from Model Test Data
Wind Velocity:
Full Scale - 15 m/sec @ 2926 meters above river level,
6 m/sec @ Liberty Borough School
Model - 0.41 m/sec § 1.22 meters above river level,
0.16 m/sec @ Liberty Pprnnph School
Sampling
Point No
(,See t^'ig
37-391
/
.
^ I-'
Glasspc
**Libertv
Note: U
tl
Ground
Elevatio
Above
River
fmeters)
-------
I
I
Full-Scale Ground Level Concentrations of S09 Calculated from Model Test Data
Wind Direction: 331°
Wind Velocity: Full Scale - 22 m/sec @ 2926 meters above river level,
8.8 m/sec & Liberty Borough School
Model - 0.60 m/sec @ 1.22 meters above river level,
0.24 m/sec @ Liberty Borough Sc
Sampling
Point No
(.see Fig
37-391
/
3
3
4
3*
6>
9
^
1
/0**
II
19
/3
/V
/\r
/6
/?
/^
/9
AO
Glasspc
*"*r v 1
Libert^
Note : U
t
Ground
Elevatioi
Above
Rivfir
(met fir q)
,TQQ/
,OJS
0
,OO£
Obf
0/Cr.
,CXjiT
.ay)
,66(3
,O03
, CO 6s?
,or>$
,003.
.J3b
,CTl(a
,009
,0^
,00.3
ig points
Lbute to
Total
nairt.nn
Clans FT
Irvin
(Fig 38)
0
,n*S
0
.£/
,lW
:O&
.cf
,rtjj
,0/
o
0
, o/
,0(#
,o/
o
, If
,0/
.o/
.01
o
are upwi
:he total
id of the
100!
source a
SO leveh..
2 |
id.
1
83
-------
Full-Scale Ground Level Concentrations of SO Calculated from Model Test Data
Wind Direction: 331° (with inversion)
Wind Velocity:
Full Scale - 15 m/sec @ 2926 meters above river level,
9 m/sec <& Liberty Borough School
Model - 0.41 m/sec @ 1.22 meters above river level,
0.25 m/sec @ Liberty Borough School
Sampling
Point No
(.bee Mg
37-391
/
,9.
^
4
sr*'
6 ..
_2_ _
X
9 ..
/o**
//
13
13
14
/s-
^
0
lie,
K
^1
b
0
ll(*
SS
o
Mt*
32 Concentration Levels
(ppm)
CJlaus
Plant
v
U
U
u
U
u
u
-U
u
u
u
i)
u
1 1
OC&
,0^)
,001
.Of>3.
.013
,061
e?
Samp! e>r
ites samp
contribu
Clairton
PT Trvin
Excluding
Glaus
.oaf
, 00^
0
.D6t
,M4
, OC 1
,00?
.OtYJ-
O
0
.DM
0
. MS
. on/
,OV(*
,£>(**
.001
,
-------
1
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
-119 FT
TEST SECTION
NOMINAL SIZE = 6 FT HIGH X 8 FT WIDE
FAN AND NOISE
ATTENUATOR
SYSTEM
FLOW
DEVELOPMENT
REGION
50 FT
TEST
SECTION
30 FT
Figure 1 THE CALSPAN ATMOSPHERIC SIMULATION FACILITY (ASF)
85
-------
CO
Q
CO
o:
O
LL
U-
C/5
LU
I
Q
LU
CO
z
LU
Q
O
CM
0)
i_
3
86
-------
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
Figure 3 CONCENTRATION SAMPLING SYSTEM
87/88
-------
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
CHANNEL
NUMBER
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
1
2
3
4
5
6
7
8
9
%DEVIATION FROM
REFERENCE CHANNEL #12
-10%
+10%
I
(NOT USED)
o
(REFERENCE)
I 1
I
I
I
-10%
I
+10%
SOLID LINE: INITIAL CALIBRATION 6/9/75 WITH 925 ppm He in AIR
DASHED LINE: FINAL CALIBRATION 8/6/75 WITH 50 ppm He in AIR
Figure 6 CALIBRATION OF SAMPLING CHANNELS
91
-------
LL
0
co
E
u.
>
t
Li
IS
\ :
\ c
\
N
LU
;LU «
CC U
LU O
LL _l
LU LU
CC >
/
'
1-
!>|z
J LU <
f s^ ^f
C LU Q
^
"*
U
"^.
E
» s
II d
Q. 0)
3 CC
,
/.
/
ro /
LU *
u /
z i
LU \
LL
LU
CC
.. .
y
en
£
LO
II
M-
OL"
5
/
/
/
'
V
' ^^^^
u
0)
vt
.
0)
3°" t
LO
II
0)
CC
*/
r
>
/
,
-
r
%
N
1
O)
»/
/
/
^^* «
1
I
T
...i...
T
1
1
-..J-.
1
1
1
1
.. 4.,
I
-J
/
1
^^
>^.
1
1
1
1
\
\
}
1
-~.
(
1
U
r
a
s
^
Q
U
a
£
c
I
1
f
0
Z
<
r
H
-J
(
i
)
) .
j
j
j
i
)
1
7 --.
0
f
i
c
L_
\/
o
o
ICALLY Sl\
x
IXHYDRAUL
1
- A-
o
N 2x
D
I
Z
LU
i
CO
CO
LU
Z
I
C3
D
O
CC
SB
o
LU
CO
CO
CC
LU
CO
D
z
CO
Q
o
z
LU
CC
LU
o
o
2
I
l-
0
LU
1
CO
CO
LU
21
I
o
o
oc
D
Z
D
O
CC
o
u
~
CO
CC
LU
CO fZ
LJJ<
?
U.LU
LUCC
(O
CM
CO
d
(O
d
HDDOH A13131dlAIOO UOd H10N31 SS3NHDnOU Ol H10N31 SS3NHDnOU JO OllVd
92
-------
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
2000
1500
se
$
0)
DC
LU
>
cc
LU
CO
1000
I
UJ
LL
500
uref = 15 m/sec (FULL SCALE AT
225° WIND
O NEUTRAL ATMOSPHERE
WITH INVERSION
HIGH-SPEED
CALIBRATION
PROFILE
0.5
1.0
u/
uref
Figure 8 VELOCITY PROFILES AT MITCHELL
93
1.5
-------
2000
1500
IT
LU
>
5E
LU
S
m
<
h-
LU
X
1000
500
ATMOSPHERE
ATM TCHELl.
uref = 15 m/sec (pULL SCALE AT
225° WIND
O NEUTRAL ATMOSPHERE
WITH INVERSION
0.5
1.0
uref
1.5
Figure9 VELOCITY PROFILES AT ELRAMA
94
-------
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
2000
1500
as
J
CC
I
<
K-
X
u
UJ
I
UJ
u
CO
1000
500
ATMOSPHERE
NEUTRAL
PROFILE AT
0.5
u/.,
Jref
Figure 10 VELOCITY PROFILES AT CLAIRTON
95
1.0
-------
DC
LU
>
DC
111
5
00
LU
I
LU
o
00
1500
1000
500
ufef = 15 m/sec (FULL SCALE
225° WIND
O NEUTRAL ATMOSPHERE
ft WITH INVERSION
ATMOSPHERE
u/,
uref
Figure 11 VELOCITY PROFILES AT LIBERTY BOROUGH SCHOOL
96
-------
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
2500
2000
01
S
LU
o
CO
C3
LU
X
o
CO
1500
1000
500
u f = 15 m/sec (FULL SCALE AT 2926 meters)
0
-14 -12 -10 -8 -6-4-20
FULL-SCALE TEMPERATURE RELATIVE TO ADIABATIC PROFILE, T - Tz (°C)
Figure 12 INVERSION TEMPERATURE PROFILE AT MITCHELL
97
-------
2500
2000
S2
o>
cc
LU
1500
LU
I
Lu 1000
z
LU
o
CO
500
uref = 15 m/sec
-------
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
2500
2000
0)
CC
UJ
1500
UJ
O
OQ
I-
I
o
V)
1000
500
i ' !
u f = 15 m/sec (FULL SCALE AT 2926 meters)
-14 -12 -10 -8 -6-4-20
FULL-SCALE TEMPERATURE RELATIVE TO ADIABATIC PROFILE, T - \\ (°C)
Figure 14 INVERSION TEMPERATURE PROFILE AT CLAIRTON
99
-------
2500
-X 2000
cc
uj
1500
O
CO
<
I-
X
u
LU
o
CO
1000
500
uref = 15 m/sec (FULL SCALE AT 2926 meters)
225° WIND
-14 -12 -10 -8-6-4-20
FULL-SCALE TEMPERATURE RELATIVE TO ADIABATIC PROFILE, T - T2 (°C)
Figure 15 INVERSION TEMPERATURE PROFILE AT LIBERTY BOROUGH SCHOOL
100
-------
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
(a) NEUTRAL ATMOSPHERE
(b) WITH INVERSION
Figure 16 SMOKE VISUALIZATION : CLAIRTON AREA : WIND SW AT 15 m/sec
AT REFERENCE HEIGHT.
101
-------
(a) MITCHELL WITH NEUTRAL ATMOSPHERE
^ »«**** *«*-l|(i
(b) ELRAMA WITH INVERSION
Figure 17 SMOKE VISUALIZATION: WIND SW AT 15 m/sec AT REFERENCE HEIGHT
102
-------
>- Z Q f
U3£ 3g
HLULU Q>
3 5s-is
S §
g.lsg^s s<|t
Suj£ H^X"; ctn: iu < ypz, ZICCLUO
*-i a^a: KU- quj3:<
LU <
^i
S
D
CD
CVJ CO
X
(OOtJ I 31V3S 1300W)
/ , ' saaiawonx aivos
HO1INOW
Honoaoa
A1U3SI1
H;.- - , - \
'- "">-< \
-------
<
<
cc
o
Q cc O
O uj S
S > 3-
O 0,3
cc -TO
" °o
Q en ^
uj <
< H^
-I (o CC
D (N UJ
O ?, CQ
I E|
& If)
^ II II
2 f £
< £C O O
cc UJ O O
^ Q. UJ UJ
O § UJ UJ
°&zz
O *< UJ LLJ
_, _| CC CC
uj < UJ UJ
> OC " u*
j ^ DC CC
Q UJ UJ uj
IQ-<<
O LU W W
O 13 u_
-------
<
I §
_J Z
LU O
O ccs
1 £5
§ =§
CC LU cc
< E cc
3 as
o en ~;
I 8 I
£ CM "i
CO ~ "
z II II
o >- >-
Pmbb
< CC O U
CC LU O O
h- X -1 J
O LU UJ
_J -J CC CC
LU < LU LU
> CC LL U.
Uj ^ LU LU
_1 D CC CC
Q Sid LU LU
-------
<
H
<
CC
O
Q _ O
o 2 s
s > x
S cc »
g -S
It > cc
Q Sg
P ^>
< E H
-I u) DC
D 8 "J
O Ol CO
" II II
£ E£
< o u
CC O O
H -1 -1
7 LJJ uj
UJ 2 > >
U o uj uj
u. u.
UU -r m m
^ I CC CC
Q UJ UJ
^ S _i _i
§ -«
;s Q tj o
O UJ CO C/3
<
O
X 03
J o -O
t m
-------
<
D
D
O
5
O
cc
cc
O
LU
> o
O m
Q eo
" = t
_i E cc
u < '
§ >>
;, ~ Ul LU
UJ E »
CJ W LU LU
Z O u o
O 5 Z Z
< "- "-
fT, ^ UJ UJ
Q O UJ UJ
= UJ -I -I
§ Z < <
O Q- co w
_ _
CM< D D
O _l LL LL
-
LU 5 9
« -
-I < -O
=J = in
D -J CM
LL LU CN
CM
CN
0)
I
LL
-------
Nouoaaia
QNIAA
\
aoiiNoiAi'
1 laodssvio..
(Ttf
aoiiNow
-Honoaoa
Aiaasn
10
<
Q
UJ g
8 E§
s > i
S ecu
g ^o
LU <
h- c >-
< E h-
_i co CC
13 Oi LU
Q Ol DQ
00
z
O
O O
O O
g < <
O -1 "-
§r
-i <
-------
" V -Tp ** "* "~ ' ' ' '''' ' \\
' ' "--4, "^ £ '"> -"=SoJ-NV1«i:f,V,T../-
^>^. ~ - .._" ^snwrv^j "- '
. « »° ;>'- VjV ^ " *"- s Ki
:'' #fV-:£s
-=>^4-,v |
'- - -- ^ v\ CTJ
-------
\ ,:?/; ,,^
\ ! ;*v- '..*"'>
> '< ,-
;' N :/ l ' v ;% '" "'*'' "
-\ V^'/N^f':,
^/VvN^r>--r^-^
.....-.:".t\\
i, > .,0V . "a2S J-NVId .--. '« \ . IHOdSSVID - sv
; Q -i -j
-------
-------
2 UJ UJ
LU Q > >
-------
-------
-------
-------
-------
Va^'i'ty) I : V-N' ?
V% r ©
Y M AV^"-"^
A,'^^, .VA 'x'0
-------
-------
<
I-
<
a:
o
Q e
O uj
o
EC
a m
LU <
_
E t-
to a:
CN LU
<
0
o
o
UJ K
O
-------
<
<
CC
o
C '^'^ " v ^': \*t\i' -
f'< ' ^fe^ij^: 'VA'^W'©"
i;
r
v"
l^-,X''"''' NIAHl
\.
5.
/ '^
y..\
'{
-.'V
-------
,- . ««'» * ' ©
{.''..r&f^^Z.
' i' fi r "a' * VT, <
.:-,M <'!<<^'v"-s* '-' -- -w, \ - ,
1''' . >?--,( '*«-' -,- ,.'..^*-' -, X ", \ / C s
'/' '' fl» W -.*--- ' .'. x ,-' \ x
it iW^-V'\ ^>""" -" - / ':o
^^^-u,iU^;,.v^,,
-------
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
o
o
I/I
s
"*
tu
CQ
CO
Q
_J
O
>-
UJ
OS
LO
CO
aj
a:
u
1
1
~ 'HIDN31 ssaNHDnoy 01 sssNHDnoy NIVHD oNvs iN3ivAinb3 do onvy
125/126
u
in
i i
a:
UJ
u
2
O
en
*
tu
CQ
to
Q
J
O
>-
a
CO
CO
ttl
3 2
O UJ
OS J
C CO
z. co
CJ3 CJ
tu O
O CH
E- a
U 2
PJ 3
a. o
tu oi
tq u
(O
f-l
-------
1
1
1
1
1
1
1
1
1
1
IB*
1
1
1.
2.
3.
4.
5.
6.
7.
8.
9
9.
10.
11.
REFERENCES
McVehil, G.E., Ludwig, G.R. and Sundaram, T.R. "On the Feasibility
of Modeling Small Scale Atmospheric Motions" Calspan Report
No. ZB-2328-P-1 April 1967
Ludwig, G.R. and Sundaram, T.R. "On the Laboratory Simulation of
Small-Scale Atmospheric Turbulence" Calspan Report No. VC-2740-S-1
December 1969
Ludwig, G.R., Sundaram, T.R. and Skinner, G.T. "Laboratory Modeling
of the Atmospheric Surface Layer with Emphasis on Diffusion"
Calspan Report No. VC-2740-S-2 July 1971
Sundaram, T.R., Ludwig, G.R. and Skinner, G.T. "Modeling of the
Turbulence Structure of the Atmospheric Surface Layer" AIAA Journal
Vol. 10 No. 6 June 1972 (Originally presented as AIAA Paper
No. 71-136 at the AIAA Aerospace Sciences Meeting, New York City,
January 1971)
Hoydysh, W.G., Diosey, P. and Yoshida, T. "Scaling Considerations
for Simulating Plume Rise and Plume Dispersion from Gas Turbine
Generators" APCA Paper No. 75-49.5 Presented at the 68th Annual
Meeting of the Air Pollution Control Association in Boston June 1975
Briggs, G.A. Plume Rise, Atomic Energy Commission Critical Review
Series, Division of Technical Information, TID 25075, 1969
Hoot, T.D., Meroney, R.N. and Peterka, J.A. "Wind Tunnel Tests of
Negatively Buoyant Plumes" EPA Report No. EPA-650/3-74-003,
October 1973.
Lumley, J.L. and Panofsky, H.A. The Structure of Atmospheric
Turbulence, Interscience (John Wiley and Sons) New York pp. 42-43
1964.
Yang, B.T. and Meroney, R.N. "Gaseous Dispersion into Stratified
Building Wakes" AED Report No. COO-2053-3 August 1970
Ogura, Y. "Diffusion from a Continuous Source in Relation to a
Finite Observation Interval" Adv. Geophysics Vol. 6 1959
pp. 149-159.
Hino, M. "Maximum Ground- Level Concentration and Sampling Time"
Atmospheric Environment Vol. 2 Pergamom Press 1968
pp. 149-165
127
-------
12. Tomback, I.H. "An Evaluation of the Heat-Pulse Anemometer for
Velocity Measurement in Inhomogeneous Turbulent Flow" Rev. Sci.
Instr. Vol. 44 No. 2 pp. 141 February 1973.
13. Schlichting, H. Boundary Layer Theory McGraw-Hill, New York
1960.
14. Cramer, H.E., Geary, H.V. and Bowers, J.F. "Diffusion-Model Calcu-
lations of Long-Term and Short-Term Ground-Level SG^ Concentrations
in Allegheny County, Pennsylvania" EPA Report No. EPA 903/9-75-OL8
March 1975.
128
-------
------- |