CONTENTS

INTRODUCTION...............................................1

CURRENT BASIS OF LEV DESIGN................................6

LIMITATIONS OF CAPTURE VELOCITY DESIGN....................15

THE CAPTURE EFFICIENCY CONCEPT............................18

CURRENT DESIGN PROCEDURE VERSUS BREATHING ZONECONCENTRATION........................................20

VORTEX SHEDDING___. .___...___..........................25

IMPORTANCE OF REVERSE FLOW PHENOMENON IN WORKER EXPOSURE..29

A SIMPLE MODEL ADDRESSING REVERSE FLOW....................32

OBJECTIVE AND PURPOSE................___.................38

METHOD OF MODEL EVALUATION................................40

Wind Tunnel Description..............................40Velocity Determination...............................41Test Object Description..............................48Sulfur Hexafluoride Generation.......................50Determination of Zone Depth: Visualization of

Test Smoke......................................51

Determination of Zone Depth: Concentration VersusDistance Curves.................................53

Determination of Zone Depth: Cbz = 0.5Co............74Determination of Zone Depth: Calculation from

Theory..........................................74

MANNEQUIN VERSUS MANNEQUIN 90 DEGREES.....................80

DISCUSSION OF MODEL EVALUATION............................87Discussion of Test Smoke Observation.................87Discussion of Actual and Theoretical Concentration

Versus Distance Curves..........................88

Discussion of Mannequin Versus Mannequin 90 Degrees..91Discussion of the Model's Ability to Predict De......92

11

Discussion of the Model's Ability to PredictMixing Zone Concentrations......................93

CONCLUSIONS..............................................112Test Smoke Observation..............................112Theoretical Model...................................112Mannequin Versus Mannequin 90 Degrees...............118

RECOMMENDATIONS..........................................119Validation of Model.................................119Effects of Hands and Arms...........................119Turbulent Diffusion Effects.........................120Study of Concentration Decrease as a Function of

Distance.......................................121

Comments Regarding Non-Uniform Flow.................136

REFERENCES...............................................143

111

INTRODUCTION

One of the most serious threats to employee health is

the inhalation of toxic airborne materials produced by

various industrial processes. Once inhaled, these

contaminants can give rise to a myriad of deleterious health

effects ranging from a simple nuisance to vital organ damage

or neoplasm. Occupational health professionals,

particularly industrial hygienists, constantly seek ways to

eliminate these exposures or at least to control them to

harmless levels.

As illustrated in figure (1), a variety of control

measures are available to assist the industrial hygienist in

achieving the goal of reduced exposure. Certain of these

methods are more effective than others. Typically,

engineering control measures such as substitution or

isolation that do not require active employee participation

are more successful than those such as personal protective

equipment (respirators) which essentially shift much of the

responsibility for protection to the employee.

If substition, change of process, enclosure, or

isolation are not feasible, other engineering controls may

be necessary. In some situations, dilution ventilation may

be adequate to reduce worker exposure. Dilution ventilation

is simply reducing the concentration of contaminant in

workroom air by diluting it with uncontaminated air.

Dilution ventilation may be sufficient if the toxicity of

the contaminant is low, the worker is far enough away from

FIGURE 7

GENERALIZED DIAGRAM OF METHODS OF CONTROL

.4------SOURCE

--------- AIR PATH

4—----------------------------------__ ^-RECEIVER

--------------------------------------^

P^ rDipTank

}'

1

/1. HOUSEKEEPING

^'v-^x/ ^^1.SU BSTITUTION VVITH A TRAINING & EDUCATIONLESS HARMFUL MATERIAL (IMMEDIATE CLEANUP) (MOST IMPORTANT)(WATER IN PLACE OFORGANIC SOLVENT) 2. GENERAL EXHAUST 2 ROTATION OF WORKERS

VENTILATION (SPLIT UP DOSE)2. CHANGE OF PROCESS (ROOF FANS)(AIRLESS PAINT SPRAYING) 3 ENCLOSURE OF WORKER

3. DILUTION VENTILATION (AIR CONDITIONED3. ENCLOSURE OF PROCESS (SUPPLIED AIR) CRANE CABS)(GLOVEBOX)

4. INCREASE DISTANCE 4 PERSONAL MONITORING4. ISOLATION OF PROCESS BETWEEN SOURCE AND DEVICES (DOSIMETERS)

(SPACE OR TIME) RECEIVER (SEMI-AUTOMATICOR REMOTE CONTROL) 5 PERSONAL PROTECTIVE

5. WET METHODS DEVICES (RESPIRATORS)(HYDRO BLAST) , 5. CONTINUOUS AREA

MONITORING (PRESET 6 ADEQUATE MAINTENANCE6. LOCAL EXHAUST ALARMS) PROGRAMVENTILATION

(CAPTURE AT SOURCE) 6. ADEQUATE MAINTENANCEPROGRAM

7. ADEQUATE MAINTENANCEPROGRAM

SOURCE: REFERENCE ?

the point of contaminant evolution, and the quantity of

contaminant is low and uniformly released [2].

However, more often, in an industrial environment, it

is desirable to remove a contaminant as close to its source

as possible, before it has a chance to escape into the

general workroom air. This is particularly true for the

more toxic materials. For this reason, local exhaust

ventilation is often employed as a viable engineering

solution to this problem.

Local exhaust ventilation is a means of inducing air

movement to capture and remove the contaminant at or near

its source. The basic elements comprising a local exhaust

ventilation (LEV) system are well documented in standard

industrial hygiene literature and are illustrated in figure

(2). These elements are a hood (or hoods), ductwork, an air

cleaning device (if necessary), and a fan to induce air

movement [1,2,3,4].

The hood is perhaps the most important part of the

system since it is the opening into the system through which

the contamimant flows after being entrained by the air

currents [3]. The hood should enclose as much of the

process as possible. If complete enclosure is not feasible,

the hood should be as close to the source as possible and

shaped and positioned so as to make use of any

directionality of contaminant release imparted to it by theprocess.

FIGURE 2

Stack

Hopper

Barrel fillingoperation

^Air cleaner

SCOURGE; REFEREWCE 3

The ductwork is the piping system through which the

contaminant-laden air flows. Its design and construction

are determined by many factors such as the type of materialconveyed, temperature, and plant layout, for example.

The function of the air cleaning device is to removethe contaminant from the air stream before it is exhausted

to the outside environment. Many different types of

cleaners exist and proper selection depends on concentrationand particle size of contaminant, degree of collectionrequired, characteristics of the air or gas stream,

characteristics of the contaminant, energy requirements, andmethod of dust disposal [2].

The fan provides the means of inducing air flow by

creating a pressure differential between the atmosphere and

inside of the system. The magnitude of this pressure

difference determines the quantity of air entering thesystem. At the end of the LEV design procedure, a specificfan is selected that will move the required amount of airagainst the necessary pressure differential.

LEV has been utilized in industry since early in the

twentieth century. However, the basic parameters used todesign these systems have changed very little over the lastfifty years. Presently, new concepts are being exploredthat may significantly improve the current state of LEVdesign with the ultimate goal of providing the best possibleprotection for the employee at the lowest possible cost.

•

CURRENT BASIS OF LEV DESIGN

The sizes, shapes, and configurations of LEV systems

are almost as varied as the industrial processes they are

designed to control. Round openings, rectangular openings,

slotted openings, booths, and cabinets, to name a few, all

have their applications to various situations. However, all

LEV systems have one particular design parameter in common;

capture velocity. Capture velocity is defined as the "air

velocity at any point in front of the hood or at the hood

opening necessary to overcome opposing air currents and to

capture the contaminated air at that point by causing it to

flow into the hood." [2]. The idea is that if you move

enough air into the hood you will also "capture" the

contaminant as well.

Historically, capture velocity has been the basis for

calculating the required air flow into local exhaust hoods.

The system designer must somewhat subjectively determine the

capture velocity necessary to entrain the contaminants given

off by the particular process he or she wishes to control.

Guidelines such as table (1) are available to aid in this

determination. In practice, the selection of hood

configuration and air flow is normally made by consulting

the ACGIH Ventilation Manual [2] for a "tried and true" VS

print that approximates the desired application. McDermott

concludes "The best way to determine the needed capture

velocity and airflow is to seek out similar equipment and

operating conditions at other plants or else build a few

TABLE /

Condition of Dispersionof Contaminant Examples Capture Velocity, fpmReleased with practically no

velocity into quiet air. Evaporation from tanks; degreasing,etc.

50-100

Released at low velocity intomoderately still air. Spray booths; intermittent container

filling; low speed conveyor transfers;welding; plating; pickling

100-200

Active generation into zone ofrapid air motion Spray painting in shallow booths;

barrel filling; conveyor loading;crushers

200-500

Released at high initial velocityinto zone of very rapid air motion. Grinding; abrasive blasting, tumbling 500-2000

In each category above, a range of capture velocity is shown. The proper choice of values depends onseveral factors:

Lower End of Range Upper End of Range1. Room air currents minimal or favorable to capture. 1. Disturbing room air currents.2. Contaminants of low toxicity or of nuisance value 2. Contaminants of high toxicity,only.

3. Intermittent, low production. 3. High production, heavy use.4. Large hood—large air mass in motion. , 4. Small hood—local control only.

SOURCE: REFERENCE Z

8

hoods (even out of cardboard) and test their effectiveness

at different airflows." [3].

Once a suitable capture velocity has been selected, the

volumetric rate of air flow to achieve it must be

calculated. The question is, "How much air, in cubic feet

per minute (cfm), must be moved into the hood to obtain the

desired capture velocity at a given distance in front of the

hood?".

Empirically determined equations for making these

calculations have appeared in literature since J.M. Dalla

Valle's [5] work in the 1930's. Working with a special

pitot tube, he mapped velocity contours for plain and

flanged round and rectangular ducts. He concluded that, for

the purpose of LEV design, the centerline velocity is the

most practical parameter. The following equations,

published in the most recent edition of the Industrial

Ventilation Manual [2], are simplified forms of Dalla

Valle's original equations:

(1) Q = V(10xH A) for plain rectangular ducts ofaspect ratios (width/length) of0.2 or greater, or round, and,

(2) Q = 0.75V(10x^+ A) for flanged rectangular ducts ofaspect ratios of 0.2 or greater,or round

where Q = quantity of air requiredto achieve the necessarycapture velocity (cfm)

V = centerline capture velocity(fpm)

X = distance from hood face to

point in ceterline (feet).

Several years after Dalla Valle, Silverman [6,7]

examined centerline velocities for flanged and unflanged

round and slotted openings. Although he considered his

equations for round openings to be an improvement on Dalla

Valle's work, they did not "catch on" with most ventilation

designers and have not been incorporated in the Industrial

Ventilation Manual [2]. However, simplified forms of his

equations for flanged and unflanged slots are presented in

the Manual as follows:

(3) Q = 3.7LVX for unflanged slots of aspectratios less than or equal to0.2, and,

(4) Q = 2.6LVX for flanged slots of aspectratios less than or equal to 0.2

where L = length of slot (feet)

V = centerline capture velocity (fpm)

X = distance from hood face to pointin ceterline where capture velocityis achieved (feet).

It is interesting to note that neither Dalla Valle nor

Silverman completely accounted for the effects of hood

aspect ratio or flanging on the centerline velocity

gradients. Silverman's equations cannot be used to

calculate velicities very close to the hood face because as

X approaches zero, V becomes indeterminate.

More recently, Fletcher [8] undertook to determine

empirical equations to calculate hood centerline velocities

as a function of volumetric flow rate, distance from hood

face along centerline, and aspect ratio. After examining

aspect ratios from 1:1 to 1:16 he concluded that the

10

centerline velocity was very much dependent on aspect ratio

and that any equation which neglected this relationship

could not be correct. His results indicate that at a given

distance, hood area, and flow rate, the centerline velocity

increases with increasing aspect ratio. Through various

curve fitting techniques he developed a rather unweildly

equation that he subsequently used to construct the nomogram

that appears as figure (3). Using this nomogram, the ratio

of centerline velocity to average hood face velocity can be

determined from the hood aspect ratio and the ratio of

centerline distance to square root of hood face area.

Fletcher also qualitatively studied how flanging

affects the centerline velocity in front of local exhaust

hoods [9]. His results demonstrate that the addition of a

flange can produce a large increase in the velocity at a

given point in front of the hood. He reported that the

optimum flange width is equal to the square root of the hood

face area. He also states that the effect of the flange

becomes increasingly more pronounced as the aspect ratio

decreases so that slot type hoods show the most benefit from

flanging. However, he offered no empirical equations to

calculate the observed effects of flanging.

Fletcher and Johnson [10] looked at the effect of an

adjacent plane on the velocity profiles around a hood. They

concluded that for a given hood, flow rate, and distance, a

much higher velocity can be produced in front of a hood on a

plane than with the same hood freely suspended.

n

FIGURE 3

TOO.

0 50.

010 J

0 05 J

O01_J

0005J

,0-05

^005

LOlO

wL

L0 50

,100

SOURCE: REFEREWCE S '

^-^-"'gyj«P^>i» -'-.j'-'i?!».rf!W!?^'!!IBgHg^^^g^--

12

Additionally, to maintain the same velocity at a givendistance in front of the hood, less air is required when thehood rests on a plane than when it has no obstruction.

Within the last decade. Garrison [11] has evaluated thework of Dalla Valle and Silverman using much smaller inletsand higher velocities, i.e., high velocity/low volumesystems. Among other things, he concluded that theempirical equations published in the Ventilation Manual [2]from Dalla Valle's and Silverman's work were generallyappropriate. However, he disagreed with the flat 33 percent increase in centerline velocity velocity attributed toflanging the circular or rectangular inlets as isrecommended in the Manual. His data indicate that a more

realistic centerline velocity increase would be on the orderof 20 to 3 0 per cent. He also recommended that Silverman'sequations be restricted to centerline distance/hood diameter(or hood width) ratios of 0.4 or greater because, as wasmentioned previously, as x approaches 0, V becomesindeterminate.

In a later paper Garrison [12] provides the followingnon-dimensional equations that describe centerline velocitygradients in terms of distance, inlet end shape, andflanging:

(5) Y(near) = a(b) ^''^(6) Y(far) = a(Xdw)^

where Y = centerline velocity/hood face velocityXdw = centerline distance/hood diameter (or hood width

for rectangular hoods)

"''.- :^jf'- '- -J-

Empirical Design Data for Nondimensional Centeriine Velocity GradientsY ^ a (b)XDW Y = a(XDw )'

SpecificY Values

at Xdw =

Nozzle

EndNozzle

Profile

Shape

0 < Xdw < 0.5 0.5 < Xdw <1.0 1.0< Xdw S XdwShape

a b a b a b a b Xdw^^ 0.5 1.0

Plain 110 0.06 .,.. 8 -1.7 8 -1.7 1.5 26 8Circular Flanged

Flared110

90

0.07

0.20 90 0.20

10 -1.6 10

18

-1.6

-1.7

1.5

2.0

30 10

40 18Rounded 98 0.50 145 0.23 --

" 33 -2.2 2.5 69 33Square Plain 107 0.09 —

.- 10 -1.7 10 -1.7 1.5 32 10(WLR-1.0) Flanged 107 0.11 ---- 12 -1.6 12 -1.6 1.5 36 12

Rectangular Plain 107 0.14 —— 18 -1.2 18 -1.7 2.0 41 18(WLR=0.50) Flanged 107 0.17 ---- 21 -1.1 21 -1.6 2.0 45 21

Rectangular Plain 107 0.18 —— 23 -1.0 23 -1.5 2.5 46 23{WLR:0.25) Flanged 107 0.22 ---- 27 -0.9 27 -1.4 3.0 50 27Narrow slot Plain 107 0.19 —.. 24 -1.0 24 -1.2 3.5 48 24(WLR=0.10) Flanged 107 0.22 ---- 29 -0.8 29 -1.1 4.0 50 29

15

LIMITATIONS OF CAPTURE VELOCITY DESIGN

In the past few years several investigators have begun

to question the usefulness of capture velocity as the

important parameter in the LEV design process. Ellenbecker,

et al [14] point out inadequacies such as the inability toaccount for the effects of crossdrafts or other air

disturbances, the uncertainty involved in shaping the hood

and distributing face velocities for the most efficient

contaminant capture, and the difficulty in determining the

optimum air flow that gives the necessary contaminant

control for the lowest energy expenditure. The authors

point out that even when the current design method is

conscientiously applied, only a qualitative prediction of

the hood's ability to capture contaminants is provided.

Esmen, et al [15] comment that design based on one

dimensional centerline capture velocities cannot include

effects due to hot sources or impediments, nor can it be

used to determine optimally designed geometric hood shapes.

Flynn and Ellenbecker [16] cite the "trial and error"

nature of the capture velocity design methodology and the

lack of a specific technique for determining how much

capture velocity is needed for a particular process. In a

subsequent article [17] they illustrate the inadequacy of

the one dimensional centerline approach by pointing out the

case when the contaminant source is not right on the

centerline or when it is so large that "centerline velocity"

• ^TSS* '-. •'^: '^^^ RT"

16

will not adequately describe the flow field over much of thecontaminant generation area.

Roach [18] states that "...it is inadvisable to designthe ventilation of an exhaust hood so as merely to produce astandard capture velocity or standard entrance velocity, as

the velocity chosen may be excessively high or, what wouldbe worse, not high enough."

Fletcher and Johnson [19] demonstrated that the removal

of gases and submicron particles released at low velocitieson the centerline of LEV hoods can be predicted by thetraditional concept of capture velocity. However, when thedirection of contaminant release is away from the hood at

velocities greater that about 0.21 m/sec the capture

velocity concept is not valid in that a higher velocity isneeded at the source to entrain the contaminant than that

published in the Ventialtion Manual [2].

Heinson and Choi [20] list the following flaws in thecurrent design methods:

(1) Contaminant concentration in the vicinity of thesource cannot be predicted.

(2) The effect of changes in design (such as systemdimensions or volumetric flow rate) on the performance of asystem cannot be estimated.

(3) Even though the performance of a particular system

is known, the effect of geometrically scaling it up or downin unpredictable.

17

(4) An engineer designing a system for a new process

(one for which a LEV design does not appear in published

literature) is left to design basically from scratch with

little knowledge of the effectiveness of the resulting

system.

(5) The idea of providing a certain velocity to capture

contaminants is inconsistant with the laws of fluid

mechanics. Contaminants move because they are suspended in

a moving medium and thus, without its fluid medium, a

contaminant has no motion to be captured.

18

THE CAPTURE EFFICIENCY CONCEPT

In an effort to significantly improve the current

practice of LEV design, a new concept, capture efficiency,

has been introduced and studied. Capture efficiency has

been defined by Ellenbecker et al [14] as "the fraction of

the airborne contaminants generated by a source that is

captured by the LEV system controlling it". This can be

stated mathmatically by the following relationship:

(7) n = G'/G

where G == contaminant generation rate (g/s)

G' = LEV contaminant capture rate

The authors state that the capture efficiency is a

function of several variables; volumetric airflow through

the hood (Q), the hood face area (A), the centerline

separation between the hood and the source (X), the

crossdraft velocity (Vc), and the source temperature (T).

Neglecting temperature, it was shown that capture efficiency

is related to the functional group, g, which can be

described by dimensionless variables;

(8) g = (Vc/Vof (X/A)^where Vo is the average face velocity (Q/A) and a and b areexperimentally determined constants.

The authors report that preliminary measurements

suggest a form such as

(9) n = (1 + Kg) or n = e

where K is an experimentally determined constant.

19

Various laboratory and field experiments were conducted

to actually measure capture efficiency using an aerosol

generator and a light scattering photometer. The aerosol

source was placed inside the hood to give a 100 per cent

photometer reading. The source was then moved to the

desired point of contaminant generation (X) and a second

photometer reading was obtained. The ratio of the two

values gives the hood capture efficiency.

The authors expressed optimism that this and subsequent

work in this area would lead to future LEV design for

capture efficiency rather than capture velocity. Capture

efficiency being the more desirable parameter because it

relates directly to airborne concentration of contaminant.

A more rigorous theoretical development of capture

efficiency as it relates to a flanged circular hood is given

by Flynn and Ellenbecker [16]. A model was developed which

can predict capture efficiency for a flanged circular hood

acting on a point source of gaseous contaminant at low

strength. The presence of a cross draft perpendicular to

the hood centerline is also handled in the model.

20

CURRENT DESIGN PROCEDURE VERSUSBREATHING ZONE CONCENTRATION

Perhaps the most fundamental deficiency in designing an

LEV system to provide a specific capture velocity at a

certain point of contaminant generation is that it tells the

designer nothing quantitatively about how effective he or

she will be in achieving the overall goal of reducing the

concentration of contaminant in the employees breathing zone

to an acceptable level. (The breathing zone is defined as a

sphere of one foot radius from the worker's nose/mouth

area.) Even when the target capture velocity is achieved,

no method exists that can relate this velocity to the

breathing zone concentration. Similarly, a designer who

wishes to achieve a certain target concentration of

contaminant (for example, one half of the OSHA Permissable

Exposure Limit) has no means of quantitatively determining

the ventilation required to do so. One cannot say, for

example, that if a particular capture velocity is provided,

a certain level of protection is achieved.

The concept of capture efficiency seeks to alleviate

this inadequacy by quantifying the amount of contaminant

that escapes the hood. This certainly is a much more useful

way of determining the efficacy of a hood in removing

contaminants from a process. However, ultimately the

effectiveness of LEV should not be measured by how

efficiently it removes contaminant but rather by its ability

to protect the worker from exposure to contaminants.

21

Consider a worker in a spray paint booth. In a well

designed booth, virtually all of the contaminant is

eventually captured, giving a capture efficiency of 100 percent. However, most spray painters are required to wearrespiratory protection because a significant quantity ofcontaminant passes through their breathing zone before beingremoved.

Clearly, a method of LEV design that somehow relatesdesign parameters to breathing zone concentration would bemost useful in protecting employees. However, before such amodel can be developed, a fundamental interaction must beinvestigated; that of the worker with the flow field.

Previous analytical models describing flow fields intohoods [16,17,20] have used potential theory as thetheoretical basis. In potential flow, the assumptions aremade that the fluid is both incompressible (the volumeexpansion is negligible) and irrotational (negligible localangular velocity) [21]. These assumptions are valid in thefree field where no object is present to obstruct the flow.While these models have certain applications, instances

arise when the worker becomes a significant obstacle in thepath of air flowing into the hood.

An object (such as a person) in the flow field

questions the validity of the potential theory approach intwo ways. First, by its very presence the object acts as anobstacle, a physical obstruction to the flow of air into the

22

hood. As such it perturbs the boundary conditions for thesolution of Laplace's equation [22].

Secondly, and most important for our discussion, when

fluid flows past a blunt body, a boundary layer is formed on

the surface of the body. A portion of the fluid adheres tothe wall of the object and thus, near the wall, the motion

of a thin layer of the fluid is retarded by frictional

forces. Within this layer, fluid velocity increases from

zero (at the wall) to the velocity of the moving fluid

stream (external frictionless flow). This thin layer is

called the boundary layer [23].

As fluid approaches a blunt object, such as a circular

cylinder, the boundary layer is formed on the upstream side

as depicted in figure (4). If the flow is frictionless,

fluid particles are accelerated on the upstream side of the

cylinder and decelerated on the downstream side. Since

acceleration of a fluid across a surface reduces pressure

and deceleration increases pressure, the pressure on the

upstream side is decreased while downstream side pressure is

increased. As fluid moves around the cylinder, pressure is

transformed into kinetic energy on the upstream side and

then back into pressure on the downstream side. Outside the

boundary layer the flow is nearly frictionless while inside

large frictional forces exist due to the large velocitygradient across the layer.

Imagine a fluid particle in the boundary layer movingaround the cylinder adjacent to the wall. Because of the

23

FIGURE 4

Thin front \boundary layer

Outer stream grosslyperturbed by broad Howseparation and wake

Separation

(«)

Separation

FIGURE 5

Broad "^wake p

Narrowwake

(b)

SOURCE: REFEREWCE 2J

24

high frictional forces inside the layer, it uses up a large

portion of its kinetic energy circumventing the upstream

side of the cylinder. Not enough kinetic energy is left to

allow it to continue on its path around the cylinder into

the area of increasing pressure on the downstream side. It

eventually stops and, because of this increasing pressure

(adverse pressure gradient) begins moving in the opposite

direction (reverse motion). A vortex is formed which grows,

separates, and moves downstream. Separation occurs more

quickly in laminar flow that in turbulent flow as is

depicted in figure (5). The adverse pressure gradient on

the downstream side of the cylinder is more effective

against laminar flow. Turbulent flow is more resistant to

the adverse pressure and separates farther along the

downstream side. This results in a large wake for laminar

flows and a smaller wake for turbulent flows [21].

25

VORTEX SHEDDING

As vortices move away from the body, a regular,

alternating pattern of shedding is noted. This alternating

arrangement of shedding is called a Karman vortex street.

Schlichting [23] states that this well defined Karman street

breaks down into complete turbulent mixing at Reynolds

numbers (Re) of about 5000. Roshko [24] reports that the

stable and well defined vortex patterns downstream of a

cylinder occur only in the Re range of 40 to 150 and undergo

a transition to turbulence at Re from 150 to 300. However,

he states that the periodic shedding of these vortices

occurs at Re of up to 100,000 or more. Vortices shed at

these higher Re quickly break down into a turbulent wake.

The frequency with which these vortices are shed is

described by a dimensionless quantity called the Strouhalnumber:

(10) S = fD/V

Where f = frequency of vortex shedding (1/min)D = diameter of cylinder (feet)V = velocity of fluid stream (feet/min.)

As figure (6) shows, the Strouhal number remains constant at

about 0.21 for Re up to about 200,000.

The downstream velocity of the vortices appear to be

somewhat slower that that of the surrounding airstream.

Fage and Johansen [25] showed that for a cylinder the speed

with which the vortices pass downstream is about 80 per cent

of the undisturbed air relative to the cylinder. This speed

increases with increasing speed of the outer vortex

26

FIGURE 6.-^0

.210

.200

.190

= ,170

.4 60

z;jrI°--B2^'^2-4>f^^^^g^=#^^^'^^^^r'^-->^—---: 0.2,2 0-^)

Best-fit line

O'.cmo a0235Q .0362O .05134 .0800V .0989

.158

.318

.635Kovosznoy"Tail" mdcotes ihoi velocity |

WQS computed from shedding 1frequency of o second cylinder !

100 200 300 400 500 600 700 800Reynolds number, /?

900 1.000 1,100 1,200 1.300

SOURCE: REFEREMCE 24

27

boundary- The authors also showed that the ratio of

longitudinal spacing between vortices to the diameter of thecylinder is about 4.27.

Bloor [26] demonstrated a stable range of vortexformation at Re below 200. That is, in this range, the flowis laminar everywhere. In the Re range of 200 - 400, thewake begins to disintegrate to turbulence. The onset ofwake turbulence moves closer towards the cylinder as Reincreases. At Re greater than 400 the separated boundarylayer becomes turbulent even before it rolls up into avortex. Thus, the vortices are turbulent upon fomnation.However, at Re between 400 and 1300 the point of transitionof turbulence remains constant relative to the cylinder.Finally, at Re of about 1300, the length of the laminar flowregion begins to decrease again until, at Re of about50,000, it is almost to the shoulder of the cylinder. Thepoint at which turbulent motion reaches the separation pointof the boundary occurs at Re of about 300,000. This pointis called the critical Reynolds number. However, a definiteshedding frequency is still observed, even at the criticalReynolds number.

Bearman [27] demonstrated regular vortex shedding at Reup to 550,000. However, at a Re of 300,000, the sheddingfrequency as described by the Strouhal number showed a sharpincrease from its relatively constant value of 0.21. TheStrouhal number leveled off to about 0.46 for Re greaterthat 400,000. The author points out that any small change

28

in the surface smoothness of the cylinder can significantlydisrupt the separation causing fluctuations in the sheddingfrequency.

Achenbach and Heinke [28] also noted the sharp increasein shedding frequency at Re of 300,000. This increasebecomes less prominent as cylinder surface roughnessincreases.

29

IMPORTANCE OF REVERSE FLOW PHENOMENON IN WORKER EXPOSURE

The practical importance of this zone of reverse flow

can be readily seen when one considers an employee working

in a typical position relative to LEV. As was mentioned

previously, employees are normally instructed to position

the work between themselves and the source of local exhaust.

In this orientation, the worker becomes the blunt body and

boundary layer separation occurs as the air flows past.

Thus this zone of reverse flow or turbulent mixing occurs

immediately downstream of the worker. If the source of

contaminant is located within this zone, it may actually be

drawn back toward the worker giving rise to significant

concentrations of contaminant in the breathing zone. Note

that this may occur even when target capture velocity is

achieved or when hood capture efficiency is 100 percent.

The effect of reverse flow on breathing zone

concentrations was previously studied by Ljungqvist [29]

using a smoke diffuser. The diffuser was placed in a

uniform air flow of approximately 50 fpm. With no

obstruction in the flow, the smoke moved directly towards

the LEV source. However, when a test person was placed

between the diffuser and the source of air flow, the smoke

was clearly directed back towards the person's breathing

zone. Ljungqvist attributes this phenomenon to the

stationary wake produced by the person in the air flow. He

states that individuals in the flow field create two kinds

of vortices; the wake caused by the body itself and that

30

arising from movements of the body. He concludes thateither of these two wake structures can completely destroythe intended beneficial effect of a LEV and that no

consideration appears to be given to this problem instandard ventilation design.

In studying push-pull ventilation systems, Hampl and

Hughes [30] also demonstrated the effect of a person in theflow field of a ventilation system. They observed thecollection of smoke by a standard LEV hood with variousorientations of air jets used as "pushing" air streams. Foreach orientation where a test mannequin obstructed thepushing jets, smoke was observed in the area in front of themannequin. However, when the jet was placed between thesmoke and the mannequin, no smoke was observed in thebreathing zone and all smoke was captured by the hood. Theyconcluded that the "push jet should be located so that theair impinging on the worker or other obstruction should beminimized".

Van Wagenen [31] studied the effects of positiveairflow (blowing rather than exhausting air) onconcentrations of various contaminants in a welder's

breathing zone. He demonstrated that when directional airflow comes from directly behind the welder, concentration offume in the breathing zone was equal to or higher than thebreathing zone concentration with no directional air flow atall. He attributed this to the eddy and convective currentsaround the welders body. He also noted that positive

31

airflow at 90 degrees to the welder's position significantly

reduced breathing zone concentration from that with no

directional airflow.

32

A SIMPLE MODEL ADDRESSING REVERSE FLOW

The ideal approach to ventilation design would be to

develop a mathmatical model capable of using LEV design

parameters to predict the concentration of contaminant in a

worker's breathing zone. Ultimately this model could be

applied by industrial hygiene engineers when designing

optimally functioning LEV systems. However, any useful

model must address the effects of this zone of reverse flow

on breathing zone concentration.

Recently, a theoretical model has been proposed by

Flynn [22] as an initial step in understanding how this zone

of reverse flow gives rise to concentrations in the

breathing zone. This model assumes this zone to be a "well

mixed" volume of specific dimensions. A steady state

concentration will be achieved within the zone with

contaminant entering from a point source within the zone and

being removed from the zone by the alternate shedding of

vortices. The following paragraphs briefly describe the

model. Please refer to figure (7) during the discussion.

Consider a circular cylinder of diameter D and height H

completely immersed in a uniform flow of air of velocity U.

Downstream of the cylinder at a distance z (measured from

the downstream edge of the cylinder) a point source of

neutrally buoyant gas is generating contaminant at a flow

rate of Qs. [Note that the flow of contaminant is

33

<riA

a

<r

5 FIGURE 7of 4

• ^e

-O 'V t

/

• -.'s "3 4i/|\ Ji ^ T^

-' , h 'V. . ^ J i

\ ' >' >v\J • o f^

~?•^ J^ «^

1 ^ C tf0 .: ^-^ 1*.

-h .- 0

0- ^u

>* o• X

•

11

Jc

^

<-

(A

U

A:r PiOlV

u

7 D -^

^ ^

SOURCE: REFEREWCE 22

34

assumed to be low enough not to affect wake formation.]

Vortices are alternately shed downstream as described

by the Strouhal number. Recall that the Strouhal number

remains constant at about 0.2 for Re up to about 200,000.

[In an industrial setting. Re around people will almost

always fall below 200,000. For example, in a paint booth,

the OSHA General Industry Standards [32] recommends booth

velocities between 50 and 250 fpm depending on booth size

and crossdraft velocities. For a person of about 20 inches

in cross section this corresponds to Re of about 8666 to

43,333 which is well within the constant Strouhal number

range.] Therefore, solving for frequency of shedding gives(11) f = 0.2U/D

The zone of reverse flow formed by boundary layer

separation around the cylinder will extend a certain

distance downstream of the cylinder. Call this distance s

which will represent the depth of the reverse mixing zone.

The mixing zone becomes significant when it extends far

enough downstream to encompass the contaminant source. That

is, z < s. When this occurs, contaminant is drawn back

toward the cylinder into the mixing zone. In this case, if

this turbulent zone is assumed to be well mixed, the steady

state concentration within the zone can be expressed as:(12) Co = Qs/Qv

where Qv = flow rate out of the mixing zone (cfm)

Qs = flow rate into the zone, i.e., flow rate of

contaminant (cfm)

35

The flow rate out of the zone is controlled by theshedding of vortices such that

(13) Qv = fV

Where V is the mixing zone volume and f is the frequencywith which this volume is removed by vortex shedding. Ifone assumes that the vortices are approximately circularcylinders of height H then the volume can be given by

(14) Vv = (pi)(De^H/4where De is the diameter of an average vortex. However, onemust account for the fact that the zone is composed of twovortices which are alterntely formed on each side and sheddownstream in accordance with the Strouhal number. Thus,

when a vortex is shed, it takes with it one half of the

volume of the zone. Therefore, the actual volume out of thezone is

(14a) V = (pi) (De^ (H)/8and the flow rate out of the zone is given by

(15) Qv = [(0.2)(U)/(D)][(pi)(De)(H)/8]

Substituting into equation (12) gives the followingrelationship for concentration within the zone

(16) Co = [3.57/De]*2[(Qs)(D)/(U)(H)]

Solving for the theoretical diameter of a vortex gives(17) De = 3.57 sq.rt.[(Qs)(D)/Co(U)(H)]

The following assumptions are made; (a) the diameter ofa vortex is essentially the same as the diameter of the zoneof reverse flow (that is, De = s) and beyond this point nocontaminant is drawn back towards the cylinder, (b) the zone

36

is well mixed, (c) the principal mechanism of contaminant

removal from the zone is that of vortex shedding, and (d)

the flow around the cylinder is essentially two dimensional.

Thus, the hypothesis can be offered that as long as the

source of contaminant is within the reverse flow mixing zone

the breathing zone concentration will remain constant. The

concentration in the breathing zone will be the same when

the source is adjacent to the cylinder (Co) as it is when

the source is at the edge of the zone. When the contaminant

source is moved out of the zone, the breathing zone

concentration drops virtually to zero almost immediately

since there is no reverse flow to bring it back towards the

cylinder. At the point where the source sits directly on

the end of the mixing zone (z = De), the breathing zone

concentration (Cbz) should equal one half of the initial

concentration since theoretically one half of the

contaminant is pulled back towards the cylinder and one half

flows away towards the exhaust source. Thus, the point

where Cbz = 0.5 Co can be considered to be the depth of the

zone (De). Figure (8) gives a graphical representation of

the theoretical concentration versus distance from cylindercurve.

.......IHHBiW

•m37

FIGURE S

A^--

CbzCo

a5

Vz

l/V

38

OBJECTIVE AND PURPOSE

The objective of this research is to study the

interaction of a separated boundary layer and subsequent

reverse flow region with a source of contaminant located

downstream of a bluff body in uniform flow. The overall

purpose is to provide additional understanding of the effect

of this reverse flow on breathing zone concentration so that

this information can ultimately be utilized in the

development of a predictive model that ties breathing zone

concentration with LEV design parameters.

Specifically, this project will;

1) Conduct a laboratory evaluation of Flynn's model to

determine its effectiveness in predicting the size of this

reverse flow zone (De) for a circular cylinder and an

anthropometric mannequin in a uniform flow of three

different velocities flow. The principal objective here

will be to evaluate the "mixing zone/vortex shedding"

concept as a useful way of predicting breathing zone

concentration.

2) Evaluate the model as to its ability to predict breathing

zone concentrations by comparing measured concentrations to

those predicted by the model.

3) Examine the difference in breathing zone concentrations

when an anthropometric mannequin is oriented in the typical

worker position with respect to LEV (airflow coming from

behind the mannequin) as opposed to the situation where the

mannequin is turned 90 degrees to the source of LEV (airflow

39

coming from the side). (In the first case the boundary

layer interacts with the contaminant source whereas in the

second case it does not.) This will also be conducted in

uniform flow at three different flowrates.

40

METHOD OF MODEL EVALUATION

The examination of Flynn's model consisted of

experimentally obtaining a concentration versus distance

curve for various points downstream of a circular cylinder

and an anthropometric mannequin immersed in a uniform flow

field of three different velocities. The model was

evaluated by;

1) Comparing the general shape of the experimental curve to

that of the theoretical curve (figure (8)) to empirically

determine whether concentration as a function of distance

behaved in such a way as to indicate uniform mixing. That

is, whether or not concentration remained constant over a

certain distance (mixing zone) and then dropped sharply as

the contaminant source moves outside the zone.

2) Using the theoretical equation (equation (17)) to

calculate the depth of the zone and then comparing this to;

(a) the actual depth of the zone as visualized by test

smoke, and

(b) the distance at which the measured concentration

actually dropped to one half its original value.

The details of the experimental process are given

in the following paragraphs.

WIND TUNNEL DESCRIPTION

The uniform air flow field was achieved by placing the

test objects in a wind tunnel 5 feet tall, five feet wide,

and 8 feet deep. The tunnel was constructed of one-half

inch plywood with a large window on top for lighting and one

41

observation window on each side. One of the observation

windows was mounted on hinges to serve as a door providing

easy access to the inside of the tunnel. The tunnel was

equipped with an airfoil at the entrance to, reduce

turbulence. A sheet metal grid of six inch squares was also

installed at the entrance to further reduce the turbulence

of the incoming air. The rear wall of the tunnel consisted

of peg board with one-quarter inch holes. This board served

to create a perforated plenum effect and provide equal air

distribution across the tunnel.

Holes were drilled in the side of the tunnel to allow

the insertion of an anemometer probe for velocity

measurements. Plugs were inserted into the holes during

experiments so that no turbulence would be introduced by air

entering through the holes. The location of the holes

allowed a velocity profile to be taken at three different

depths inside the tunnel.

VELOCITY DETERMINATION

The average velocity of the air moving through the wind

tunnel was determined by obtaining a velocity profile at

three different depths within the tunnel. Each profile

consisted of twenty equally spaced points for a total of

sixty measurements. The arithmetic mean of all velocity

measurements was taken as the average tunnel velocity.

Measurements were taken with a calibrated TSI "hot wire"

anemometer [33].

42

A blast gate and a large venturi meter were installed

in the duct leading to the tunnel to allow regulation and

measurement of air flow (see figure (9). The anemometer

probe was inserted into the tunnel and the blast gate was

adjusted until the desired velocity (as measured by the

anemometer) was achieved. At this point the pressure drop

across the venturi was recorded. Then the velocity profiles

were obtained and the average velocity computed. Thus a

specific venturi pressure drop corresponded to a particular

average tunnel velocity. Then the blast gate was adjusted

until the second experimental velocity was reached and the

process repeated. This was done for all three experimental

velocities. In this manner the velocity could be accurately

regulated by simply adjusting the blast gate until the

proper venturi pressure drop was achieved.

In this project, experiments were made at three

different wind tunnel velocities; 46 fpm, 143 fpm, and 249

fpm. Figures (10), (11), and (12) illustrate the velocity

profiles and corresponding venturi pressure drops for each

of these velocities respectively. Figure (13) shows the

excellent linearity achieved by plotting the volumetric flow

(rate calculated from tunnel area and velocity) against the

square root of the venturi pressure drop.

It should be noted that the wind tunnel was calibrated

without the test object in place. The presence of these

objects increases the velocity through the tunnel because

their cross sectional area effectively "blocks" a portion of

91

43

1-—»

UJ

S

6"245 255 255 245 240

12«240 260 270 240 235

12"240 230 240 230 230

12"230 255 240 230 250

6"

44

FIGURE 10 - WIND TUNNEL VELOCITY PROFILE

DATE: 11 MAY 1988 TIME: 1200 TEMP: 70 DEGREES FTUNNEL DIMENSIONS: 5' X 5' X 8'

FRONT CROSS SECTION - 18" FROM FACE

6" 12" 12" 12" 12" 6" MAXIMUM = 270 FPM

MINIMUM = 230 FPM

RANGE = 40 FPM

MEAN = 243 FPM

S.D. = 11.4 FPM

C.V. = 4.7 %

MIDDLE CROSS SECTION - 51" FROM FACE

MAXIMUM =260 FPM

MINIMUM =230 FPM

RANGE =30 FPM

MEAN = 251 FPM

S.D. = 8.9 FPM

C.V. = 3.5 %

REAR CROSS SECTION - 83" FROM FACE

6" 12" 12" 12" 12" 6" MAXIMUM =260 FPM

MINIMUM =240 FPM

RANGE =20 FPM

MEAN =254 FPM

S.D. = 7.2 FPM

C.V. = 2.8 %

MAXIMUM = 270 FPM RANGE = 40 FPM S.D. = 10.3 FPMMINIMUM = 230 FPM MEAN = 249 FPM C.V. = 4.1 %TOTAL FLOW RATE = 6225 CFM VENTURI PRESSURE DROP = 2.40"

6" 12 II 12 II 12 II 12 II gii

6"250 255 255 235 240

12"250 260 260 255 250

12"250 260 250 255 260

12"230 240 260 260 250

6"

6"

1 o II260 260 260 260 255

1 "D II255 260 250 260 260

±Z

240 260 240 250 24012"

6"250 255 250 250 260

SUMMARY DATA

45

FIGURE 11 - WIND TUNNEL VELOCITY PROFILE

DATE: 30 MAY 1988 TIME: 1000 TEMP: 75 DEGREES FTUNNEL DIMENSIONS: 5' X 5' X 8'

FRONT CROSS SECTION - 18" FROM FACE

6" 12" 12" 12" 12" 6" MAXIMUM = 160 FPM

MINIMUM =130 FPM

RANGE = 30 FPM

MEAN =142 FPM

S.D. = 7.7 FPM

C.V. = 5.4 %

MIDDLE CROSS SECTION - 51" FROM FACE

6"150 140 145 140 130

12"140 150 160 150 135

12"140 140 140 145 135

12"135 145 155 135 135

6"

6" 12 1 12 1 12 1 12 1 6" MAXIMUM =150 FPM

6"135 145 145 150 130

MINIMUM =125 FPM

12"140 150 150 150 130

RANGE =25 FPM

12"125 150 150 150 130

MEAN =143 FPM

12 •»135 145 150 145 150

S.D. = 8.7 FPM

6" C.V. = 6.1 %

REAR CROSS SECTION - 83" FROM FACE

6" 12" 12" 12" 12" 6" MAXIMUM = 155 FPM

MINIMUM = 130 FPM

RANGE =25 FPM

MEAN =145 FPM

S.D. = 7.3 FPM

C.V. = 5.1 %

MAXIMUM = 160 FPM RANGE = 35 FPM S.D. = 7.9 FPMMINIMUM = 125 FPM MEAN = 143 FPM C.V. = 5.5 %TOTAL FLOW RATE = 3575 CFM VENTURI PRESSURE DROP = 0.74"

6"145 150 145 155 135

12"140 155 150 155 135

12"130 145 145 150 135

12"150 150 140 145 150

6"

SUMMARY DATA

46

FIGURE 12 - WIND TUNNEL VELOCITY PROFILE

DATE: 30 MAY 1988 TIME: 1130 TEMP: 75 DEGREES FTUNNEL DIMENSIONS: 5' X 5' X 8'

FRONT CROSS SECTION - 18" FROM FACE6 1 12" 12" 12" 12" 6" MAXIMUM

MINIMUM

50 FPM

6" 30 FPM40 45 50 45 40

12"40 45 40 43 38

RANGE = 20 FPM

12"40 50 48 40 30

MEAN = 42 FPM

12"45 40 50 35 40

S.D. = 5. 2 FPM

6" C.V. = .L2. 3 %

MIDDLE CROSS SECTION - 51" FROM FACE

6 1 12" 12" 12" 12" 6" MAXIMUM

MINIMUM

55 FPM

6" 30 FPM50 45 45 50 45

12 ••30 45 55 53 45

RANGE = 25 FPM

12"50 45 50 52 42

MEAN = 47 FPM

12"40 45 50 55 45

S.D. = 5.6 FPM

6" C.V. = 12. 3 %

REAR CROSS SECTION - 83" FROM FACE

6" 12" 12" 12" 12" 6" MAXIMUM = 55 FPM

MINIMUM = 30 FPM

RANGE =25 FPM

MEAN = 48 FPM

S.D. = 6.3 FPM

C.V. = 13.0 %

MAXIMUM = 55 FPM RANGE = 25 FPM S.D. = 6.1 FPMMINIMUM = 30 FPM MEAN = 46 FPM C.V. = 13.2 %TOTAL FLOW RATE = 1150 CFM VENTURI PRESSURE DROP = 0.09"

6"45 50 50 50 50

12"40 50 55 40 40

12"30 50 50 55 45

12"45 55 48 53 50

6"

SUMMARY DATA

4T

riGURE 13

a: o

0'-'

VENTURI CALIBRATION CHECKFLOW RATE BASED ON ANEMOMETER READING

SQUARE ROOT OF PRESSURE DROP (in. H20)a DATA POINTS + REGRESSION POINTS

48

the tunnel cross section. This increases the velocity inproportion to the amount of tunnel area blocked by theobject. Therefore, a "blockage ratio" must be calculated.A blockage ratio is basically a factor by which theunblocked velocity must be multiplied to get the true tunnelvelocity. To determine the blockage ratio, the amount oftunnel cross section blocked by the object must beestimated. The following formula can then be applied:

(18)Blockage = Tunnel Cross SectionRatio Tunnel Cross section - Object Cross Section

(19)Corrected = Measured X BlockageVelocity Velocity Ratio

Notice from table (3) that by appling the blockageratio the corrected velocities are 265 fpm, 152 fpm, and 49fpm for the mannequin and 292, 167, and 54 fpm for thecircular cylinder.

The Re for air flow around the objects at thesevelocities is also given in table (3). These velocitieswere selected because the Re are in the same range as thosefor air flowing around an industrial worker in a uniformflow such as a spray paint booth.TEST OBJECT DESCRIPTION

The circular cylinder used in this project wasconstructed of sheet metal and was 48 inches tall and 12inches in diameter. Two holes were drilled in the cylinder,one in the front about 15 inches from the top and another inthe back about 6 inches from the bottom. One end of a one

49

TABLE 3

CORRECTED VELOCITIES AND REYNOLDS NUMBERS

TUNNEL CROSS SECTION: 25 SQ. FT.

APPROXIMATE CYLINDER CROSS SECTION: 3.66 SQ. FT.APPROXIMATE MANNEQUIN CROSS SECTION: 1.52 SQ. FT.CYLINDER DIAMETER: 1 FT.

�MANNEQUIN DIAMETER: 0.67 FT.

ALL VELOCITIES IN FPM

MEASURED VELOCITIES: 46 143 249CORRECTED VELOCITIES, CYLINDER: 54 167 292REYNOLDS NUMBERS, CYLINDER: 5616 17,368 30,368CORRECTED VELOCITIES, MANNEQUIN: 49 152 265REYNOLDS NUMBERS, MANNEQUIN: 3414 10,539 18,465

*The mannequin diameter is not actually a diameter becausethe mannequin cross section is more elliptical than circularin shape. The value reported here as the diameter is thebreadth of the mannequin chest as measured just under thearmpits and at the same distance from the floor (27") as thesource of SF6.

50

quarter inch rubber tube was connected to a calibrated

Mobile Infrared Analyzer or MIRAN (see appendix I). The

other end of the tube was inserted through the hole in the

back of the cylinder, pulled up through the inside and

mounted in the hole in the front. In this manner the MIRAN

could be used as a means to sample the "breathing zone" ofthe cylinder.

The anthropometric mannequin used was a typical

commercial type mannequin 41 inches tall (including base)

and 8 inches wide at the chest. The hose from the MIRAN was

inserted through the back of the mannequin's head and

mounted in the mouth (about 4 inches from the top of the

head). Both the cylinder and the mannequin were placed

approximately on the centerline of the tunnel about 2 feetfrom the tunnel face.

SULFUR HEXAFLUORIDE GENERATION

Sulfur Hexafluoride (SF6) was the test gas selected for

this experiment. The gas was metered from a compressed gas

cylinder of 10 percent SF6 through a one-quarter inch

diameter ceramic sphere. The sphere was mounted on a ring

stand approximately 27 inches from the floor of the wind

tunnel. Pores in the sphere allowed the gas to diffuse inall directions.

An SF6 flow rate of 0.0005 cfm (corresponding to a

velocity of 15 fpm) was chosen. This velocity was high

enough to give detectable readings yet low enough so as not

51

to interfere with zone formation. Appendix (2) describes

the SF6 metering system calibration.

DETERMINATION OF ZONE DEPTH: VISUALIZATION OF TEST SMOKE

To evaluate a model predicting the the depth of the

turbulent mixing zone, some method of judging the "true"

depth of the zone was necessary. In the study conducted by

Ljungquist [29], a cloud of smoke was generated to make the

reverse flow phenomenon easily observable. A similar

technique was used in this experiment to visualize themixing zone.

A continuous source of smoke was achieved by the

apparatus depicted in figure (14). As room air was blown

into a suction flask containing titanium tetrachloride

(TiCL4), a smoke (titamium dichloride) was forced out of the

flask, through several feet of tygon tubing, and out of one-

quarter inch diameter glass tube mounted on a ring stand.

Thus, a dense cloud of white smoke could be continuously

generated.

When the smoke source was placed downstream of the

cylinder or mannequin, the turbulent mixing could be easily

observed. Very close to the object, the majority of the

smoke was drawn back toward the object before eventually

being removed. As the smoke source was moved farther

downstream, more of the smoke was drawn directly downstream

and away from the object. Finally, a point was reached where

approximately one half of the smoke was drawn back toward

the object while the other one half was directly removed.

FIGURE U

VACUUM PUMPTITANIUM

TETRACHLORWE

\

SMOKEOUTLET

Ww

T<J1

53

This point was considered to be the edge of the mixing zone

or the "true" De. Beyond this point, most of the smoke was

drawn directly away from the object. In this manner, an

estimation of the actual size of the zone was made for each

of the three velocities for both test objects. Table (4)

provides this data for both the cylinder and the mannequin

at all three velocities.

TABLE 4

ZONE DEPTH AS VISUALIZED BY TEST SMOKE

54 FPM 167 FPM 292 FPM

CYLINDER 14.0 16.0 22.0

49 FPM 152 FPM 265 FPM

MANNEQUIN 10.0 9.0 13.0

DETERMINATION OF ZONE DEPTH: CONCENTRATION VERSUS DISTANCE

CURVES

The curve of concentration versus distance was obtained

with the experimental set-up depicted in figure (15). The

circular cylinder was placed inside the wind tunnel along

the centerline approximately 2 feet from the tunnel opening.

The SF6 diffuser was also placed on the centerline of the

tunnel at a distance of 0.5 inches downstream of the

cylinder. The tunnel velocity was set at 292 fpm (corrected

velocity) and the SF6 at 15 fpm. The system was allowed to

equilibrate for 10 minutes. After equilibration, the

breathing zone concentration, as measured by the MIRAN, was

logged and integrated over a 10 minute period by a

FIGURE J 5

OBJECT

/ r\

SV6 PIFFUSER

MI RAW

ROTAMETER

DATALOGGER

SF6TAWK JNCIAMEV

MAWOMETEK

55

Metrosonics dl 714 Data Logger. (Please refer to appendix

(IV) for a description of the Data Logger.) In this manner,

a time weighted average concentration was obtained over that

10 minute period for the distance 0.5 inches. After the

logging period, the SF6 source was turned off and the tunnel

was allowed to purge for 10 minutes. After the purging

period the SF6 diffuser was moved to 1 inch downstream and

the process was repeated. Thus, concentrations at 0.5, 1.0,

2.0, 3.0, 4.0, 6.0, 8.0, 10.0, 12.0, 14.0, 16.0, and 18.0

inches were obtained.

After these points were generated the velocity was

lowered to 167 fpm and the procedure was repeated. The same

was done at 54 fpm.

Once the values for the cylinder were obtained, the

entire procedure was repeated with the mannequin at 265 fpm,

152 fpm, and 49 fpm.

Table (5) lists the data and figures (16), (17), and

(18) illustrates the concentration versus distance curves

generated for the cylinder. Figure (19) depicts all the

graphs. Figures (20) through (23) provide the same graphs

for the mannequin data.

A repeat of the experiment was conducted with the

following changes:

1) Equilibrium time was reduced to 7 minutes.

2) Logging time was reduced to 5 minutes.

3) Purging time was reduced to 2 minutes.

4) The SF6 source was not turned off during purging but

56

TABLE 5

SF6 CONCENTRATIONS AT INDICATED DISTANCES

EXPERIMENT #1CYLINDER MANNEQUIN

CONCENTRATION CONCENTRATION

(ppm) (ppm)

DISTANCE 54 167 292 49 152 265

fINCHES) FPM FPM FPM FPM FPM FPM

0.5 18.0 12.4 7.6 23.2 7.1 6.0

1.0 16.7 15.0 4.8 26.0 8.6 6.5

2.0 18.5 11.1 4.5 40.0 13.7 7.2

3.0 14.8 10.7 4.0 35.8 10.6 7.7

4.0 22.3 9.3 2.7 23.5 8.3 6.6

6.0 19.7 7.0 2.4 11.7 5.9 4.8

8.0 10.1 6.0 1.8 11.0 3.8 3.1

10.0 9.9 4.4 1.5 11.4 2.1 1.9

12.0 10.4 3.8 1.4 7.0 1.7 1.4

14.0 4.3 2.8 1.0 5.0 1.1 1.0

16.0 4.3 2.3 0.9 3.8 0.8 0.7

18.0 4.9 1.8 0.7 1.9 0.4 0.5

20.0 4.7 1.4 0.6 1.1 0.3 0.5

TABLE 6

SF6 CONCENTRATIONS AT INDICATED DISTANCES

EXPERIMENT #2CYLINDER MANNEQUIN

CONCENTRATION CONCENTRATION

(ppm) (ppm)

DISTANCE 54 167 292 49 152 265

fINCHES) FPM FPM FPM FPM FPM FPM

0.5 22.4 17.1 14.8 11.9 7.1 6.0

1.0 28.8 10.6 9.9 17.1 7.0 5.9

2.0 29.1 6.9 11.1 20.6 10.4 8.4

3.0 19.4 7.7 5.0 12.3 9.4 7.9

4.0 19.1 7.5 3.7 17.2 7.2 6.4

6.0 13.9 4.4 3.7 10.6 6.7 5.0

8.0 10.9 4.9 3.5 9.2 3.8 2.9

10.0 10.1 4.9 2.7 5.3 2.3 1.7

14.0 7.1 2.7 2.0 4.5 1.2 1.1

18.0 4.5 2.0 1.1 1.9 0.4 0.2

57

FIGURE 16

SF6 CONCENTRATION VS. SOURCE DISTANCE

Eaa\^

zO

Iuuz

8Iflu.m

12" CYUNDER IN WIND TUNNEL

DISTANCE OF SOURCE FROM CYLINDER (IN.)D 54 fpm

5»

FIGURE 17

SF6 CONCENTRATION VS. SOURCE DISTANCE

Eaa

Zo

toz

810k.

12" CYUNDER IN WIND TUNNEL

DISTANCE OF SOURCE FROM CYUNDER (IN.)D 167 FPM

59

FIGURE n

SF6 CONCENTRATION VS. SOURCE DISTANCE12" CYLINDER IN WIND TUNNEL

Ea.a.\^

Xo{=

uOz

810b.H

DISTANCE OF SOURCE FROM CYUNDER (IN.)a 292 FPM

60

FIGURE 19

SF6 CONCENTRATION VS. SOURCE DISTANCE

Eaa

zO

oz

01

12" CYUNDER IN WIND TUNNEL

0 2

aDISTANCE OF SOURCE FROM CYUNDER (IN.)

292 FPM +167 FPM O 54 FPM

61

FIGURE 20

SF6 CONCENTRATION VS. SOURCE DISTANCE

Eaa

8h.10

MANNEQUIN IN WIND TUNNEL

DISTANCE OF SOURCE FROM CYUNDER (IN.)O 49 FPM

62

FIGURE 21

SF6 CONCENTRATION VS. SOURCE DISTANCE

zo

ioz

8IDii.

MANNEQUIN IN WIND TUNNEL

DISTANCE OF SOURCE FROM CYLINDER (IN.)D 152 FPM

63

FIGURE 11

SF6 CONCENTRATION VS. SOURCE DISTANCEMANNEQUIN IN WIND TUNNEL

Eaa

Zo

uOz

8u>u.V)

7 -

6 -

5 -

4 -

3 -

1 -

1------1------r

21------1------1------1------r6 8 10

1-----1-----1-----1-----1-----1-----1-----1-----r12 14 16 18 20

DISTANCE OF SOURCE FROM CYLINDER (IN.)D 265 FPM

63

FIGURE 22

SF6 CONCENTRATION VS. SOURCE DISTANCEMANNEQUIN IN WIND TUNNEL

Eaa.

ZoB

8IDk.M

DISTANCE OF SOURCE FROM CYUNDER (IN.)D 265 FPM

64

FIGURE 23

SF5 CONCENTRATION VS. SOURCE DISTANCEMANNEQUIN IN WIND TUNNEL

aa.s^

Zo

uoz

8

40

35

30 -

25 -

20 -

15 -

10 -

5 -

DISTANCE OF SOURCE FROM CYLINDER (IN.)D 265 FPM + 152 FPM O 49 FPM

65

simply moved to the rear wall of the tunnel.

The data and curves generated in the second experiment

appear as table (6) and figures (24) through (31).

Notice from tables (5) and (6) that the values measured

in experiment 2 differ considerably in some areas than those

from experiment 1. The probable reason for this is the

difference in methods between the two experiments. The most

significant differences were likely that the reduction in

purging times between measurements and the fact that the SF6

was not secured during purging in experiment 2. Because of

additional logging, equilibrium, and purging times and

because the SF6 was turned off during purging for the first

experiment allowing a more thorough clearing of the tunnel,

the measurements made in the first experient should be

considered more accurate.

66

FIGURE 24

SF5 CONCENTRATION VS. SOURCE DISTANCECYLINDER IN WIND TUNNEL

Ea.a.>w

SII-

8IDla.01

DISTANCE OF SOURCE FROM CYLINDER (IN.)D 54 FPM

67

FIGURE 25

SF5 CONCENTRATION VS. SOURCE DISTANCE

Ea.

Zo

I-zuuz

810Ik

CYUNDER IN WIND TUNNEL

DISTANCE OF SOURCE FROM CYUNDER (IN.)D 167 FPM

6i

FIGURE 26

SF6 CONCENTRATION VS. SOURCE DISTANCE

Eaa

Oz

810b.n

CYUNDER IN WIND TUNNEL

DISTANCE OF SOURCE FROM CYLINDER (IN.)D 292 FPM

69

FIGURE 27

SF6 CONCENTRATION VS. SOURCE DISTANCE

Ea

1UiOZ

8U)

in

CYUNDER IN WIND TUNNEL

aDISTANCE OF SOURCE FROM CYUNDER (IN.)

292 FPM + 167 FPM O 54 FPM

10

FIGURE 2g

SF6 CONCENTRATION VS. SOURCE DISTANCE

Eaa.

IiuOZ

m

MANNEQUIN IN WIND TUNNEL

DISTANCE OF SOURCE FROM CYUNDER (IN.)a +9 FPM

71

FIGURE 29

SF6 CONCENTRATION VS. SOURCE DISTANCE

ao.w

Zo

ioz

ID

MANNEQUIN IN WIND TUNNEL

DISTANCE OF SOURCE FROM CYUNDER (IN.)a 152 FPM

72

FIGURE 30

SF5 CONCENTRATION VS. SOURCE DISTANCEMANNEQUIN IN WIND TUNNEL

Eaa.

IIruz

810

in

DISTANCE OF SOURCE FROM CYLINDER (IN.)a 265 FPM

73

FIGURE 31

Ea

IuOZ

810

M

SF6 CONCENTRATION VS. SOURCE DISTANCEMANNEQUIN IN WIND TUNNEL

DISTANCE OF SOURCE FROM CYUNDER (IN.)265 FPM +152 FPM O 49 FPM

m

74

DETERMINATION OF ZONE DEPTH: Cbz = 0.5Co

The model predicts that the edge of the zone is reached

when Cbz = 0.5 Co (figure (8)). In other words, the

distance at which the concentration drops to one-half of its

original value should be the distance that equals the depth

of the zone (D = De).

To find this distance from the curve, it is necessary

to determine a value for Co. For these calculations, two

values for Co were evaluated; Co equals the breathing zone

concentration measured by the MIRAN at a distance of 0.5

inches and Co equals the maximum concentration point on the

concentration versus distance curve.

Once Co is established it remains to find the distance

at which the concentration drops to one half Co. This was

determined by plotting log concentration versus log distance

and obtaining a regression equation for the line. The value

for distance was calculated from the equation when

concentration was 0.5 Co. The use of log-log plots to find

this distance was a matter of practicality. The regression

lines obtained provided an efficient way to make this

calculation. Appendix (4) provides the regression data from

the log-log plots. Table (7) summarizes the values obtainedfor De.

DETERMINATION OF ZONE DEPTH: CALCULATION FROM THEORY

The theoretical equation for predicting zone depth as a

function of Co, U, Qs, D, and H (equation (17)) was also

75

TABLE 7

ZONE DEPTH AS DETERMINED FROM LOG-LOG PLOTS OFCONCENTRATION VERSUS DISTANCE

EXPERIMENT ONE - CYLINDER

VELOCITY(FPMl

De USING MAXIMUMCONCENTRATION AS Co

De USING INITIALCONCENTRATION AS Co

* 54

167

292

8.3"

3.9"2.1"

7.8"

4.3"2.1"

EXPERIMENT ONE - MANNEQUIN

VELOCITY

fFPM)De USING MAXIMUM

CONCENTRATION AS CoDe USING INITIAL

CONCENTRATION AS Co

49

152

265

4.4"

4.2"5.8"

4.6"

4.3"4.2"

EXPERIMENT TWO - CYLINDER

VELOCITY(FPM)

De USING MAXIMUMCONCENTRATION AS Co

De USING INITIALCONCENTRATION AS Co

54167

292

5.2"1.9"1.9"

6.9"

1.9"1.9"

EXPERIMENT TWO - MANNEQUIN

VELOCITYrFPM)

De USING MAXIMUMCONCENTRATION AS Co

De USING INITIALCONCENTRATION AS Co

49

152

265

5.0"4.8"4.7"

10.1"4.9"

4.3"

* VELOCITY ADJUSTED FOR BLOCKAGE RATIO

76

evaluated. Each of the above listed values was known for

the cylinder and the mannequin. Once again, the same twovalues for Co mentioned above were used in this calculation.Table (8) gives the calculated values of De from thepredictive model.SUMMARY

Table (9) summarizes the results of the modelevaluation experiment by comparing the depth of the zone asdetermined by;

1) the observation of test smoke,

2) the concentration versus distance curves, and3) the theoretical equation.

77

TABLE 8

ZONE DEPTH AS DETERMINED FROM THEORETICAL EQUATION

EXPERIMENT ONE - CYLINDER

VELOCITYfFPM)

De USING MAXIMUMCONCENTRATION AS Co

De USING INITIALCONCENTRATION AS Co

54

167292

13.8"9.6"

10.2"

15.4"10.5"10.2"

EXPERIMENT ONE - MANNEQUIN

VELOCITYfFPM)

De USING MAXIMUMCONCENTRATION AS Co

De USING INITIALCONCENTRATION AS Co

49

152

265

9.6"9.3"9.4"

12.6"12.9"10.6"

EXPERIMENT TWO - CYLINDER

VELOCITY

fFPM)De USING MAXIMUM

CONCENTRATION AS CoDe USING INITIAL

CONCENTRATION AS Co

54

167

292

12.1"9.0"7.3"

13.8"9.0"7.3"

EXPERIMENT TWO - MANNEQUIN

VELOCITY(FPM)

De USING MAXIMUMCONCENTRATION AS Co

De USING INITIALCONCENTRATION AS Co

49

152265

13.3"10.7"9.0"

17.6"12.9"10.6"

78

TABLE 9

SUMMARY TABLE OF ZONE DEPTH BY VARIOUS METHODS

[ALL MEASUREMENTS FOR ZONE DEPTH GIVEN IN INCHES]

EXPERIMENT ONE - CYLINDER

VELOCITYfFPM^

SMOKE

VISUALIZATION�LOG-LOGPLOTS

�THEORETICALEQUATION

54

167

292

14.0

16.0

22.0

7.8/8.34.3/3.92.1/2.1

**15.4/13.810.5/ 9.610.2/10.2

EXPERIMENT ONE - MANNEQUIN

VELOCITY

fFPM)SMOKE

VISUALIZATION�LOG-LOG

PLOTS�THEORETICALEQUATION

49

152

265

10.09.0

13.0

4.6/4.44.3/4.24.2/5.8

12.6/ 9.612.9/ 9.310.6/ 9.4

EXPERIMENT TWO - CYLINDER

VELOCITYfFPM)

SMOKEVISUALIZATION

�LOG-LOGPLOTS

�THEORETICALEQUATION

54

167292

14.0

16.0

22.0

6.9/5.21.9/1.91.9/1.9

13.8/12.19.0/ 9.07.3/ 7.3

EXPERIMENT TWO - MANNEQUIN

VELOCITY

fFPM)SMOKE

VISUALIZATION�LOG-LOGPLOTS

�THEORETICALEQUATION

49

167292

10.0

9.0

13.0

10.1/5.04.9/4.84.3/4.7

17.6/13.312.9/10.710.6/ 9.0

� FIRST VALUE REPORTED CONSIDER INITIAL CONCENTRATION AS Co.SECOND VALUES CONSIDER MAXIMUM CONCENTRATION AS Co.

79

TABLE 9 fcont.)

** The theoretical value for zone depth (De) was calculatedfrom equation (17) as follows:

De = (3.57)sq.rt.[(QS)(D)/(Co)(U)(H)]

De = (3.57)sq.rt._____f f0.0005 cfm) (1 ft)1[(1.8 X lOE-5)(54 fpm)(4 ft)]

De = 15.4 inches

80

MANNEQUIN VERSUS MANNEQUIN 90 DEGREES

Another way of examining the effect of boundary layerseparation on breathing zone concentration is to monitorconcentration in the breathing zone under conditions inwhich the boundary layer will interact with the contaminantsource and compare the result to a situation in which itdoes not. As has been mentioned, in the typical orientationof a worker with respect to LEV, the separated boundarylayer can possibly extend downstream far enough to cause thecontaminant to be drawn back into the breathing zone.However, if the worker stands at right angles to the flow(figure (32)), whatever reverse flow zone is formed will beless likely to reach out and interact with the contaminantsource but will rather extend downstream.

To evaluate the effect of a 90 degree orientation withrespect to ventilation on breathing zone concentration, amethod similar to that previously described was used. Themannequin was again placed in the wind tunnel. However,this time it was oriented at 90 degrees from the flow. Thatis, the wind was flowing from its side rather than aroundits back. A set of SF6 concentration versus distance curves

was obtained in exactly the same manner as before for 2 65fpm, 152 fpm, and 49 fpm.

The curves obtained with the mannequin at 90 degreesare graphically compared to the curves obtained in theprevious experiment in figures (33), (34), and (35). A

JS^SS--:-

S1

>

->

WPICAL ORIENTATION

FIGURE 32

>

>

"^^^o

9(? VEGREES TO FLOW

B2

FIGURE 33

SF6 CONCENTRATION VS. SOURCE DISTANCE

Eaa

\Oz

8If

MANNEQUIN - 49 FPM

DISTANCE OF SOURCE FROM CYLINDER (IN.)MANNEQUIN/90 + MANNEQUIN

S3

FIGURE 34

SF6 CONCENTRATION VS. SOURCE DISTANCE

Eaa.^^

Zo

uOz

IDU.in

MANNEQUIN - 1 52 FPM

DISTANCE OF SOURCE FROM CYLINDER (IN.)MANNEQUIN/90 + MANNEQUIN

S4

FIGURE 35

SF5 CONCENTRATION VS. SOURCE DISTANCE

aa.

ItOz

810Ik

MANNEQUIN - 265 FPM

DISTANCE OF SOURCE FROM CYUNDER (IN.)a MANNEQUIN/90 i- MANNEQUIN

85

table of these values is provided as table (10)

86

TABLE 10

SF6 CONCENTRATIONS AT INDICATED DISTANCES

MANNEQUIN/9 0 MANNEQUIN

CONCENTRATION CONCENTRATION

(ppm) (ppm)

DISTANCE 49* 152 265 49 152 265

rINCHES1 FPM FPM FPM FPM FPM FPM

0.5 1.68 1.04 1.05 11.9 7.1 6.0

1.0 1.11 0.63 0.68 17.1 7.0 5.9

2.0 0.90 0.63 0.99 20.6 10.4 8.4

3.0 0.41 0.37 0.44 12.3 9.4 7.9

4.0 0.40 0.32 0.34 17.2 7.2 6.4

6.0 0.37 0.31 0.33 10.6 6.7 5.0

8.0 0.38 0.33 0.33 9.2 3.8 2.9

10.0 0.34 0.31 0.32 5.3 2.3 1.7

14.0 0.33 0.31 0.31 4.5 1.2 1.1

18.0 0.35 0.31 0.31 1.9 0.4 0.2

*Note that these velocities are not entirely accurate sincethey were based on the blockage ratio at the maximummannequin cross section. However, for comparitive purposesthey are quite close.

87

DISCUSSION OF MODEL EVALUATION

DISCUSSION OF TEST SMOKE OBSERVATION

The observation of test smoke supports the findings of

Ljungqvist and others that there is, indeed, a turbulent

mixing zone formed downstream of a bluff body in an air

flow. The smoke was actually drawn back towards the

cylinder or mannequin when the outlet of the smoke generator

was within about four inches of the object. As the distance

increased, more of the smoke was drawn downstream but still

a significant portion was "swirled" around and back towards

the object. As recorded in table (4), a point was reached

where it appeared as if approximately one half of the smoke

was directly removed while the other half was drawn back

towards the object before being removed. This distance was

recorded as the depth of the mixing zone, De.

Using the distance at which half the smoke was drawn

back while half was directly removed is consistent with

Flynn's theoretical model. However, this turbulent,

swirling motion continued for several more inches

downstream.

It should be noted that determining the zone depth in

this fashion is appropriate. However, it does require a

subjective assessment of a swirling cloud of smoke to

determine when approximately one half of the cloud moves in

one direction while the other half moves in the other.

However, the smoke vividly illustrates the presence of this

zone and clearly demonstrates that this phenomenon can have

88

a significant impact on breathing zone concentration at

downstream distances at least as large as those reported intable (4).

Note from table (4) that the zone depth increases with

increasing airstream velocity. For the cylinder, the zone

increases slightly from 54 fpm to 167 fpm and then shows a

much larger increase between 167 fpm and 292 fpm. The

difference in zone depth between 49 fpm and 152 fpm is also

very small for the mannequin. In fact, the recorded value

is slightly smaller for 152 fpm than for 49 fpm. However,

as with the cylinder, the zone depth makes a relatively

large jump from 152 fpm to 265 fpm.

The difference in zone depths between the mannequin and

the cylinder is most probably due to the difference in

diameter of the two objects. The cylinder, with a diameter

of 12 inches has a deeper zone than the mannequin whose

shape is more elliptical with a cross section of 8 inches

facing the flow.

DISCUSSION OF ACTUAL AND THEORETICAL CONCENTRATION VERSUS

DISTANCE CURVES

Flynn's model assumes a well mixed-turbulent zone in

which a steady state concentration (Qs/Qv) is achieved

throughout. Under this assumption, the theoretical

concentration versus distance curve (figure (8)) predicts a

constant concentration when the contaminant source is

anywhere within the zone and a rapid drop in concentrationonce the source moves outside the zone.

89

A comparison of the theoretical curve with the

experimental curves (figures (16) - (31)) does not support

the idea of complete, uniform mixing throughout the zone.For the cylinder at the highest velocity (292 fpm - figure(18)), very little mixing seems to occur. The concentrationappears to fall off exponentially with increasing distancethroughout the zone, dropping to essentially zero by thetime the edge of the zone (as determined from test smoke) isreached. However, at 167 fpm (figure (17)), experiment 1

data shows that a little more mixing is occurring, givingconcentrations that are relatively uniform up to 4 to 6

inches before falling off. The best mixing within the zoneseems to occur at 54 fpm (figures (16) and (24)). Here, asomewhat uniform concentration can be observed out to

approximately 8 to 12 inches before a sharp decrease occurs.

The mannequin data indicate some mixing in the zone out toabout 6 to 8 inches for all velocities (figures (20), (21),

(22), (28), (29), (30)). Note that the mannequin provides amore realistic view of the effect of the turbulent zone

phenomenon on human exposure because of the hands and armswhich extend into the zone. The hands extend a distance of

approximately three inches from the body of the mannequin.This may account for the better mixing observed for themannequin at lower flows. Perhaps some smaller scaleturbulent zone is formed downstream of the arms themselves.

The motion of air around arms and hands is an important

do

determinant in breathing zone concentration for workers atlaboratory hoods [34].

In addition to the assumption of a well mixed zone, the

model under study also postulates that the edge of the

mixing zone is considered to be the point at which the

initial concentration (Co) falls to one half of its originalvalue (see figure (8)).

In table (9), the zone depth as visualized by the testsmoke is compared with the distance at which the

concentration actually fell to one half of its initial value(as determined from the log-log plots of concentration

versus distance). Recall that Co was defined in two ways;

first as the breathing zone concentration at a source

distance of 0.5 inches and second as the maximum breathing

zone concentration value on the curve. Note from table (9)

that in all but one case the zone depth as determined from

Cbz = 0.5Co is considerably less than that visualized by thetest smoke. The difference ranges from a factor of about

two to a factor of more than ten. The one exception is in

the second mannequin experiment at a velocity of 49 fpm whenCo is defined as the breathing zone concentration at a

source distance of 0.5 inches. This zone depth determinedfrom this point (10.1") is almost exactly equal to that

visualized by smoke (10"). However, note from table (9)that this value is considerably out of line with pointsobtained from the other log-log plots. Observation of

figure (28) indicates a very low initial concentration for

91

this particular curve when compared to the curves at thisvelocity for the first mannequin experiment and bothcylinder experiments. Therefore, it is concluded that theclose agreement at this single point is not significant.

On the basis of these experiments, it appears that thefundamental assumption of a well mixed turbulent zone maynot be accurate. The data seems to indicate that evenwithin the zone a definite variation in concentration withdistance is observed. Perhaps the periodic shedding ofvortices from this zone is not the only contaminant removingprocess involved. Other phenomenon, such as turbulentdiffusion, may also affect concentration, giving rise togradients within the zone.DISCUSSION OF MANNEQUIN VERSUS MANNEQUIN AT 90 DEGREES

Perhaps the most dramatic indication of thesignificance of the reverse flow phenomenon on workerexposure can be seen in the comparison of breathing zoneconcentration between the mannequin at typical orientationwith respect to LEV and the mannequin oriented at rightangles to the flow. From figures (33) through (35) it isobvious that standing 90 degrees to the direction of airflow significantly reduces the concentration of contaminantin the breathing zone. Concentrations measured with airflowcoming from behind the mannequin ranged from 6 to 43 timeshigher than those measured at corresponding distances withthe mannequin at 90 degrees to the air flow.

92

DISCUSSION OF THE MODEL'S ABILITY TO PREDICT De

Recall that a mathmatical model was proposed that

predicts the depth of the turbulent zone as a function of

initial concentration, air stream velocity, contaminant flow

rate, and object diamenter and height (equation (17), page

35). This relationship is based on the removal of

contaminant from the mixing zone by the periodic shedding ofvortices.

As was demonstrated, the fundamental assumption

involved in the formualtion of this model, that of a steady

state concentration within the turbulent zone, does not

appear to be correct. That is, Co does not equal Qs/Qv.

However, it can be seen from table (9) that the

theoretical equation does show some agreement with the smoke

visualization at the lower velocities (Reynolds numbers).

Observe figures (16), (20), (24), and (28), and table (5).

Particularly note the concentration variation over the

distances that have been previously considered to be the

"true" zone depth (from smoke visualization). At these

lower Reynolds numbers it appears that even though the

entire mixing zone is not completely well mixed, some mixing

is indeed occurring giving concentrations that are more

consistent at least through the first few inches of thezone.

93

DISCUSSION OF THE MODEL'S ABILITY TO PREDICT BREATHING ZONE

CONCENTRATIONS

In the theoretical model, the following relationship

was derived for calculating concentrations within the zone:

(16) Co = [(3.57/De)^2][(Qs)(D)/{U)(H)]

In this case, Co would be the breathing zone concentration

when the source is located anywhere within the zone. Using

the zone depth determined from test smoke as De and the

experiment one data, actual values for Co can be obtained

for each velocity. Table (11) compares the breathing zone

concentrations actually measured at various distances within

the zone with those predicted from the model as calculated

with equation (16) above. Figure (36) plots measured versus

calculated concentrations illustrating the line of perfect

correlation.

Note that at the lower velocities (54 fpm-cylinder, 49

fpm-mannequin), the breathing zone concentration predicted

from the model agrees with the measured concentrations to

within a factor of five in all cases and in most cases the

factor is less than three. Even at the higher velocities

the agreement is within a factor of about five although

under these conditions the predicted value tends to

underestimate the measured value. The use of a mixing

factor is not uncommon in mathmatical models based on a well

mixed steady state concentration. For example, the dilution

ventilation model presented in reference [2] recommends a

mixing factor between three and ten. Therefore, it would

94

TABLE 11

USING THEORETICAL EQUATION TO PREDICT

CONCENTRATION WITHIN ZONE

CYLINDER: EXPERIMENT #1 DATA

DISTANCE Cbz (ppm) Co/CbzriNCHES) @ 54 FPM Co fppm) fMIXING FACTOR)

0.5 18.0 *21.7 1.21

1.0 16.7 21.7 1.30

2.0 18.5 21.7 1.17

3.0 14.8 21.7 1.47

4.0 22.3 21.7 0.97

6.0 19.7 21.7 1.10

8.0 10.1 21.7 2.15

10.0 9.9 21.7 2.19

12.0 10.4 21.7 2.07

14.0 4.3 21.7 5.05

*Co calculated from equation (20) using De from test smokeobservarion (table (4)) as follows:

Co = (3.57/De)^2[(Qs)(D)/(U)(H)]

Co = (3.57/1.17')*2[(0.0005 cfm)(l')/(54 fpm)(4')

Co = 21.7 ppm

DISTANCE Cbz (ppm) Co/CbzfINCHES) P 167 FPM Co (ppm) miXING FACTOR)

0.5 12.4 5.4 0.44

1.0 15.0 5.4 0.36

2.0 11.1 5.4 0.47

3.0 10.7 5.4 0.50

4.0 9.3 5.4 0.58

6.0 7.0 5.4 0.77

8.0 6.0 5.4 0.90

10.0 4.4 5.4 1.23

12.0 3.8 5.4 1.42

14.0 2.8 5.4 1.93

16.0 2.3 5.4 2.35

95

TABLE 11 rcont.)

DISTANCEriNCHES)

Cbz (ppm)0 292 FPM Co (ppm)

Co/Cbz(MIXING FACTOR)

0.5 7.6 1.6 0.21

1.0 4.8 1.6 0.33

2.0 4.5 1.6 0.36

3.0 4.0 1.6 0.40

4.0 2.7 1.6 0.59

6.0 2.4 1.6 0.67

8.0 1.8 1.6 0.89

10.0 1.5 1.6 1.07

12.0 1.4 1.6 1.14

14.0 1.0 1.6 1.60

16.0 0.9 1.6 1.78

18.0 0.7 1.6 2.29

20.0 0.6 1.6 2.67

MANNEQUIN: EXPERIMENT #1 DATA

DISTANCE

(INCHES)Cbz (ppm)0 49 FPM Co (DDm)

Co/Cbz(MIXING FACTOR)

0.5 23.2 36.7 1.58

1.0 26.0 36.7 1.41

2.0 40.0 36.7 0.92

3.0 35.8 36.7 1.03

4.0 23.5 36.7 1.56

6.0 11.7 36.7 3.14

8.0 11.0 36.7 3.34

10.0 11.4 36.7 3.22

DISTANCE

(INCHES)Cbz (ppm)@ 152 FPM Co (DDm)

Co/CbZ(MIXING FACTOR)

0.5 7.1 14.6 2.06

1.0 8.6 14.6 1.70

2.0 13.7 14.6 1.07

3.0 10.6 14.6 1.38

4.0 8.3 14.6 1.76

6.0 5.9 14.6 2.47

8.0 3.8 14.6 3.84

96

DISTANCE CI:)Z (ppm) Co/Cbz(INCHES) P 2 65 FPM Co (ppm) (MIXING FACTOR)

0.5 6.0 4.0 0.67

1.0 6.5 4.0 0.62

2.0 7.2 4.0 0.563.0 7.7 4.0 0.52

4.0 6.6 4.0 0.61

6.0 4.8 4.0 0.83

8.0 3.1 4.0 1.2910.0 1.9 4.0 2.11

12.0 1.4 4.0 2.86

fIGURE 36

MEASURED VS. PREDICTED CONCENTRATIONSFROM ORIGINAL THEORETICAL EQUATION

£aa

UizOP

UJOzoo

QUa:

i

THEORETICAL CONCENTRATION (ppm)

98

seem that a mixing factor of about five for this model would

not be unreasonable.

It is of intrest to examine this mixing factor

(henceforth called "K") for any trends that may be evident.

Taking the data in table (11) and plotting In K versus a

dimensionless distance term (Z/D) gives a linear

relationship. This is particularly true if a separation is

made between data collected at Re greater than 10,000 and

that collected at Re less than 10,000. Figures (37) and

(38) illustrate the plots for the cylinder and mannequin

respectively for Re less than 10,000. Figures (39) and (40)

provide these same plots for the cylinder and mannequin for

Re greater than 10,000.

The capability to calculate De from known parameters is

all that is lacking to complete a model incorporating the K

factor and allowing the use of equation (16) to predict

concentrations within the mixing zone. Note from table (4)

that a linear relation appears to exist between De and Re.

In general, as the Re increases, the depth of the mixing

zone (as visualized with test smoke) also appears to

increase. By plotting a dimensionless zone depth (De/D)

versus Re, figure (41) is obtained. Using the regression

equation from this plot, an estimate of De can be acquired

by knowing object diameter and the Re of the flow around the

object.

Table (12) provides a complete theoretical model for

predicting concentration of contaminant in the turbulent

99Q r e? s s i o n CJ u t i::) u t if

Std Err ot Y iEst

R SquaredH o., a -f O ta s e r vat i o n s

D e c:j 1- e? e B ! j f F r e e ci o fn

-v„ (j;.:::/4!.j

0.. 286667

:i. o

a

; Cosrf t :i c: i en t v;;;;; J. „ 0:37444

]td Err of Coef „ ij,. 240713

FIGURE 37

In MIXING FACTOR VS. Z/D

5£

iozX

2

CYLINDER — 54 FPM

D DATA POINTSZ/D (DIMENSIONLESS DISTANCE)

-I- REGRESSION POINTS

FIGURE 37

100

C a n s t a n t O „ i.56> 3 O iEjtd Err of Y EZst 0„ 303773R S q I ..I a r • e ci ('!;.. 5 3 Ei 9 2 4No. of Observations ISDegrees of Freedom 13

Std E"'T Q-f Coef a C>„214437

I§

In MIXING FACTOR VS. Z/DMANNEQUIN----49 FPM & 152 FPM

n DATA POINTSZ/D (DIMENSIONL£SS DISTANCE)

+ REGRESSION POINTS

FIGURE 38

««f

Keg re

R 5Qi..iare?dN a,. i;:i t 0 !::> b e r vat :i. a n;;Degrees of Fr eeci om

702:put"

0. 2-.:i.6i:,:i:l.("1, Q "/ 3 Q 9 9

A I.;, ("i e T 'f-11:: i e n c i s j .;.„.!. •.::! / *.. ' a .

IozX

2

In MIXING FACTOR VS. Z/DMANNEQUIN-----265 FPM

-0.7

D DATA POINTSZ/D (DIMENSIONLESS DISTANCE)

+ REGRESSION POINTS

FIGURE 4(;

103

FIGURE 41

DIMENSIONLESS De AS A FUNCTION OF Re

Q

>Q

1.9ZONE DEPTH DOWNSTREAM OF OBJECT

DATA POINTS

U 18(Thousands)

REYNOLDS NUMBER+ REGRESSION POINTS

R e g r e s s i o n 0 i.i t p:) i..( t. nConstant 1„029587Std Err of Y Est 0„144758R Squared 0«7S7622M o, o f 0 b s e r V a t :i. g p. s 6Degrees of Freedom 4

X Coe-f -f i c i en t < s) 0, 000025S t. d E r r o t C o e f . O.. O O O 0 0 6

^

104

TABLE 12

COMPLETE THEORETICAL MODEL

CO = rf3.57/De)^2]r(C)s)fD)/fU) (H) 1K

De = D[(0.000025)(Re) + 1.03]

MIXING FACTOR (K):

CYLINDER

- Re < 10,000;

- Re > 10,000!

MANNEQUIN

- Re < 10,000:

- Re > 10,000:

K = exp[(1.04)(Z/D) - 0.03]

K = exp[(1.37)(Z/D) - 1.10]

K = exp[(0.93)(Z/D) + 0.16]

K = exp[(1.14)(Z/D) - 0.82]

Co = concentration in the mixing zone

Qs = flow rate of contaminant into zone (cfm)

D = diameter of object (ft)

U = velocity of air around object (fpm)

H = height of object (ft)

De = diameter of the mixing zone (ft)

Re = Reynolds number

K = correction (mixing) factor

105

mixing zone formed downstream of a circular cylinder and an

anthropometric mannequin at high and low Re. Figures (42)

through (47) are plots of the measured versus calculated

concentrations with the line of perfect correlation being

illustrated.

106

FIGURE 42

MEASURED VS, PREDICTED CONCENTRATION;

bJO

J!

i

o

iI)z

FROM CYUNDER MODEL (Re > 10,COO)a -

X7 _ y^6 -

5 -

y^D

4 -

a y^

'O.y^Vi y^

1 -

r'T —r 1 1 1 i 1 1 1

MEASURED cONCEfJTRATION fpptT,1Q VELOCITY - 292 FPM

107

FIGURE 43

MEASURED VS. PREDICTED CONCENTRATIONSFROM CYUNDER MODEL (Re > 10,000)

Eaa.

^-•

_iUi£iO

O

i0

MEASURED cQNcENTRATiQN fppt-n)D '/ELOCrrf - 16? FPM

TOS

FIGURE 44

MEASURED VS. PREDICTED CONCENTRATIONSFROM OOJNDER MOCCL (Re < 10,000)

aQ.

_luQO2

0

Iu

z

MEASURQJI CuNCENTftATION (ppm)D VELOCITY" - 54 FPM

'09

FIGURE 45

MEASURED VS, PREDICTED CONCENTRATIONSFROM MANNEQUIN MODEL (R« > 10,0)0)

8:-Iud02

zO

IiuOzoo

MEASURED cuNcENTRATiON f'ppm^D VELOCITY" - 266 FPM

10

FIGURE 46

MEASURED VS. PREDiCTED CONCENTRATIONSFROM MANNEQUIN MODEL (Re < 10,000)

Eaa

hia02

z0e

uUzo

MEASUKkl) CUNCtNTRATluN (ppmiD VELOCITt' - 152 FPM

msm

m

FIGURE 47

MEASURED VS. PREDICTED CONCENTRATIONS

i:aa.

-IuaO

zo

0

8

FROM MANHEOUiN MODEL (Re < 10.1X30)60

y'^

^

50 -

aAQ -

d"

Q ,

30 -

y

2D -

a

y'

10 -^

/'1 \ 1 1 \

&i 40 60

MEASURED CUNcENTRATItJN (ppmia VELOCITY - 45 FPM

.iUMtf

112

CONCLUSIONS

TEST SMOKE OBSERVATION

The effect of the interaction of a separated boundary

layer with a downstream contaminant source on concentrations

within the breathing zone can be easily visualized with testsmoke. A turbulent mixing zone is quite obviously formed

downstream of a bluff body in uniform flow and contaminant

sources located within this zone can potentially be drawn

back towards the body. The impact of this phenomenon on

worker exposure is noted by considering a situation in whicha worker is immersed in a uniform flow (such as a paint

booth) with the contaminant source between himself and thesource of ventilation.

THEORETICAL MODEL

Although this turbulent mixing zone obviously exists,

the results of this experiment do not support the

theoretical idea that it is completely well mixed

throughout. Considering the true zone depth to be that

visualized by test smoke, the actual concentration versus

distance curves indicate that concentration drops off

significantly before the source moves out of the zone. If

the zone were completely well mixed with a steady state

concentration equal to Qs/Qv the concentration would be moreuniform throughout.

Considering the edge of the zone to be the point at

which the initial concentration drops to one half itsoriginal value also appears to be incorrect since Cbz =

113

0.5CO well before the end of the zone. Again, the observed

behavior is inconsistent with the concept of a well mixedzone.

To explain these results in light of the theoretical

model it is necessary to examine the original assumptions

made. One principal assumption was that the depth of the

reverse flow or mixing zone is equal to the diameter of a

vortex and that beyond this depth no contaminant is drawn

back into the zone. Another critical assumption was that

vortex shedding is the only mechanism of contaminant removal

from the zone and that the reverse motion of the vortex is

the only phenomenon responsible for pulling contaminant back

toward the object.

However, by observing the actual concentration versus

distance curves and comparing these to the zone depths as

visualized by test smoke one notices that smoke is pulled

back into the zone at distances well beyond the point at

which concentration drops to one half of its original value.

If the relatively uniform concentrations over distances of

six to eight inches (at low Re) are indications of a

reasonably well defined vortex, then it appears that the

actual mixing zone extends farther downstream than just the

diameter of a vortex and that perhaps some other mechanism

is moving contaminant around in the zone, causing

concentration gradients and pulling contaminant back into

the zone. Considering the literature describing vortex

114

formation, it does not seem unreasonable to assume that this

is the case.

Recall that the well defined, stable, laminar vortex

street occurs at Re much lower than any used in this

experiment. A completely laminar vortex is observed at Re

less than 200, while the lowest Re for this experiment was

approximately 3500. Above a Re of 200, the wake begins to

deterioriate into turbulence. As Re increases, the point at

which the wake becomes turbulent moves back towards the

cylinder. Thus, a laminar boundary layer exists out to a

certain distance, beyond which it undergoes transition to

turbulence. Therefore, the boundary layer that is rolling

up into a vortex is becomming turbulent closer and closer to

the cylinder as Re increases, particularly at Re greater

than 1300. At a Re of 50,000 the transition to turbulence

is almost to the shoulder of the cylinder [26]. Thus, at

this point, the boundary layers are turbulent almost as soon

as they are separated, even though the turbulent vortices

that are formed are still shed with a periodic frequency

[24].

Since all Re studied in this experiment are greater

than 1300, it is postulated that what is being observed is

that the region of a laminar boundary layer is undergoing

transition to turbulence at distances very close to the

cylinder and the vortices are becomming less well defined.

At the lower Re (cylinder at 54 fpm for example) a

relatively well mixed zone may be observed for six to eight

115

inches. This may be indicative of a reasonably well defined

vortex. However, if it is indeed a somewhat stable vortex

that is causing the relatively uniform concentration (and K

factor of about one) out to say eight inches at this

velocity then one would expect to be able to use this

diatance as De in equation (16) and solve for concentration.

If this substitution is made, a concentration of

approximately 66 ppm is calculated. When this is compared

to the measured value of 10.1 ppm (table (5)) it appears

that even though vortex shedding may be removing a portion

of the contaminant, some other contaminant removal process

is involved, even at the lower Re where stable vortices are

formed farther in front of the cylinder.

At higher Re, no such uniform concentration is observed

over any distance. This is possibly because the boundary

layers and subsequent vortices are becomming turbulent very

close to the shoulder of the cylinder.

These conclusions are somewhat consistent with Bloor's

work [26] which demonstrated that for 2000 < Re < 8500 the

point of transition to turbulence decreases from 1.4 to 0.7

diameters downstream of the cylinder center. For the length

scales in our experiment this would mean that at Re equal

2000, transition to turbulence occured approximately 8.4

inches downstream of the back edge of the cylinder. This

distance decreased with increasing Re such that at Re equal

8500, transition to turbulence occurs approximately 2.4

inches downstream of the cylinder edge. On the other hand.

at 20,000 < Re < 45,000, Bloor reports that the flow

degenerates very suddenly to turbulence in less than one

radius downstream of the cylinder center. For our

experiment this would mean that the boundary layer wasB

turbulent even before it passed the back edge of the

cylinder.

If the periodic shedding of stable vortices were the

only mechanism pulling contaminant back towards the object,

one would not expect the see the zone depth extend as far

downstream as was noted in the visualization of test smoke.

In fact, one might expect the zone to decrease as the

vortices deteriorated. Therefore, some other mechanism must

be involved in the process of contaminant movement in the

turbulent mixing zone. It is possible that this other

mechanism of contaminant transport within the zone is

turbulent diffusion. In fact, many researchers studying the

transport of contaminant in the wake downstream of a bluff

body consider a turbulent diffusion parameter to be the most

important consideration [35,36,37,38,39].

If turbulent diffusion is the additional mechanism of

contaminant removal from the zone, then the K factor

calculated for the complete model is, in effect, accounting

for this mechanism. This may explain the why the In K

versus Z/D plots gave much better relationships at higher Re

and why the complete model (including K factor) worked much

better at predicting zone concentrations at higher Re than

at lower Re. At Re less than 10,000, vortex shedding may

117

play a significant role in the flow rate of contaminant outof the mixing zone. Therefore, at these Re, the K factor,which in effect is accounting for turbulent diffusion, doesnot work as well at predicting the concentration (figures(44) and (47)). On the other hand, at Re greater than10,000, turbulent diffusion may predominate since thevortices are becomming more and more turbulent and thus themodel (including the K factor) predicts concentration withinthe zone very nicely (figures (42), (43), (45), and (46)).

There are also indications in the literature that the

distance over which the turbulent zone entrains contaminant

may increase with increasing Re. Gerrard [40] suggests thatthe entrainment flow of the turbulent wake is governedmainly by the length of the turbulent shear layer as itcrosses the axis of the wake and that this flow increases

with increasing Re. Gerrard terms the thickness of thisshear layer the "diffusion length". This could explain theincreasing zone depth with increasing Re observed with thetest smoke.

To summarize, it appears that the two crucialassumptions involved in the development of the model are notcompletely correct. The distance downstream to which theturbulent mixing zone extends seems to be greater than thediameter of a vortex. The turbulent wake appears to effectcontaminant concentrations in the zone at distances beyondthe region of vortex formation. Consequently, the alternateshedding of vortices do not appear to be the sole mechanism

118

involved in removing contaminant from the mixing zone.

Perhaps at lower Re (less than 10,000) vortex shedding does

make a signifcant contribution but at higher Re (greater

than 10,000) turbulent mechanisms seem to dominate.

MANNEQUIN VERSUS MANNEQUIN AT 90 DEGREES

Consideration of breathing zone concentrations for the

mannequin with its back to the flow as compared to the

mannequin at 90 degrees with respect to flow emphasizes this

potential for the interaction of the reverse flow zone with

a contaminant source downstream. The much higher

concentrations measured with the back to the flow shows that

the separated boundary layer can interact with the

contaminant source in such a way as to increase

concentrations in the breathing zone. This experiment

demonstrates that in situations such as paint booths where a

worker is immersed in a uniform flow, a much higher level of

control can be achieved by standing beside the workpiece

than standing with the back to the flow with the workpiece

between the worker and the source of suction.

119

RECOMMENDATIONS

VALIDATION OF MODEL

Before the complete model presented in table (12) is

ready for field use, it should be validated in the

laboratory at different velocities, object diameters, objectheights, and contaminant flow rates. Of particular need is

a validation of the linear relationship between De/D and Re.

Several other points are needed to adequately describe this

relationship. Additionally, the breakpoint of Re equals

10,000 between the high and low K factors should also be

validated. There were large gaps between Re in this

experiment and these gaps must be filled in before the modelis considered generally applicable.

EFFECTS OF HANDS AND ARMS

As was noted throughout this report, the presence of

the hands and arms of the mannequin present both a

theoretical and practical problem. In the plots of measured

concentration versus calculated concentration (figures (45)

through (47)) the effect of the arms and hands is obvious.

The possible presence of additional boundary layers around

the arms causes an effect on the mixing zone that could not

be accounted for in the present model. When one considers

that in a practical situation with a worker in a spray paint

booth for example, not only would arms and hands be present

but would also be moving back and forth and changing

position. Consideration of some means to incorporate these

120

effects into the model is essential if this work is to

applied to actual situation in the workplace.TURBULENT DIFFUSION EFFECTS

Since turbulent diffusion is suspected to be asignificant mechanism involved in the transport ofcontaminants in turbulent wakes, it is desirable to

incorporate this process into the model in a meaningful,theoretical way rather than to simply fit it to the datawith a mixing factor. To do this it will be essential tostudy the current literature on this subject to gain a morethorough understanding of this phenomenon. To proceed withthis research without attempting to account for turbulencewould seem to be unreasonable.

121

STUDY OF CONCENTRATION DECREASE AS A FUNCTION OF DISTANCE

Another entirely different approach to predictingbreathing zone concentration from LEV parameters issuggested by noting the decrease in concentration as thesource moves farther and farther away from the samplingprobe. Figures (48) through (59) illustrate log-log plotsof concentration versus absolute distance for the data

collected in this experiment. [Absolute distance is definedas the distance from the tip of the sampling probe to thepoint source of contaminant.] Note from these graphs thatthere appears to be a reasonably strong inverse relationshipbetween log concentration and log distance. Regressionlines give values for R squared ranging from 0.84 to 0.99.

The 95% confidence interval for the slopes of theseregression lines (table (12)) indicate that there is nostatistically significant difference between the slopes ofthe lines at 54 fpm, 167 fpm, and 292 fpm for the cylinderdata. The average value of the slope being approximately -1.6. Likewise for the mannequin data, slopes at all threevelocities are statistically the same with the exception ofthe 49 fpm line from experiment two data. For the mannequinthe average slope is about -2.4. Thus, for a given objectgeometry, the slope appears to be independent of velocity.

122

TABLE 13

95% CONFIDENCE INTERVALS FORTHE SLOPES OF THE REGRESSION LINES

CYLINDER

EXPERIMENT VELOCITYNUMBER rFPM) SLOPE 95% CONFIDENCE INTERVAL

1 54 -1.27 -1.61----0.922 54 -1.48 -1.72----1.251 167 -1.71 -1.83----1.592 167 -1.42 -1.92----0.921 292 -1.71 -1.95----1.452 292

AVERAGE-1.83

---1.57-2.47----1.19

MANNEQUIN

EXPERIMENT VELOCITYNUMBER rFPM) SLOPE 95% CONFIDENCE INTERVAL

1 49 -2.35 -2.81----1.902 49 -1.72 -2.19----1.251 152 -2.74 -3.13 — -2.362 152 -2.48 -3.07----1.901 265 -2.23 -2.53----2.042 265

AVERAGE-2.74

= -2.34-3.49----2.00

123

FIGURE 48

SF6 CONCENTRATION VS. SOURCE DISTANCECYUNDER IN TUNNEL 5+ FPM

3.2

J.I

3

2.9

2.8

Ea. 2.7a.>-• 2.6Zo 2.5p

$ 2.4Hz 2.JU

^ 2.2

8 2.1

if 2U)

z1,9

_l

1.8

1.7

1.6

1.5

1.4

P DATA POINTSLM ABSOLUTE DISTANCE (IN.)

+ REGRESSION POINTS

R e g r e s s i q n 0 u t p i.i. t .1Constant. 5n 294948Std Err at Y Est 0., 250954t< -Squared 0.853593No. of Observations 13Degr'ees o-f Freedom 11

X C;oef 11 c: i en t (s) .....1 „ 2657£BStd Err o-f Coef „ 0., 158058

124

FIGURE 49

SF5 CONCENTRATION VS. SOURCE DISTANCECYLINDER IN TUNNEL 1 67 FPM

2.8

2.6

2.4

2.2Ea.

•5 2

5 1.8H

i^ 1.62Ulo 1.4z

8 1.2U)h.at 1

z_i

0.8

0.6

0.4

0.2

a DATA POINTSLN ABSOLUTE DISTANCE (IN.)

+ REGRESSION POINTS

R e g r e s s i D n 0 n t p u t. sConstant _, 5,. 650873Std Err o-f Y Est 0.084051R Squared 0.. 989556!n|o. of 0bser^ Vat i ons 13Deqrees of Freedom 11

X CoG-ft i ci ent (s,Std Elrr of Coef

-•1 » 70903!"i fi ' ! '7 Q '" , "f

Eaa

Zg

Iz

Oz

m

z

?25

FIGURE 50

SF6 CONCENTRATION VS. SOURCE DISTANCECYLINDER IN TUNNEL 292 FPM

LN ABSOLUTE DISTANCE (IN.)DATA POINTS + REGRESSION POINTS

Reg r e s s1on 0ut p u t sConstant 4„660508Std Err a+ Y Est 0„174467F^ Squared 0.956482No., of Observations 13Degrees of Freedom 11

X L;oef + 1 c 1 en t (s ) — 1 „ 7()fS5SStd Err of Coef „ 0.109£!84

J 26

FIGURE 5/

Eaa

iijuz

810u.in

3

SF5 CONCENTRATION VS. SOURCE DISTANCEA

MANNEQUIN 1 N TUNNEL 49 FPM

3.5 -

D

nn

D

3 - \^^^2.5 - n ^\{n n

2 - ^V^^ n1.5 -

a

D

1 - "^

D ^0.5 -

n 11 1 1 1 1 1 1 1 1 1 1 1 1

1,7 1.9

D DATA POINTS

2.1 2.3 2,5 2,7

LN ABSOLUTE DISTANCE (IN.)+ REGRESSION POINTS

2.9 3.1

Regression Output.,';Constant 7» 79156 :LGitd Err o-f Y Est . 0,326626R Squared 0,922.394No,, of Qtaservati ons 13Degrees o-f F'resdom 11

X Coe-f-f i cient (s)Std lErr of Coef „

--2., 352240.205718

127

FIGURE 52

SF6 CONCENTRATION VS. SOURCE DISTANCE

aa.

Zo

f-Z111uz

8

H

z

2,5

-0,5 -

MANNEQUIN IN TUNNEL 152 FPM

D DATA POINTSLN ABSOLUTE DISTAN(£ (IN.)

+ REGRESSION POINTS

Regression Output;:C o n s t ant 7„429680Std Err of Y Est 0.277235

R Squared 0.957294i'^-Jo. of (Jb seI" Vat i on s 13

Dsqrees of F.'"-eedom t 1

X l; o e f f :i. c i e n t (e.) - 2.741 El 7Std Eirr of Coef . 0.174610

128

FIGURE 53

SF6 CONCENTRATION VS. SOURCE DISTANCEMANNEQUIN IN TUNNEL 265 FPM

Eaa^^

zo

I.ZUuz

810h.U)

2,2

2

1.8

1.6

1.4

1.2

1

0.8

0.6

0.4

0.2

0

-0.2

-0.4

-0,6

-0.8 ------1----------,------

1.7 1.9

D DATA POINTS

—T----------1----------1----------r

2.1 2.3

lU ABSOLUTE DISTANCE (IN.)+ REGRESSION POINTS

3.1

R e g r a B s i o n 0 u. t p u t ;iConstant 6.221454Stci Err o+ Y Est 0.179319R Squared 0.973832No. of Observations 13Degrees of Freedom '^ 11

X Coef f i c i en t (s) --2. 2851 1Std Err of Coef„ 0.112940

J29

FIGURE 54

SF6 CONCENTRATION VS. SOURCE DISTANCE

Eaa

iZ

z

8il.

J.4

3.3

3.2

3.1

3

2.9

2.8

2.7

2.6

2.5

2.4

2.3

2.2

2.1

2

1.9

1.8

1.7

1.6

1.5

1.7

D aCYUNDER IN TUNNEL 54 FPM

D DATA POINTS LN ABSOLUTE DISTANCE (IN.)+ REGRESSION POINTS

Regress.! on OutputsConstant 5„899558Std Err of Y Est 0,123820R Squarsd 0„9638761^1 o. ot Observations 10Degrees of Freedom 8

X Uaett :i. ci ent (s )br.d ot Coef

-1 . 482320.101457

130

FIGURE 55

£aa\^

zo5

uOz

810h.V)

SF6 CONCENTRATION VS. SOURCE DISTANCECYUNDER IN TUNNEL 167 FPM

D DATA POINTSLN ABSOLUTE DISTANCE (IN.)

+ REGRESSION POINTS

'Regression Output.!;Constant 4., 860895Std Err o-f Y E^st 0.264082R Squared 0,843299No,, of Ubservations 10Deqrees of F-'reedom 8

X C;oBt t i c: i en t (s) -1. 419£! 1Std Birr o+ Coe+ .

/37

FIGURE 56

Eaa

^-zuoz

to

M

z-I

SF5 CONCENTRATION VS. SOURCE DISTANCECYUNDER IN TUNNEL 292 FPM

DATA POINTSUN ABSOLUTE DISTANCE (IN.)

+ REGRESSION POINTS

Regr ess .1 on 0ut p ut:Constant 5.465901Std Err of Y Est 0„33£i653R S q u a r e d 0« 8 4 4 -319ix!o., o-f Observations 10Degrees of Freedom ,8

X Coef f i c i en t(s) .....1„82779Std Err of Coef. 0.277489

j^'<!

J32

FIGURE 57

Eao.»•

zo

iUlOz

810b.»

SF6 CONCENTRATION VS. SOURCE DISTANCEMANNEQUIN IN TUNNEL 49 FPM

D DATA POINTSLN ABSOLUTE DISTANCE (IN.)

+ REGRESSION POINTS

Regression Output:Constant 5.970544Std Err of Y Est 0,, 249771F; SquaiTE^d 0.898118Mcj . of 01:) seI- Vat :i. on s , 10Degrees o-f Freedom 8

X LJaet t :i. c: i en t (s) -1,. 71 £368Std Err of Coef. 0.204660

J '

J33

FIGURE 58

Ea.a.

Zo

kiOz

840ll.

(n

SF6 CONCENTRATION VS. SOURCE DISTANCEMANNEQUIN IN TUNNEL 152 FPM

HD,2

D DATA POINTSIH ABSOUUTE DISTANCE (IN.)

+ REGFESSION POINTS

Regressi c:)n Output:Con st an t 6 „ 815163Std Err of Y Est 0„310511f< Squared 0.922532No., of Observations 10Degrees of F-'reedom 8

X Coe-f t i. c: i en t (s)Std Err of Coef,

•-2.463380,. 2 5 4 4 -3 0

134

FIGURE 59

SF5 CONCENTRATION VS. SOURCE DISTANCE

Eaa

5oz

8ii.V)

z

2.5

2 -

1.5 -

1 -

0.5 -

-0.5 -

-1 -

-1.5 -

-2

MANNEQUIN IN TUNNEL 265 FPM

D DATA POINTS LN ABSOLUTE DISTANCE (IN.)t- REGRESSION POINTS

Regrss?iion OutputsConstant 7,.1.33664Std EErr of Y Est " 0„:390552R Squared 0„901783N o,. o -f t!) ta s e r v a t :i. o n s 10Deqrees o-f Freedom 8

5v

135

These results suggest a possible relationship between

concentration and distance of the form:

(21) Concentration = Ka/r'Kb

where r is the absolute distance and Ka and Kb are

constants. Kb equals about 1.6 for the cylinder and about

2.4 for the mannequin according to the data from this

experiment.

Future research along these lines could concentrate on

Ka and Kb to determine the effect on these variables brought

about by changing various parameters such as object

geometry, velocity, and source flow rate. The goal of this

research would be to develop a mathematical model that could

predict breathing zone concentration as a function of

distance and these other known parameters.

136

COMMENTS REGARDING NON-UNIFORM FLOW

Another and even more theoretically complex problem

relating to the effect of the separated boundary layer on

breathing zone concentration is introduced when the flow

around the object is no longer uniform but is rather

accelerating. This is the case in the majority of

industrial situations involving LEV.

Air flowing into a local exhaust hood is accelerating

rapidly as it approaches the hood, giving rise to a steep

acceleration gradient adjacent to the hood face. The entire

field of flow into the hood begins to fall off quickly, even

at short distances away from the hood face. The

acceleration and localized flow field phenomena were not

addressed in the uniform flow model.

Additionally, air flows into a hood from all

directions, causing velocity vectors in three dimensions as

opposed to the two dimensional uniform flow situation.

Therefore, the studies mentioned in this paper which

describe the wake downstream of a bluff body in uniform flow

are not directly applicable to a wake formed in an

accelerating flow.

Although not as well understood theoretically, a

separated boundary also occurs in accelerating flow around

an object, giving rise to a zone of reverse flow similar to

that in uniform flow. The separation is caused by a similar

mechanism as that described for uniform flow; that is, the

decelerating boundary layer does not have enough energy to

137

enter the area of increasing pressure on the downstream side

of the object and thus it separates. However, in front of a

hood it would be expected that the additional accelerationgiven to the air as it moves around the object would alterthe characteristics of the turbulent zone. It is possible

that this additional acceleration given to the decelerating

boundary layer may act to decrease the size and influence ofthis downstream vortical mixing zone [22].

A preliminary experiment was designed by Flynn [22] as

a first step in understanding the effect of this reverse

flow area on breathing zone concentrations of workers in

front of local exhaust hoods (accelerating flows). The

basic set-up of the experiment is illustrated in figure(60).

Basically, this experiment was designed to test the

effect of object size relative to hood size on the formation

of the reverse flow zone. A smooth cylinder of diameter Do

was placed in front of a local exhaust hood of diameter Dhon the hood centerline a distance z from the hood face. The

hood was operated at flow rate Q. A source of test smoke

(titanium tetrachloride as described previously) was also

placed on the centerline of the hood between the hood face

and the object. As with uniform flow, when the smoke source

was close to the hood face, all the smoke was drawn directly

into the hood. However, when the source was close to the

-f

J3«

FIGURE 60

|_-B„ H-DK

\^-

/

11

5

^-

Dh = diameter of hood opening (inches)Do = diameter of object placed in hood flowfield along

centerline (inches)Q = volumetric flow into hood (cfm) based on calibrated

orifice plateVf = face velocity of hood (fpm)z = distance from hood face to object (inches)s = distance from point of smoke generation to object when

smoke first begins to be drawn back toward object byvortex (inches)

139

cylinder, the smoke was pulled back towards the cylinder bythe reverse flow. By moving the source slowly back andforth,the point at which the smoke first began to be drawnback toward the cylinder could be estimated. This distancefrom the source to the cylinder at which backflow starts tooccur is called s. This was considered to be the depth ofthe reverse flow zone.

Appendix V tabulates values of s for variouscombinations of one hood diameter (6"), three cylinderdiameters (1.5", 6.0", and 12.0"), and five flow rates (1075cfm, 995 cfm, 910 cfm, 810 cfm, and 700 cfm). At higherflow rates, values for s were measured at z distances(centerline distance from cylinder to hood face) rangingfrom 2" to 18". However at lower flow rates, the z distanceranged from 2" to only 10" because the influence of the hoodflow field does not extend as far away from the hood facefor the lower flow rates.

A complete analysis of the data contained in Appendix Vhas not yet been conducted. However, preliminary analysissuggests that the size of the turbulent mixing zone may varyas the size of the object relative to the size of the hoodvaries. Figures (61) and (62) illustrate s/Do versus z forflow rates of 1075 cfm and 995 cfm respectively. Thisindicates that, relatively speaking, the size of the zone ismuch larger when the object is small relative to the hooddiameter and becomes smaller as the object gets largerrelative to the hood. This relationship holds true for the

140

FIGURE 61

RELATIVE ZONE DEPTH VS. DISTANCE

oa

0,7

0,6 -

0.5 -

6" HOOD — 995 CFM

a Do-1 ,5"Z INCHES

+ Do-6" O Da-12"

141

FIGURE 62

RELATIVE ZONE DEPTH VS. DISTANCE

0Q

n

0.7

0.6

0.5 -\

0.4

0,3

0.2

0.1

6" HOOD — 1075 CFM

n Do-1.5"Z INCHES

+ Do-6" O Do-12"

142

other flow rates as well. In an industrial situation, theeffect of the reverse flow phenomenon on breathingzoneconcentrations may be more significant for employeesworking in front of a large hood as opposed to a smallerhood.

Perhaps as the hood gets large relative to the object,the zone formation approaches the uniform flow situation asdiscribed previously whereas when the object is much largerthan the hood, most of the flow comes from the side ratherthan around the object. However, this explanetion isspeculative. Much empirical and theoretical work is yet tobe done to adequately explain this phenomenon.

143

REFERENCES

1. National Safety Council, Fundamentals of IndustrialHygiene. 2nd Ed., New York (1979).

2. American Conference of Governmental IndustrialHygienists, Industrial Ventilation - A Manual ofJ^ecommended Practice. 19th Ed., ACGIH, Cincinnati, OH(1986).

3. McDermott, H.J., Handbook of Ventilation forContaminant Control. Ann Arbor Science, Ann Arbor, MI(1976).

4. National Institute for Occupational Safety and Health,The Industrial Environment - Its Evaluation andControl, U.S. Government Printing Office, Washington,D.C. (1973).

5. Dalla Valle, J.M., and Hatch, T., Studies in the Designof Local Exhaust Hoods, presented at the 6th AnnualWood-Industries Meeting, Winston-Salem, NC, Oct. 15-16, 1931 of the American Society of MechanicalEngineers.

6. Silverman, L., "Centerline Velocity Characteristics ofRound Openings Under Suction", Journal of IndustrialHygiene and Toxicology. Vol. 24, No. 9, pp. 257-266.(Nov. 1942)

7. Silverman, L., "Centerline Velocity Characteristics ofNarrow Exhaust Slots", Journal of IndustrialHygiene and Toxicology. Vol. 24, No. 9, pp. 267-276.(Nov. 1942)

8. Fletcher, B., "Centerline Velocity Characteristics ofRectangular Unflanged Hoods and Slots Under Suction",Annals of Occupational Hygiene. Vol. 20, pp. 141-146(1977).

9. Fletcher, B., "Effects of Flanges on the Velocity inFront of Exhaust Ventilation Hoods", Annals ofOccupational Hygiene. Vol. 21, pp. 265-269 (1978).

10. Fletcher, B., and Johnson, A.E., "Velocity ProfilesAround Hoods and Slots and the Effects of an AdjacentPlane", Annals of Occupational Hygiene,. Vol. 25, pp.365-372 (1982).

11. Garrison, R., "Centerline Velocity Gradients for Plainand Flanged Local Exhaust Inlets", American IndustrialHygiene Association Journal^ Vol. 42, No. 10, pp. 739-746 (1981).

144

12. Garrison, R., "Velocity Calculation for Local ExhaustInlets - Empirical Design Equations", AmericanIndustrial Hygiene Association Journal. Vol. 44, No.12, pp. 937-940 (1983).

13. Garrison, R., "Velocity Calculation for Local ExhaustInlets - Graphical Design Concepts", AmericanIndustrial Hygiene Association Journal. Vol. 44, No.12, pp. 941-947 (1983).

14. Ellenbecker, M.J., Gempel, R.F., and Burgess, W.A.,"Capture Efficiency of Local Exhaust VentilationSystems", American Industrial Hygiene AssociationJournal. Vol. 44, No. 10, pp. 752-755 (1983).

15. Esmen, N.A., Weyel, D.A., McGuigan, F.P., "AerodynamicProperties of Exhaust Hoods", American IndustrialHycfiene Association Journal. Vol. 47, No. 8, pp. 448-453 (1986).

16. Flynn, M.R. and Ellenbecker, M.J., "Capture Efficiencyof Flanged Circular Local Exhaust Hoods", Annals ofOccupational Hygiene. Vol. 30, NO. 4, pp. 497-513(1986).

17. Flynn, M.R. and Ellenbecker, M.J., "The Potential FlowSolution for Air Flow into a Flanged Circular Hood",American Industrial Hygiene Association Journal. Vol.46, No. 6, pp. 318-322 (1985).

18. Roach, S.A., "On the Role of Turbulent Diffusion inVentilation", Annals of Occupational Hygiene. Vol. 24,No. 1, pp. 105-112 (1981).

19. Fletcher, B. and Johnson, A.E., "The Capture Efficiencyof Local Exhaust Ventilation Hoods and the Role ofCapture Velocity", Ventilation '85 - Proceedings of the1st International Symposium on Ventilation forContaminant Control, Toronto, Canada, Oct. 1 - 3, 1985.

20. Heinson, R.J. and Choi, M.S., "Advanced Design Methodsin Industrial Ventilation", Ventilation '85 -Proceedings of the 1st International Symposium onVentilation for Contaminant Controll Toronto, Canada,Oct. 1-3, 1985. y

21. White, F.M., Fluid Mechanics. 2nd Ed., McGraw - HillBook Company, New York (1986). \

22. Flynn, M.R., "The Impact of Separation on Exposure andHood Capture", Grant Proposal, National Institute forOccupational Safety and Health (1987).

145

23. Schlichting, H., Boundary-Layer Theory. 7th Ed.,McGraw-Hill Book Company, New York (1979)

24. Roshko, A., "On the Development of Turbulent Wakes fromVortex Streets", NACA Rep. 1191 (1954).

25. Fage, A. and Johansen, F.C., "The Structure of VortexStreets", Phil. Magazine. Vol. 5, No. 28, pp. 417-441(1928) .

26. Bloor, M.S., "The Transition to Turbulence in the Wakeof a Circular Cylinder", Journal of Fluid Mechanics,Vol. 19, part 2, pp. 290-304 (1964).

27. Bearman, P.W., "On Vortex Shedding from a CircularCylinder in the Critical Reynolds Regime", Journal ofFluid Mechanics. Vol. 37, part 3, pp. 577-585 (1969).

28. Achenbach, E. and Heinke, E., "On Vortex Shedding fromSmooth and Rough Cylinders in the Range of ReynoldsNumbers 6X10'3 to 5X10*6", Journal of Fluid Mechanics.Vol. 109, pp. 239-251 (1981).

29. Ljungqvist, B., "Some Observations on the InteractionBetween Air Movements and the Dispersion of Pollution",Document D8, Swedish Council for Building Research,Stockholm, Sweden (1979).

30. Hampl, V. and Hughes, R.T., "Improved Local ExhaustControl by Directed Push-Pull Ventilation Systems",American Industrial Hygiene Association Journal, Vol.47, No. 1, pp. 59-65 (1986).

31. Van Wagenen, H.D., "Assessment of Selected ControlTechnology Techniques for Welding Fumes", ResearchReport, National Institute for Occupational Safety andHealth, Cincinnati, OH (1979).

32. Occupational Safety and Health Administration GeneralIndustry Standards 29 CFR 1910.94c6i, Table G-10,(1985).

33. Fitzgerald, M.L., "Comparison of Three DimensionalVelocity Models for Flanged Rectangular Hoods",Master's Technical Report # 1087, University of NorthCarolina, Chapel Hill, NC

34. Lilley, D.B., "Using A Tracer Gas Technique to EvaluateLaboratory Hood Effectiveness", Master's TechnicalReport # 1061, University of North Carolina, ChapelHill, NC

"x

\

146

35. Humphries, W. and Vincent, J.H., "An ExperimentalInvestigation of the Detention of Airborne Smoke in theWake Bubble Behind a Disk", Journal of Fluid Mechanics.Vol. 73, part 3, pp. 453-464 (1976).

36. Humphries, W. and Vincent, J.H., "Experiments toInvestigate Transport Processes in the Near Wakes ofDisks In Turbulent Air Flow", Journal of FluidMechanics. Vol. 75, part 4, pp. 737-749 (1976).

37. Humphries, W. and Vincent, J.H., "Near Wake Propertiesof Axisymmetric Bluff Body Flows", Applied ScientificResearch. Vol. 32, pp. 649-669 (1976).

38. Vincent, J.H., "Scalar Transport in the NearAerodynamic Wakes of Surface-Mounted Cubes",Atmospheric Environment, Vol. 12, pp. 1319-1322 (1978).

39. Vincent, J.H., "Model Experiments on the Nature of AirPollution Transport Near Buildings", AtmosphericEnvironmentf Vol. 11, pp. 765-774 (1977).

40. Gerrard, J.H., "The Mechanics of the Formation Regionof Vortices Behind Bluff Bodies", Journal of FluidMechanics. Vol. 25, pp. 401-413 (1966).

APPENDIX I

CALIBRATION OF THE WILKS MIRAN-I GAS ANALYZER

INTRODUCTION

A Wilks Mobile Infrared Analyzer, Model 5652, was used

in this experiment to detect concentrations of the test gas.

The MIRAN is a portable, single beam spectrophotometer that

operates in the infrared region of the electromagnetic

spectrum. It is equipped with a circular variable filter

which can scan the spectrum from 2.5 to 14.5 microns in

either a manual or an automatic mode. The volume of the

Teflon coated sample cell is 5.5 liters and its base length

is 0.75 meters. However, by using a system of gold coated

mirrors, pathlengths of up to 20.25 meters (in multiples of

0.75 meters) can be obtained.

The instrument is well suited for quantitative analysis

because of its ability to accurately measure sample

absorption at a particular wavelength. Due to individual

molecular make-ups, each compound gives a unique absorption

pattern when a plot of transmission (or absorbance) versus

wavelength is obtained. The most suitable specific

wavelength to use in the analysis of a particular compound

is determined from various factors such as strength of

absorption at that wavelength and resolution of the

transmission peak. A table of optimum analytical

wavelengths for a host of compounds is published by the

manufacture of the instrument. The most suitable pathlength

-^to use for a certain concentration range can also be

determined from these tables.

The basic method of analysis is to measure the

transmission of infrared radiation through the sample cell

first with sample present and then with no sample in the

cell.. The ratio of these two measurements can be used to

find the concentration of sample. This is typically

accomplished by injecting a series of known sample

concentrations into the cell and constructing a calibration

curve for the sample. The calibration curve should be the

straight line predicted by the Beer-Lambert Law:

A = l/lo = e'^^^where I = signal with sample

lo = signal with no sample

A = sample absorbance

1 = cell pathlength

c = sample concentration

°C= absorption coefficient (characteristic of sample

and units chosen for c and 1)

The concentration of sample in the cell should be

proportional to the reading on the absorbance scale of the

meter.

It should be noted that the linear relationship of the

Beer-Lambert Law breaks down at higher sample concentrations

as molecules begin to shield each other from the light

source. This causes a flattening of the absorbance versus

concentration curve.

^

[The Wilks MIRAN Operations Manual was consulted as the

source for the general information included in this

Appendix]

PROCEDURE

For the purposes of this experiment, sulfur

hexafluoride (SF6) was selected as the test gas. One

calibration curve was obtained for each of the followingconcentration ranges; 0-1.5 ppm, 0-10 ppm, 0 - 100 ppm.

For the 0-1.5 ppm curve, sixteen points were obtained.

For both the 0-10 ppm and the 0 - 100 ppm curves, twelvepoints each were plotted.

For each point a series of three injections of known

concentration of SF6 was made with a gas tight syringe. Theabsorbance was recorded for each injection. The mean of the

three injections was taken as the average absorbance for

that concentration. The sample cell was flushed completelybetween each injection.

Attachments (1), (2), and (3) illustrate the

calibration curves for the ranges 0-1.5 ppm, 0-10 ppm,and 0 - 100 ppm respectively. Due to the curve flattening

phenomenon mentioned earlier, the latter two curves were

plotted as log concentration versus log absorbance in order

to obtain straight lines. A linear regression was carried

out on the data so that equations for concentration as a

function of absorbance could be utilized to easily calculateconcentrations.

a.a.

Oz

8

MIRAN CALIBRATION CURVE - SF6ABSORBANCE < 0.1 A.U.

DATA POINTSABSORBANCE (A.U.)

+ REGRESSION POINTS

Ml RAN CALIBRATION CURVE - SF6ABSORBANCE - 0.1 - 0.3 A.U.

1.1 -

1 -D

0.9 - r^ 10.8 -

'•^

?aa

0.7 -y^^

0.6 -

0.5 -^/ n

g 0.4 ->/^a

O

0.3 -

If 0.2 -m

0.1 -

0 - y^

n y^-0.1 -

__ri o —

^—U.z ~ 1 1 1 1 1 1 r 1 1

-1.3 -1.1 -0.9 -0.7 -0.5

a DATA POINTSABSORBANCE (A.U.)

+ REGRESSION POINTS

Eaa

oz

8

-0.6

MIRAN CALIBRATION CURVE - SF6ABSORBANCE > 0.3 A.U.

DATA POINTSABSORBANCE (A.U.)

+ REGRESSION POINTS

APPENDIX II

CALIBRATION OF THE SULFUR HEXAFLUORIDE fSF6) METERING SYSTEM

The flow rate of sulfur hexafluoride was measured by a

rotameter connected in line between the cylinder and the

diffuser. The system was calibrated by inserting the

diffuser into the end of a short section of rubber hose.

The other end of the hose was placed over an inverted 50 ml

buret (soap bubble meter). Thus, the flow rate of SF6

through the rotameter was calibrated with the same system as

used in the experiment. Please note attachment (1) for a

schematic of the system. The rotameter calibration curve is

provided as attachment (2).

y.

ROTAMETER CALIBRATION CURVEAPPARATUS CAUBRATED IN TACT

E

E

t

c111

111-IoaoDm

800

700 -

600 -

SOO

400 -

300 -

200 -

100 -

DATA POINTSROTAMETER READING

+ REGRESSION POINTS

APPENDIX III

DESCRIPTION OF THE METROSONICS dl-714 ANALOG DATALOGGER

INTRODUCTION

The voltage output of the MIRAN was logged and

integrated by a Metrosonics dl - 714 Analog Datalogger. The

Datalogger is a multi-purpose, microcomputer based

instrument capable of logging voltage, current, and/or

temperature data. The Datalogger has a variety of features

and functions. The features utilized in this project will

be briefly described.

PROGRAMMABLE CHANNEL INPUT

The Datalogger can be programmed in either a four

channel differential mode or an eight channel single ended

mode. That is, it can simultaneously receive input from

four or eight sources. Each channel may be programmed

independently to receive voltage, amperage, or temperature

data (such as from a thermocouple). For this project, only

one channel was utilized. The Datalogger was set to match

the MIRAN output of 0 - 1 volt.

PROGRAMMABLE LOGGING PERIOD

The logging period is the period of time during which

the Datalogger is actually recording samples. In the manual

logging mode, the Datalogger begins logging when the

LOG/STBY key is pressed and ceases logging when it is

pressed again. In the autostop logging mode, the Datalogger

is programmed to log samples for a specific user defined

time period and then automatically stop.

PROGRAMMABLE SAMPLING RATE

The sampling rate is the rate at which samples are

recorded when the Datalogger is in the logging mode. The

programmable options are: 2/sec, 1/sec, 4/min, l/^in, 4/hr,and 1/hr.

PROGRAMMABLE AVERAGING PERIOD

Perhaps the most useful feature of the Datalogger is

its ability to average the samples recorded over a

programmable period of time. The averaging period can be

programmed in one second intervals anywhere from one second

to twelve hours. During the averaging period the Datalogger

considers each sample taken, averages them, and reports a

minimum, maximum, and average value for that period.EXAMPLE

To illustrate the above mentioned features, consider

the following parameters as programmed in this experiment:

Sampling Rate: 1/sec

Averaging Period: 10 sec

Logging Time: 10 minutes

For these values, the Datalogger records MIRAN output

for ten minutes once the LOG/STBY key is depressed. During

that ten minutes the Datalogger samples and records the

MIRAN output every second. For every ten second period

during that ten minutes (a total of 60 periods) the

Datalogger will have ten values that it has recorded, one

for each second. It will average these samples and reportthe minimum, maximum, and average values.

SAMPLE OUTPUT

When the appropriate software is used, the Dataloggercan "dump" the recorded data into a personal computer for asubsequent hard copy report. The following report types areavailable; overall statistics, time history, amplitudedistribution, raw data, and multiple channel.

For this project, the time history report was the mostuseful. This report gives the minimum, maximum, and averagevalues for each averaging period of the logged data. Agraphical representation is also provided by the computerwhich graphs a minus (-) sign for the period's minimumvalue, a plus (+) sign for the period's maximiim period, andan asterisk (*) for the period's average value. The graph'saccuracy and resolution depend on the variation (spread) ofthe test data. Please note attachment (1) for an example ofa time history report.

"TEST START DATE:

"TEST START TIME:

" TEST LOCATION:" EMPLOYEE NAME:

"EMPLOYEE NUMBER:DEPARTMENT:

^j^~iMMENT FIELD 1:W'MMENT FIELD 2:" NUMERIC CODES:

08/03/1988"

13:30"

CHADYRU - PUBI..IC HEALTH"

DENNIS K. GEORGE"

419-aB-t-J297"

ESE ~ INDUSTRIAL HYGIENE"

REPEAT OF CYLINDER AT 150"

SF6 FLOW = 15"

"METROSONICS dl~714 SN 1222 VI.8 11/87"

"CURRENT DATE: 08/04/89"

"CURRENT TIME: 08:41:44"

"TEST STARTING DATE: 8/ 3/88""TEST STARTING TIME: 13:31:21"

"ELAPSED TIME: 0 DAYS 0:50: 0""SAMPLE RATE: 1/sec"

TIME HISTORY FOR CHANNEL 1

CALIBRATION POINTS0. 0000 V —

1.0000 V =

ALARM VALUES

LOWER: 1.192E-07 ???UPPER: 1.192E~07 ???

0.0000

1.0000

PERIOD LENGTH: 0: O:10

!=• PERIODS COMBINED:

MIN AVG MAX

DATE:

o

0

o

0

o

0

0

o

0

8/ 3/88

3270

3290

3259

3230

3206

,3145

3055

2978

, 2886

TIME: 13:31:21 TAG #:

O.2809

2754

2717

2711

2732

2746

2736

0.2712

3334

3338

3274

3240

O. ?69;

n.75

:.094

liOlO

^047

0.2780

0.2740

2726

2728

2737

2755

0.2748

0.2725

0.2714

3374

3376

3297

3263

3248

3201

3133

3057

2968

2885

2812

2765

<? -y t; Cy

0.2741

O,. 2748

0.2765

0„2766

0.2766

0.2737

0.2729

o,

o,

0.

0

o

0

0

0

o

o

0

0

o

- *

-*+

—V

—*+

—*+

8/ 3/88 TIME: 13:

0.2693

0.2668

0.2631

0.2571

0.2499

0.2426

0. 2671

0. 2654

0. 2612

0. 2551

0. 2461

0.2396

34:41 TAG

0.2712

0.2688

0,2650

0.2601

0.2550

0.2451

#:

—h

-* +

-* +

@

0.24260.251.3

0.24560.25430.2621

O.24870.25870.2642

DATE: 8/ -3/88 TIME: 13:50:18 TAG #:U.

O.

0.0.0.o.0.o.0.o.0.

o,0,0,00,0oo

25782597"^ 6 3 '^-'26092574

2487243824012404

0.258628002948303030353006.;. 96629192885

0-25960.25950.26170.26390.26200.25880.25510.250800O

246324182500268128862989303730473018298629492913

u

o0o0oo0oooC)000o000o

26152612263426462627260725662534248024362579277529463028305330563029301829732934

---+•

— -*-+

--* -+•

—#

DATE: 8/ 3/88 TIME: 13:

@

0,28530-28520.28290. 280-30.27970.29280-30770-3215O.32520.3283

0.28700.28620.28500.28230.28420-29950-3166O-32340.32760.3326

5:38 TAG0-28830-28770.28620-28540,2915O.3060O.32120.32580.32960,3348

-*

-»-+-

DATE:I 8/ 3/880.26690.26110.25600,25430-25190-25010-24940-24890.24720-24670.24580,24640,24760-24600.24430.24760.2595O.26930-27080.2695

TIME:O.26850-26410.25860.25560.25330.25140.25010.24920.24880,24780-24740.24770.24850,24730-2463

14:

0. ?5310-26430-2717

0,2707

8:56 TAG0.2700

267126032569

0.25440.25300.

o.

o.o.

u514501

0.24970,24910,24900.24870.2498

2494Z480

o,

O u0.0.o.0.0.272;

269127412731

—^

--+

-*

•+

DATE: 8/ 3/880.26460.26100,25770.2569

TIME: 14: i:0.26710,26360,25870,2581

!: 16 TAG #:0.26850,26600.26170-2589

APPENDIX IV

CONC VS DISTANCE

REBRESSION DATA

EXPERIMENT ONE - CYLINDER

54 FPM INITIAL CONC. AS Co

Regression Output:Constant

Std Err of Y Est

R SquaredNo. of Observations

Degrees of Freedom

X Coefficient(s)

Std Err of Coef.

3. 06(:>496

0. 4i: 909

0. 60Z.649

s 13

11

-0.41807

0. lo;:i42

EXPERIMENT ONE - MANNEQUIN

152 FPM INITIAL CONC. AS Co

Regression Output:Constant

Std Err of Y Est

R SquaredNo. of Observations

Degrees of Freedom

2.577530

0.774584

0,666629

13

11

X Coefficient(s)

Std Err of Coef,

-0.89865

0.191610

EXPERIMENT ONE ~ CYLINDER

292 FPM MAXIMUM CONC. AS Co

Regression Output:Constant 1.

Std Err of Y Est 0.

R SquaredNo. of Observations

Degrees of Freedom

810686

225870

0.9: 7061

13

11

X Coefficient<s)

Std Err of Coef,

-0.66066

0.055874

EXPERIMENT TWO - CYLINDER

54 FPM INITIAL CONC. AS Co

Regression OutputsConstant. 3.313068Std Err of Y Est 0.314470

R Squared 0.766993Ho. of Observations 10

Degrees of Freedom 8

X Coefficient(s) -0.46364

Std Err of Coef, 0.090349

EXPERIMENT ONE - CYLINDERl<b7 FPM INITIAL CONC. AS Co

Regression Output:Constant 2,710482Std Err of Y Est 0.354691R Squared 0.814014No- of Observations 13Degrees of Freedom il

X Coefficient(s) -0.60879Std Err of Coef, 0.087740

EXPERIMENT ONE - MANNEQUIN265 FPM INITIAL CONC. AS Co

Regression Output:Con st an t 2.201013Std Err of Y Est 0,604198R Squared 0,702925No. of Observations 13Degrees of F'reedom 11

X Coe-ff icient (s) -0.76251Std Err of Coef. 0.149461

EXPERIMENT ONE - MANNEQUIN49 FPM MAXIMUM CONC. AS Co

Regression Output:Constant 5,060071Std Err of Y Est 0.413796R Squared 0.882762No. of Observations 11Degrees of Freedom 9

X Coefficient(s) -1.39697Std Err of Coef. 0.169698

EXPERIMENT TWO - CYLINDER167 FPM INITIAL CONC. AS Co

Regression Output:Constant 2.482728Std Err of Y Est 0,211916R Squared 0,899092No- of Observations 10Degrees o-f Freedom 8

X Coefficient's) -0.51403Std Err of Coef. 0.060884

EXPF£RIMEIMT ONE - CYLINDER

292 FF'II INITIAL CONC. AS Co

Regression Output:Constant 1.810686

Std Err o-f Y Est 0.225870

R Squared 0.927061No. o+ Observation 13

Degrees o-f Freedom 11

X Coe-f-f i ci ent (s) -0.66066

Std Err of Coef- 0.055874

EXPERIMENT ONE -- CYLINDER

54 FPM MAX IHUM CONC, AS Co

Regression OutputsConstant

Std Err of Y Est

R SquaredNo. of Observations

Degrees of Freedom

4.736741

0.239658

0.878546

s 9

7

-1.12424

0.157992

X Coefficient(s)

Std Err of Coef.

EXPERIMENT ONE - MANNEQUIN

152 FPM MAXIMUM CONC. AS Co

Regression Output:Constant 4.272520

Std Err of Y Est 0.397921

R Squared 0,917495No. of Observations 11

Degrees of Freedom 9

X Coef fici ent (s) --1.63257

Std Err of Coef. 0,163188

EXPERIMENT TWO - CYLINDER

292 FPM INITIAL CONC. AS Co

F;egre5sion Output:Constant 2.410440

Std Err of Y Est , 0.254802

R Squared 0.911868No. of Observations 10

Degrees of Freedom 8

X Coefficient(s) -0.66603

Std Err of Coef. 0.073206

EXPERIMENT ONE - MANNEQUIN49 FPM INITIAL CONC. AS Co

Regression OutputsConstantStd Err of Y EstR SquaredNo. o-f ObservationsDegrees o-f Freedom

X Coefficient(s)Std Err of Coef.

EXPERIMENT ONE -167 FPM MAXIMUM

y, .6:13S6S0 . 6B6S.390 .6?16S36

ns

m

1311

-0.779610.169904

CYLINDERCONC. AS Co

Regression Output:Constant 3,072344Std Err of Y Est 0.267357R Squared 0,892396No. of Observations 12Degrees of Freedom 10

X Coefficient<s) -0.76992Std Err of Coef, 0.084543

EXPERIMENT ONE ~ MANNEQUIN265 FPM MAXIMUM CONC. AS Co

Regression Output;Constant 4. lOrStd Err of Y EstR SquaredNo. of ObservationsDegrees of Freedom

;8160.2259420.957649

108

X Coefficient(s)Std Err of Coef.

-1.567390,116536

EXPERIMENT TWO - MANNEQUIN49 FPM INITIAL CONC. AS Co

Regression Output:Constant

Std Err of Y Est 0R Squared ONo. of ObservationsDegrees of Freedom

2.882279,520805557042

108

X Coefficient(s)Std Err of Coef.

-0.474600.149630

EXPERIMENT TWO - MANNEQUIN152 FF'M INITIAL CONC. AS Co

Regression Output:Constant 2.337873Std Err of Y Est 0,744445R Squared 0„554721No. of Observations 10Degrees of Freedom 8

X Coefficient(s) -0.67521Std Err of Coef. 0.213883

EXPERIMENT TWO - MANNEQUIN292 FPM MAXIMUM CONC. AS Co

Regression Output:Constant 2.410440Std Err of Y Est 0.254802R Squared 0.911868No. of Observations 10Degrees of Freedom 8

X Coefficient(s) -0,66603Std Err of Coef. 0.073206

EXPERIMENT TWO - MANNEQUIN265 FPM INITIAL CONC. AS Co

Regression Output:Constant 2.190597Std Err of Y Est 0.841548R Squared 0.543977No. of Observations 10Degrees of Freedom 8

X Coefficient<s) -0.74690Std Err of Coef. 0.241782

EXPERIMENT TWO - MANNEQUIN49 FPM MAXIMUM CONC. AS Co

Regression Output:ConstantStd Err of Y EstR SquaredNo. of ObservationsDegrees of Freedom

X Coefficient(s)Std Err of Coef.

0.. 8707940. 3202230. 857881

s 8h

-0 .95.:3730. 158476

3. 8086680, 4706880. 854493

s 86

-1 .382720. 232941

EXPERIMENT TWO - CYLINDER54 FPM MAXIMUM CONG. AS Co

Regression Output:Constant 3.943893Std Err of Y Est 0.128772R Squared 0.960585No. of Observations SDegrees of Freedom 6

X Coef f i c i en t < s > -0.77063Std Err of Coef. 0.063728

EXPERIMENT TWO - MANNEQUIN152 FPM MAXIMUM CONC. AS Co

Regression OutputsConstantStd Err of Y Est

R SquaredNo. of Observations

Degrees of Freedom

X Coefficient(s)Std Err of Coef.

EXPERIMENT TWO - CYLINDER167 FPM MAXIMUM CONC. AS Co

Regression Output:Constant 2.482728Std Err of Y Est 0.211916R Squared 0.899092No. of Observations 10Degrees of Freedom 8

X Coefficient(s) -0.51403Std Err of Coef. 0.060884

EXPERIMENT TWO - MANNEQUIN265 FPM MAXIMUM CONC. AS Co

Regression Output:Constant 3.775568Std Err of Y Est 0.585499R Squared 0.818931No, of Observations 8Degrees of Freedom 6

X Coefficient(s) -1.50944Std Err of Coef. 0.289761

APPENDIX V

RAW DATA FROM CYLINDER AND HOOD EXPERIMENT

*

Dh (in.) Q (c-fm) Do (in.) z (in.) s (in. )

6 1075 1.5 2 0.438h 1075 1.5 3 0.5636 1075 1.5 4 0.375

6 1075 1.5 5 0.625h 1075 1.5 6 0.56 1075 1.5 7 0.6256 1075 1.5 8 0.6256 1075 1.5 9 0.6896 1075 1.5 10 0.8756 1075 1.5 12 1

6 1075 1.5 14 0.875

6 1075 1.5 16 1

6 1075 1.5 18 1

6 1075 6 3 0.5

6 1075 6 4 0.625

6 1075 6 5 0.75

6 1075 6 6 0.75

6 1075 6 7 0.813

6 1075 6 8 0.813

6 1075 6 9 1.125

6 1075 6 10 1.125

6 1075 & 12 1.125

6 1075 6 14 1.25

6 1075 6 16 1

6 1075 12 3 0.813

6 1075 12 4 0.75

6 1075 12 5 0.875

6 1075 12 6 0.813

6 1075 12 7 0.938

6 1075 12 8 0.875

6 1075 12 9 0.8136 1075 12 10 1

6 1075 12 12 0.938

6 1075 12 14 0.875

6 995 1.5 2 0.75

6 995 1.5 3 0.688

6 995 1.5 4 0.625

6 995 1.5 5 0.813

6 995 1.5 6 0.875

6 995 1.5 7 0.813

6 995 1.5 8 0.875

6 995 1.5 9 0.875

6 995 1.5 10 0.938

6 995 1.5 12 0.938

6 995 1.5 14 0.875

6 995 6 3 0.688

6 995 6 4 0.75

6 995 6 5 0.875

6 995 6 6 0.938

6h6h66hh66666666a6h666666666666

66666

666

66

&

6666666

995995995995995995995995995995995995995995995995910910910910910910910910910910910910910

910910910910910910910910910910910910910910910910910910810810910810810810810810

666666

12121212121212121212

1.51.51,51.51.51.51.51.51.51.51.5

666

h6666612121212121212121212

1.51.51.51.51.51.51.51.5

7 18 19 1. 06310 1. 18812 1.37514 1. 3753 0.8134 1. 1255 16 0.757 0.8753 1, 0639 1

10 1.2512 1.37514 1.52 0.75•2' 0. 6884 0.755 0.6886 0. 9387 0.758 0.759 0.875

10 0.93812 0. 93814 0.875O' 0.6884 0.755 0.8136 0.9387 0.8758 1. 0639 1. 12510 1. 06312 1. 12514 1.25

~Z 0.754 0.755 0.756 0.8757 0.8758 19 0.87510 0.87512 1. 12514 1 . 063

0.8133 0.754 0. 755 0.756 0.757 0.758 0.8759 1

^p^

6

6

6

6

6

666

6

66

6

6

h6

66

6

6h

6

66

6t)

6

6

6

6

6

6

6

6

666

h666

66h

b

666

810810810810810810810810810

810810810

810

810810810810

810810

810700

700

700

700700

700700700

700

700

700700700

700700700700

700700700700

700700700

700

700

700700

700

1.5

1.51.5

6

6

6

666666

12121212121212

1.51.5

1.51„51.5

1.51

1

1

11

. .J

.5

.5

.5,5666666

66

6a

121212

1212

12

12

12

10 0.87512 1. 06314 1. 188y. 0,754 0.755 0.8136 0.8137 0,8758 0.8759 0.87510 0. 93812 1. 06314 1.253 0.754 0.8755 0.8756 0.8137 0.8138 19 0.875r^ 0.813-^r 0.6254 0.8135 0.8756 1

7 0.8758 0. 759 0.87510 1

12 0,62514 0.6883 0.8754 0.8755 1. 1256 1. 1257 18 0. 9389 1.12510 1. 06312 1. 12514 13 0.754 0, 755 0,75

6 0.9387 0.875a 0.8759 0.87510 0,75

m