United States
Environmental Protection
Agency
Atmospheric Sciences Research
Laboratory
Research Triangle Park NC 2771 1
EPA/600/3-85/022
April 1985
Research and Development
Modeling
Demonstration of
Good-Engineering-
Practice Stack
Height in Complex
Terrain
-------
FLUID MODELING DEMONSTRATION OF GOOD-ENGINEERING-PRACTICE
STACK HEIGHT IN COMPLEX TERRAIN
by
William H. Snyder
and
Robert E. Lawson, Jr,
Meteorology and Assessment Division
Atmospheric Sciences Research Laboratory
U.S. Environmental Protection Agency
Research Triangle Park, NC 27711
ATMOSPHERIC SCIENCES RESEARCH LABORATORY
OFFICE OF RESEARCH AND DEVELOPMENT
U.S. ENVIRONMENTAL PROTECTION AGENCY
RESEARCH TRIANGLE PARK, NC 27711
-------
NOTICE
The information in this document has,been funded by the United
States Environmental Protection Agency. It has been subject to the
Agency's peer and administrative review, and it has been approved
for publication as an EPA document. Mention of trade names or
commercial products does not constitute endorsement or recommen-
dation for use.
The authors, William H. Snyder and Robert E. Lawson, Jr.,
are physical scientists in the Meteorology and Assessment Division,
Atmospheric Sciences Research Laboratory, U.S. Environmental Pro-
tection Agency, Research Triangle Park, NC. They are on assignment
from the National Oceanic and Atmospheric Administration, U.S.
Department of Commerce.
11
-------
ABSTRACT
A demonstration study using fluid modeling to determine the good-
engineering-practice (GEP) stack height for a power plant installation in
complex terrain is discussed. The site chosen for this demonstration
study was the Clinch River Power Plant in southwestern Virginia, and a
1:1920 scale model of surrounding terrain was const "tinted. Measurements
are presented that describe the simulated atmospheric boundary layer
structure, plume-dispersion characteristics in that boundary layer, and
the maximum ground-level concentration (glc) of effluent downstream from
the source, both in the presence of all significant terrain surrounding
the plant and in the absence of "nearby" upwind terrain. Analysis of the
maximum glc showed that, in this case, a stack height of 326 m meets the
current GEP criteria under 50% plant-load conditions, i.e., the nearby
upwind terrain effected an increase of 40% in the maximum ground-level
concentration.
IV
-------
PREFACE
This report was prepared for the purpose of demonstrating the
application of the fluid modeling approach to the determination of good-
engineering-practice stack height for a power plant in complex terrain.
The approach follows the general guidance set forth in Guideline for
Fluid Modeling of Atmospheric Diffusion (Snyder; I'.^'I) and the specific
recommendations set forth in the Guideline for Use of JFluid Modeling to
Determine Good Engineering Practice Stack Height (EPA, 1981) and the
Guideline for Determination of Good Engineering Practice Stack Height
(Technical Support Document for the Stack Height Regulations, Revised
Draft) (EPA, 1984).
Data plots are provided throughout the text; however, tabulated data
are not included because of the excessive printing costs that would be
involved with this widely distributed report. Such data should be
submitted to the reviewing agency for actual studies. Listings or magnetic
tapes of the data from the current study are available upon request from
the authors.
m
-------
CONTENTS
Preface Ill
Abstract 1v
Figures vl
Acknowledgements ix
1. Introduction 1
2. Technical Approach 3
3. Examination of Topography, Meteorological Parameters,
Selection of Area to be Modeled, and Plant Operating
Characteristics 4
3.1 Topography 4
3.2 Meteorological Parameters 5
3.3 Selection of Modeled Area 7
3.4 Plant Operating Characteristics 9
4. Evaluation and Justification of Modeling Criteria 10
4.1 Similarity Criteria 10
4.2 The Model 13
4.3 Relationship Between Model and Field Concentrations . 16
5. Results of Atmospheric Dispersion Comparability Tests .. 19
5.1 Boundary-Layer Characteristics 19
5.2 Dispersion Comparability Test 21
6. Determination of GEP Stack Height 27
6.1 Flow Structure Over Topographic Model 27
6.2 Dispersion Over Topographic Model 30
6.3 Further Discussion of Results 36
7. Summary 39
References 41
Tables 43
Fi gu res 45
Appendices
A. Description of Facilities and Instrumentation 70
B. Concentration Measurements for Other Stack Heights at
Half- and Full-Load Conditions 77
C. Data Listings 89
-------
LIST OF FIGURES
1. Topographical area modeled in wind tunnel. "T" marks location
of Tower site.
2. Cumulative frequency distribution of neutral wind speeds for the
22.5° sector centered on 264.5° at the 30 m level at the Tower
site.
3. Photograph of assembled model inside wind tunnel - view from
downstream.
4. Photographs of (a) nearby upwind terrain, which was removed and
replaced by (b) faired terrain.
5. Sketch of model layout in test section of wind tunnel. Dimensions
in cm. Not to scale.
6. Mean velocity profiles of the simulated atmospheric boundary layer
over rough flat terrain.
7. Turbulence intensity and Reynolds stress profiles of the simulated
atmospheric boundary layer over rough flat terrain.
8. Crosswind profiles of simulated atmospheric boundary layer over rough
flat terrain: (a) mean velocity, (b) longitudinal turbulence intensity.
9. Surface longitudinal concentration profiles of simulated atmospheric
boundary layer over rough flat terrain compared with Gaussian
predictions using Pasquill-Gifford stability categories C and D, and
HGS category D (z0 = 100 cm).
10. Vertical concentration profiles of simulated atmospheric boundary
layer over rough flat terrain compared with Gaussian predictions
using Pasquill-Gifford stability category C.
11. Lateral concentration profiles of simulated atmospheric boundary
layer over rough flat terrain compared with Gaussian predictions
using Pasquill-Gifford stability category C: (a) elevated profiles
through maxima in vertical distributions, (b) surface level.
12. Plume widths measured in simulated atmospheric boundary layer over
rough flat terrain compared with Pasquill-Gifford and HGS (z0 = 100 cm)
sigmas: (a) vertical, (b) lateral. Open symbols represent plune
widths at surface.
13. Comparison of mean velocity profiles over topographical model with
those over rough flat terrain.
14. Comparison of turbulence intensity profiles over topographical
model with those over rough flat terrain. Open symbols: longitudinal
turbulence intensity; closed: vertical.
VI
-------
15. Comparison of Reynolds stress profiles over topographical model
with those over rough flat terrain.
16. Lateral profiles of mean velocity over terrain at (a) x = 0 and
(b) x = 6.1 km downstream of the plant. Terrain cross sections are
shown at bottom of graphs.
17. Lateral profiles of turbulence intensity over terrain at (a) x = 0
and (b) x = 6.1 km downstream of the plant. Terrain cross sections
are shown at bottom of graphs. Open symbols: longitudinal turbulence
intensity; closed: vertical.
18. Surface longitudinal concentration profiles displaying Reynolds number
independence.
19. Sampling port locations (decimal points) on topographical model. Note
that upper and right-hand boundaries correspond with model boundaries;
others do not.
20. Surface concentration distributions observed with GEP stack height
(326 m) and all terrain in place. Numbers displayed are 10 times
actual values.
21. Surface concentration distributions observed with GEP stack height
(326 m) and nearby upwind terrain replaced by faired section. Numbers
displayed are 10 times actual values.
22. Surface lateral concentration profiles taken with GEP stack to locate
maximum ground-level concentration.
23. Surface longitudinal concentration profiles taken with GEP stack to
locate maximum ground-level concentration at y = 96 m.
24. Vertical concentration profiles taken at various positions downwind
of GEP stack.
25. Lateral concentration profiles taken at various positions downwind
of GEP stack.
Al. Typical calibration curve of hot-wire anemometer.
A2. Sketch of source and flow measurement apparatus used for injecting
gases into the wind tunnel.
Bl. Surface concentration distributions observed with existing stack
height of 138 m, at 50% plant load. Numbers displayed are 10 times
actual values.
B2. Surface concentration distributions observed with stack height of
300 m, at 50% of plant load. . Numbers displayed are 10 times actual
values.
vi i
-------
B3. Surface concentration distributions observed with GEP stack height of
336 m, at 50% plant load. Numbers displayed are 10 times actual values.
B4. Surface concentration distributions observed with GEP stack height of
326 m, but at 100% plant load. Numbers displayed are 10 times actual
values.
B5. Excess maximum ground-level concentration as a function of stack
height for 50% plant load conditions.
vm
-------
ACKNOWLEDGEMENTS
The assistance and cooperation of the entire staff of the Fluid
Modeling Facility are gratefully acknowledged. Particular thanks are due
to: Mr. Paul Bookman (Northrop Services) and Mr. Ralph Seller (NOAA) for
their painstaking efforts in constructing the model; Mr. Mike Shipman
(Northrop Services) for his programming support which permitted efficient
data retrieval, analysis, and graphical displays; to Ms. Anna Cook (NOAA)
for her patience in typing this report; Mr. Joe Smith (NOAA) for various
and sundry tasks; and to Mr. Larry Truppi (NOAA) for processing the field
meteorological data.
IX
-------
1. INTRODUCTION
Section 123 of the Clean Air Act Amendments of 1977 defines Good-
Engineering-Practice (GEP) Stack Height as "the height necessary to insure
that emissions from the stack do not result in excessive concentrations
of any air pollutant in the immediate vicinity of the source as a result
of atmospheric downwash, eddies and wakes which may be created by the
source itself, nearby structures or nearby terrain obstacles". A previous
report (Lawson and Snyder, 1983) provided a demonstration of the use of
fluid modeling to determine the GEP stack height for a power plant
installation where the plume was downwashed due to the presence of a
nearby structure, the power plant building itself. The purpose of the
current study is to demonstrate the use of fluid modeling to determine
the GEP stack height for a power plant installation where the plume is
downwashed due to the presence of nearby terrain. The model was based on
an existing facility, the Clinch River Power Plant, for which plant
operating conditions, meteorological parameters, and detailed topographical
maps were available. Every installation will have unique features, but
the fluid modeling approach is practical and, if applied properly, is a
useful tool for determining GEP stack heights.
Whereas a formula (Hg = H + 1.5L, where HQ is the GEP stack height,
H is the building height, and L is the lesser of the building height or
width) is available for determining the GEP stack height where the stack
is in the immediate vicinity of a building or structure, no such formula
exists for the case where downwash may be caused by nearby terrain features.
"The GEP creditable stack height, based on nearby terrain, must be determined
on a case-by-case basis through the use of appropriate field or fluid
-------
modeling studies" (EPA, 1981). The GEP stack height is that needed to
prevent excessive pollutant concentrations in the vicinity of the source.
The maximum ground-level concentration (glc) measured in a model that
includes nearby terrain obstacles is termed "excessive" when it is 40% or
more in excess of the maximum glc measured in a model that does not
include downwash, wake or eddy effects produced by the nearby terrain.
"Nearby" terrain is defined as that within 0.8 km (or 0.5 mi) of the
stack. The procedure specified by EPA (1984) for existing sources is
that the nearby and distance limitations would apply with respect to the
terrain feature(s) inserted and removed while fluid modeling. Further,
the lesser of lOHy or 2 miles is selected as the upwind extent of the
nearby terrain feature(s) for fluid modeling as long as such feature(s)
achieves a height (Hy), at or within 0.8 km (0.5 mi) from the stack, that
is greater than or equal to 40% of the GEP stack height (Hg) determined
by the above equation. The specific steps undertaken will be discussed
in the following section.
The document that provides specific definitions of terms and specifies
general procedures is referenced as EPA (1984) (hereafter referred to as
the Technical Support Document). The document that stipulates requirements
for fluid modeling GEP studies is referenced as EPA (1981) (hereafter
referred to as the "Guideline"). A more detailed reference (Snyder,
1981) provides technical standards for evaluation of various aspects of
this study.
-------
2. TECHNICAL APPROACH
Bearing in mind that the height of the stack is creditable as GEP if
the maximum glc in the presence of the nearby terrain is 40% greater than
that measured in its absence, the ultimate objective of this study was
simply to examine maximum glcs as functions of stack height, in the
presence and absence of nearby terrain. Other criteria specified in the
Guideline must be met in order to validate the fluid-modeling approach.
Certain steps specified in the Guideline must be followed when conducting
any GEP fluid modeling study:
1. Examination of the topography and meteorological parameters,
and selection of the area to be modeled.
2. Evaluation and justification of modeling criteria.
3. Evaluation of the test facility in the absence of buildings,
other surface structures, or large roughness and/or elevated
terrain (atmospheric dispersion comparability tests).
4. Determination of the GEP stack height.
5. Documentation of the facility operation, instrumentation
used in the study, and associated parameters.
These steps were followed and are reported in the following sections.
-------
3. EXAMINATION OF TOPOGRAPHY, METEOROLOGICAL PARAMETERS, SELECTION
OF AREA TO BE MODELED, AND PLANT OPERATING CHARACTERISTICS.
3.1 TOPOGRAPHY
The site for the study was the Clinch River Power Plant near
Bristol, VA. This particular site was chosen after a survey of several
potential sites primarily because (a) topography of the site showed
excellent potential for demonstrating a relatively large GEP stack height,
(b) details on plant operating parameters and an extensive set of
meteorological data were available from a previous field study (Koch et
al, 1979; Pickering et al, 1980), and (c) a large portion of the scale-
model topography was available from a previous wind-tunnel study (Thompson,
1979).
The plant has two stacks of height 138 m. It is located near the
base of a U-shaped bend in the Clinch River, which flows generally WSW
(Figure 1). A prominent hill is located due west of the plant, with a
peak reaching 210 m above stack-base elevation (461 m MSL) at a distance
of 0.6 km. A lesser hill lies ENE of the plant, with a peak reaching 161 m
at a distance of 0.7 km. In the range of 10 to 20 km from the plant,
steep hills rise in all directions to maximum heights in the vicinity of
400 m. The hills are generally covered by trees and rock outcroppings.
The primary plant structure is approximately square, 122 rn on a
side, with a height of 59 m, so that the "formula" GEP stack height is
HQ = 148 m. According to the Technical Support Document, then, the terrain
height must exceed 40% of HQ, or 59 m, within a distance of 0.8 kn of the
stack in order to meet the requirement of "nearby" terrain. This
requirement is easily satisfied, since the terrain reaches 210 m within
0.8 km.
4
-------
From study of the topographic maps and especially a 1:1920 scale
model of the terrain still available from the previous wind tunnel study
(Thompson, 1979), a wind direction of 264.5° was selected as likely to
show the highest GEP stack height.
3.2 METEOROLOGICAL PARAMETERS
In an earlier field study, meteorological measurements were made at
a series of 8 fixed sites over a period of 16 months (Koch et al, 1979).
That report established that the most common wind direction was from WSW.
Of the 8 possible field sites, the "Tower" site (shown in Figure 1) was
selected as the most appropriate for analysis and determination of the
"98th percentile" wind speed, because of all the sites, it was closest to
the plant, the 30 m level (164 m above stack base) was closest to plume
elevation, and it was a relatively unobstructed site located on an elevated
plateau 3.4 km northeast of the plant.
According to the Guideline, the design wind speed should be less than
the speed that is exceeded 2% of the time for the given wind direction.
This wind speed was obtained as follows. A magnetic tape of the meteoro-
logical data was available from the previous field study. Hourly values
of wind speed at the 30 m level of the Tower site were sorted, selecting
out only those winds in the 22.5° sector centered on 264.5°. That subset
was then sorted by stability class through the use of a bulk Richardson
number RB, selecting out only those cases where Rg indicated neutral stability
(Pasquill-Gifford class D). This was a slight modification of a technique
shown to be reasonable in the report by Pickering et al (1980). In that
report, the bulk Richardson number was determined using hourly Tower site
measurements as follows:
-------
RB = g(AT/Az + y)z2/Tu2,
where g is acceleration due to gravity (9.8 m/s), AT/Az is the vertical
temperature gradient determined from the 30 m and 0.5 m levels, 7 is
taken as 14.75 m, the 4 m temperature is used as T, u is the 10 m wind
speed, and y is the dry adiabatic lapse rate (-0.00976° C/m). The range of
RB values used for classification of D stability was -.01 < RB <_ 0.015.
This range is slightly wider than that used by Pickering et al (1980)
(-.005 < RB j< 0.01), in order to obtain a more reasonable proportion of
hours for D stability class. If this method errs, it does so conservatively,
i.e., in the direction of a lower design wind speed. Values provided on
the data tape were "resolved" to 0.1° wind direction, 0.1° C in temperature,
and 0.01 m/s in wind speed.
The selected subset of neutral stability data is plotted as a cumulative
frequency distribution in Figure 2, where the 98th percentile wind is
observed to be 12.2 m/s. The fraction of winds in the chosen sector was
greater than the average fraction per sector (1/16). Of all winds in the
chosen sector, 35% were classified as neutral (Pasquill-Gifford D Stability)
using the above method.
Ideally, a one-year record (or an integral number of years) should
be used in the determination of the 98th percentile wind. The field
study lasted 16 months, but because of lightning damage and floods, this
record contained some large gaps. Because of this, the full data set was
used, about 11 months of record, but data for months of August and October
were missing, and June months for two successive years were included.
Note that similarity criteria (see Section 4) require a matching of the
effluent speed to wind speed ratio. Generally, this requirement refers
-------
to the wind speed at stack-top elevation. Note that in the current study,
the wind speed used in the ratio is the 98th percentile wind at the Tower
site, 3.4 km distant from the plant, and the actual wind speed at stack
top is, in fact, unknown. However, we may reasonably assume that, since
the various similarity criteria are satisfied, the ratios of wind speeds
at corresponding points in model and prototype are independent of the
position where measured, so that the ratio of effluent speed to the wind
speed at the Tower site (30 m level) is just as appropriate a parameter
to match as is the ratio of effluent speed to wind speed at stack top.
The Guideline requires that the model boundary layer be scaled to
represent 600 m in the field, but points out that this depth is not
critical. The pibal measurements from the field suggest that the boundary-
layer depth is in the neighborhood of 1 km under neutral conditions
(Thompson, 1979). This value appeared more reasonable under the current
circumstances, especially as the terrain heights frequently exceed 200 m,
and therefore was used for design purposes.
3.3 SELECTION OF MODELED AREA
An area 12.8 km long and 7.0 km wide was modeled in detail. This
provided an upwind fetch of 6.5 km and a downwind fetch of 6.3 km. The
upwind fetch was sufficient for development of appropriate boundary layer
characteristics at the source, and the downwind fetch extended beyond the
expected distance to the maximum in the glcs. For further details, see
Section 4.2.
-------
The study was conducted with near-westerly winds (264.5°) under
conditions of neutral atmospheric stability. This wind direction was
chosen because it results in the largest nearby hill being directly
upwind of the stack and therefore should yield the highest GEP stack
height. Because of the irregular shape of this upwind hill, some latitude
is allowed in the selection of the precise wind direction to model.
Indeed, the direction chosen for the present study was approximately 10°
different from the direction toward the highest peak; this direction was
chosen from a consensus of the persons involved in this study. Hence,
the question arose as to what value to use for the terrain height Hj in
the determination of the distance limitation (see Section 1, or for more
details, the Technical Support Document, Option C, which allows an exclusion
of nearby terrain, i.e., smoothing and sloping of nearby terrain features,
falling within a "distance limitation" of 10 Hy or 2 miles (3.2 km),
whichever is less, for establishment of the model baseline). Pursuant to
discussions with EPA's Office of Air Quality Planning and Standards, the
height Hy was established to be the height of the highest upwind terrain
(above stack base) within 0.8 km and within a 10° sector centered on the
chosen wind direction. This value was Hy = 191 m in the current study.
Hence, all terrain remained in position at all times in the GEP
portion of the study, except for a roughly semicircular arc of upwind
terrain (to 10 Hy) which was removed and replaced with sensibly-faired
terrain in order to establish the model baseline for each stack height
tested. The creditable stack height was thus the one where the maximum
glc in the presence of nearby terrain was 40% in excess of that with
faired terrain (same stack height in both cases).
-------
3.4 PLANT OPERATING CHARACTERISTICS
The Clinch River Plant has a maximum generating capacity of 712 MW.
Exhaust gases are emitted through two 138 m high stacks, 46 m apart. The
diameters are 4.75 m and 3.8 m. Effluent conditions were specified by
Koch et al (1979) as shown in Table 1. The plant operates essentially as
a base load plant, with normal diurnal variability of its output load
from a day-time peak of 90% to a nighttime low around 60%. Hence, other
questions that arose concerned (a) whether one or both stacks should be
used, and (b) what plant load to use in establishing the GEP stack height.
Although not necessarily always true, it is most likely the case that the
lowest plant load (even in the absence of stack-tip downwash) will result
in the largest creditable GEP stack height, primarily because of the
reduced effluent momentum rise.
Pursuant to discussions with EPA's Office of Air Quality Planning and
Standards (OAQPS), 50% plant-load conditions were simulated and, once the
GEP stack height was determined, 100% load conditions were simulated
using that stack height to determine if a higher stack might be justified
under the higher load conditions. These results are shown in Appendix B.
Also pursuant to discussions with OAQPS, only the larger of the two
stacks was used in the model tests to establish the GEP stack height.
Note that exceedance of ambient air quality standards was not a factor to
be considered in these tests. Only the 40% excess concentration criterion*
was used to establish the GEP stack height. Hence, modeling of emissions
from the smaller stack was deemed unnecessary.
*The term "excess concentration" as used in this report refers only to the
increase in maximum concentration associated with wakes, eddies, and down-
wash due to nearby terrain compared to the case in the absence of nearby
terrain.
-------
4. EVALUATION AND JUSTIFICATION OF MODELING CRITERIA
4.1 SIMILARITY CRITERIA
As specified in the Guideline, there are five parameters in addition
to geometric similarity that are relevant when simulating dispersion in
atmospheric flows. These are the Rossby number, Peclet number, Reynolds-
Schmidt product, Froude number, and Reynolds number.
The Rossby number can be ignored here, because it represents the
effects of Coriolis forces in comparison with local or advective
accelerations and is significant only when modeling large prototype
distances. Snyder (1981) suggested a length-scale cut-off of about 5 km
in neutral conditions in relatively flat terrain. In complex terrain,
local and advective accelerations are greatly enhanced, so that this cut-
off distance can likely be increased quite significantly. In the current
study, terrain was modeled to a distance of only 6.2 km downwind of the
plant, and the Rossby number may be safely ignored.
The Peclet number and Reynolds-Schmidt product are indicators of the
importance of turbulent diffusivity in comparison with molecular diffusivities
(thermal and mass diffusivities, respectively). According to Snyder (1981)
molecular diffusivities are negligible provided that the Reynolds number
(Re) is large enough; i.e., Re is large enough that turbulent motions are
the primary mechanisms for dispersion.
The Froude number indicates the relative importance of inertia! and
buoyant forces. A neutral atmospheric boundary layer may be simulated by
an isothermal boundary layer in the wind tunnel, so that buoyant forces in
the approach flow are irrelevant in the current study.
10
-------
For precise scaling of buoyant effluent releases, the Froude number
of the effluent from the model stack should match that of the prototype.
However, the Guideline specifies that the ratios of stack diameter to
building height (or other characteristic length scale), effluent density
to ambient air density, and effluent speed to crosswind speed must be
matched between model and prototype. The Guideline does not require a
matching of Froude number (effluent buoyancy), so that the stack Froude
number was ignored.
The Reynolds number is the ratio of inertial to viscous forces acting
on an air parcel. For a given fluid, strict matching of the model and
prototype Reynolds numbers requires that the model velocities be increased
in direct proportion to the decrease in scale. As this is impractical
for large reductions in scale, the principle of Reynolds-number independence
is invoked to enable modeling under such conditions. Basically, the
principle of Reynolds-number independence states that the structure of
turbulent flow is similar at all sufficiently large Reynolds numbers.
Three Reynolds numbers were considered in this study, the roughness Reynolds
number (Rep), the terrain Reynolds number (Rej), and the effluent Reynolds
number (Res).
The terrain Reynolds number is defined as:
ReT = UTHT/v,
where Uj is the wind speed at the general elevation of the hill tops, Hy
is a length scale generally characteristic of the heights of the hills
surrounding the plant, and v is the kinematic viscosity of air. For
sharp-edged structures, the critical Reynolds number is 11,000 (Snyder,
1981). For the present model, the characteristic length scale was about
10 cm (200 m full scale) and the velocity at hill top elevation was
11
-------
approximately 3 m/s, so that Rej ~ 20,000. Because the terrain is not,
generally speaking, sharp-edged, but relatively smooth and rounded, the
critical Reynolds number could be considerably larger than the 11,000
value stated above, and Reynolds-number independence tests were therefore
conducted (see Section 6.2.1).
The roughness Reynolds number is defined as:
RBR = u*z0/v,
where u* is the friction velocity and z0 is the roughness length. These
parameters are very difficult to evaluate in complex terrain, and it is
perhaps meaningless to do so (see Section 6). In the atmospheric dispersion
comparability tests (Section 5), however, an attempt was made to simulate
an atmospheric boundary layer of the same depth over flat terrain with the
same roughness characteristics as the terrain model. In those tests,
u* = 0.19 m/s and z0 = 0.35 mm were obtained. Hence, we estimate
Rep ~ 4.4, well above the value of 2.5 required to insure an aerodynamically
rough surface (Snyder, 1981).
The effluent Reynolds number is defined as
Res = WsD/v,
where Ws is the effluent speed and D is the internal diameter of the stack.
The prototype and model values of this parameter (half load) were
approximately 9 X 10^ and 1 X 10^, respectively. The Guideline specifies
that if this parameter is below 2000, the flow should be tripped to insure
a turbulent flow at the stack exit. A thin washer with ID of 1.59 mm was
inserted 10 stack diameters (ID = 2.48 mm) upstream of the stack exit to
trip the flow and ensure a fully turbulent exhaust, which was verified
using smoke and flow visualization.
12
-------
4.2 THE MODEL
Values of various parameters in the prototype and in the model are
listed in Table 2. The scale ratio selected, 1:1920, was based on several
considerations. Compromises were necessary to meet the opposing requirements
of large Reynolds number, fully developed boundary layer, adequate up- and
downwind fetch, and limited length of the wind-tunnel test section. To our
advantage was the fact that a previous wind-tunnel study, albeit at a
different wind direction, had shown quite favorable comparisons with
field data, both in terms of flow structure and dispersion patterns
(Thompson, 1979). That study was done at much lower model wind speeds in
order to properly simulate the buoyancy (Froude number) of the effluent.
Considerations for determining the amount of fetch required in the
model were as follows. The Guideline specifies that a three-dimensional
hill upstream of the source should be included if its height exceeds
l/20th of the distance from the source, whereas an obstruction with
crosswind dimension large compared to its height should be included if
its height is greater than l/30th of its distance from the source. The
tallest hills upwind of the source were in the neighborhood of 200 m high
(peaks to local troughs) in the field or approximately 10 cm in the model.
Hence, the amounts of upwind fetch required were about 4 km and 6 km in the
field or 2 m and 3 m in the model for three- and two-dimensional hills,
respectively. Perusal of the topographic map (Figure 1) shows that,
generally speaking, the upwind hills were three-dimensional and conical in
shape. More recent wind-tunnel measurements of the wakes of conical
hills (Gadiyaram, 1984) suggest similar values. The actual amount of
upwind fetch used in the model was 3.4 m (6.5 km in the field).
13
-------
Enough downwind fetch was desired to include the region of maximum
glc. In flat terrain and neutral stability, the maximum glc may be
expected to occur at approximately 15 stack heights downstream. As the
tallest stack anticipated was 180 mm (350 m full scale), the required
downwind fetch was 2.7 m (5.2 km full scale). The actual amount of
downwind fetch used in the model was 3.27 m (6.3 km full scale).
The terraced model was constructed from 0.64 cm (1/4 in) plywood
sheets; each thickness of plywood corresponded to the 12.2 m (40 ft)
contour intervals of U.S. Geological Survey topographic maps to provide
undistorted geometric scaling in all directions of 1:1920. At this scale,
the 7 km x 12.8 km area surrounding the power plant was constructed to
occupy a 3.65 m x 6.7 m long portion of the tunnel test section. Because
of the previous wind-tunnel study, a portion of the model was reusable.
Because of the different wind direction, however, additional triangular
sections were constructed. The various subsections were assembled inside
the tunnel and bolted to the tunnel floor. Joints were filled with
architectural putty. For additional details of model construction, see
Thompson (1979). A view of the model assembled in the wind tunnel is
shown in Figure 3.
To avoid abrupt elevation changes at the up- and downwind edges
of the terrain model, gravel of approximately 5 to 10 mm diameter was
used for fairing from the model to the gravel-covered tunnel floor over a
length of approximately 1 m.
Other similarity criteria considered in the model design were the
ratios of roughness length to boundary-layer depth, and boundary-layer
depth to terrain height (Table 2). As mentioned previously, the boundary-
layer depth in the field was observed to be approximately 1 km. Thus, at
14
-------
a scale ratio of 1:1920, the design boundary-layer depth in the wind
tunnel was 52 cm. The measured value was 55 cm (see Section 5.1).
The roughness length z0 of the flow at the upwind edge of the model
should presumably match that characteristic of a heavily forested area.
From the Guideline, that value may be expected to lie in the neighborhood
of 0.9 m, suggesting a model design value of 0.47 mm. The value of z0
evaluated in the boundary layer in the absence of the terrain (but with
similar roughness) was found to be 0.35 mm (see Section 5.1), or about
0.7 m full scale.
For the baseline measurements in the absence of nearby terrain, a
roughly semicircular arc of terrain with radius lOHy or 99 cm (1.9 km full
scale) and with straight side perpendicular to the wind direction was cut
out and replaced by a faired section. The actual shape of the cut-out
was a polygon for ease of construction and is shown in Figure 4. The
replacement section was constructed to match terrain elevations at its
edges and faired down to flat terrain upwind of the plant. A photograph
shows the contours of the faired terrain in Figure 4b.
Brief descriptions of the wind tunnel and instrumentation are provided
in Appendix A. Further details may be obtained from Snyder (1979) and
Lawson (1984).
The boundary layer for the atmospheric dispersion comparability
tests was initiated using the basic Counihan (1969) system of fence and
elliptic-wedge vortex-generating fins, followed by a rough surface. In
the present case, the fins were 46 cm in height and were spaced on 23 cm
centers across the span of the test section. A castellated barrier with
base and top heights of the castel lations of 7.2 and 8.5 cm, respectively,
preceded the trailing edge of the fins by 61 cm. The fins were followed
15
-------
by Sanspray, a commercially available construction material of nominally
1 cm size gravel epoxy-cemented onto plywood sheets. The identical
system was used to generate the approach flow to the topographic model.
A sketch showing the layout of the model in the wind tunnel is provided
in Figure 5.
4.3 RELATIONSHIP BETWEEN MODEL AND FIELD CONCENTRATIONS
For the stack effluent, ethane was used as the tracer gas for the
concentration measurements and helium was mixed with it in such proportion
to provide the required effluent density. The concentration C (mass per
unit volume) of a pollutant at a position downwind of the stack is
proportional to the effluent rate Q (mass per unit time) and inversely
proportional to the mean wind speed U and the plume cross-sectional area
A. That is, C « Q/UA. To obtain a relation between the full scale
concentration Cf and the model concentration Cm, we assume similarity in
plume shape and spread between field and model,
Af/Am = L2/£2,
where L and i are characteristic lengths in the field and model,
respectively; the ratio L/£ is, of course, the geometric scale ratio of
1920:1. Hence, model concentrations can be used to calculate full scale
values according to
cf (Qf/Qm)(um/uf)U2/L2) cm. (i)
By defining a nondimensional concentration x = Cl)h2/Q, this equation may
be simplified to
xf = Xnr (2)
Air pollution meteorologists frequently use the parameter CUS/Q,
where Us is the wind speed at stack top (Turner, 1970). This form is
16
-------
also suggested in the Guideline, and is used in Section 5, where the
purpose is to make direct comparisons with atmospheric dispersion
parameters. Elsewhere in this report, however, concentration data are
reported in the nondimensional form
X = CUoohZ/Q,
where C is the measured concentration, Uoo is the free-stream wind speed,
h is a characteristic length scale, henceforth taken as 100 mm (192 m full
scale) to be characteristic of the height of the nearby terrain, and Q is
the volume flow rate of tracer. Particular note should be taken that the
wind speed used is the free-stream wind speed, not that at stack top as
used for CUS/Q in Section 5. The reason for usage of the nondimensional
concentration x and free-stream speed Uoo is that the "excess concentration"
definition requires calculation of the ratios of concentrations; the form
CUS/Q is not appropriate because the nearby terrain is quite likely to change
Us, the wind speed at stack top. Hence, ratios of the parameters in the
form of CUS/Q would not be concentration ratios. Since Uoo, h, and Q are
identical with and without nearby upwind terrain, the ratios of x are
identically equal to the ratios of concentrations.
Another reason for using the nondimensional x form is that this
parameter is identical in the model and in the field (Equation 2). Hence,
given this parameter in the model, it is not difficult to calculate the
field concentration using Equations 1 and 2, i.e.,
Cf = (Qf/U»fhf2) xm.
As an example, we will assume the full scale effluent rate (for purposes
of this example calculation only) is 500 g/s. From above, we use the
characteristic length scale of 192 m. The free-stream velocity in the
17
-------
model is 3.9 m/s and the velocity scale ratio is 5.55 (see Section 6.1),
so that U» = 3.9 X 5.55 = 21.6 m/s. Hence,
fi o
c _ (5GOgS02/Sx (1 X 10" m S02) (1Q6)
(21.6m/s) (192m) (2.93 X lo" gS02)
= 0.214 Xm,
where the units of Cf are parts per million (ppm) S02 and xm is the
dimensionless model concentration. Of course, once the field concentration
is obtained, calculation of CUS/Q is straightforward. Further discussion
pertaining to the relationship between field and model concentrations is
provided by Snyder (1981).
18
-------
5. RESULTS OF ATMOSPHERIC DISPERSION COMPARABILITY TESTS
One of the requirements of the Guideline is that dispersion in the
fluid model in the absence of buildings, other surface structures, or
large roughness and/or elevated terrain must be shown to be comparable to
that described for the atmosphere by the basic Gaussian plume distribution
(Turner, 1970). The results of these tests are described in this section.
The same basic boundary layer was used both here and as the approach flow
to the topographic model (in the GEP stack height determination tests).
5.1 BOUNDARY-LAYER CHARACTERISTICS
The system used for generating the deep boundary layer was described
at the end of Section 4.2. The free-stream velocity was set at 4.0 m/s.
The tunnel ceiling height was adjusted to provide a non-accelerating free-
stream flow. The topographic model was not installed, and the entire floor
was covered with Sanspray.
The Guideline requires that the following flow measurements be made:
1. Vertical profiles of mean velocity, longitudinal and vertical
turbulence intensities and Reynolds stresses at the position
where the stack would be placed (x = y = 0), downwind at the end
of the planned study area (x = 3270 mm, or 6.3 km f.s.) and
midway between these two positions (x = 1655 mm, or 3.1 km f.s.).
These measurements were made and additional profiles were measured
at the beginning of the planned study area (x = -3410 mm, or 6.5 km;
see Figure 5).
2. Lateral profiles of mean velocity and longitudinal turbulence
intensity along the model surface, and two elevated profiles
bracketing the range of plume heights evaluated in the study
19
-------
near the position where the stack would be placed, and near the
end of the planned study area. These parameters as well as
vertical turbulence intensities and Reynolds stresses were
measured at elevations of z = 38, 75, and 150 mm (73, 144, and
288 m f.s.}, both at x = 0 and 3270 mm (6.3 km f.s.).
Figures 6 to 8 show the measured boundary-layer characteristics.
Figure 6a shows that the mean velocity profile over the study area is
essentially nondeveloping, as little variation is observed between the
various profiles and all fit the 0.23 power law quite well. The boundary-
layer depth <5 is found to be 550 mm (1050 m in the prototype), very close
to the design value. A semilogarithmic plot of the mean velocity profiles
was used to determine the roughness length and surface shear stress
(Figure 6b). Note that the origin of the vertical coordinate was the
tops of the gravel stones of the Sanspray, an effective "displacement
height" of approximately 10 mm. All data points are shown in Figure 6b,
but only those below 50 mm (100 m in the prototype) were used to determine
the best-fit logarithmic law. The roughness length z0 was found to be
0.35 mm (70 cm in the prototype), and the friction velocity u*/Uoo was
0.049. These values are suitably close to the design values of z0 ~ 90
to 100 cm and u*/U«> = 0.05.
In Figure 7a, the longitudinal and vertical components of turbulence
intensity are plotted as functions of height. Again, these profiles show
the boundary layer to be fully developed at the upstream end of the
planned study area (topographic model). The general shapes of the turbulence
intensity profiles are reasonably consistent with examples in the Guideline,
and comparisons of surface layer values are excellent. Above the surface layer,
20
-------
however, the model values are somewhat smaller than those suggested in
the Guideline. The ratio of the vertical to the longitudinal component
in the surface layer is 0.5, consistent with values typically found in
the atmosphere. Figure 7b shows the shear stress normalized by that
determined from the mean velocity profiles. Although there is considerable
scatter, the data tend to collapse near the surface around a value of 0.9.
The relatively small difference between this boundary layer and the
naturally grown boundary layer presented in the Guideline is that this
one had a somewhat thicker constant stress region.
Lateral profiles of mean velocity and longitudinal turbulence
intensity are presented in Figures 8a and b, respectively. The maximum
deviation in mean velocity from the average was approximately 10% of the
average at that level; on a similar basis, maximum deviations in turbulence
intensity were about 20%. The homogeneity shown by these measurements is
regarded as excellent, since the variations observed correspond quite
closely with thoso that might be expected fom a series of repeat measurements
at the same point, i.e., normal scatter.
5.2 DISPERSION COMPARABILITY TEST
The Guideline requires that dispersion be measured from a model
stack of height 104 mm (200 m full scale). It further specified the
effluent flow rate and stack diameter. However, as the purpose of these
measurements was to test the dispersive properties of the boundary layer
in the absence of plume rise or stack downwash, neutrally buoyant tracer
gas (ethane) was released at the 104 mm elevation through a hollow,
perforated plastic ball 10 mm in diameter. This type of source allows
for injection of a suitable quantity of tracer gas, while at the same time
21
-------
minimizing the effluent momentum in any preferred direction; it simulates
a point source and avoids any plume rise or stack downwash. The specific
concentration measurements required by the Guideline were:
1. Vertical and lateral concentration profiles through the
plume centerline near the quarter intervals between the
source and the end of the planned study area. Additional
lateral and vertical profiles were required to clearly
show an elevated plume centerline height. Hence, these
profiles were measured at x = 409, 818, 1635, and 2453 mm
(0.8, 1.6, 3.1, and 4.7 km full scale). Vertical profiles
were measured first; then lateral profiles were measured
at the elevation of the maximum concentration determined
from the vertical profile.
2. Ground-level centerline longitudinal profile downwind of
the source to the end of the study area. Determination of
the surface ground-level centerline should be supported by
several lateral profiles of glc. The ground-level longitu-
dinal profile was actually measured to the end of the test
section of the tunnel, and surface lateral profiles were
measured at x = 1635, 2453, and 5437 mm (3.1, 4.7, and
10.4 km full scale); the last position is the end of the
test section.
Figures 9 through 12 present the concentration measurements for the
104 mm (200 m full scale) stack. These measurements were converted to
equivalent full-scale concentrations in the form CUS/Q (m~^) (see Section
4.3) for comparison with dispersion estimates using Pasquill-Gifford (PG)
22
-------
stability categories (Turner, 1970), where C is the measured concentration,
Us is the wind speed at stack-top elevation, and Q is the volume flow
rate of tracer. Figure 9 compares the measured surface concentration
profiles (longitudinal) with estimates using PG C and D stability classes.
PG class C provides the best fit to the data upstream of the maximum glc,
but the data appear to be asymptotic to the class D curve well downstream
of the maximum. The location of the peak xmx falls between those for C
and D stabilities, and the value of the maximum glc Cmx is approximately
20% larger than that for C stability.
Familiar expressions that have long been used as the basis on which
to interpret data on concentrations from elevated sources are (Pasquill
and Smith, 1983, p. 269):
x = xmx when Hs/az = /2
and Cmxus _ 0.24 fz
Q -H7 ./
Because xmx falls between those suggested by the class C and class D
curves, we may surmise that az in the model is larger than typical
atmospheric values for neutral stability (class D) and smaller than those
for slightly unstable (class C) conditions. Because Cmx was larger than
that suggested by the class C curve (and considerably larger than that
suggested by the class D curve), we may surmise that ay in the model is
less than or at best equal to typical atmospheric values for neutral
stability. More definitively, the ratio oz/ay is slightly larger than
that typical of a slightly unstable atmosphere (class C).
In the above discussion, it must be remembered that the PG curves
were derived on the basis of a 10-cm roughness length, whereas the present
23
-------
roughness length is equivalent to 70 cm full scale. Better estimates of
the vertical dispersion coefficients for the larger roughness length may
be made using a distance-dependent roughness-correction-factor technique
attributable to Hosker (1974), Gifford (1975) and Smith (1973). The
Hosker, Gifford, Smith (HGS) class D curve for a 100-cm roughness length
is close to the PG class C curve (z0 = 10 cm) near the source, but tends
toward the PG class D curve far downwind; at 6 km from the source, the
HGS class D curve falls about halfway between the PG class C and D curves.
Also shown on Figure 9, then, is the longitudinal glc profile derived
using the HGS technique for the vertical dispersion coefficient (oz, class
D, z0 = 1 n) and the PG system for the lateral dispersion coefficient
(°y, class C). Both the location and value of the maximum glc are much
closer to the measured values, and the general shape of the curve is
quite close to the measured data.
Figure 10 shows the vertical concentration distributions and compares
them with predicted profiles using PG stability class C for a 200 m
effective stack height. The use of 200 m as the effective stack height
is supported by the profile nearest the source. Plume widths near the
source fit the PG class C plume widths reasonably well, whereas those
well beyond xmx were somewhat smaller, i.e., they tended toward the class D
widths. The maxima in the PG distributions ranged from about 10% smaller
at 0.8 km from the source to 65% smaller at 4.7 km downstream. The
mixing of the plume into the surface layer is quite apparent, but is
somewhat slower than predicted by class C curves.
Lateral concentration distributions are shown in Figure 11 and are
again compared with those expected using PG stability class C. As with
the vertical profiles, the class C curves fit the elevated profiles
24
-------
(Figure lla) reasonably well near the source, whereas the plume widths
tend toward those of class D farther downwind. Still farther downwind,
the use of PG stability classes to predict surface lateral distributions
is even poorer (Figure lib). At 10 km downwind, the PG class C curve is
much too broad and the maximum predicted is only 30% of that observed.
Note that the centroid of the distributions is offset slightly from the
centerline. This suggests that the mean flow in the tunnel deviated in
direction from the tunnel axis by approximately 1°, which is regarded as
inconsequential.
Figure 12 shows the variations of vertical and lateral plume widths
(az and 0y, respectively) with distance. These were derived from the
concentration profiles by assuming Gaussian and reflected-Gaussian
distributions, as appropriate. The solid lines represent PG stability
categories C and D. Both az and ay closely approximate those obtained
with C stability close to the source and, especially Oy, approach those
for D stability farther downstream. Also included in Figure 12a is the
curve of the HGS dispersion coefficients (for the larger roughness length),
This prediction matches the data much better than does that using the PG
dispersion coefficients for either class C or D stability.
In summary, the boundary layer dispersive characteristics were most
closely approximated by PG class C (slightly unstable) in the vertical
direction and by PG class D (neutral) in the lateral direction. Both ay
and crz tended from slightly unstable near the source toward neutral
farther downwind. The larger az values close to the source are primarily
attributable to the large value of z0. The larger-than-normal o2/ay
ratio is due primarily to the larger value of az, and secondarily to the
25
-------
inability to simulate the large-scale horizontal fluctuations (wind meander)
in the wind-tunnel flow. Overall, the dispersive characteristics of this
boundary layer are regarded as highly comparable to those which may be
expected in a neutral atmospheric boundary layer over flat terrain with a
roughness length in the range of 70 to 100 cm.
26
-------
6. DETERMINATION OF GEP STACK HEIGHT
The determination of GEP stack height was based on the effect of
the terrain immediately upwind of the stack. As the basis for comparison,
a section of nearby upwind terrain was removed and replaced by a section
constructed to match existing terrain at its boundaries, but faired down
smoothly to flat terrain immediately upwind of the stack. The difference
in maximum glc measured in the presence of the terrain section and that
with the faired section inserted is defined as the excess concentration.
The GEP stack height is that which results in an excess concentration
equal to 40% of the maximum glc measured in the absence of the nearby
upwind terrain.
The effect of the nearby terrain was initially examined by flow
visualization using smoke sticks (titanium tetrachloride). In the presence
of the nearby terrain, a recirculation region clearly existed over the
plant location, as smoke was frequently swept upstream from the source.
This recirculation was initiated by flow separation near the top lee side
of the steep hill immediately upwind of the plant. The position of the
reattachment point fluctuated back and forth across the plant, as the
direction of the smoke flow was intermittently upstream, then downstream.
In the absence of the terrain (faired section inserted), the recirculation
region was absent, and the smoke moved continuously downstream.
6.1 FLOW STRUCTURE OVER TOPOGRAPHIC MODEL
The following flow measurements were made over the topographic model:
1. Vertical profiles of U, u1, w1, and "uw at the plant location
(x = y = 0), and at x = -3200, 1600, and 3200 mm (-6.1, 3.1 and
6.1 km f.s.), and at the Tower site (see Figure 1).
27
-------
2. Vertical profiles of the above quantities at x = 0 and at
x = 3200 mm (6.1 km) with the faired terrain section in position.
3. Lateral profiles of the above quantities at elevations of Hmx/2
and 3Hmx/2, both at x = 0 and at x = 3200 mm (6.1 km), where Hrix
is the height of the highest terrain feature in the particular
lateral cross section. These measurements were made with the
"nearby" terrain section in position.
The vertical profiles are shown in Figures 13 to 15. Note that the
origin of the vertical coordinate in all these graphs is the stack base
elevation. The mean velocity profiles (Figure 13) show quite strong
influences of the terrain; strong reductions in speed (compared with the
flat terrain case) are observed near the surface and, except at the upwind
edge of the topography, some reduction in speed is observed throughout
the entire depth of the boundary layer. Note that the depth of the
boundary layer is essentially constant. Comparison of the profiles at
the plant location with and without nearby terrain clearly shows the
strong influence of the upwind terrain to a height of approximately 325
m. Recall that the maximum height of the nearby terrain was Hj = 191 m,
so that the influence extends to 1.7 Hj. The flow measurements below
about 200 m with the nearby terrain in place are not highly accurate
because here the probe was within the recirculation region. The hot-wire
anemometer used cannot resolve the direction of flow and will indicate
higher mean velocities than actually occur (Khurshudyan et al, 1979).
The measurements are included here to provide qualitative indications and
to show areas of substantial flow distortion.
The profile shape at the Tower site (Figure 13c) compares favorably
with that at (x,y) = (3.1 km,0); the slight overspeed through most of
28
-------
the boundary layer depth is presumably related to the fact that the Tower
is located on an elevated plateau 133 n above stack base elevation whereas
the profile at (3.1,0) is near the river. Recall that the wind speed at
the 30 m elevation was analyzed to obtain the 98th percentile wind. This
value is 0.56 U«, or 2.2 m/s in the model and corresponds to the 98th
percentile wind of 12.2 m/s. Hence, the velocity ratio (model to full
scale) is 1:5.55. At the downwind edge of the model, the profiles with
and without the nearby upwind terrain are essentially identical.
The turbulence intensity (Figure 14) and Reynolds stress (Figure 15)
measurements correspond quite well with the mean velocity measurements.
Generally, strong increases in turbulence are observed near the surface
and some increase is observed over the full depth of the boundary layer.
Note that at the plant location, the indicated turbulence intensities
exceed 60%; as indicated above, such measurements with a hot-wire
anemometer cannot be trusted, but are nevertheless included to provide
qualitative and semi-quantitative understanding. Comparison of the
profiles at the plant location with and without nearby terrain again
clearly shows the strong influence of the nearby terrain to a height of
nearly 350 m. Comparison of the profiles at the end of the model again
shows that the presence of the hill just upwind of the plant had no
influence on the mean velocity or turbulence intensity profiles 6.1 km
downstream, but a small perturbation to the Reynolds stress profile is
still evident. These observations provide an indirect confirmation of
our earlier ideas concerning the amount of fetch required upwind of the
power plant to develop the appropriate flow structure at the plant site.
The lateral profiles are shown in Figures 16 and 17 along with the
terrain cross sections. In general terms, the mean velocities and
29
-------
turbulence intensities are fairly uniform at the higher elevations
(z = 3Hmx/2); i.e., they do not show effects of individual terrain features
but rather display characteristics more appropriate to a large-scale but
randomly distributed roughness beneath. The profiles at the lower
elevations (z = Hmx/2) are strongly inhomogeneous, reflecting the effects
of individual terrain features; the variations in the data appear to be
weakly related (if at all) to the shape of the terrain directly underneath.
They seem to correlate much better with terrain features immediately
upwind. For example, the very small mean velocities and very large
turbulence intensities in the river valley at the plant location
(x = y = 0) are caused by the nearby terrain just upwind of the plant,
which is not evident from the terrain cross section at x = 0. It is
interesting to note that the vertical turbulence intensity is approximately
2/3 of the longitudinal intensity everywhere, but the reason for this
observation remains unexplained.
6.2 DISPERSION OVER TOPOGRAPHIC MODEL
6.2.1 Reynolds Number Independence Test
The Guideline requires that a simple Reynolds-number-independence
test be conducted because of the rounded and relatively smooth-shaped
terrain (compared with, for example, sharp-edged buildings). For this
test, the hollow, porous plastic ball source was placed at the plant site
at the elevation of Hs = 86 mm (165 m f.s.). Neutrally buoyant ethane
was emitted from the source and a longitudinal surface-level profile of
concentration was measured along the downwind direction. The freestream
wind speed was then doubled (to 8 m/s), and the longitudinal glc profile
measurement was repeated. The two glc profiles are compared (Figure 18)
30
-------
using the respective freestream wind speed as the reference speed in the
normalization of concentrations x = CUJi^/Q (see Section 4.3).
While the surface profiles were being measured, a second tube simul-
taneously sampled concentrations at a distance of 40 mm (77 m) above the
first tube (on the surface). These results are also shown in Figure 18.
Reynolds number independence is clearly shown, as the maximum difference
in normalized concentrations at any position was approximately 10%, which
may be accounted for through expected scatter in repeated measurements of
the same variable. The irregularity in shape of the glc profile is
obviously attributable to the terrain. For example, the relative minimum
at x = 0.73 km is located on the upstream side of a shallow ravine that
is diagonal to the free-stream flow, and the relative maximum just downwind
of that is located on the downstream side of that same ravine. The
elevated profiles show a much reduced effect of the ravine and are much
more regular, i.e., similar to flat-terrain profiles.
The Guideline also requires a measurement of the mean velocity
profile at the plant location at the higher wind speed (8 m/s). This
profile compares very well with that measured at 4 m/s (Figure 13b),
again confirming our Reynolds-number-independence hypothesis.
6.2.2 Determination of GEP Stack Height
After the Reynolds-number-independence tests, a number of lateral
glc profiles was measured using the same stack height (104 mm or 200 m f.s.),
Further longitudinal and lateral glc profiles were also measured using
stack heights of 145 and 186 mm (275 and 375 m f.s.). The purpose of
this set of measurements (not shown) was to select permanent sampling
port locations to be used in the GEP portion of the study. Subsequently,
31
-------
an array of 59 sampling ports (brass tubes) was installed in the model;
these are shown on the map of Figure 19. These sampling tubes protruded
2 mm (4m f.s.) above the model surface. Sampling ports were selected in
banks of five by a Scanivalve system, from which the samples were routed
to five separate hydrocarbon analyzers (for further details, see Appendix
A). The porous ball source was then replaced by the scale model stack.
A mixture of helium and ethane was emitted from the model stack as
described in Section 4.1 to simulate the density and velocity ratios at
one-half plant load.
To find the GEP stack height, data were collected for stack heights,
of 72, 156, 170, and 175 mm (138, 300, 326, and 336 m f.s.). Note that
the existing stack height is 138 m. For each stack height, both with and
without the nearby upwind terrain, surface concentration maps were constructed
from the sampling port data. In each case, the maximum glc was determined
and the ratio of the maxima was calculated. If this ratio exceeded 1.4
(40% excess concentration), the stack height was increased; if less, it
was decreased until the value 1.4 was obtained. At that point, further
lateral and longitudinal glc profiles were measured near the location of
the maximum glc in each case (with and without nearby terrain), so that
the location and value of the maximum glc would be unquestionably determined.
The GEP stack height was found to be 326 m, under 50% plant load conditions.
Finally, at this same stack height, the plant load was increased to 100%
to determine whether a higher stack could be justified under higher load
conditions. It could not. Documentation for this GEP stack height is
included within this section. Additional data taken in search of the GEP
stack height are included in Appendix B.
32
-------
The surface concentration maps with and without nearby upwind terrain
are shown in Figures 20 and 21, respectively. Even with this very tall
stack (326 m), the location of the maxima were clearly within the range
of the model. The distance to the maxima were 2.6 km (about 8 stack heights)
with all terrain included, and 4.2 km (about 13 Hs) with nearby upwind
terrain removed. The ratio of the maximum measured concentrations in
this case was 1.39. Detailed surface lateral profiles were then measured
at x = 3.8 and 4.4 km with nearby terrain removed (Figure 22). Both
these lateral profiles showed maxima at y = 96 m, so that a surface
longitudinal profile was measured through the line along the surface at y
= 96 m (Figure 23). This showed that the maximum glc was indeed located
at x = 4.2 km, and had a value of xmx = 0.152. (The curves drawn through
the somewhat scattered data points in Figures 22 and 23 represent the
authors' estimates of the best faired lines through the data. For a
discussion of possible errors, see Section 6.3).
With the nearby terrain inserted, a detailed surface lateral profile
was measured along x = 2.6 km (Figure 22). This showed a maximum at
y = 96 m, so that a surface longitudinal was measured along y = 96 m (Figure
23). This showed a maximum at x = 3.0 km, so that another lateral was
measured along x = 3.0 km (Figure 22). This verified that the glc was
essentially constant for 96 m < y < 288 m. The conclusion from this set
of profiles was that the maximum glc in the presence of the nearby terrain
was located at x = 2.7 km and had a value of xmx = 0.214.
The effect of the nearby upwind terrain was two-fold: the location
of the maximum glc moved closer to the source by approximately 30%, and
the value of the maximum glc increased by 41%. A stack height of 326 m,
then, is GEP in terms of excess concentration.
33
-------
6.2.3 Detailed Plume Behavior
Numerous additional concentration profiles were measured in accordance
with the Guideline in an attempt to understand the plume behavior and to
relate the increased maximum glc to anticipated effects of downwash,
wakes or eddies. Additional vertical concentration profiles were measured
at one-quarter and one-half the distance to the end of the model, at the
end, and at the location of the maximum glc, both with and without the
nearby upwind terrain. These are shown in Figure 24. Several interesting
features are to be noted. First, in the absence of nearby terrain, the
profile closest to the source (1.6 km) is clearly elevated; the bottom of
the plume has just begun to touch the ground (note that the vertical
coordinate is referenced to stack base elevation so that the bottom of the
profile is the local ground surface) and the elevation of the maximum
concentration is very near to stack-top elevation. By contrast, the
corresponding profile with the nearby terrain in place clearly shows that
the plume has reached the surface and the elevation of the maximum
concentration is considerably lower, about 85% of stack-top elevation.
Note that the location of these profiles is just downwind of the first
hill downwind of the source. The wind speed and turbulence intensities
at (and above) the top of the stack were found to be essentially identical
with and without the nearby upwind terrain (Figures 13b and 14b), so that
we might expect the plume rise to be the same in both cases. From Figure
24, however, the plume is obviously higher in elevation in the absence of
the nearby terrain than it is in its presence. The cause of this disparity
may be plausibly explained as follows. In the presence of the faired
terrain upwind of the plant, the streamlines at the plant location are
34
-------
nearly parallel to the underlying surface (which is flat) because of
the (now) large distance from the nearest upwind hill, or indeed, they
may be slightly ascending in anticipation of rising over the first hill
downwind. However, in the presence of the nearby upwind terrain, the
flow structure observed was that of flow separation from near the top of
the upwind hill, and the reattachment point fluctuated in position across
the plant location. Under these circumstances, the mean streamlines at
stack-top elevation are likely to be descending. Hence, the effluent
from the stack would be rising, but relative to a descending mean flow.
The result is a smaller net plume rise due to the downwash and recirculation
region caused by the nearby upwind terrain.
The two concentration profiles measured at 3 km from the source
(Figure 24) are essentially identical above stack top elevation, which is
consistent with the fact that the flow structure was essentially identical
above that elevation (Figures 13, 14, 15). Below stack-top elevation,
however, the downwash caused by the nearby upwind terrain has resulted in
an increase in the glc at this location by a factor of approximately 2.
This location is near the location of the maximum glc in the presence of
the nearby terrain, but the location of the maximum glc in the absence of
the nearby terrain is considerably farther downwind, at 4.2 km.
At 6.2 km downstream, the two profiles are, for practical purposes,
identical. Note that the locations of the maxima in the glcs are in the
relatively flat river valley in the case with the nearby terrain, but
farther downwind and on the windward side of a hill in the case with the
faired upwind terrain.
Additional lateral profiles were also measured at one-quarter and
one-half the distance to the end of the model, at the end, and at the
35
-------
location of the maximum glc, both with and without the nearby upwind
terrain. These are shown in Figure 25. They were measured at the elevation
of the maximum concentration determined from the corresponding vertical
profile. At the 1.6 km position, the two plume shapes are very similar,
the main difference being the value of the maximum, which is about 16%
smaller in the case with nearby upwind terrain. This corresponds with
the vertical profiles (Figure 24).
At the 3 km position, the lateral profile in the presence of nearby
terrain was measured along the surface, whereas the profile in the
presence of the faired terrain was measured at an elevation of 307 m.
The slight lateral shift of the two profiles (~100 m) is unlikely to be
caused by the presence or absence of the upwind terrain, but rather to
the channeling effect of the river valley in the lowest levels toward the
positive y-direction. At 6.2 km, the profiles are practically
indistinguishable from one another.
6.3 FURTHER DISCUSSION OF RESULTS
The plume rise near the source in an adiabatic atmosphere may be
predicted from the Briggs (1975) formulation to be
Ah3 = (3/62)Lm2x + 4.17LBx2,
where S = 1/3 + USWS,
Lm = (1/2) (p^^/p^)1/^,
and LB = (1/4) (WS/US)3 (l/Fra2)D.
For the conditions in the current study (half load, GEP stack height),
Lm = 4.0 m, LB = 0.158 m, and the plume rise at the first measurement
station downwind (1.6 km) is predicted to be 121 m. However, since the
buoyancy per se (Froude number) was not matched, we may use these same
equations to predict a model plume rise of 52 m (full scale). The
36
-------
measurements, however, showed that plume rise was negligible in the case
with faired upwind terrain and, in fact, negative (-50 m) with all upwind
terrain present (Figure 24). None of these values of plume rise (predicted
or observed) is large because of the relatively large wind speed at the
source. The negligible and negative rises observed are postulated to
occur as a result of the large turbulence intensities at the source, the
effects of the hill immediately downwind of the source and, in the case
with nearby upwind terrain, to the descending streamlines at the source
due to the recirculation induced by the nearby upwind terrain. Because
the plume elevation was lower in the case with nearby upwind terrain,
that plume passed very close to the crest of the hill immediately downwind
of the source, and its lower edge was quickly mixed to the ground in the
wake of the hill. In the case with the faired upwind terrain, the plume
elevation was high enough to completely miss the first downwind hill, and
the maximum glc was reached considerably farther downwind (indeed, on
the next hill in line with the plume axis downstream).
The scatter in the maximum concentration values was on the order of
± 9% peak to peak (see, for example, Figure 23). This results in a relatively
large range of scatter in the excess concentration. For example, using
the extremes of +9% error in the presence of nearby terrain and -9% in
the absence of nearby terrain, a worst-case excess concentration of 66%
(as opposed to 40%) is indicated. This corresponds to a possible worst-
case 6EP stack-height error of about ±28 m. A more realistic and
certainly more reasonable way to estimate the error in excess concentration
is to assume that the errors in maximum glc are normally distributed.
The standard deviation is then on the order of one-sixth of the peak-to-
peak value, or about 3%. Using this standard error in the maximum
37
-------
concentration, the worst-case excess concentration is 49% (as opposed to
40%), which corresponds to a GEP stack-height error of about _+ 10 m.
This estimate of error in the measurement of excess concentration is
based on a relatively small sample size and may be statistically
questionable; however, it does provide an idea of the accuracy with which
the excess concentration and, hence, GEP stack height can be determined,,
Although not a direct confirmation, it is interesting to note that
Pickering ett al (1980) observed that nearby terrain features (within 1 to
2 km) effected pronounced downwash of the stack plumes on windy days.
Unfortunately, their study was not designed to examine these near-source
effects - only one of their sampling sites (Tower) was within the area
modeled, but it was far off to the side of the plume axis. Also, Koch et
al (1979) presented a photograph (their Figure 3) in which the hilltop
wind direction appears to be westerly, the lower level winds are easterly,
and the plume path is a semi-circular arc, reminiscent of the recirculation
region observed in the wind-tunnel experiments.
It is useful to compare the present results with other wind tunnel
measurements. Lawson (1984) made measurements of terrain amplification
factors (TAFs) for sources placed at a matrix of heights and distances
downstream of an idealized, axisymmetric hill with maximum slope of 25°.
His results suggest that, for the current plant location (about 600 m
downwind of the crest of the nearby upwind hill), the GEP stack height
(one for which the TAF = 1.4 or the excess concentration is 40%) is 1.7
Hy, or 325 m. The nearly perfect agreement between the current results
and those of Lawson is probably fortuitous, but it does show the usefulness
of the generic-type wind tunnel studies.
38
-------
7. SUMMARY
A fluid modeling study was conducted in a wind tunnel to determine
the Good-Engineering-Practice (GEP) stack height for a power plant located
in complex terrain. A stack height of 326 m was shown to meet the current
GEP criteria.
The meteorological conditions simulated were westerly winds (264.5°)
and neutral stability. The background dispersion characteristics in the
absence of the model were shown to conform most closely to Pasquill-Gifford
stability class C (slightly unstable) near the source, and tended toward
class D (neutral) farther downwind, probably because of the large roughness
of the surface (forest). A topographical model of a 7 km x 12.8 km area
surrounding the power plant was modeled at a scale of 1:1920. The ratios
of effluent density to ambient density and effluent speed to wind speed
were matched between model and prototype. Separate tests showed that the
flow over the topographical model was independent of Reynolds number.
For baseline measurements (against which to measure excess concentrations),
a roughly semicircular arc of terrain extending to 1.9 km upwind of the
plant was removed and replaced by a section which matched terrain contours
around the perimeter, but faired into flat terrain just upwind of the
plant. The effect of the nearby upwind terrain was shown to be a decrease
in the plume rise, a decrease in the downstream distance to the point of
maximum ground-level concentration (glc), and an increase in the magnitude
of the glc by 41% (for the GEP stack height).
Vertical and lateral concentration profiles both with and without
the nearby upwind terrain were provided to show that the maximum glc in
each case was determined beyond a reasonable doubt. The error (standard
39
-------
deviation) in the measurement of excess concentration was approximately
3%, which corresponds to a possible difference in the GEP stack height of
±10 m (out of 326 m). The observed differences in maximum glc with and
without the nearby upwind terrain were shown to have resulted from the
influence of the nearby upwind terrain.
40
-------
REFERENCES
Bearman, P.W., 1971: Corrections for the Effects of Ambient Temperature
Drift on Hot-Wire Measurements in Incompressible Flow, DISA Information,
no. 11, p. 25-30.
Briggs, G.A., 1975: Plume Rise Predictions, In: Lectures on Air Pollution
and Envir. Impact Analysis, Amer. Meteorol. Soc., Boston, MA, p. 59-104.
Counihan, J., 1969: An Improved Method of Simulating an Atmospheric Boundary
Layer in a Wind Tunnel, Atmos. Envir., v. 3, p. 197-214.
EPA, 1981: Guideline for Use of Fluid Modeling to Determine Good Engineering
Practice Stack Height, Envir. Prot. Agcy. Rpt. No. EPA-450/4-81-003, Res.
Tri. Pk., NC, July, 47p.
EPA, 1984: Guideline for Determination of Good Engineering Practice Stack
Height (Technical Support Document for the Stack Height Regulations), Rpt. No.
EPA-450/4-80-023 (Revised Draft, 11/1/84), Envir. Prot. Agcy., Res. Tri. Pk.,
NC, 60p.
Gadiyaram, P., 1984: Flow and Dispersion over Three-Dimensional Axisymmetric
Hills: A Wind Tunnel Study, M.S. Thesis, Dept. Marine, Earth, and Atmos.
Sci., NC State Univ., Raleigh, NC, 126p.
Gifford, F.A., Jr., 1975: Atmospheric Dispersion Models for Environmental
Pollution Applications, Lectures on Air Pollution and Environmental Impact
Analysis, Amer. Meteorol. Soc., Boston, MA, p. 35-58.
Hosker, R.P., 1974: Estimates of Dry Deposition and Plume Depletion over
Forests and Grassland, Proc. Symp. on Phys. Behavior of Radioactive Contam-
inants in Atmos., Paper No. IAEA-SM-181/19, p. 291-308, IAEA, Vienna.
Khurshudyan, L.H., Snyder, W.H. and Nekrasov, I.V., 1981: Flow and Dispersion
of Pollutants over Two-Dimensional Hills: Summary Report on Joint Soviet-
American Study, Envir. Prot. Agcy. Rpt. No. EPA-600/4-81-067, Res. Tri. Pk.,
NC., 143p.
Koch, R.C., Biggs, W.G., Cover, D., Rector, H., Stenberg, P.F. and Pickering,
K.E., 1979: Power Plant Stack Plumes In Complex Terrain: Description of an
Aerometric Field Study, Envir. Prot. Agcy. Rpt. No. EPA-600/7-79-010a, Res.
Tri, Pk., NC, p. 157.
Lawson, R.E., Jr., 1984: Standard Operating Procedures for the EPA Fluid
Modeling Facility, FHF Internal Document, U.S. Envir. Prot. Agcy., Res. Tri.
Pk., NC, 132p.
Lawson, R.E., Jr., 1984: Effect of Terrain-Induced Downwash on Determination
of Good-Engineering-Practice Stack Height, FMF Internal Rpt., Envir. Prot.
Agcy., Res. Tri. Pk., NC, July, 21p.
41
-------
Lawson, R.E. Jr. and Britter, R.E., 1983: A Note on the Measurement of
Transverse Velocity Fluctuations with Heated Cylindrical Sensors at Small
Mean Velocities, J. Phys. E., Sci. Instrum., v. 16, p. 563-7.
Lawson, R.E. Jr. and Snyder, W.H., 1983: Determination of Good-Engineering-
Practice Stack Height: A Fluid Model Demonstration Study for a Power Plant,
Env. Prot. Agcy. Rpt. No. EPA-600/3-83-024, Res. Tri. Pk., NC, 70p.
Pasquill, F. and Smith, F.B., 1983: Atmospheric Diffusion, 3rd Ed,,, Ellis
Horwood, Chichester, England, 437p.
Pickering, K.E., Woodward, R.H. and Koch, R.C., 1980: Power Plant Stack Plume
in Complex Terrain: Data Analysis and Characterization of Plume Behavior,
Envir. Prot. Agcy. Rpt. No. EPA-600/7-80-008, Res. Tri. Pk., NC, 333p.
Shipman, M.S., 1984: Fluid Modeling Facility Computer Program Guide, FMF
Internal Document, Envir. Prot. Agcy., Res. Tri. Pk., NC, Feb, 115p.
Smith, F.B., 1973: A Scheme for Estimating the Vertical Dispersion of a Plume
from a Source near Ground Level, Chapt. 17, Proc. 3rd Mtg. Expert Panel on
Air Poll. Modeling, N.A.T.O. CCMS, Paris, France, Oct., 1972, Proc. No. 14,
Air Poll. Tech. Info. Cntr., U.S.E.P.A., Res. Tri. Pk., NC.
Snyder, W.H., 1979: The EPA Meteorological Wind Tunnel: Its Design, Construc-
tion, and Operating Characteristics, Envir. Prot. Agcy. Rpt. No. EPA-600/4-79
051, Res. Tri. Pk., NC, 78p.
Snyder, W.H., 1981: Guideline for Fluid Modeling of Atmospheric Diffusion,
Envir. Prot. Agcy. Rpt. No. EPA-600/8-81-009, Res. Tri. Pk., NC, 200p.
Thompson, R.S., 1979: Dispersion of Sulfur Dioxide from the Clinch River
Power PlantA Wind-Tunnel Study, Envir. Prot. Agcy. Rpt. No. EPA-600/4-79-
052, Res. Tri. Pk., NC, 74p.
Turner, D.B., 1970: Workbook of Atmospheric Dispersion Estimates, Office
of Air Programs, Pub. No. AP-26, U.S. Envir. Prot. Agcy., Res. Tri. Pk., NC.
42
-------
TABLE 1. CLINCH RIVER PLANT EFFLUENT CONDITIONS FOR STACK # 1
(from Koch et al, 1979)
LOAD VELOCITY TEMPERATURE
50% 29.3 m/s 139° C v
100% 39.9 m/s 150° C
43
-------
TABLE 2. PROTOTYPE AND MODEL PARAMETERS FOR CLINCH RIVER PLANT
PARAMETER
Scale
Terrain height, Hj
98th percentile wind speed
Boundary layer depth, <5
Roughness length, z0
Friction velocity, u*/lL
Power law index
V6
ZO/HT
Stack diameter, D
Plant load
Effluent speed, Ws
Effluent density ratio ps/Pa
PROTOTYPE
1
\
191 m
12.2 m/s
1 km
0.9 m
0.052
0.23
9 X 10'4
0.0047
4.75 m
50%
29.3 m/s
0.70
MODEL
1/1920
99 mm
2.2 m/s
550 mm
0.35 mm
0.049
0.23
6 X ID'4
0.0035
2.48 mm
50%
5.23 m/s
0.70
Effluent to wind speed ratio
at stack exit (Hs = 326 m)
2.5
2.5
44
-------
c
o
(B
O
O
4-
I O
O)
(_
3
CD
45
-------
1UU -
90 -
80 -
70 -
X
*
UJ
O
1 60 -
=3
O
O
o
fe 50 -
o-
UJ
C£
U.
UJ W '
tt
_J
1 30 -
0
20 -
10 -
0 -
0
1 1 1 1 1 1 1 1 1 1 1 1 ^^.f^ 1
A
A
I \
A
A
A
A
A
/
A
A
A
: 1 ;
A
A
A
: 1 -
I
A
; / ;
A
: / ;
i i i i i i i i i i i
5 in i
WIND SPEED, m/s
Figure 2. Cumulative frequency distribution of neutral wind speeds for the
22.5° sector centered on 264.5° at the 30 m level at the Tower
site.
46
-------
01
c
OJ
a
co
c
eu
o
o
a
0)
.
io ai
s_ t_
cn+->
o
M C
O 5
^ o
a_ -o
47
-------
r-"W&
Figure 4. Photographs of (a) nearby upwind terrain, which was removed and
replaced by (b) faired terrain. The ruler is 1.5 m long.
-------
r
OJ
oo
00
00
OC
O. .v.x.
-
mum
O)
E
O)
c
c
3
-i-j
T3
C
3
u_
O
c
O
o>
0)
-I-J
c:
O OJ
rrj fC
i O
O) O
c v
O
E^
u- ^
O
J^
0 E
4-> '->
0)
^ c:
oo -r-
LTI
o;
49
-------
3
O
-Q
(J
Q.
(/I
O
(O
"O
01
0)
cr
_i
DC
LU
3
^ LD *-< OO CO
*- . O . .
! 3
O) O
> 1-
-------
o
0)
(O
3
Ol
.c
Ol fO
, c_
I- (-
4- O)
O 4->
(_
Q.4->
CO 1-
O>
t_ J=
^_> en
CO 3
o
CO t-
-o
"o a;
c >
>> o
(LI
di t-
Ol
TD >,
c
t_
3
O)
9/Z
51
-------
: g
.
z rJ< O CO :* 03
^ r- =? o r» =r oo
z
o oo .
(£ CO 03
;
- tf a
: 11
i °^
' i i i i i i i i
-
>»
.E +->
"" t-
- 0 ^ O QJ
*» e£
at a)
_ re 3
- I r- -Q
C- +->
CM "O i
i c re
3 C
o -i-
JD TD
CO O 4->
ll en
ai c
x: o
a.!
=1- to
o *~~-
W - 0 J£,
^^
re
n/,n v £
CU +J
TO o
i O
rs
E QJ
I i 1
^^ I i - -[-"""'," i i1 [ , | i -, i
-
cm 4 -
ai <^ :
IB 04
UM 4
^^^
D 4M
n <4
_
[JJ <^4j
n 4 :
n 4i ;
^4 "
m <4 -
M < ^ :
i i i i i.i i i i i i i i
- ,T) T >
C
o a;
E
-CM CO
QJ -
i re
q_
O
- t_ c
Q.-1-
re
^C3 ^-
o re
i. r-
.7 o <+-
00
.~ £
3
cn
co £
31
" 1
°n/n
52
-------
ICr6
e\j
E
*
o-
10-7 .
10
-8
100
x, km
Figure 9. Surface longitudinal concentration profiles of simulated atmospheric
boundary layer over rough flat terrain compared with Gaussian
predictions using Pasquill-Gifford stability categories C and D, and
HGS category D (z0 = 100 cm).
53
-------
700 F
600
500
100
300 -
200
100 --
.00001
.00002 .00003
CIL/Q, m-2
.00004
.00005
Figure 10. Vertical concentration profiles of simulated atmospheric bounda
layer over rough flat terrain compared with Gaussian prediction
using Pasquill-Gifford stability category C.
54
-------
C 4-> (-
3 O CLr
O -i- Ol
J2 TD T3 >
O)
03
T3
: O
x D/sno
o
E ciJ 0) .-
r- t_ 4-> ^
(/) (O ,£
O 4-> 4->
(^ -i-t/l
O C ' 'r
03
O
C j_> £-
O (T3 O
4-> C7> I
C 3 i >)'-'-
1^-3^
OJ
3
CD
CM
55
-------
o
o
o ii e
!- O.
OJ Nl O-
o
o
o
o
o
Ul
r- 00 C
O QJ
>>:c to
t- 4- O
1) tD >>
JC I
E 3 O.
4-> O"O
iTJ CO
(O
O Q.
0)
r- -r- O)
C 1C -O
I- Q. p
no o **
'3J O i
c_ 10
3 C O
-I -r OJ
(O (D -l-J O
D &. t- 'O
E t- 0» *-
O) > t-
4^ 4-J ^
T3 (O -t->
., , (O
E cn
r O -I- !-
Q- t_ tO S
C\J
tI
tl)
c_
3
C7I
O
O
O
o
o
Ul s2D
56
-------
CE > CC
CC 31 LU
CC (X. I
LU cr
i cc a
CD LU
I O CC
CE 0- i
_i o cr
. U- h- la_
Q o a
: co ~^ -«
9
- = - CC LU
CC X LU 0-
CC CL. \~ U~l
LU cr
1 CC Q LU
O LU _l
1 O CC CD
CE CL. >-> ID
_J O ff ED
. Lu I U- Q
o o o
01 - $-
CE >-
cc IT
cc Q_
LU CC
h- O
CE Q_
I O
- CO CO
; I I
T <
a r
O)
a
o
ID
CL
as
o
a,
o
O)
o
01
o
t_
a.
o a;
O (->
f_
O) »->
> ^
CU
c «-
o a>
in >
i- O
c_
ra OJ
Q. CO
3
CD
o
o
o
C\J
o
o
in
o
o
o
o
Q
in
Ul 'Z
57
-------
CTl.
O O
a. -a
o E
+-> >>
in
t_
QJ C
> CU
O Q.
O
(/>
CD
^«
r- C
14- -I-
O
Zf -v^
"s
n >
" =>
'*+^
. "s
r- 4-> 0)
4-> -a
c -c QJ
i CD ,
3 > +j
J3 O -
t- in
3 ai c
+j t/> a>
o +j
<*- i: c
o »-> -
c -C ai
o (-> o
in -i- c
.- $ a;
OJ
3
en
-------
a
a
0
to
o
Q.
(0
a>
o
Q.
0
O)
O M-
£_ 0)
CO 4->
0)
t- 4->
+J O
-a"-a' a i a1
1 1 1 1 1 1 "1 1 1 ' 1
- z
cr >-
cc x:
: cc a. .
uj cr
I i- or J0
LO
1 O
cr o_
_l D
I u. t
s^
CD CD
- ^ co 01
- x ' '
:
-------
en
0)
(-
ro C
C
O
Q.
+j at
r- -C
o +->
o
O) I- fO
E -i-> t-
cn en
<4- C
O f <4-
o o
I/) TJ
O) E
r- E O
O i-H O
t_ ^3
Q.IO
4->
r II ID
«3
t- X C
0) 2
4-> ^- O
(O J3 ^
J - in
a>
3
cn
n/n
60
-------
n/,M 'n/,n
i,y.
E
A
>>
0)
O
to c
o c tu
O r
II -i- 3
+-> ^J
X O l-
4->
(O
(/) 1
CO It]
+j o c
ID t- -r-
u -o
C 3
i- C -M
IB -i- .-
t-
O
I C
: o>
,- . ' o.
c o
c
cu
O E to
c «a -C
0) t- O
.Q to CT)''-
l_ C (->
3 S <+- «-
+-> O O
i+- =
O E
CO
QJ
+-> T3
4-> O)
O to
o O
O II IQ
L_ * **
Q. X C >>
^- ^~ O -r-
IO ^3 -E tO
t- tO C
O) O)
I ^ "^J ^y ^ *
03 C t- C
_j 10 nj -t-
a>
3
Ol
.w «n/,n
61
-------
10
1 -
CD-
CM
II
X
.1 -
Uco
A
D
M/S
8
OPEN: SURFfiCE
FILLEDi 77M flBOVE SURFRCE
.01
.01
x, km
10
Figure 18. Surface longitudinal concentration profiles displaying Reynolds nui
independence.
62
-------
C\J
o
o
EC
O
Q_
CC
CJ
to
2
ED
cn
LJ
o
az
o
a.
ai
4-j a>
o T-
13
i C
[ \ r*
fXJ CD
O -r-
O t-
o
(-> C 4->
C_ (O O
O C
Q. t-
O) O
CD Q.X3
C CL
i- 3 CO
i t_
O)
ai
=3
CD
-------
I-J
-C CO
CD CO
^ o
O rH
> LO
t_ i_
O> QJ
CO JO
-Q E
O 3
CO
c
o
r- O
3 TO
J2
i- Q.
t_
4-> C
CO T-
5 c
r
C Itf
O t-
i- C-
4_> C1J
C i
OJ 1TJ CO
o ai
c -a 3
O C r
O re (O
>
0) *->
o E
(O (O
4- IO 3
l_ CM +j
3 CO O
o
C\J
3
CD
64
-------
co
t_
OJ
JO
E
3
QJ
J* O
O <-
re 4-)
-1-) O
CO QJ
to
D-
UJ "O
O QJ
+J re
QJ
> "O
S- QJ
OJ O
co re
o Q- .
cu co
co t- QJ
c 3
O C
r 'r re
i- OJ
t- (->
>,
o E re
re <
14- to a.
t_ CM co
3 PO -r-
tn o
CD
L_
3
O)
65
-------
.2
.16 --
o-
cvj
ZJ
o
II
X
.12 -
.08 -
X KM
A 3.6
D 4.4
2.6
* 3.0
FILLED: WITH TERRflIN
OPEN: FHIREL) TtRRHIN
D
D
A4
A
n A
+#+!
i i 1_
a
_i 1 1 i 1 i
A^- 4
_iii|. i,i_
-1200
-800
-liOO
o
y, m
400
800
1200
Figure 22. Surface lateral concentration profiles taken with GEP stack to locate
maximum ground-level concentration.
66
-------
.214
.2
.16 -
Cf
CM
3
o
II
X
.08
FILLED! WITH TERRflIN
OPEN: FRIRED TERRHIN
x, km
Figure 23. Surface longitudinal concentration profiles taken with GEP stack to
locate maximum ground-level concentration at y = 96 m.
67
-------
1500
1000 -
rsl
500 -
FILLED: WITH TERRRIN
OPEN: FfllRED TERRRIN
-0.1 0
.2 .3 .11
X = CUJl2/Q
Figure 24. Vertical concentration profiles taken at various positions downwind
of GEP stack.
68
-------
.7
cy
evi
"*!
o
it
x
FILLED! HITH TERRRIN
OPEN: FRIRED TERRRIN
1200
Figure 25. Lateral concentration profiles taken at various positions downwind
of GEP stack.
69
-------
APPENDIX A
DESCRIPTION OF FACILITIES AND INSTRUMENTATION
\
A.I THE EPA METEOROLOGICAL WIND TUNNEL
This study was conducted in the Environmental Protection Agency's
Meteorological Wind Tunnel. It is an ultra-low speed, open-return wind
tunnel with a test section 2.1 m high, 3.7 m wide and 18.3 m long. Air
enters the test section through a flow-straightening honeycomb and four
turbulence-reducing screens. A plenum chamber just upstream of the 2.8:1
contraction allows turbulence in the wake of the screens to decay. An
adjustable ceiling allows compensation for blockage effects of models and
achievement of a zero-pressure-gradient flow in the test section.
Transparent windows form the sides of the test section to facilitate
flow visualization. An instrument carriage provides the capability for
positioning a probe anywhere in the test section with an accuray of ±1 mm.
Controls and readout for the carriage are conveniently located at an
operator's console. Downstream of the test section, the air passes
through an acoustic silencer, a rectangular-to-round transition section,
the fan, a diffuser, and another acoustic silencer before being exhausted
back into the room. The tunnel is driven by a 75 kilowatt AC motor with
eddy-current coupler for speed control of the 1.8 m diameter fan. This
apparatus provides steady speeds in the test section of 0.3 to 8 m/s.
The motor and fan assembly is enclosed in an acoustic silencer to provide
a low noise level in the laboratory. Further details of the wind tunnel
and its operating characteristics are described by Snyder (1979).
70
-------
A.2 INSTRUMENTATION
A.2.1 Velocity Measurements
Mean velocity, turbulence intensity, and shear-stress profile data
were obtained with TSI, Inc. model 1053B constant-temperature anemometers
in conjunction with model 1243-T1.5 x-array hot-wire probes (boundary-
layer style). Calibrations were performed in the free stream with the
sensor mounted on the instrument carriage. The reference velocities for
calibration were obtained with a Dwyer model 160-24 pitot-static tube;
the differential pressure was monitored with an MKS Baratron capacitance
manometer (model 310BH sensor head with model 170M electronics unit).
Yaw-response corrections were made to the anemometer output according to
a scheme developed by Lawson and Britter (1983). Temperature in the test
section was continuously monitored and the anemometer output was corrected
following the technique of Bearman (1971).
Temperature near the sensor location was monitored both during
calibration and routine operation by a Yellow Springs Instruments 4320
thermister. Analog output from the anemometers was converted to digital
form by a 12 bit analog-to-digital converter. These voltages were
converted to velocities through the use of a King's Law form of equation.
The resulting data were processed on Digital Equipment Corp. PDP-11/40
and PDP-11/44 minicomputers. Two-minute averages at a sampling rate of
500 samples per second yielded reasonably stable mean values (± 1% on
mean velocity). Further details of the hot-wire and data-processing
systems are given by Snyder (1979) and Lawson (1984).
71
-------
A.2.2 Concentration Measurements
A hydrocarbon tracer technique was used to measure concentrations
downwind of the source. The technique employed a mixture of helium and CP
grade ethane as the source. Ethane provided the tracer and helium provided
the required density ratio. Concentrations were measured with Beckman model
400 flame ionization detectors (FIDs) operated in the continuous sampling
mode.
The FIDs were calibrated using 0.8994% certified (Scott Environmental
Technology, Inc.) "span" gas; zeroing was accomplished with "zero" air
(< 1 ppm hydrocarbons). The FIDs were shown in a separate series of tests
to respond linearly over four decades of concentration. The samples to
be analyzed were drawn either from a "rake" of tubes which was mounted on
the instrument carriage to allow convenient positioning or from the sample
ports on the model surface (total of 59) through a Scanivalve. The ethane
flow rate was 973 cm3/min and the helium flow rate was 543 cm3/min (half
load), providing the required effluent to ambient density ratio of 0.7
and effluent speed to wind speed ratio (Tower) of 2.38. Five analyzers
were used simultaneously to speed the data acquisition process. Analog
outputs from the FIDs were also digitized to 12-bit precision for processing
by the minicomputer. Additional details may be obtained from Lawson
(1984).
A.2.3 Data Acquisition System
All laboratory data were collected and analyzed using Digital
Equipment Corp. PDP-11/40 and PDP-11/44 minicomputers. Anemometer
calibrations were performed over the velocity range of interest (typically
6 to 9 points over the range 1 to 5 m/s). The computer was used to fit
72
-------
a King's Law form of equation to the calibration data. This best fit
relation was then used to generate a "look-up" table for conversion of
voltage to velocity during routine operation. A typical calibration
curve is shown in Figure Al. The hot-wire anemometer was typically sampled
at 500 samples per second, and data reduction took place between samples;
hence, real-time outputs of velocity, intensity, and shear stress were
available. Temperature compensation was accomplished using the method of
Bearman (1971), which required occasional modification of the look-up
table as temperature in the test section (room temperature) changed.
As the time constant of the FIDs is on the order of 0.5 second, these
units were sampled at a rate of ten samples per second. Two-minute averages
again provided stable mean values. The FIDs have linear response, so that
the generation of mean values was straightforward. Zero and span values
were recorded at the beginning and end of each test to assure that analyzer
drift was not a problem. Background values were subtracted from each
sample to account for background drift, assuming a linear change in back-
ground with time between samples.
All data files were stored on disk for later processing and preserved
on magnetic tape. Data reduction was facilitated by the use of software
developed at the Fluid Modeling Facility (Shipman, 1984).
A.2.4 Volume Flow Measurements
Ethane and helium were mixed prior to injection into the model
stack in order to obtain correct density and velocity ratios. The flow
rates of these gases were measured and continuously monitored using Meriam
laminar-flow elements (LFEs); the differential pressure was observed on
73
-------
Meriam micromanometers. Figure A2 shows the typical apparatus for in-
jecting gases into the tunnel. Calibration of the LFEs was accomplished
using a volumetric flow calibrator (Brooks model 1050A Ul), which had a
>
rated accuracy of 0.5%.
74
-------
2.55
PROBE CflLISRflTION
E VRS U
20-MRR-85
2.14 -
2.25
o
2.1 -
1.95 -
1.8
PROBE
SLOPE =
INT
RLPHfl =
CflLIB.TEMP.
CnLIBRnTION POINTS
ZERO FLOW VOLTflGC
'1.5
U, m/s
Figure Al. Typical calibration curve of hot-wire anemometer.
75
-------
MANOMETER r"l
A '
"I'!
MANOMETER
B
PRESSURE
REGULATOR
LAMINAR
FLOW
ELEMENT
HOLLOW
i PERFORATED
I PLASTIC
SPHERE
DIA. * 15mm
Figure A2. Sketch of source and flow measurement apparatus used for injecting
gases into the wind tunnel.
76
-------
APPENDIX B
CONCENTRATION MEASUREMENTS FOR OTHER STACK HEIGHTS
AT HALF- AND FULL-LOAD CONDITIONS
Figures Bl to B3 present the surface concentration maps obtained
through measurements using stack heights of 138 m, 300 m, and 336 m at
half-load plant operating conditions. For each stack height, two maps
are presented: (a) in the presence of nearby upwind terrain and (b) in
its absence. These measurements were made during the process of determining
the GEP stack height, and were conducted in the same manner as those for
the GEP stack height described in Section 6.
With the existing stack height of 138 m (Figure Bl), the nearby upwind
terrain clearly showed downwashing of the plume. Indeed the location of
the maximum glc was near the base of the first hill downstream (0.4 km).
Although measurements were not made upstream of the plant, tracer was
clearly transported there as observed with the smoke flow visualization
and suggested by the isoconcentration contours of Figure Bla. In the
absence of the nearby upwind terrain, downwash of the plume was not
evident (Figure Bib). The maximum glc was reduced by a factor of 2.5 and
its location was farther downstream, at 0.9 km. Notice that the effect
of the downwashing caused by the nearby upwind terrain was to broaden the
surface distribution near the source, but to result in a narrower
distribution farther downwind (in terms of the crosswind widths of the
isoconcentration "ovals"). The plumes show very slight tendencies to
follow the river valley, being diverted slightly toward the north.
At the higher stack height of 300 m, maximum glcs were reduced
dramatically, of course, both with and without the nearby upwind
77
-------
terrain (Figure B2). It is interesting to note for comparison purposes
that in flat terrain an increase in stack height from 138 to 300 m would
be expected to result in a decrease in the maximum glc by a factor of 4.7
(i.e., the maximum glc is proportional to the inverse square of the stack
height). In the present case, the factors are 6.3 in the presence of
nearby upwind terrain, and 4.6 in its absence. The ratio of the maximum
glcs, however, still exceeds 1.4; the excess concentration is 80%. The
location of the maximum was on the lee side of the first hill downwind
(1 km) in the presence of the nearby upwind terrain, but much farther
downwind (4.4 km) in its absence. Indeed, the maximum glc with the
faired terrain was located at the crest of the next hill in line with the
plume axis downwind. As discussed in Section 6.2.3, the higher surface
concentrations were caused by the downwash effected by the nearby upwind
terrain.
With the tallest stack (336 m, Figure B3), two (essentially equal)
relative maximum glcs were observed, one in the middle of the river valley
at 2.6 km, and the second at the foot of a hill (at 3.5 km) in the presence
of the nearby upwind terrain. The effect of the nearby upwind terrain
was still evident, as only one maximum was observed (at 4.4 km) in its
absence, but the excess concentration (32%) was insufficient to justify
this stack height as GEP.
Finally, with the GEP stack (326 m), the plant load was increased to
100%. The surface concentration maps in this case are shown in Figure 84.
Whereas the effect of the nearby upwind terrain was still evident, the
excess concentration was only 16%. This value suggests that the effective
stack height exceeded that for the 336 m stack at half plant load (Figure B3),
78
-------
As a summary graph, Figure B5 shows the excess glc as a function of
stack height at 50% plant load conditions. The excess decreases monotonically
with an increase in stack height and would appear to reach zero excess
at a stack height near 350 m (about 1.8 Hj). This value is in good agreement
with the flow-structure measurements presented in Section 6.1, where
strong nearby upwind terrain influences were observed to an elevation of
350 m, and weak ones were observed above that (cf., Figures 13b, 14b, and
15b). The faired curve shown in Figure B5 suggests that the GEP stack
height of 326 m is slightly conservative; however, the values associated
with the other-than-GEP stack heights are somewhat more uncertain because
only the surface ports were used to obtain the values of the maxima. The
GEP value is more certain because detailed surface probing was conducted
to unquestionably determine the locations and values of the maximum glcs,
both with and without nearby upwind terrain.
Table Bl provides the values and locations of the maximum measured
surface concentrations for all stack heights tested.
TABLE Bl. VALUES AND LOCATIONS OF MAXIMUM SURFACE CONCENTRATIONS
Stack Ht.
(m)
138
300
326(GEP)
336
326 (100% load)
With All Terrain
Xmx
1.83
0.290
0.214
0.196
0.174
xmx(km)
0.4
1.3
2.7
3.5
3.2
With Faired Terrain
Xmx
0.735
0.160
0.152
0.149
0.150
xmx(km)
1.0
4.4
4.2
4.4
4.4
79
-------
O
,
01 (SI
T3 to
CL) L.
> O)
CO 3
jQ Z
O
co
C T3
O to
r- a
4-> i
3
JD +J
i- C
i- 10
!-> r
CO Q.
o
O
in
CO
s_
80
-------
to
o>
J«J !-
O 4->
ro
4-> O
(/) r-l
O) O)
c t-
i- to
4->
t/) T3
- <1J
X >,
o> 3
X) Z
O
in
c -O
O r-
3
J3 4->
r- C
(/) Q.
5 6^
O
c tr>
+J E
c
O) CO 01
o c*°> cu
C -< 3
O i
O 4-
O 4-> I
>*- CD 3
t_ - +->
3 Ol O
OO -C
-------
H
8
r>
OJ
r- T3
S 01
0)
o
<- O E
rc _o
t_ T-
5- T3
4_> ro
to O
r T3
< 4->
03 C C
o re
o,
=J O
00 CO
C\J
CO
O)
CD
82
-------
If- 3
O +J
U
cn to
i- (U
O) E
O)
r- T3
T3 "3
0)
. > Q-
Q, t- <"
O Of-
* ^^
Q. O £
c irt O
a; *«
o o
c i«
o
o »->
ITS
a; .
o c/>
(O E QJ
>*- 3
t_ O r-
3 O rtJ
00 CO >
CM
co
-------
01
o
-------
85
-------
86
-------
T3
.C O
CD
(O
4-> O
CO ^-1
a. ai
LU t-
O
> !-
c_ -a
to ifl
J3 t-
o r
to
to
C i
o a.
!-> s-a
(O O
t_ o
+J -<
c
a> )->
o
O 3
J3
a>
o "
_ i
3 CVJ (O
CO CO >
CQ
0)
L.
3
CT
87
-------
200
ISO --
CC
cc
LU
CJ
LJ
tn
en
LU
o
X
LU
100 --
50 -
100 150 200 250 300
STflCK HEIGHT, M
350
'100
Figure B5. Excess maximum ground-level concentration as a function of stac
height for 50% plant load conditions.
88
-------
APPENDIX C
DATA LISTINGS
In order to minimize printing costs, the data listings have not
been included with th^s report, but are available from the authors upon
request.
U S GOVERNMENT PRINTING OFFICE- 559-013/20010
89
------- |