ML17055C372

Wind Tunnel Tests of Gas Diffusion from Leak in Shell of Nuclear Power Reactor & from Nearby Stack

ML17055C372
Person / Time
Site:	Nine Mile Point
Issue date:	04/01/1963
From:	Golden J, Halitsky J, Halpern P NEW YORK UNIV., NEW YORK, NY
To:
Shared Package
ML17055C371	List: ML17055C369 ML17055C372 ML17055C374 ML17055C375 ML18038A692
References
63-2, NUDOCS 8609120404
Download: ML17055C372 (140)

Text

t )f

.,t,

~

~

~,'l

~ v ll a

(i

~

l v'

v l

I 1

l

> ~ I

~ I

-, tt.t t

~

~

~ *"1 I

gE.ET PR~

P) v Y

NEW YORI( UNIVERSITY

>CCCXX+

i

~

lt!II College of Engineering ATTACHMENT j.

't ~

I Ii tl

~e~

~ ~

C I

\\

~

I

~

y ~

~ I l. ~ ~;

t tt

~

I I

1 g

~

RESEARCH DIVISION.:,i Uniyersify Heights, NqIjt York 53, N. Y.

I Department of Meteorology and Oceanography GeophysicqiSciences

>laboratory Report No. 63,r. 2,

<rs t ~

t (t I tlyl

~

P,

~ $

tl t.

~

~

" l', II,

~

t Il II> tt tv

~

~

'l tt,t,

'I pa

~.t,: <):.

)/ ) i l ( 'I '

~, i'; "! >I I

'4 r

~

~ tv, I

v

v. ~

v'tl.'

I.

t I

~ I I',.

~';

I

~,

~ll I,

l v.

~

vt

~"..

(

s WIND TUNNEL TESTS OF G$S".DIFFUSION t',

I ~

A LEAK IN THE SHELL'F'A'UCLEAR POWER REACTOR AND FROM A NEARBY STACK

,~,

~I'l~ ~j I

I t

~

~ '

't'

~

I l.t! v James Halitsky Jack Golden I Halper Paul Wu t

I"

~ ~ v t

I

~

~

I II I Prepared for,.

t I

I Environmental Meteorological Research Pro ject;

'nited States Weather Bureau'ashington 25, D. C..'

I Contract No. Cwb. -. I032I

~

~

I

~ ~

~ I I

April I, l963 t

I I

~

I t

1 I

~

I I

a 8+09120404 860rtl'05 PDR ADOCK 05000410 PDR

I

E 0

NEW YORK UNIVERSITY COLLEGE OF ENGINEERING RESEARCH DIVISION Department of Meteorology and Oceanography Geophysical Sciences Laboratory Report No. 63-2 WIND TUNNEL TESTS OF GAS DIFFUSION FROM A LEAK IN THE SHELL OF A NUCLEAR POWER REACTOR AND FROM A NEARBY STACK James Halitsky Jack Golden Paul Halpern Paul Wu Prepared for Environmental Meteorological Research Project United States Weather Bureau Washington 25, D. C.

Contract No.

Cwb - 10321 April 1, 1963

I I

~

0

ABSTRACT Tests were conducted in the NXU 3$ x 7 ft low speed wind tunnel to determine the distribution of gas concentrations resulting from a gas'leak in the shell of,an industrial nuclear power reactor, The shell was patterned after the EBRII reactor at the NRTS at Idaho Falls.

The basic tests were made on the she'l alone in an adiabatic atmosphere having a logarithmic velocity profile corresponding to an average profile at NRTS.

Additional tests showed the effect of a uniform velocity pro-file, change of absolute wind velocity, inclusion of auxiliary buildings and heating the shell.

The effect of proximity of the stack top to the shell was studied briefly.

'1

~ '

~

I 1

TABLE OF CONTENTS ABSTRACT

~

~

~

~

~ l

~

~

~

~

~

~

~

~

LISTS OF FIGURES AND TABLES LIST OF SZSOLS

~

~

~

~

~

~

. ~

1

~

~

~

~

~

~

~pa e ii iV V

l.

INTRODUCTION

~ I

~

~

~

~

t

~

~

~ 1

~

~

~

~

2 ~

TEST APPARATUS AND PROCEDURE

~

~

~

~

~

~

~

~

~

~

~

~

~

~

3 2.1 Wind Tunnel 2.2 Model 2.3 Wind Velocity Profiles 2.4 Gas Tracer Technique 2.5 Photographic Technique

~

~

~

~

~

~

~

~

~

4

~

~

~

~

~

~

~

~

~

~

~

~

~

~

~

~

~

~

~

~

~

~

~ l

~

~

~

~

~

~

~

~

~

t

~

~

~

~

~

~

~

~

~

~

~

~

~

~

~

~

~

~

~

~

~

~

~

~

~

~

3 3.4' 6

3.

TEST PROGRAM AND RESULTS 3.1 Test Program 3.2 Test Results

~

~

~

~

~

~

~

~

~

~

~

~

~

~

3.21 General Nature of Flow Fields Near Objects.

3.22 Smoke Photographs 3.23 Gas Diffusion in Non>>Uniform Flow Fields.

3,24 K-Isopleths 4,

EVALUATION OF RESULTS

~

~

~

~

~

~

~

~

~

~

~

~

~

~

5, APPLICATION TO FULL SCALE DIFFUSION CALCULATIONS.

W 6.

RECOMMENDATIONS

~

~

t

~

~

~

~

~

~

~

~

~

~

~

REFERENCES TABLES FIGURES 7

7 7

8ll 12 16 22 26 29

'(

1 I

~

~

T I

t

~

~

I I

LIST OF FIGURES Fig l Fig 2 Fig 3 Fig 4 Fig 5 Fig 6 Fig 7 Photograph of NRTS Complex Photo'graphs of Model of NRTS Complex in the Wind Tunnel Arrangement of NRTS Buildings in the Tunnel in a SW Wind Comparison of Model and Actual Building Arrangements at NRTS EBRII Model Dimensions'nd Wind Velocity Profile Photographic Sequence Showing Diffusion of Smoke Released at Various Points in the Longitudinal Centerplane Through the Reactor Shell Photographs of Smoke Diffusion Downwind of Reactor Shell With and Without Boundary Layer Control 8 to 28, K-Isopleths for Gas Released Through Reactor Shell (see Table l for details).

Fj.gs 29 to 3(

K-Isopleths for Gas Released Through Stack (see Table 2

for details)

Fig 37 Fig 38

, Maximum Value of K vs Distance from the Shell Dependence of Maximum Value of K on Wind Velocity Table l Table 2

Table 3

LIST OF TABLES Summary of Shell Release Tests Summary of Stack Release Tests Shell Surface Temperature Excess Above Tunnel Ambient Temperature

~

~

v A

c cav Df H

KK' s

S up vpw u

U V

Ve x,y,z x,r,z 0B 6q

<s 6g LIST OF SYMSOLS area of the pro)ection of the shell on a plane transverse to the, background flow direction

~ plume cross-section area 2(( (x jsq/V)

~ local time-mean concentration Q/A' shell diameter in plan view ~ 80 ft subscript denoting full scale shell height

~ 98 ft experimentally determined distribution function cAV/9

~ Hay-Pasquill distributi~n function for a plume from a cont1nuous potut source exp j -r /2(x 6'</V)2 I reference length subscript denoting model

~ general veloc1ty component in 1sotropic turbulence volume flow rate of nent of effluent radial coordinate, z

subscript denoting distance in 1sotropic turbulence distance measured radially outward from shell surface

~ general velocity components in the x, y, z directions local mean velocity in the x direction

~ friction velocity ~

shear stress density refe'rence velocitv for tunnel airstream or full scale wind

~ general reference velocity velocity of.effluent at exhaust aperture

~ general coordinates:

downwind, lateral, up~ respectively distances in the tunnel along coordinate axes

~ surface roughness parameter azimuth angle from upwind direction r.m.s. veloc1ty fluctuation in isotropic turbulence

~ r,m,s, distance of concentration distribution

~ r.m.s. gust angle (radians)

'l

~

~

0

1, ItiTRODUCT10N Nuclear power reactors are generally enclosed in a gas-tight shell which serves to prevent the uncontrolled release of radioactive gas or air to the atmosphere.

In normal operation, the air within the shell becomes, irradiated and must be replaced periodically.

In order to protect people residing or working in the vicinity of the reactor, the contaminated air is usually discharged through a stack when meteoro-logical conditions are favorable.

In the event of a power excursion, the shell will retain the fission products until such time as they may be released safelv through the stack, There is the possibility, however, Chat the shell willbe breached during an excursion, resulting in the es-cape of radioactive gas directly to the atmosphere.

In this case personnel must be evacuate'd from areas likely to become contaminated.

In order to realistically prescribe safe gas or air'release rates from the stack, and to delineate danger areas resulting from gas leaks, it is necessary to calculate the field. of gas concentration created, by a source located arbitrarily at any point on the surface of the reactor shell or at the top of a stack located near the reactor, Un-fortunately, the well-known continuous point-source diffusion formulas cannot be used for gas leaks or for stacks that do not rise considerably above the reactor and auxiliary buildings.

These formulas were derived under the assumption that the flow field into which the gas is released has straight, parallel mean streamlines and homogeneous turbulence.

This assumption is evidently invalid near buildings where the streamlines curve strongly near the upwind face, and almost disintegrate into a con-fused field of variable, high turbulence near the downwind face.

The experimental approach is the only one that gives any promise of producing reasonably accurate estimates of gas diffusion= near buildings at the present time.

Two types of experiments are possible:

full scale and wind tunnel model tests.

In a given configuration, the full scale test will give the most direct and reliable data,

However, the cost of a comprehensive test series is considerable, particular+ if one wishes to systematically evaluate the effect of changes in source location and wind direction.

Practically, the wind tunnel test is the more desirable

approach, but its use can be justified only if questions of flow simulation and scaling can be resolved.

These are discussed in detail in Ref l~ where it is concluded that diffusion patterns around sharp-edged buildings in an adiabatic atmosphere can be modeled if geo-metric similarity of the model and dynamic similarity of the free stream are preserved, and the Reynolds Number is not too low.

Ref 1 also pro-poses a method of. presenting concentration data in non-dimensional form, thus enabling the scaling up of model results.

Confirmation of the validity of model diffusion testing by direct comparison with full-scale tests has not received much attention, However, in reporting on one such correlation test of gas released in the streets of a model city, Ref 4 states that the concentrations predicated

'y the wind tunnel tests fall within the range measured in the field under a not-too-steady atmcsphedc condition, thereby implying that tunnel tests are, at least, a good approximation to field tests.

I

The principal purpose of the tests described in this report was to determine characteristic diffusion,patterns around gas leaks at arbitrary locations on the surface of a typical nuclear reactor shell, The EBRII reactor at the National Reactor Testing Station at Idaho Falls, Idaho was used as a model.

Most of the tests were conducted with the reactor alone on the bare floor of the wind tunnel in a wind stream having a logarithmic mean velocity profile scaled down from an average profile measured at NRTS.

A few exploratory tests were made to determine the effects of. surrounding the reactor with the'buildings that exist at NRTS, changing the velocity profile from logarithmic to uniform, in-creasing and decreasing the wind velocity, and raising the surface temperature of the reactor shell.

A few tests were also made to determine ground concentrations resulting from gas released through a stack near tie

shell, The tests were sponsored by the Environmental~ Meteorological Research Prospect of the U. S. Weather Bureau, in collaboration with the Environmental and Sanitary Engineering Branch, Division of Reactor Develop-

ment, U, S.

A. E. C., Germantown, Md.

Personnel of the W. B. Research

.Station at NRTS cooperated in providing physical dimensions and wind data at the Station.

t 1

e h

0

~

i c

1 2.

TEST APPARATUS AND PROCEDURE 2.1 Wind Tunnel The New York University 3>~x7 ft wind tunnel, in which these'ests were conducted~

was designed specifically for the study of air flow phenomena associated with the dispersal of stack gases.

The tunnel is of the open return type~

the air being drawn from within the laboratory building and exhausted out through'he roof.

The three main parts of the tunnel are:

(1) inlet section~

(2) test section and (3) air driving and exhaust section, The function of the inlet section is to produce an air stream of uniform velocity and temperature at the entrance to the test section.

The inlet section contains fans and thermostatically controlled heaters to raise the temperature of the intake air to about 10F above building ambient temperature.

The fans prevent thermal stratification, Three fine-mesh screens in an expansion section~ followed by a contraction cone, reduce air turbulence to an extremely low value and produce a

sub-'tantially constant velocity distribution.

The test section is seen in Fig 2, It is basically a horizontal

'ectangular duct 7 ft wide x 3$ ft high x 40 ft long.

A grid of horizontal electric heating wires spaced 10 per inch vertically is stretched across the air stream 1 ft upwind of the test section.

Although not used in these

tests, the heating grid provides a means for introducing a vertical tempera-ture gradient in the air stream as it moves into the test section.

The floor and ceiling of the test section are of sheet steel construction and are temperature-controlled for their entire lengths, The left wall of the test'section (looking upwind) has a series of windows beginning 10 ft from the upwind end.

The right wall is of plywood construction, Two rails on the ceiling sup'port a survey carriage which positions instrument probes in three dimensions by remote control, The driving and exhaust section is located downwind of the test section.

It contains an electrically driven fan that produces continuously-variable air velocities in the range 0 - 20 fps.

2.2 Model auxiliary The basic the model sagebrush an aerial The model consisted of the reactor shell~

the stack and the buildings constructed to a linear scale of 1 inch 8 ft,or 1:96.

feature of the topography~ its flatness, was reproduced by placing on the tunnel floor.

The surface roughness, consisting of 20-inch covering 20$ of (he area, was not reproduced.

Figs 1 and 2 show view of the prototype and two views of the model in the tunnel.

H The model wa's built to dimensions taken from a plot plan of the NRTS area, DWG. EBRII-101-1DO-1 Rev l~ dated 5/28/58.

Subsequently~

an aerial photograph, taken July 28~ 1961, was provided, from which the size and location of actual buildings could be determined.

Some discrepancies were found in the locations of several of the buildings.

Fig 3 shows the model as constructed, Fig 4 shows the actual situation, with the model H

C 1

1 0

0 f

~

~

1 I

building outlines superimposed as dashed lines.

Auxiliarybuildings were constructed of 1/8 inch cardboard; only the gross features of the build-ings were reproduced, The dimensions of the reactor shell, in terms of full-size feet and fraction of reactor diameter~

are shown in Fig 5.

Two" models of the shell having identical external contours were used during the test, The first of these,

unheated, consisted of a cylindrical lower section of brass tubing having a 10 inch O.D, and a hemispherical solid wood dome of 5. inch radius.

The entire model shell stood 12.25 inches high.

Aper-tures of 0.063 inch diameter~

were provided at 0.75 inch and 7,75 inch above the base of the cylinder (along a vertical line) and at the top of the dome, to serve as gas leakage points.

Individual sections of 0.094 inch I,D. copper tubing connected each aperature to an S02 source outside the tunnel.

Only'one line was used at a time~ the remaining lines being capped to avoid leakage into the tunnel.

The heated shell was similar to the unheated sheU., but its dome was of

~ inch thick brass.

Heating elements were mounted within the shell, and thermocouples were placed in the shell to enable monitoring of the shell temperature.

The current input to the heating coils was adjustable~

and the sheU.'s temperature could be maintained at any value from tunnel ambient temperature (85 F) to 400 F above ambient.

The leakage points and cdnnections in this model were idbntical with those of the unheated model.

The prototype stack is 100 ft high and has an I,D. of 5 fC These dimensions were approximated by a 12.5 inch length of 5/8 inch O.D. tubing (0o578 inch I,D.).

Stack extensions consisted of additional lengths of identical tubing.-

H 2,3 Wind Velocit Profiles The wind velocity in the empty test section is uniform in each cross-section except in the boundary layers~

which are about 4 inches thick at the model location.

Therefore, the full scale equivalent vertical velocity profile would consist of a ground boundary layer about 30 ft high~

above which the velocity is uniform to a height of about 300 ft.

For con-venience, this profile willbe designated "uniform",

The uniform'rofile can be modified by inserting horizontal plates in the upwind end of the test section, The desired profile is obtained by

'djusting the plate spacing and adding roughness strips to each plate.

The profile so produced decays over the length of the test section~ but is reasonably stable over the 15 ft length occupied by the model.

In most of the tests in this series~

the plates were adjusted to reproduce the profile measured at NRTS in moderate winds and adiabatic conditions.

According to data from NRTS~ representative full scale wind parameters are ~ ~ 25 cm/sec (0,82ft/sec) and Zo 1.5 cm (0.0492 ft).

When reduced. in a velocity scale of 1:3 and a linear scale of 1:96~ these parameters become ~

0.273 ft/sec and Zo Q QQQ513 ft.

Assuming k

0,4~ the equations of the full scale and model profiles become

5 l

1

'I

'll J

full scale:

uf = 2.05 ln(Zf/.0492) model:

u 0,683 ln(Zm/0.000513) ~

(2) where u

mean velocity,ft/sec, at height Z

Z elevation, ft, above ground

'm and f

~ model and full scale For convenience, this profile will be designated "logarithmic".

The full scale logarithmic profile is shown in Fig 5, together with the equivalent location of the tunnel ceiling and its boundary layer.

The ad)ustment of the profile.plates was guided by vertical velocity traverses with a hot-wire anemometer probe at the location of the reactor shell in the absence of the mo'del.

After the desired pro-file was achieved~

the plate arrangement was not changed for the entire series of logarithmic profile tests.

Although the tunnel was shut down overnight, the profile was recovered without difficultyby ad)usting the propeller speed to produce the same velocity at a given elevation.

This elevation was arbitrarily chosen as 20 inches in the tunnel (160 ft full scale);.

the corresponding mean veloc itywas 5$4$t/sec (1h.7+sec full scale),

When the model was placed in the tunnel, the traverse ro'dwas re-located 3 ft upwind and 1,67 ft to the side of the reactor, as shown in Fig 2, A vertical velocity traverse was made at the start of each test day.

The velocity at the 20-inch elevation was monitored constantly during the tests.

2.4 Gas Tracer Techni ue After the flow in the tunnel stabilized, S02 was released from the model~

and sanples af air from the vicinity of the model were withdrawn from the tunnel and analysed.

The apparatus used for dispensing the gas and analysing the sample consisted of a metering bench and a

Consolidated Titrilog, somewhat modified to suit the particular needs of the tunnel.

A complete description of the gas system may be found in Ref l.

In this test, samples were taken in the air around the model, along the surface of the model~

and along the, ground.

At the model, the tip of the 1/16 inch I,D. probe was placed 0.1 in (0,8 ft full sire) from the surface, When measuring ground cohcentrations, the probe tip was placed 0.5 inch (4 ft full size) above the tunnel floor.

The probe was mounted on an out-rigger extending forward of the control carriage in order to minimize flow disturbances that might be introduced by the carriage.

The sample aspiration rate was maintained constant at 900 cc/min,

I t

~

I

~

~

I The pure gas release rate from the shell orifices varied from 6.8 to 300 cc/min~ according to the location of the sampling probe~

the larger rate was used fo'r traverses considerably downwind of the shell~

for configurations involving the shell in the complex~

and for the heated shell tests.

The emission velocities from the 0.063 inch diameter orifices, c'orresponding to the above flow rates~

were 0.2 to 8.1 ft/sec, respectively, In the stack tests~

the gas component of the effluent mixture

. varied from 100 to 1010 cc/min, while,the emission velocity was maintained constant at 4.6 ft/sec, 2.5 Photo ra hic Techni ue In order to obtain a visible record of gas and air motions in the vicinity'of the reactor, smoke was released at various points near the reactor surface, and single-flash photographs were taken with cameras mounted above and to the side of the reactor.

The oil-fog smoke was pro-duced by a special generator located outside the tunnel~

and brought through the tunnel wall by means of a rubber tube terminating at a nozzle having an inside diameter of 0.45 inch, The tip of the nozzle was placed within 0.5 inch of-the reactor surface, and smoke was released at a speed of about 5 ft/sec while the tunnel velocity was maintained constant at 2 ft/sec.

These conditions were optimum for the smoke photographs but were not used in~ and do not represent, the gas-tracer test conditions.

The smoke nozzle was positioned for each photograph so that the smoke was re-leased normal to a solid surface~

thus dissipating the momentum of the get and allowing the smoke to assume the local air velocity quickly.

To provide the necessary lighting~ two special strobe tubes were mounted under the tunnel ceiling directly over the model, A Leica IF camera with a 3,5 cm wide-angle lens was used with Flux X film and a

shutter speed

'of f/16.

The camera is shown mounted in its ceiling loca-tion~ but without the strobe lights~ in Fig 2.

~

I

7 3.

TEST PROGRAM AND RESULTS 3.1 Test Pro ram The test program consisted of (1) a qualitative study of the field around the reactor shell by visual observation of the travel of smoke released near the shell and (2) a quantitative study of gas con-centrations produced by'he release of gas through orifices in the shell and through a nearby stack.

The test conditions for the latter study are summarized in Tables 1 and 2, Locations in the tunnel are referred to a right hand coordinate system originating at the center of the base of 'the reactor shell on the tunnel floor.

The coordinate axes are oriented parallel to the tunnel axes:

X longitudinal~ 'ositive downwind; Y ~ lateral, positive to left looking downwind; Z

vertical~ positive up.

Linear dimensions are expressed as equivalent full scale feet, or as non-dimensional ratios referred to the reactor shell diameter D.

80 ft.

In the case of stack releases only~ the non-dimensional ratios are referred to the reactor shell height'

~ 98 ft.

Unless otherwise noted, the term wind velocity refers to the

'velocity at the reference elevation of 20 inches in the tunnel, and is designated by U, with appended subscript m or f to indicate model or full

-scale.

The wind velocity" profiles are designated as either logarithmic or uniform~

the profile details are given in Sec 2,3.

The surface temperature differential listed in Table 2 is the nominal excess of shell temperature over. the uniform temperature of the tunnel wind stream.

The differential is actually the average of four

readings, two in the dome surface and two in the cylinder surface,

~

symmetrically disposed on the upwind and downwind centerlines.

Due to poor heater location~ the dome temperature was considerably higher than the cylinder'temperature.

The actual temperature distributions during the two heated shell tests are given in Table 3.

3.2 Test Results The test results consist of photographs showing the general nature of air flow and diffusion in the vicinity of the reactor (Figs 6 and 7),

and graphs of concentration coefficient isopleths (Figs 8 to 36),

It willbe helpful if the results are viewed against a background of a general understanding of cavity and wake flow behind objects.

The subject is covered in considerable detail in Ref 1, but some of the material pertinent to flow around spherical domes willbe repeated here for the reader's convenience.

~

g

~

3.21 General Nature of Flow Fields Near Ob ects The flow field around an object in a'wind stream contains several zones having markedly different characteristics:

a)

Adjacent to each surface~

and completely surrounding the

object, there exists a thin boundary layer in which the mean velocity increases asymptotically from zero at the object surface to a slowly-varying value in the outer portion of the boundary layer.

b)

Outside of the boundary layer and immediately downwind of the object, there exists an ellipsoidal region called a cavity in which the velocities and pressures are low and the turbulence is very high, c)

Surrounding the cavity and extending a considerable distance downwind from the object, there exists a paraboloidal region called a wake, characterized by ambient pressures and velocities lower than free-stream

velocity, d)

Surrounding the wake and the upwind boundary layer~ there exists a region called a displacement zone in which the fluid is displaced laterally as it flows around the object and the wake.

The flow in the displacement zone is substantially potential~

and is characterized by well-

defined, curved streamlines~

low turbulence~

and pressures and velocities related through Bernoullis Law along a streamline.

e)

The object and its boundary layer~ cavity~ wake and dis-placement zone are all immersed in the background flow~ which~ in the case of a building resting on the ground~ is the earth's boundary layer.

The sketch on the following page shows how these zones are arranged about the reactor shell.,The sketch is approximate~

since no velocity measurements were taken other than in the background flow.

The zone boundaries were established as follows')

between background and displacemeat flo or between back<<

ground and wake flows:

the surface at which the local velocity deviates more than 5$ in magnitude or direction from the velocity of the back>>

ground flow in the absence of the object, b) between displacement flow and upwind boundary layer or between displacement and wake flows; the surface at which the velocity deviates more than 5C from the (theoretical) potential flow velocity around the object~

c) between wake and cavity flows the streamline surface within the wake~ separating the circulatory flow within the cavity from the consisteritly downwind flow in the balance'f the wake.

1 I

~

L I

4 L

0

displacement zone background flow cavity '

>~qc wake stagnation points a)

Flow in a horizontal plane near the ground upwind boundary layer separation cavity poiirr-~

boundary cavity displacement background flow wake primary flow IB ground vortices g gC secondary flow b)

Flow ln the longitudinal centerplane c) Velocity profiles in the longitudinal centerplane Sketch of Flow Patterns near the Reactor Shell

1

~

~

~

10 A very important characteristic of flow around objects is the existence of the cavity.

The cavity is generated when the upwind boundary layer separates from the surface of the object, and prevents the potential flow from following the object contours, The dimensions of the cavity, and the velocities and pressures within it~ depend principally on the velocity and pressure of the potential flow at the line af separation, On an object with rounded surfaces~

the line of separation is near the inter-section of the object surface and a transverse plane through its maximum cross-section.

Along this line~ the potential velocities are a maximum (25-5(g higher than the background flow) and the static pressures are a minimum for the entire flow field.

The streamlines originating at the line of separation, and forming the upwind portion of the cavity boundary~

are only slightly concave toward the axis of the object~

therefore~

the velocities and pressures along this part of the cavity boundary are fairly constant and equal to the velocities and pressures at the line of separation'.

The pressure within the cavity is basically the same as at the boundary, At the downwind end of the cavity, the boundary collapses toward the axis, and the streamlines converge to a common point at the ground~

forming a stagnation point (at C in the sketch),

where the velocity is zero and the pressure is somewhat higher than the background flow pressure.

Thus the pressure gradient along the ground is such as to induce a flow from the stagnation point C toward the object~

as well as from the stagna-tion point downwind.

The upwind flow near the axis of the cavity canbines with the downwind flow near the cavity boundary to form a toroidal vortex within the cavity.

All flow downwind of the cavity is in the same direction, producing the characteristic wake velocity pattern of low velocity near the axis increasing to the background flow velocity radially outward and longi-tudinally downwind, Velocity profiles at various longitudinal stations are shown in the sketch, If one is interested only in the flow pattern in the immediate vicinity of the object, one may ignore the downwind part of the cavity and view the object as being exposed simultaneously to a strong primary (down<<

wind) flow and a weak secondary (upwind) flow.

These two flows produce stagnation zones of local high pressure and zero velocity at the upwind and downwind portions of the object surface(at A and B in, the sketch),

The high pressures cause the ground boundary 'layers of the primary and second-ary flows to separate from the ground near the base of the object, and form grourid vortices directed down along the base of the object and radially out-ward along the ground. If the ground boundary layer i.s thick, the kinetic energy of the primary and secondary flows in the lowest layers willbe very small, allowing the radial outward 'flow caused by the vortices to extend a

considerable distance from the base of the object.

Both primary and secondary flows accelerate around the object from stagnation regions A and B to high velocity regions near the lateral periphery of the object, where they combine and leave the surface.

The line of separation is the juncture of the surface boundary layers produced by the primary and secondary flows.

The position of the separation line is quite sensitive to the relative kinetic energies of the two flows, The distribution of kinetic energy within each flow is different.

The primary flow has principally high velocity and low turbulence, and therefore, has a

~

E

'V 1

0

h1gh proportion of ordered

energy, The secondary flow has low velocity and extremely high turbulence;,

therefore the ordered energy is smaller, This imbalance of orderedbut oppositely-directed energy always produces boundary layer separation downwind of the max&urn cross-section of,a, rounded ob)ect,

>1hen conditions are such that the ordered kinetic energy in one of the two boundary layers is increased~

that boundary layer will remain'attached to the surface for ageeater distance.

An increase of turbulence in the background flow willcarry ordered energy from the primary flow into the primary boundary layer, thus driving the line of separation downwind~ thereby reducing the size of the cavity, increasing the cavity pressure and weakening the toroidal vortex.

3.22 Smoke Photo ra hs The smoke sequence of Fig 6 shows t'op and side views of smoke-patterns produced when smoke is released at points successively farther downwind in the longitudinal centerplane but along the solid surface of the ground or'hell.

The length of the side'f each square in top view is equal to the shell diameter D.

The proximity of the camera makes the dome seem oversized.

In side view, true distances may be scaled along the tunnel centerline through the base of the shell.

Top and side views were not photographed simultaneously; the pictures do not correspond ex-actly due to the turbulent nature of the flow, In Frame l the flow at the ground appears to move as in potential flow around the shell.

In Fx'arne 2, however, the clockwise-rotating ground vortex shows up clearly in side view, and the top view shows that the vortex carries gas almost D/2 upwind from the shell

surface, In Frame 3 the effect of the ground vortex is even more ap-parent.

The vortex core has bent into a horseshoe around the shell~

inducing a downflow along the lower upwind surface; therefore~

smoke from the nozzle moves diagonally downwind and down.

Frame 4 shows the effect of introducing buildings upwind of the shell without moving the probe.

The cavities created by these buildings completely destroy the flow pattern~

and draw smoke upwind to the boundaries of the building cavities.

The flow over the upwind part of the dome and the upper part of the cylinder is essentially potential (Frames 5 and 6),

However~ Frames 6 and 7 clearly show that separation occurs at about 90>> ll0

~ marking the beginning of the cavity.

Frame 8 shows the powerful effect of the background flow in overcoming the secondary flow near the lee surface of the dome.

Although a thin haze appears as far upwind as 90, the main flow in the upper part of the cavity has a strong downwind component which leave's the dome surface radially at an angle of about l50o.

The top view shows no lateral spread near the nozzle; in fact the sharp smoke outline through an azimuth angle of 270 indicates that the flow converges toward the nozzle.

The side view shows a similar convergence from below.

In Frame 9 the nozzle is partly in the 3-dimensional flow region around the dome and partly in the 2-dMnsioralregion around the cylinder.

The flow is upward and laterally'utward.

Smoke moving upward departs as in Frame 8, but smoke moving laterally outward travels around the sides of the cylinder to the cylindrical separation poin't near 90,

~

T 0

12 h

In Frame 10 the probe is clearly in the two-dimensional flow zone created by the cylinder, In side view we again see a sharp downwind outline of the smoke cloud near the tip of the nozzle~ indicating the strong secondary flow of the cavity vortex impinging on the lee surface of the cylinder. 'n top view we find an arc of denser smoke hugging the she 11.

Frame 11 shows the downwind ground vortex spreading smoke down-wind along the ground despite the general reverse flow.

Frames 12 and 13 show that the cavity extends at least as far as 2D downwind.

In Frame 13 the smoke is indecisive.as to direction.

Frames 15 and 16 show only downwind flow. It is clear that the cavity boundary is very close to 2.25 D from the shell center, Generally, we observe that very little smoke is visible down-wind of x ~~

2D~ or higher than z ~~

1H in side view.

These may be taken as the effective length and height of the cavity.

The maximum total width of the cavity between x = 0 and 2D is ~~ 1.5D, 3.23 Gas Diffusion in Non-Uniform Flow Fields The conventional point-source diffusion equations can not be used for estimating diffusion near ob)ects in a wind stream because they were derived for~ and apply only to~ uniform flow fields containing straight, parallel streamlines, and homogeneous velocity and turbulence.

There are no theoretical solutions available for diffusion in the highly non-uniform flow fields around objects.

At the present time, the experimental approach is the only one that shows any promise of providing useful data, The wind tunnel model test is evidently a very practical method of testing a large number of configurations, but data obtained in such tests must be capable of extrapolation to full scale in order to be of use, In thi.s section~

we shall describe a procedure for non-dimensionalizing test data for subsequent use on any scale.

In Ref 1~

the problems of similarity and scaling are discussed, and it is concluded that diffusion patterns in model and full scale flow fields willbe similar if dynamic similarity of the fields exists.

Dynamic similarity means that velocities in one flow field can be made to equal the velocities at corresponding points in another flow field by application of a single multiplying factor.ln turbulent flow, similarity of instantaneous velocities is not possible, but similarity of mean velocities and the r.m.s.

values of the velocity fluctuations are believed to be attainable a'nd adequate for most scaling problems.

To achieve dynamic similarity one must provide geometric and dynamic similarity of the boundary conditi.ons, and Reynolds Number and~

sometimes, Froude Number scaling must be used in the flow field.

The boundary conditions are, for the most part~ controllable.

They consist of a) physical dimensions; including the object, the size, shape and location of the exhaust aperture, and the terrain,

1 I

~

I

~

~

1 13 b) velocities; including the mean velocity gradients and turbulence distribution of the background flow, and the mean velocity and turbulence of the effluent at the point of release~

and c) temperatures; including temperature gradients in the back-ground f'low and temperature excess of the effluent over the background.

\\

All of the above conditions, except turbulence in the background flow and in the effluent~ were controlled during the shell tests.

The Reynolds Number was not observed; the Froude Number was observed in the heated surface tests.

The effect of these lapses on dynamic similarity will be discussed in a later. section.

For the present~ let us assume that the model and full-scale flow fields have similar mean velocity and turbulence distributions. It follows therefore, that the concentration patterns in the two fields should also be similar.

I I

~

I, I

0 0

e y

e e

e 14 A concentration pattern can be described completely by specifying the local concentration at each point in the field. If two concentration patterns are similar, each should be reducible to the same non-dimensional form by an appropriate scale factar,Let us determine, first, the non-dimensionalizing constants for the 'simple case of continuous point-source diffusion in a wind stream having uniform mean velocity and isotropic turbu-lence everywhere.

The Hay-Pasquill equation for the special case of linear plume expansion over a short distance~

and remoteness from solid boundaries is r2 c ~ exp s

V (3) where:

c ~ gas volume concentration vol. gas/vol. mixture gas volume release rate (vol./time)

V uniform mean wind velocity (length/time) x downwind distance (length) 6s r.m.s. distance of concentration distribution (length).

If it is assumed that small eddies travel,in straight lines for short distances, the concentration distribution can be related to the at-mospheric turbulence through the expression where:

<, - x6q/V r.m.s. velocity fluctuation Eq (3) then becomes:

2 c

exp r

2+(x 6 /V) V 2(x6 /V)

Eq (5) can be paraphrased thus:

where:

c K')V A~>> 2+(x 6 /V)

K'xp

- r /2(x 6'V)

~

I 0

0

Since K's an exponential function~ it must be non-dimensional, and the argument must also be non-dimensional.

Therefore, the non-dimensionalizing length factor must be ~2x 5 /V, This factor also appears as the radius of a circle of area A', oriente3 normal to the wind.

The quantity Q/A'V can then be interpreted as an average concentration produced by depositing Q of gas into a duct of cross-section area A'arrying an air stream of uniform mean velocity V:

cav

~ Q/A'V (7)

From Eqs (6) and (7), we can write K

c/c ave and the non>>dimensionalizing factor for concentrations becomes c

. Q/2 g (x +q/V) V.

The quantity cav determines the general magnitude of the concentrations and, therefore, it maybe said to define the'scale of the

problem, The quantity K'imply describes the relative magnitudes of the concentrations in the field, independent of scale.

It willbe caU.ed the distribution function. K's wholly dependent on the coordinates x and r, and the dynamic property 6'q/V~ assumed uniform everywhere in the flow field.

The uniformity of Pc/V automatically provides dynamic similarity in all fields to which Eq (5$ applies.

We now wish to apply the same type of reasoning to diffusion in the non-uniform, non-homogeneous flow field about an object.

Un-fortunately~

no theoretical solution for the concentration distribution is available.

However, we have specified that the mean velocity and turbulence distributions in model and full scale are similar.

Therefore, we may ex-pect ghat an equation similar to Eq (6) can be written, containing a scale term 9/AU and a distribution function K:

(9)

In Eq (9)~

c and 9 have exactly the same meaning as in Eq(6)~

but A~

U and K are analogous to, but not the same as, A', V and K~,

The difference lies in the fact that A', V and K'or a.uniform field are uniquely defined, whereas A,

U and K in a non-uniform field are not.

Nevertheless~

Eq (9) can be made to serve the same function as Eq (6) by arbitrarily selecting an area A and a velocity U in the flaw field, and calculating K at each point in the field using experimental determinations of c for a gas release rate Q,

By measuring c and calculating K at a sufficient number of points, the function K can be determined in the entire flow field, In this procedure, the numerical value of K is meaningless

='ithout a statment as to how A and U are to be measured, but once this

~

1 0

4 t

16 statement is made, K is as firmly established as K'.

Evidently, the functional form of K will not be affected by the arbitrary choice of the constant AU, although the magnitude will. In the present tests, U

was taken as the mean wind velocity at an elevation of 160 ft (20 inches in the tunnel).

The area A was chosen as the cross-section area of a projection of the shell in a plane transverse to the mean wind direction.

The area choice was suggested. by the concept of an average concentration in the wake equal to cav 9/AU, analogous to Eq (7),

In some of its properties, K resembles the lift and drag co-efficients used in aerodynamics, and the friction factor used in hydraulics.

It is a non-dimensional coefficient'which maintains a more or less constant magnitude over a wide range of values of the independent parameters to which it is related.

One important difference between K and a liftco-efficient, however, is that K is not a single-valued quantity for a total configuration, as CL is for a given airfoil shape.

K varies 'from point to point in space and must be viewed as a field:

K K(x/L~ y/L, z/L)~ where x, y and z are space coordinates and L is a reference

length, However~

each dynamic configuration has a unique K field which is independent of the magnitude of 4, U and A.

K values in the field are, of course, time-means.

3.24 K-Iso leths The value of K at each sample point was calculated according to the formula:

1,7 cm"mUm K ~

(10) where:

cm sample volume concentration (ppm) shell frontal.y projected area 0.776 ft2 Um mean wind velocity (ft/sec) at Zm 20 inches Q

gas release rate (cc/min)

~ subscript to designate "model" The point values were then plotted on graphs corresponding to a surveyed plane~

and isopleths of oonstant K were drawn through the points.

The results are presented in Figs 8 to 36, Each Figure re-presents a different configuration.

The tests will be discussed in groups, following their arrangement in Tables 1 and 2.

18 8-4 This series is the most comprehensive in the program.

The basic configuration is the shell alone at ambient temperature in an isothermal wind having a logarithmic velocity profile.

The variable is the source location.

The isopleths of Figs 8 - 12 may be compared with the photo-graphs of Fig 6 as follows:

~ '

'7 0

0

Release point 17 bottom mid-height mid-height bot tom

'pwind upwind downwind downwind Isopleth Fig No.

8 10 12 Photograph 6, Frame No.:

2 Each Figure contains isopleths in the longitudinal centerplane, in a horizontal plane 4 ft above ground, in transverse planes at X/D 1,25 and 2.5~

and along the shell surface.

Fig 8~ bottom upwind release, is characterized by the splitting

'f the gas into two highly concentrated symmetrical streams moving down-wind along the base of the-shell and separating near 8 90 After the 0

streams leave the shell, the line of maximum concentration appears to follow the cavity outline at a lateral distance of ~ D/2 from the axis.

However~

gas iq found along the ground up to~D/2 upwind and~D laterally from the base of the shell.

This is believed to have been caused by the upwind ground'ortex.

Vertical diffusion is negligible along the upwind shell surface, but diffusion in the cavity causes contamination of the entire le'e surface of the shell.

In Fig 9, mid-height upwind re3sase, the air flow near the re-lease point is principally in the upward and lateral directions~ pro-ducing strong contamination of the upwind dome surface and general con-tamination of the entire lee surface of the shell; The upwind cylindrical surface is not contaminated.

Gas concentrations in the cavity are highest near the top.

Fig 10, top release, is similar to Fig 9, except that the. upwind dome surface is clear.

Fig ll~ mid-height downwind release, shows high concentrations along the lee surface of the dome and in the top of the cavity.

Ground concentrations are about the same as in. Figs 9 and 10, Fig 12, bottom downwind release, shows high concentrations along the lee surface of the shell from ground to mid-height~

and high ground concentrations in the cavity for a distance of w D from the base of the shell.

Fig 13, bottom side release, shows an asymmetric pattern with high concentrations near the release point, of the same order of magnitude as in the symmetrical bottom upwind release of Fig 8.

The high ground concentrations are found closer to the axis than in Fig 8.

Fig 14, mid-height side release, shows low ground concentrations everywhere in the cavity.

Apparently the gas stream separates from the side of the sheU. and undergoes considerab3a diffusion before Joining the secondary flow, There appears to be some indication of a local high con-centration region near the top of the dome on the side opposite the re-lease point.

I

~

I I

~

I

It is of some interest to determine the maximum concentration in the vicinity of the shell as a function of distance from the shell.

Such'curves.are shown in Fig 37, based on the data of Figs 8 - 13.

The abscissa S/D is a non-dimensional straight line distance from the shell surface to the farthest point of a given K isopleth~

measured as shown in the sketch, The numerals on the curves designate the Figure Number.

The letters C and G indicate whether the maximum*value of K was found within the cavity or on the ground, respectively.

The heavy dashed line approximates the maximum value of K measured in these configurations at the distance S/D.

The line may be expressed by the formula

(~/O)

Maximum concentrations occur at the ground for bottom release points~

and in the cavity space for mid-height and top release

points, The maximum ground and cavity concentrations Are about equal, The bottom upwind and bottom side releases are critical for ground ccntaninaMn.

The mid-height downwind release is critical for cavity contamination.

Fis 15-17 These tests were made to determine the effect of wind velocity profile on concentrations in the longitudinal centerplane of the cavity.

The following configurations are comparable except for profile:

Release point:

bottom upwind top bottom downwind Log Profile Fig No.:

8 Uniform Profile Fig No.:

10 16 12 17 There is very little difference in the K isopleths for the top release.

The bottom upwind release is affected markedly, concentrations in a uniform profile being less than in a logarithmic profile everywhere in the cavity by a factor of 4 or more.

The bottom downwind release is affected to a lesser

degree, the concentrations being lower by a factor of ~ 1.5.

These tests were made to check the effect of Reynolds Number by varying the wind velocity, The tests were run on+ with a uniform profile.

Isopleths may be compared to Fig 16.

The shapes of the isopleths in the longitudinal centerplane of the cavity do not change much as the wind velocity is varied from 3 to 15 ft/sec.

The variation af maximum concentration with distance is shown

I I

~

\\

~

in Fig 38, with wind velocity as a parameter.

K is found to reduce by a factor of 2 over the tested velocity range, the bulk of the reduction occurring in the step from 10 to 15 ft/sec.

It is not clear from the data whether this is a Reynolds Number effect or experimental error.

F s 21 to 23 These tests were made to evaluate the effect of heating the shell surf ace.-

Customarily, tests involving elevated temperature are run with Froude Number scaling~ which requires that the velocity be re-duced by a factor equal to the square root of the linear scale.

Thus, the velocity scale factor is ~9

~ 9.22 and the reference velocity

~

Um 16.7/9.22 1.8 ft/sec.

The tests were actually run at 1,7 ft/sec.

Fig 21 may be compared with Fig 10 to observe the effect of reducing the absolute wind velocity while maintaining a logarithmic profile.

In Fig 21, the isopleths between the shell surface and X/D 1.5 are believed to have been based on questionable test data, and should be

ignored, At distances greater than X/D ~ 1.5, the isopleths for U

1.7 ft/sec appear to be smaller than those at 5.54 ft/sec by a factor oi about 2.

This is contrary to the trend shown in Figs 18 - 20.

Actually, insufficient data were taken to draw detailed isopleths at 1.7 ft/sec; therefore Fig 21 will be disregarded~

Fig 10 will be taken as representative of the isopleth pattern at a temperature differential of 0 F, Mhen the surface temperature is raised to 150 F above ambient (Fig 22)~ the limited data indicate that convection currents created by heat loss from the shell carry most of the ga's stream out of the cavity, and produce only very small concentrations on the ground and at the shell surface.

At a differential of 300 F, none of, the gas reaches the gr'ound.

Fi s 24 to 28 As seen in Frame 4 of Fig 6, the concentrations around the shell are affected strongly by, the presence of upwind auxiliary buildings, These buildings. create their own displacement fields, cavities and wakes~

which~ in effect, become the background flow for the shell.

In SN and SSK'inds, the principal upwind building is the 40 ft high~ L-shaped power plant.

The lower part of the 98 ft high shell is immersed in the wake of the power plant.

The upper part is exposed to the normal background flow, The general flow pattern at low elevations is very'omplicated, It consists, roughly, of two elements:

one is a characteristic cavity vortex between the power plant and the shell~

the other is a smooth air stream that splits laterally around the power plant, the westerly branch swinging around the edge of the building and being diverted toward the shell by the 46 ft high sodium boiler building to the North.

This westerly branch impinges on the shell, flows around it~ and leaves the shell vicinity through the NE passage between the sodium boiler building and the fuel cycle facility (16 ft high),

I I

~

~

0 0

20 In Figs 24 to 28, the isopleths are in a horizontal plane 4 ft above the ground; the x<<marks are individual sample locations on the building roofs.

In Fig 24, bottom upwind release, most of, the gas is caught in the power plant cavity~

some of it even moving upwind over the power plant roof.

Generally~

the concentrations between the shell and the power plant are higher than the concentrations downwind of the shell, In Fig 25,top release, the gas concentration pattern on the ground is similar to the pattern around a chimney.

There are little or no concentrations at the base of the shell, and the maximum concentration is in the vicinity of 5D downwind, This condition is caused by the channel-ing of air into the shell cavity by'he sodium boiler and fuel cycle buildings, partial+ destroying the shell cavity vortex and reducing down-wash and secondary flow.

In Fig 26, mid-height downwind release, conditions are about the same as in Fig 24.

In Fig 27, bottom downwind release, the release point is in the partial shell cavity, and fairly high concentrations are found downwind of the shell.

Fig 28 shows the worst configuration with respect to contamination of the yegion upwind of the shell. It occurs in a SSM wind, It is interest-ing to note that the values of K 8 and 10 o'n top of the power lant are found at distances that satisfy Eq (U.),

For example~S/D 20 10~1.4 at K=10.

Fis 29 to 32 These figures show the effect of the presence of the shell on the concentration downwind of a stack whose height is equal to the shell height.

The wind and stack parameters were maintained constant for all,four tests.

The wind contained the standard logarithmic velocity profile; its mean velocity at the elevation of the stack opening was 5.2 ft/sec.

The effluent was discharged at a velocity of 4.6 ft/sec.

The effluent/wind velocity ratio was 0.9 at the stack opening.

The variable was the location of the shell; it was placed in several locations with its center 1H oz 98 ft from the center of the stack The configurations may be evaluated by comparing the maximum ground concentrations~

summarized in the following table:

Fig No.:-

29 30 31 32 Shell location:

none Kmax at ground:

0.01 X/H 'at Kmax, 17

side, 0,12 20?

downwind 0.35 12 upwind 0 7

I P

1

~

21 The greatest ground contamination occurs with the upwind shell

location, K az being 70 times greater than for the stack alone.

This is explainable by noting that the stack aperture is located'pproximately on the cavity boundary, whose streamlines curve downward, carrying the effluent to the ground with a minimum of dilution (see sketch on Page 9).

Mhen the shell is downwind of the stack, the aperture is in the displacement flow, and the effluent is carried over the cavity, de-scending farther downwind and at the same time~ experiencing greater dilution.

The increase of 35 times in K az over the stack alone condition is only half as much as in the upwind shell location~ but it is still considerable.

The side location of the shell produces a maximum ground con-centration 10 times as high as the stack alone, The question marks in the table indicate that these measurements, taken in a longitudinal plane through the stack, may not be true maxima, since the streamlines passing through the region near the stack opening take on a lateral curvature and may descend to the ground with higher concentrations at some lateral distance from the centerline.

In interpreting these results with respect to full-scale conditions, it must be noted that the fuU.-scale exhaust velocity is 35 mph (51 ft/sec),

and therefore, the corresponding full-scale wind velocity at the top of the stack must be 35/0.9 39 mph.

A smaller wind velocity will create a higher effluent/wind velocity ratio, which wiU. raise the plume to the top~ or out, of the cavity, and reduce ground concentrations.

Fis 33 to 36 These tests were similar to the previous series except that the stack height was increased 50$.

The stack opening was, therefore, well above the cavity for all locations of the she13..

In comparing the con-figurations, we are not able to use Kmaz at the ground as in the previous series~

since Kmaz occurs farther downwind than the observations, For convenience, we shaU. use K at X/H 17, Z/H ~ 0.2.

The following table summarizes the results:

Fig No,:

33 34 35 36 Shell location:

none K L17,0.2]:1 0.004 side 0,0022 downwind 0.05 Upwind K at this point is increased by a factor of 45 for the upwind shell and 12 for the downwind shell, The results for the side location of the shell must be regarded,

again, as uncertain.

I I

)

1

22 4,

EVALUATION OF RESULTS The calculation of diffusion may be regarded as a two-part process~

one is the establishment of an average concentration in the field,. the other is the determination of a function that describes the variation of concentration from point to point in the field.

The former quantity is specified, for continuous gas release in a moving air stream~

by the gas release rate, the wind velocity, and some representative area.

There is no ambiguity in these terms~

and they need no evaluation.

The latter quantity~ called the distribution function, appears tobe wholly dependent upon the mean and turbulent velocities of the flow field. If the distribution function is found, either analytically or experimentally~

for one flow field, then the same function will apply to a33. dynamically similar fields, The evaluation of the test results, therefore, need concern itself only with the distribution function and the nature of the flow field to which it applies.

In distorted flow fields~

such as occur around objects in a wind stream~

the distribution function K is determined experimentally by a series of simple measurements and a simple calculation.

The precision of the measurements is far greater than the concentration variation with time at any given point~

and so we may assume that the principal error in measurements would appear in the estimation of the mean concentration from a record. that shows very high fluctuations at many points in the field.

It is believed that the resulting maximum error in K is no greater than 10$ for any individual sample determination.

However, the unsteadiness of flow around objects~ particularly; the long period fluctuations, may give rise to variations of up to 255 ih the K determinations for groups of points.

This effect is noticed mostly in regions where the concentration gradients'are high.

The characterization of the flow field that exists during the

. determination of a set of.K-isopleths is more of a problem, The classical hydrodynamic treatment of isothermal~

viscous, non-turbulent flow leads to

'he conclusion that, if the boundary conditions and Reynolds Number are specified, the flow field is also specified~

even if an analytical solution of the equations of motion can not be obtained.

In non-turbulent flow~ the boundary conditions are zero velocity at all solid surfaces and a prescribed velocity variation at large distances from the source of disturbance, in this case the shell.

Generally~

these conditions can be met, although the Reynolds Number would have to be very low to produce completely laminar flow.

Reynolds Numbers encountered in model testing and in full-'scale are always higher than the critical Reynolds Number for the onset of turbu-

lence, and turbulence is always present in the flow field.

There are two sources of turbulence~

however.

One is the turbulence in the background flow and the other is the turbulence in the wake downwind of an object.

The relationship between the two has not been clearly established.

Wake turbu-lence willoccur at a high enough Reynolds Number even if the background flow is laminar, but a turbulent background flow induces wake turbulence at a lower Reynolds Number.

Clearly, the boundary conditions for turbulent flow must include the distribution of turbulence as well as mean velocity in the background flow.

23 Assuming that the turbulent background flow boundary conditions are met, we must now evaluate how the Reynolds Number affects the flow field~ if we wish to estimate its effect on K.

When the Navier-Stokes equations of f3uid motion are non-dimensionalized~ it is found that the inverse of the Reynolds Number appears as a coefficient of the shear stress term.

Thus~ if shear stresses are absent~

Reynolds Number willnot be

relevant, and if Reynolds Number is high, the shear term willbe numerically small; in either case, the shear term may be neglected.

This leaves the Navier-Stokes equations devoid of any scaling parameter, and the flow pattern becomes potential and independent of scale.

>le conclude, therefore, that Reynolds Number scaling is important wherever shear stresses are high.

Since shear stresses are proportional to velocity gradients, we may expect that Reynolds Number effects will show up in the boundary layer~

and in those regions that are controlled by boundary layer activity.

It has been found that the boundary layer thickness on a surface of finite length varies inversely as the Reynolds Number.

Therefore, we may expect that boundary layers on the full-scale shell willbe proportion-ately thinner than on the model.

Reduction in boundary layer thickness is equivalent to an increase in velocity at the same distance from the s'urface, Consequently, ventilation is greater, and gas concentrations should be'maller.

An indication of this was found in the unifom profile tests, Figs 15 to 17, where the boundary layer of the background flow was only 4 inches. thick~ compared to the logarithmic profile boundary layer that occupied the entire tunnel height.

The important role of the boundary layer in the formation of the wake and cavity was described in Section 3.21. It was shown that the size of the cavity depends strongly on the location of the line of separation, which, in turn, depends on the energy in the boundary layer.

High Reynolds Number flows have thin~ high-energy boundary layers which tend to remain attached to rounded surfaces.

Therefore, the line of separation moves downwind as Reynolds Number is increased>and the cavity size decreases.

The effect of increasing the energy in a boundary layer by introducing a disturbarae at the surface, and thereby replacing low energy air from the boundary layer with high energy air from the potential flow, is shown in Fig 7. The photographs at the right were made with the shell in the normal test configuration.

The photographs at the left were made with a $ x '~'angle taped to the surface of the shell with its outstanding leg along Q

-45.

The angle, or tripper~

causes the boundary layer to separate farther downwind, as seen in side elevation.

The half-width of the cavity on the tripper side (upper left photograph) is seen to be somewhat smaller than in the other configurations, and the cavity boundary appears to be sharper.

If gas is released in the cavity~ the available air volume for dilution is reduced, and concentrations within the cavity should increase.

However, gases released outside the cavity should have a greater chance to dilute.

These considerations are, of course, qualitative.

Unfortunately, no experimental data are available at present to evaluate the net effect of

I 0

E

fg i

l j

24 background turbulence and Reynolds Number on concentrations in the cavity, Research on this aspect is now in progress (Ref 2).

However, we would like to call the reader's attention to several correlation studies which bear upon, and tend to confirm~ the above line on reasoning, although they can not be cited as conclusive evidence.

Refs 3 and 4 are a correlation study of wind tunnel and full-scale tests of gas released. in the streets of a model c1tyo The results of both phases were non-dimensionalized in a manner similar to the one used in this report (although the reference area Awas selected differ-ently),. and the distribution functions found in the two phases were in reasonable agreement.

Ref 5 compares pressure differences on a model and a full-scale building in a.wind stream.

It was found that the pressure pattern on the building could be duplicated only when the atmospheric mean velocity pro-file was reproduced in the tunnel.

The similarity of pressure patterns indicates that the velocity fields outside the boundary layer were probably quite, similar. If diffusion depends only on the velocity field, then the requirement for uniqueness of a K field would demand scaling of the mean velocity of the background flow.

Ref 6 is the often-cited Rock of Gilbraltar study.

The full-scale flow:field was studied by balloon tracking and observations of cloud formation, The tunnel flow field was studied b7 visual observation of threads attached to wires. It was concluded that qualitative+, at least, the flow fields in the lee of the mountain were similar as to mean velocity, and turbulence distribution.

In summary, it is believed that the K-isopleth patterns presented in this report are basically independent of scale~ but willbe subject to some variation as a result of changes in Reynolds Number and turbulenc'e in the background flow. Sensiti'vity to these variables is attributed to the rounded surface of the shell, which permits movement af the separation line and~ therefore~ variation of the cavity size.

This condition does not exist for buildings with sharp

edges, since the separation line is fixed at the edges for aU. Reynolds Numbers (see Ref l).

To some extent an increase 'of turbulence and an increase of Reynolds Number have opposite effects, the former providing greater dilution by eddy activity~ and the latter providing less dilution due to shrinkage of the cavity.

The net effect is unknown.

The tunnel turbulence during the test was not measured, but subsequent measurements in flows produced by a grid of 2-inch wide slats on 6-inch centers plus a length of very rough terrain~

pr'oduced an average value of turbulence intensity 6 /V 1Cg. It

, is bel1eved that a similar value was, achieved during the shelf test series.

For comparison, the turbulence intensity in the empty tunnel is of the order of O.Ig.

One aspect of diffusion scaling, that sometimes is not given due

~ttention, is the implicit assumption that the flow f1eld is independent of the source strength g.

This is true when the gas molecules merely re-place air molecules in,the effluent, and the total veloc1ty of emiss1on is

~'

l c

l 0

25 unchanged, Evidently~ this condition can be achieved only if the gas concentration in the effluent is less than 1, When pure gas is released from a fixed aperture, the emission velocity willvary with ( and a new flow configuration willbe established for each value of 9, The K-field willbe very sensitive to 4 in a small region close to the emission aperture~ but the sensitivity decreases rapidly with distance, This is the justification for use of the point-source equations in stack diffusion calculations~

and the reason for the lack of validity of these equations close to the source.

In'the present test series, the gas release rate for measurements near the source produced a negligible emission velocity (Ve~ 0.2 ft/sec).

The high release rate (V

8 ft/sec) was used on+ for measurements at'large distanc6s from 'the shell, Effectively then, the shell releases can be propert'classified as "leaks",

In the stack tests, the gas concentrations in the effluent were too anall to noticeably affec't the effluent velocity.

~

1

)

I 0

0 d

M

26 5.

APPLICATION TQ~ULL SCALE DIFFUSION CALCULATIONS If exact dynamic similarity of a model and a full-scale con-figuration can be found, the full-scale concentration may be calculated by:

Qf cK K

(12) 1,7 AfUf where:

cf ~ concentration (ppm)

Af shell frontally projected area 7160 ft Uf mean wind velocity(ft/sec)at Zf 160 ft c

Qf ~ gas release rate (cc/min) subscript to denote full scale.

The factor 1.7 may be omitted if Qf, Af and Vf are in consistent units and Cf is dimensionless.

The value of K at any point CX/D~ Y/D~ Z/Dj,may be taken from the appropriate Figure.

The formula may be modified by the introduction of a numerical constant if the velocity at a different refer-ence elevation is used.

or, since in this case, In practical situations, one is often interested in determining the horizontal distance from the base of a shell~ within which the concen-trations will be higher than a given value, irrespective of the location of a leak in the shell.

Eq (11) is suitable for such estimates.

Substituting (ll) into (9)~ we derive C

20 (U)

AU (S/D)

A' 1.12 D ~.

4 2

</<<allow where S, Q,

U and chill~~ are in consistent units, Eq (14) produces conservative values of S, since the numerical constant 20 in Eq (11) was taken high enough to include all test data.

From Fig 37, one may see that a numerical constant of 10 would be a better average fit for the test data.

If the average fit is acceptable, the c'oefficient of Eq (3)I) should be reduced to 4.2/~2 3.0.

Mhen gas is released on the upwind face of the shell, the primary flow carries a,concentrated gas stream downwind along the cavity surface toward stagnation point C~ where the gas stream splits~ part moving downwind and part moving toward the shell in the secondary flow. 'Thus~ the

~

~

I 0

lI 0

I

27 downwind face of the shell is bathed in a contaminated stream whose base concentration can not be less than the concentration at point C.

We may estimate K at point C, since we have found that the cavity length is X/D 2.25 measured from the center of the shell~ or S/D

~ 1.75 measured from the shell surface.

Therefore K at point C

10/1.75

~ 3.

An exam1nation of the Figures shows that an average value of K along the downwind part of the cylindrical face of the sheU. is also about 3 for top, upwind and side releases.

Downwind of the cavity, the flow field first acquires the wake velocity profile and then the background flow profile.

Zt would be con-venient to be able to compute a continuous maximum ground concentration through all these regions, starting from the shell surface, At large distances from a continuous po1nt source at ground level~ the maximum concentration at ground level may be expressed by and where c

(see Ref 7, sec 5.3)

~l 5 y c"sV 6

x6e x{ 6v/V) x 6g /2 x(ts /V)/2 (15)

(+is the lateral r.m.s.

gust angle

'is the lateral r.m,s. fluctuation velocity v

The recommended value of Cg for a neutral atmosphere is given as a function of distance from the source, as followsi x (feet)'30 990 1980 3300 6z (radians):.076

.065

,056

,049 Expressing Eq (15) in terms of Q, we have c 'or x )330 ft 0.64Q x

6g V

(16)

We 'may transform the empirical Eq (13) into the form of Eq (16) by using A

1.12D, and a numerical constant of 10 instead of 20, in which case, Q

10D 8.9Q c ~. ~ ~ fcr S/D(5 112DV S

SV (17)

~

p

~

'I 0

0

28 Eq (16) will predict the same concentration as Eq (17) if s

~ x and 6e 0.27.

Therefore we find that the rate of diffusion for at least 2 cavity lengths domwind of the shell is about 4 times the rate in a neutral atmosphere

( g~ ~ 0.07).

The transition from cavity turbulence to normal turbulence in the atmospheric boundary layer depends on the rate of decay of the wake.

Experiments with flat plates nozmal to a laminar air stream in a wind tunnel show that the wake is quite persistent, having been measured as far as 680 disk diameters

downwind, and that the turbulence decreases as (x/d)" /3 for the range 10< x/d(1000.'he

-2/3 law may also be pre-dicted from momentum considerations, For lack of better data, we suggest the use of the following values of Cg in con)unction with Eq (16) for the calculation of diffusion in wakes in the atmosphere:

for 0 < S/D(5

~

6g constant at 0.27 for 5 C'/D($0, g+

0,8 (S/D) for 404 S/D, gg constant at 0.06 These values conform to our measurements close to the shell, to the Hay-Pasquill suggested values at large distances and to the theoretical (and experimentally confirmed) variation between the cut-off levels.

I I

'I e

0

la 4

),

e e

a 29 6.

RECOMMENDATIONS The concentration measurements made during this study have been reduced to a non-'dimensional coefficient form designated by the symbol K.

'As such, they are believed to be independent of scale in the same. context that the aerodynamic drag coefficient GD and the hydraulic friction factor f are independent of scale.

Practically, this means that the coefficients are not absolutely constant over a range of scales, but are sufficiently so to allow order of magnitude estimates in most of the range.

Evidently, more precise estimates require knowledge of the variability of the co-efficients, Under the, pressure of engineering economics~

considerable effort has been expended in the study of the aerodynamic and hydraulic coefficients; their nature and variability is now well-understood and well-documented.

On the other hand, very little is known about the concentration coefficient K.

Research in'atmospheric diffusion from 1935" to 1945 was done primarily in response to military demands, and was directed toward the solution of simple diffusion problems in uniform flows.

Those important studi,es led to formulas involving diffusion coefficients, of which a number have been, and still are being, suggested, Since 1945, attention has turned from military needs to environmental sanitation, and the focus is now on the very numerous sources of contamination near industrial buildings However, the classic solutions are not valid for the distorted flow fields near. such buildings, and a-more generalized form of the diffusion coefficients is required.

The concentration ccegficient K

g fx,y,zunis a general coefficient of this type.

There is ample room for research into the nature of K and its dependence on known and, as yet unknown, parameters, Two types of information are required:

one is the delineation of K distribution patterns around build<<

ings or groups of buildings other than the prisms of Ref 1 or the shell of this report; the other is a study of the variation of these patterns with Reynolds Number~ Froude Number and atmospheric turbulence, at the least.

The former may be carried out most conveniently in the wind tunnel, the latter will require some tests in the full-scale atmosphere.

The two studies are not dis)oint, since proper wind tunnel procedure will depend on precise know-ledge of the circumstances under which certain scaling procedures are mandatory, Taking into consideration the need for more order-of-magnitude

data, the uncertainty of extrapolation procedures and the available test facilities~ it would appear that two avenues of research should be pursued.

Some fuU.-scale'tudies should be made at the EBRII site to determine whether the K-isopleths found in the tunnel are in reasonable agreement with field studies, and model studies'f diffusion in built-up areas should be undertaken.

The KRII study would supplement the correlation research now in progress (Ref 2) by introducing a rounded building configuration and ex-tending the range of Reynolds Number.

The built-up area study would be more representative of practical configurations, The EBRII complex would be ideal as a prototype built-up area for gas released from sources located on buildings or at various stack elevations.

1

I

~

g I

'I 1

0 1

'C

RIPERENCES 1.

Halitsky, J.,

1963:

Gas Diffusion Near Buildin s.

N.Y.U. Dep't.

of Met. and Ocean.,

Geophys; Sci. Lab. Rept. No. 63-3.

2.

U.S.P.H.S.

Research Grant No. AP-59(R1) to James Halitsky at New York University, 1962:

Effect of Turbulence on Gas Diffusion Near Obstacles.

3.

Kalinske, A. A., Jensen~

R. A. and Schadt~

C. F1945:

Mind Tunnel Studies of Gas Diffusion in a T ical Ja anese r an D strict,

OSRD, NDRC, Div, 10 Informal Rep.

No.

~s*-

4.

Kalinske~ A. A., Jensen,R.

A. and Schadt, 1945:

Correlation of

'Mind Tunnel Studies with Field Measurements of Gas Diffusion.

OSRD, NDRC, Div, 10 informal Rep.

No. 10.3A-a, available from Library of Congress, Mash.,

D. C.

5.

Jensen, M. 1958:

The Model-Law for Phenomena in Natural Mind, Xngenigren 2, ansk ngeni rforening.

Denmark.

pp 121-128 1

6.

Field, J.

H. and Warden, R1933:

A Survey of the Air Currents in the Ba of Gibraltar 1929-30.

Air Ministry~ Geophys.

Mem.

o, 9, London, 7.

Pasquill, F.~ 1962:

Atmos heric Diffusion.

D. Van Nostrand Co,~

London, 297 pp.

t

~

i a

1 0

0

TABLE 1.

SUMMARY

OF SHELL RELEASE TESTS Fig,No, of Tes Results Config-uration Model urf.Temp.

iff.

F Gas Release Loc.

XD YD Re'f.Vel.

ft sec 3'lind Profile Direc-tio 10 12 13 Shell Alone 0

0 0

0 1

0 1

0 0

-1 0

-1 o.o7 5.54 o.'78 1.22 o.78 0,07 t

0. 07

. o.78 Logo 0

0 0.07 5.54 Un'rm 16 17 18 19 20 0

0 0

+1 0

1.22 0.07 1.22 10 Uniform 21 22 23 26 27 28 Shell and Auxilia Buildings

+0

+300 0

0

-1 0

0 0

1 0

1 0

1.22 1.70 o.o7 5.54 1.22 o.v8 0.07 0.07

Log, Logo SN D ~ Reactor shell diameter

= 80 ft WOJ t'ai I ~

x

TABLE 2,

SUMMARY

OF STACK RELEASE TESTS Fig.No.

of Test Results 29 ac Height eac or Location none Configuration Wind ro e

e.

e.

ft/sec, Log e 30 31 32 33 34 35 36 1.5H I

1H to side lH downwind 1H upwind none 1H to side 1H downwind 1H upwind 5.54 Logo H

Reactor shell height

~ 98 ft TABLE 3.

SHELL SURFACE TEMPERATURE EXCESS ABOK TUNNEL AMBIENT TEMPERATURE Tem erature Excess F

Wind Location Dome upwind Dome downwind Cylinder upwind Cylinder downwind Average Test Fig 22 185 188 119 115 152 Test Fig 23 373 371 241 3O4 direction dome.

ylinder 0 ft

)

~

1 0

Fig.

I Photograph of NRTS Complex Dimensions are distances in feet

s I

4'

gfti Fig.

I Photograph of NRTS Complex Dimensions are distances in feet

t I

looking downwind 94ewpy looking upwind I

l g

11I<~

} %

Fig. 2 Photographs of Model of NRTS Complex ln the Wind Tunnel

~

g 1

looking downwind

'j'q-N

'('lpfI(oq<<";

looking upwind n

i~M~

Fig.

2 Photographs of Model of NRTS Complex In the WInd Tunnel

E L

)

k 1

j

-700'M'400>

300'00

-l00' l00'00'00'00'00'00'00'00'00'00'W WIND

//

///

/

/'

/////

/////'

//

/

REACT

!0 I

I I

I TUNNEL WALL 4D 3D 2D ID 0

-ID

-2D

-200'E

/

TUNNEL WALL~

-3D

-4D

-8D

-7D

-6D

-SP

-4D

-3D.

-2D

-ID 0

ID 2D 3D 4D 5D 6D 7D 8D Fig. 3 Arran0ement of NRTS Suildinas In the Tunnel in a SW Wind D > Reactor shell diameter ~ 80 ft.

1

~

I I

l l

1 0

800'00'00'00'00';

Notes Distances are ln full scale feet

---

Model

Actual I

I I

IL I

I I

I I

I I

I I

I L

I r

I I

I I

I I

I I

I I

I I

I J

300'00'00'00'00'00'00'00'00'00'00'00' 0

Rg. 4 Comparison of Model and Actual Bulldlng ArrangenIents at NRTS

yl I

\\

, ~

~

s TUNNEL CEILING WIND 300 s

s

~

250

-X 40' 0 SD I

ORIGIN I

I I

Zfr ft.

I60 Referen ce Velocity Uf ~ l6.6 fps GAS RELEASE POINTS ISO

-Y top mid-hefght bottom 2.05 InlZI/O.0492)

IOO 40 9

I 56 I ORIGIN 0.7ID 0.78D I.22D

0. 07D 50 IO TUNNEL FLOOR 20 40'.5D vp fps Fly. 5 ESR II Model Dlmenslons and Wind Velocity Profile Dlmenslons are In equivalent full scale feet D < Reactor shell dfameter < 80 ft.

I I

\\

I

c

~

WIND 3

A

~II

) <<c I g~ t'r gC hg

<<E4 "

E

. I t

18' Fig. 6 Photographic Sequence Showing Diffusion of Smoke Released. at Various Points ln the Longitudinal Centerplane through the Reactor Shell

l

~

J 1

I I

'h

~

1 Wl SD h <,

~ IC

~<

'lr I

I

)

I )1 C

)

', $>xQ~ '

<!'1<

i'I I

II<)

I III

+I P

~tk~'<

I I7'M

~L ~ ~

l~B I

I<

4

'Pl lP~W

'rP AI<

!<,'t)'I4

'I EH Fig.

6 Photographic Sequence Showing Diffusion of Smoke Released at Various Points ln the Longitudinal Centerplane through the Reactor Shell

~

~

I,<

~

I,,<I~

~ <,+)<< I, k4 r

4'

~

~

7 I

t I'

Wl ND IO l2 I

I O'- M~2 r

t

. ~

. Y DEKE IJ V ptlp l3 l4 l6 I

J 11 I A

I hatt" n

'o 4K+I I

t'.I I Fig. 6 ( cont. )

Photographic Sequence Showing Diffusion of Smoke Released at Various Points in the Longitudinal Centerplane through the Reactor Shel I

~

I II L

I

, ~

'I t

~

Wl ND l0 "BÃ~

!2 K(g/i) l3 l4 I6 I

I I

I N) 'I I

Fig. 6 ( cont. )

Photographic Sequence Showing Diffusion of Smoke Released at Various Points In the Longitudinal Centerplane through the Reactor Shell

~

~

~

~

~

WIND with tripper without tripper I

y h) g I' APE~

~'

tq,.

at m-L I n

li, f',

Fig. 7 Photographs of Smoke Diffusion Downwind of Reactor Shell With and Without Boundary Layer Control

~

~

~

f 4

0 0

with tripper without tripper

,k i,

~

~ k

~ P)k tgkkvg (r v ~..W N"4"-

ink' "gg i

i k,'ig.

7 Photographs of Smoke Diffusion Downwind of Reactor Shell With and Without Boundary Layer Control I

~

~ k k

~ v I

~

X,'( v

~

v '.-

ik, i,Pii iv t. ~

~..

v V"

t

'I

~

f 0

I.C Lonyltvdlnol Cent ~rplone S ~ ctlon 1.4 12 I.O 0.8 2.0 0

r R ~ Seat ~

-Polit

~ WD Grovnd Svrlac ~

I8 Y/0 I.2 I.O OB ac 0.4 0.2 R ~ leot ~

Paint 0

3 2I/O I4 y

Iron ~ v ~ r ~ ~

5 ~ C II at X/0 1.25 14 o

Ir o n r v ~ I ~ ~ I ~ c I I0 II at X/D 2.5 12 Z/D I.2 2, D10 08 OII Od 04 re 02 0 0 02, 0.4 06 08 10 Y/D 1.4 I.C I 8 I.O 00 o2 04 OC OB 10 I'L I4 I.C 18

~ Y/D LO ZID QB Oeveloped Skell Svrl ~ c ~

06 R ~ lear ~

Point 0'5'OO IO ISO'80 sty, 5 K -4op4<IC lor Got Retraced ttvovyl Reactor Stall Cordlyvratlon t Releor ~ polre e

<Nlnd t

~tall ~Iona hot tera vpvlrd Ioy profile, 5.51 Ipr AIR',>,<v

~

~

0

rl I

~

I I

1 1.4 lasaltvdlsol C ~ *I~rplose 5 ~ ctlos 1.2 i/0 I.o as Relear n Point aC 04 0.2 0

3 XD~

Graved 5vrf~ co 1,4 Y/o l.2 lo 08 aa 0.2 2 ~ I ~ ~ < ~

Pa tnt Z/0 1,2 I

0 Iran<var< ~

5 ~ ctlon Jat X/D I ~ 25 1.4 1:2 Zio

. ID

~tXD 2

5 Iran<ver< ~

~ ctlon 04 QZ OZ O

OZ O4 OC 08 10 IZ.

H IC I.b 222 Y/o o

az o4 oG o.s l,o Iz I4 I.c I.s Lo '

'(/D 1.2 10 Z/0 IO Developed S<>>ll Seri ~ c ~

oa 04 2 ~ I ~ o ~~"

pa IIII 0.2 Oe ISS'80 a'a, t

X l<oplette Iar Ge< tele<<ed

<Ivevat< Reactor Sl>>ll Cerdlp>>ellen r kale<no point t Wlrd

~I>>II ~lese

<<Md>>ly<d vpwtsd lap fr<Ill~, 5.54 fte

~

~

I,c 14

~ Laayftvdleel Coat ~ rpleeo 5 ~ etio*

1.2 kola at ~

Pol et IO l5 o.c 04 a2 0 0 X/D 4 18 Graved 5v ~ fec ~

I.C 1.4 Y/42 I,2 1,0 0.8 ac Og I.$

0.2 kola ~ t ~

pcIAI X/D~

I.c Ir ~ rrr v ~ rr ~

$ ~ c IIa e at X/G I.25 I/v 1.4

~tX/0 2$

2 ~ erv ~ ~ I ~

5 ~ ~ flee 1.2 z/o 10 IO I.2 z/o ID 04 04 0,4 aR 00 02 04 OC OS ID 11 l4 IG Ib ZO Y/D 0

o1 O.4 O.C 0.8 ID 12 14 I.C 15 20

Y/o 12 1.0

'>Ob 2 ~ I~ et ~ politi 20 20.

Oovalopod St>>ll Svrlec ~

06 0.4 0.2 0

Oo 90 135 160

e Ffrf. IO X, lropfotte For Ger Roloarod ttravrttr lractor 5f>>II Cordtyvratfoe t Rofocao polot t INfrrd r

tf>>ll alar>>

top lop praflie, 5.54 Fte

1 I

~

1

~

1 I

'I c

~

I.C 1.4 Loeplsvdtoel C ~ OI ~ IPI ~ rl~

5 ~ ctloll IX Z/II IO tp 08 Rel ~ ~ s ~

Polet OC a4 02, ar 0 2.0 S

X/0 4'roved Svrlec ~ I v/b I,X

. 0 0 2 ~ I ~ ~ I ~

Pplot

)</D 1Io o ~ v ~ II ~

5 ~ c \\ I0 rs at X

D w I 25 14 5Io III v ~ I~ ~

5 ~ ~ II0 II

~ t X/D 2 ~ 5 I.O Z/0 a

06 Zjb la OC 04 0.4 0.2 Ot OO 02.

O4 06 08 IO 5'X 14 I C Id 2O Y/O O 0 m

a%

0.6 Ob 10 I L 14 IC IIS dt Y/D 10-

<~an IOO 2ct 5

~ kelees ~

Dee ~ loped 5sell Svrlec ~

04 155 150' e

Sty. Il X

lsoptette lor O<<r tel<<oed steovctr peoctor 55ell Grrdtpvrottoo s

Pet<<oo potre I Wled

~

I'tell ~ loco

<<tdketptrt dewewtrd lop proftl~, 5.5I Ito

~

~

l

~

Losyltvdts ~ I C ~<< ~ rpl ~ s ~

Sec tlos

).4 I.O 05 0,4 02 l~ I ~ ~ I ~

PQIAI rp sp 0CI 4 ~ x/D ZO Grove y Svrloc ~

).0 05 04 ro 0.2 o

l~ I~ ~ I ~

Io

~ols 4 ~xp

).4 reste ~ rr ~

5 ~ ~ Iles

~ t X/D <<I ~

S

'ir~ sse ~II~

S ~ c Itos

~ I X/D ~

5 5

I,'K Z/)5 IO

)2 Qu OB O.C CI4 O.C Og 00 02 04 OZ ISB

)0

)2

)4 Il

).8 20 Y/D 00 a2 a4 aa OB

)ct n 54 IC fSS io Yrtr o

5 Doeeloperl Swell 5<<rfoc ~

06 04 az l~ le ~ I ~

Polo~

I55'80' e

fty. Il K

Iseplesto lor Ges 4loessf tteovytr 4<<csee St<<II Co<<fly<<sotto<<'I 4lome petro s

INtref stroll close Sorrsrs 4<<s<<lsf loy profile, 5,5l fp

I 1

~

C

IC See ~Itell

~ I C ~ ~ I ~ e ~ lee ~

ecol ~

IO Ob 0.4 02 0 0

~ ~ I ~ ~ I~

lelol le 0 ~ Sech Se ~I~I~

l2 IO 0.4 02

-at d4 2 ~ Iooo ~

~Iel

-Of

-OB IO

- I4

- I.B

~ ~

~ e ~ I ~ ~ I~ ~ II~ e I ~ 25 I h D

IO 2/0 1,4 le ~ *~ ~ one I~ ~ ~ I ~

~ h/D ~ 2. I 1,2 IO O.S OO

~ I ~ol~

2 ~lel 0

I

.I4

~ St

~lo Nb hIC.Of C21 O

CSt O4 CSf Ob IIS lt v/r I2 2/f2 IO I ~ I ~ ~I

~Oleo b

IO H4 Wt 80 Ob WC noh nDW 0 DT 04 Ofo Ob IO IT l4 14 Y/0 e

Doe ~ l~I~ 2 Soo'2 Seel ~ I~

S ~

IS fe 01

~ ~ I ~I~ ~

~ oleo I 5' bo 170'fb 6l~

1 CO

Slo, SS h

SCOIoho lee Gee Select heehh Seeeeee Deh Cwaevehee e

~eleeee Ieee I I

~leh ~loco

~eeoc IISC Ieh lnfhee S.SI llo

4

~

'I

~

~

0 I

L

Ie*~ lsvdleel C ~ eiei ~ I ~ oi 5 ~ c ~ lee l4 I '2 2JO I ~ I ~ ~ se, ref*I OC 04 02 0

9 X/D Gite

~ 5vsltce l4 I2 04 5'2 Ot 04 5 ~ Icos ~

~ Iel 12 3

X/0 2,4 ZI525 Is ~ 1 ~ v ~ ~ ~ ~

5 ~ ~ sl ~ e

~I X/D I.

S I4 2P It lives

~ is ~ 5ecslee

~ i X/D S.S 0.6 pl dC

~ ~I I ~le

~i 0/5 0,2 IO Mb OC -04 &2 0 Ol O.

OC 05 I.O sic/II~

OA 0.2

~

IC.I4.It.IO,OD ~C.OI ~g 0 Gt 04 dCi o ID It l4 v/0 IO 4 ~ v ~ (eyed 54 II Svil~ c ~

I~

e ~ i ~

~ lei 0,4 02 0 0 155'SO'25'70

~

5 I 5'D t

IIS, Id X

lsttltdo I Goe I loosed dvevocs Seecsei Scell leleele soles c

Wed s sell

~lees

~cd4+i olde Ito SceAles S.SC llc

A C

~

s I

I,C f le. I5 Gos reteose points bottom vpnind I,O O.B D,C 0 ~ leos ~

oint 2

I 5

fic, II Gos releor ~ anoints top I2, LO R ~ I ~ os ~

Point 0,5 0.4 Q2 lk z/u I,2 floi tf Gos relents points bottenn downwtnd 0,'X Release Point x/c fttts, IS lo If K

Isoptette for Ges Relented tteovstr keoctor Shell Contlgwotlon s

%eicos ~ point s Wind Sweets s

~toll clone sorlobte unllorrn prosit ~, S.SC lps lonotNdtnol centerplone

0 V

~

0 j

t W

s s

,s t

s

~

IA 7/P

/.i 2 ~ leos ~

point 2$

0 4

2 Fla.

IO U~ e 2FpS 0(

x/g (4

z/u l.2 FIO.

I U~ e IOfta 2oleos ~

Point t$ 8 O.L 0 0 X/D I.C Zjo i.4 Ffg. 20 V

e ISfps I.O 0.8 0 ~ I oat ~

Point 2$

IO 0.4 0.1 O 0 Figs.

IS to 20 K

lsaptetlo fa Gos Relesaerf ttvavyh peectar Shell x/z Cantlyvratton I Release point t Wing s

Surveys s

shell alone tap unlfarrn profll~, $ r IO, ond I$ fps lanpftudlnal conterptene

~'

I 0

N

/

s

~

s s

ftyc 21 Svrfoce Te<<p. Ditf. <<OPf I.O

~ leos ~

Point 0S 0.'K 0

iS lf 7./p, l.O 0.8 X.02

$.80 5I I2.5 X.20 05

'<<Sc.ly Release Point ffy. 22 Svrface Temp. Diff. <<+150 f fiy. 23 Svrfoco Temp. Diff. <<<<3 f

/.0 TTS

.0$

.Id 0 ~ leos ~,IO Point

'O flys. 2I to 23 K - froptettolor Gas ysfeored tieovytr yeoctor 54II Conf to<<ration Release point s

Wind s

Svrf. te<<p. diff.s Serrate s

~Tall alone top Io1 proAle, I.P lps s0 f, sl50 fr s3 KPf lonyltvdlnol co<<terpin<<a

I

at ground level X

at roof leve l Wind Release Point I

.6

.8 l

lhx 5

.og.

x +$'

SOdtufft Boiler

-8 4

l ower Plant lo 5

I Puef Cyclt l.t x

.8 l

I Fig. 24 K - hopleths for Gas Released through Reactor Shell Configuration:

Release point:

Wind entire complex bottom upwfrd SW, tog profile, 5.54 fps

I J

~

~

I

l I

I I

Release Point 0W

.Os go

.I 3.5

~ 7 L

a t ground l eve I X

at roof level 0

0 Fig. 25 K - lsopleths for Gos Released through Reactor Shell Configuration s

Release point:

Wind entire complex top SW, log profile, 5.54 fps

~

~

at ground level

)C at roof level Release Point

./ 0

. 0 I Wind

.0 Ffg. 2d K

Isopfeth for Gas Released throvgh Reactor Shell Conffgvrotfon t Release point r Wind entire complex midnight downwind SW, log profile, 5.54 fps

I s

r I

0

Release Point I

I I

I at ground level at roof level Ql 2

0 l0 Wind I

I I

I I

I Fig. 27 K -.isopfeths for Gas Released through Reactor Shell Configuration:

Release point s

Wind entire complex bottom downwind SW, log profile, 5.54 fps

C

~

4 I

f 4

at ground I eye l X

at roof level Release Point l.5 Wind

~3X 8

X IO X

5 10 l.5

.5 l.5

.9

.7

.6

.5 I

I I

I O I

.2 Fig. 28 K - lsopleths for Gas Released through Reactor Shell Configuration Release point Wind entire complex bottom upwind

SSW, log profile, 5.54 fps

~

~

0 t.

l0 Rg. 29 Stack atone

.Ol lo i5 y/H Flg. 30 Reactor IH to Side of Stock

.Ol lo i5 lO Flg. 31 Reactor IH Downwind of Stack

,5,6 7

lO i5

)C/g Flg, 32 Reactor IH Upwind of Stack I.O 8

.7

.8

,5

,4 IO ts xQ Figs. 29 to 32 K - Isoplelhs for Gas Released through 98 ft. Stack Conflgvratlon Release poInt Wfnd Or!gin of coord.

r H > Reactor height ~ 98 stack with and without reactor shell top of stack log profile, 5.54 fps base of stack ft.

l

~

~

I 8

@Il

i.O Fig. 33 Stock Alone o.r

(.o IO 2/H Fig. 34 Reactor IH to the Side of Stack ZO (0

Fig. 35 Reactor IH Downwind of Stack

.Ol

.I5 IO i5 X/H

~ ZO I.O Fig. 36 Reactor IH Upwind of Stack

.Ol

.05

.I

. I5 5

jo I5 X/8 Figs. 33 to 36 K - Isopleths for Gas Released through 147 ft. Stack Configuration Release poInt Wind Origin of coord.s H 1 Reactor height ~

stack with and without reactor shell top of stack log profile 5.54 fps base of stack 98 ft.

I

~

r

~

0

100 90 8o 70 80~e S

5o 4o 30 20 lo-C 12-6 9

C D

K-isopleth 0

< lo 9

r 8

K 20 (S/D) 3 O,l Oi2 0~3 0,4 0,6 0,8 1

2 3

4 5

6 Non-dimensional distance from shell surface, S/D Fig. 37 Maximum Value af K vs Distance from the Shell

I I

100 90 80 50 5.54 f /sec 40 30 20 10 ft/sec+

K 20 (2/o) 10 9

~

8 d

7 ID

~P 4 35 f /se 3ft/s c 0,1 0,2 0,3 0,4 0,6 0,8 1

3 4

5 6

Non-dimensional distance from shell surface, S/D Fig. 38 Dependence of Maximum Value of K on Wind Velocity

~

iP.V I

~I