Velocity Determination............................... 41 Test Object Description.............................. 48

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 ZONE

CONCENTRATION........................................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..............................40
Velocity Determination...............................41
Test Object Description..............................48
Sulfur Hexafluoride Generation.......................50
Determination of Zone Depth: Visualization of
Test Smoke......................................51
Determination of Zone Depth: Concentration Versus
Distance Curves.................................53
Determination of Zone Depth: Cbz = 0.5Co............74
Determination of Zone Depth: Calculation from
Theory..........................................74
MANNEQUIN VERSUS MANNEQUIN 90 DEGREES.....................80
DISCUSSION OF MODEL EVALUATION............................87

Discussion of Test Smoke Observation.................87
Discussion of Actual and Theoretical Concentration
Versus Distance Curves..........................88
Discussion of Mannequin Versus Mannequin 90 Degrees..91
Discussion of the Model's Ability to Predict De......92
11
Discussion of the Model's Ability to Predict
Mixing Zone Concentrations......................93
CONCLUSIONS..............................................112
Test Smoke Observation..............................112
Theoretical Model...................................112
Mannequin Versus Mannequin 90 Degrees...............118
RECOMMENDATIONS..........................................119
Validation of Model.................................119
Effects of Hands and Arms...........................119
Turbulent Diffusion Effects.........................120
Study 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
SOURCE AIR PATH RECEIVER

.4------ ---------ͨ 4—----------------------------------__ͨ ^---------------------------------------^
Dip }
P^
'
r
Tank
/ ^'v-^x/ ^^
1.SU BSTITUTION VVITH A 1. HOUSEKEEPING 1 TRAINING & EDUCATION
LESS HARMFUL MATERIAL (IMMEDIATE CLEANUP) (MOST IMPORTANT)
(WATER IN PLACE OF
ORGANIC 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 CONDITIONED
3. ENCLOSURE OF PROCESS (SUPPLIED AIR) CRANE CABS)
(GLOVEBOX)
4. INCREASE DISTANCE 4 PERSONAL MONITORING
4. ISOLATION OF PROCESS BETWEEN SOURCE AND DEVICES (DOSIMETERS)
(SPACE OR TIME) RECEIVER (SEMI-AUTOMATIC
OR REMOTE CONTROL) 5 PERSONAL PROTECTIVE
5. WET METHODS
DEVICES (RESPIRATORS)
(HYDRO BLAST) , 5. CONTINUOUS AREA
MONITORING (PRESET 6 ADEQUATE MAINTENANCE
6. LOCAL EXHAUST ALARMS) PROGRAM
VENTILATION
(CAPTURE AT SOURCE) 6. ADEQUATE MAINTENANCE
PROGRAM
7. ADEQUATE MAINTENANCE
PR OGRAM
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 the
process.
FIGURE 2
Stack
Hopper
Âir cleaner
Barrel filling
operation
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 material
conveyed, temperature, and plant layout, for example.
The function of the air cleaning device is to remove
the contaminant from the air stream before it is exhausted
to the outside environment. Many different types of

cleaners exist and proper selection depends on concentration
and particle size of contaminant, degree of collection
required, characteristics of the air or gas stream,
characteristics of the contaminant, energy requirements, and
method 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 the
system. At the end of the LEV design procedure, a specific
fan is selected that will move the required amount of air
against the necessary pressure differential.
LEV has been utilized in industry since early in the
twentieth century. However, the basic parameters used to
design these systems have changed very little over the last
fifty years. Presently, new concepts are being explored
that may significantly improve the current state of LEV
design with the ultimate goal of providing the best possible
protection 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 Dispersion
of Contaminant
Examples
Capture Velocity, fpm
Released with practically no
velocity into quiet air. Evaporation
etc.
from tanks; degreasing, 50-100
Released at low velocity into

moderately still air. Spray booths; intermittent container 100-200
filling; low speed conveyor transfers;
welding; plating; pickling
Active generation into zone of Spray painting in shallow booths;
rapid air motion 200-500
barrel filling; conveyor loading;
crushers
Released at high initial velocity Grinding; abrasive blasting, tumbling 500-2000
into zone of very rapid air motion.
In each category above, a range of capture velocity is shown. The proper choice of values depends on
several factors:
Lower End of Range Upper End of Range

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

aspect ratios (width/length) of
0.2 or greater, or round, and,
(2) Q = 0.75V(10x^+ A) for flanged rectangular ducts of

aspect ratios of 0.2 or greater,
or round
where Q = quantity of air required

to achieve the necessary
capture 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 aspect

ratios less than or equal to
0.2, and,
(4) Q = 2.6LVX for flanged slots of aspect

ratios less than or equal to 0.2
where L = length of slot (feet)
V = centerline capture velocity (fpm)
X = distance from hood face to point

in ceterline where capture velocity
is 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.
,0-05
^005
0 50.
LOlO
010 J
w
L
0 05 J
L0 50
O01_J
,100
0005J
SOURCE: REFEREWCE S '

ͣ^-^-"'gyj«P^>i» -'-.j'-'i?!».rf!W!?^'!!IBgHg^^^g^--
12
Additionally, to maintain the same velocity at a given

distance in front of the hood, less air is required when the
hood rests on a plane than when it has no obstruction.
Within the last decade. Garrison [11] has evaluated the
work of Dalla Valle and Silverman using much smaller inlets
and higher velocities, i.e., high velocity/low volume
systems. Among other things, he concluded that the
empirical equations published in the Ventilation Manual [2]
from Dalla Valle's and Silverman's work were generally
appropriate. However, he disagreed with the flat 33 per
cent increase in centerline velocity velocity attributed to
flanging the circular or rectangular inlets as is
recommended in the Manual. His data indicate that a more
realistic centerline velocity increase would be on the order

of 20 to 3 0 per cent. He also recommended that Silverman's
equations be restricted to centerline distance/hood diameter
(or hood width) ratios of 0.4 or greater because, as was
mentioned previously, as x approaches 0, V becomes
indeterminate.
In a later paper Garrison [12] provides the following

non-dimensional equations that describe centerline velocity
gradients in terms of distance, inlet end shape, and
flanging:
(5) Y(near) = a(b) ^''^

(6) Y(far) = a(Xdw)^
where Y = centerline velocity/hood face velocity
Xdw = centerline distance/hood diameter (or hood width
for rectangular hoods)
"''.- :^jf'- '- -J-
Empirical Design Data for Nondimensional Centeriine Velocity Gradients

Y ^ a (b)XDW
Y = a(XDw )'
Nozzle Nozzle Specific
End Profile 0 < Xdw < 0.5 0.5 < Xdw <1.0 Y Values
1.0< Xdw S Xdw
Shape Shape at Xdw =
a b a b a b a b
Xdw^^ 0.5 1.0
Plain 110 0.06 .,
..
8 -1.7 8 -1.7
Flanged 110 1.5 26 8
Circular 0.07 10 -1.6 10 -1.6 1.5
Flared 90 0.20 90 30 10
0.20 18
Rounded -1.7 2.0 40 18
98 0.50 145 0.23 --
"
33 -2.2 2.5 69 33
Square Plain 107 0.09 —
.-
10 -1.7 10 -1.7
(WLR-1.0) Flanged 107 0.11 --
1.5 32 10
--
12 -1.6 12 -1.6 1.5 36 12
Rectangular Plain 107 0.14 —
—
18 -1.2 18 -1.7
(WLR=0.50) Flanged 107 0.17 --
2.0 41 18
--
21 -1.1 21 -1.6 2.0 45 21
Rectangular Plain 107 0.18 —
—
23 -1.0 23 -1.5
{WLR:0.25) Flanged 107 0.22 --
2.5 46 23
--
27 -0.9 27 -1.4 3.0 50 27
Narrow slot Plain 107 0.19 —
..
24 -1.0 24 -1.2
(WLR=0.10) Flanged 107 0.22 --
3.5 48 24
--
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 to
account 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 the
contaminant generation area.
Roach [18] states that "...it is inadvisable to design

the ventilation of an exhaust hood so as merely to produce a
standard capture velocity or standard entrance velocity, as
the velocity chosen may be excessively high or, what would
be worse, not high enough."
Fletcher and Johnson [19] demonstrated that the removal

of gases and submicron particles released at low velocities
on the centerline of LEV hoods can be predicted by the
traditional concept of capture velocity. However, when the
direction 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 is
needed at the source to entrain the contaminant than that
published in the Ventialtion Manual [2].

Heinson and Choi [20] list the following flaws in the
current design methods:
(1) Contaminant concentration in the vicinity of the

source cannot be predicted.
(2) The effect of changes in design (such as system

dimensions or volumetric flow rate) on the performance of a
system cannot be estimated.
(3) Even though the performance of a particular system

is known, the effect of geometrically scaling it up or down
in 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 are
experimentally 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 VERSUS

BREATHING 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 per
cent. However, most spray painters are required to wear
respiratory protection because a significant quantity of
contaminant passes through their breathing zone before being
removed.
Clearly, a method of LEV design that somehow relates

design parameters to breathing zone concentration would be
most useful in protecting employees. However, before such a
model can be developed, a fundamental interaction must be
investigated; that of the worker with the flow field.
Previous analytical models describing flow fields into
hoods [16,17,20] have used potential theory as the
theoretical basis. In potential flow, the assumptions are
made that the fluid is both incompressible (the volume
expansion is negligible) and irrotational (negligible local
angular velocity) [21]. These assumptions are valid in the
free 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 the
path of air flowing into the hood.
An object (such as a person) in the flow field

questions the validity of the potential theory approach in
two ways. First, by its very presence the object acts as an
obstacle, a physical obstruction to the flow of air into the
22
hood. As such it perturbs the boundary conditions for the

solution 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 to
the 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 velocity
gradient across the layer.
Imagine a fluid particle in the boundary layer moving

around the cylinder adjacent to the wall. Because of the
23
FIGURE 4
Thin front \ Outer stream grossly

perturbed by broad How
boundary layer separation and wake
FIGURE 5
Separation
Separation
Broad "^ Narrow

wake p wake
(«) (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 Strouhal
number:
(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
Best-fit line
.210
z;jrI°--B2^'^2-4>f^^^^g^=#^^^'^^^^r'^-->^—---
.200
: 0.2,2 0-^)
.190
= ,170 O'.cm
o a0235
Q .0362
O .0513
4 .0800
.4 60
V .0989
.158
.318
.635
Kovosznoy
"Tail" mdcotes ihoi velocity |
WQS computed from shedding 1
frequency of o second cylinder !
100 200 300 400 500 600 700 800 900 1.000 1,100 1,200 1.300
Reynolds number, /?
SOURCE: REFEREMCE 24
27
boundary- The authors also showed that the ratio of

longitudinal spacing between vortices to the diameter of the
cylinder is about 4.27.
Bloor [26] demonstrated a stable range of vortex

formation at Re below 200. That is, in this range, the flow
is laminar everywhere. In the Re range of 200 - 400, the
wake begins to disintegrate to turbulence. The onset of
wake turbulence moves closer towards the cylinder as Re
increases. At Re greater than 400 the separated boundary
layer becomes turbulent even before it rolls up into a
vortex. Thus, the vortices are turbulent upon fomnation.
However, at Re between 400 and 1300 the point of transition
of turbulence remains constant relative to the cylinder.
Finally, at Re of about 1300, the length of the laminar flow
region begins to decrease again until, at Re of about
50,000, it is almost to the shoulder of the cylinder. The
point at which turbulent motion reaches the separation point
of the boundary occurs at Re of about 300,000. This point
is called the critical Reynolds number. However, a definite
shedding frequency is still observed, even at the critical
Reynolds number.
Bearman [27] demonstrated regular vortex shedding at Re

up to 550,000. However, at a Re of 300,000, the shedding
frequency as described by the Strouhal number showed a sharp
increase from its relatively constant value of 0.21. The
Strouhal number leveled off to about 0.46 for Re greater
that 400,000. The author points out that any small change
28
in the surface smoothness of the cylinder can significantly

disrupt the separation causing fluctuations in the shedding
frequency.
Achenbach and Heinke [28] also noted the sharp increase

in shedding frequency at Re of 300,000. This increase
becomes less prominent as cylinder surface roughness
increases.
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 that

either of these two wake structures can completely destroy
the intended beneficial effect of a LEV and that no
consideration appears to be given to this problem in

standard ventilation design.
In studying push-pull ventilation systems, Hampl and

Hughes [30] also demonstrated the effect of a person in the
flow field of a ventilation system. They observed the
collection of smoke by a standard LEV hood with various
orientations of air jets used as "pushing" air streams. For
each orientation where a test mannequin obstructed the
pushing jets, smoke was observed in the area in front of the
mannequin. However, when the jet was placed between the
smoke and the mannequin, no smoke was observed in the
breathing zone and all smoke was captured by the hood. They
concluded that the "push jet should be located so that the
air impinging on the worker or other obstruction should be
minimized".
Van Wagenen [31] studied the effects of positive

airflow (blowing rather than exhausting air) on
concentrations of various contaminants in a welder's
breathing zone. He demonstrated that when directional air

flow comes from directly behind the welder, concentration of
fume in the breathing zone was equal to or higher than the
breathing zone concentration with no directional air flow at
all. He attributed this to the eddy and convective currents
around 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
5 FIGURE 7
of
4 (A
<r
iA
^ U
ͣ•ͣ
^
e
V /
-O ' t
•ͣ-.'s "3 4i
/
|\ Ji ^ T^
i A:r Pi OlV
-'
\
,
'
h 'V. .
>' ^ J
a >v
\J • o
f^
~?
«^
1
•^ J^ tf
^0 .:
C
-^ 1*. ^
-h .- 0
0- ^
u <- u
>*
o
<r •
1
1
Jc
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 the

shedding of vortices such that
(13) Qv = fV
Where V is the mixing zone volume and f is the frequency

with which this volume is removed by vortex shedding. If
one assumes that the vortices are approximately circular
cylinders of height H then the volume can be given by
(14) Vv = (pi)(De^H/4
where De is the diameter of an average vortex. However, one
must account for the fact that the zone is composed of two
vortices which are alterntely formed on each side and shed
downstream 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 the
zone is
(14a) V = (pi) (De^ (H)/8

and 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 following
relationship 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 of

a vortex is essentially the same as the diameter of the zone
of reverse flow (that is, De = s) and beyond this point no
contaminant 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 cylinder
curve.
.......IHHBiW
37
•m FIGURE S
A^--
Cbz
Co
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
44
FIGURE 10 - WIND TUNNEL VELOCITY PROFILE
DATE: 11 MAY 1988 TIME: 1200 TEMP: 70 DEGREES F

TUNNEL DIMENSIONS: 5' X 5' X 8'
FRONT CROSS SECTION - 18" FROM FACE
6" 12" 12" 12" 12" 6" MAXIMUM = 270 FPM

6" MINIMUM = 230 FPM
245 255 255 245 240
12« RANGE = 40 FPM
240 260 270 240 235
12" MEAN = 243 FPM
240 230 240 230 230
12" S.D. = 11.4 FPM
230 255 240 230 250
6"
C.V. = 4.7 %
MIDDLE CROSS SECTION - 51" FROM FACE

II II
6" 12 12 12 II
12 II gii MAXIMUM =260 FPM
6"
MINIMUM =230 FPM
250 255 255 235 240
12" RANGE =30 FPM
250 260 260 255 250
12"
MEAN = 251 FPM
250 260 250 255 260
12" S.D. = 8.9 FPM
230 240 260 260 250
6" C.V. = 3.5 %
REAR CROSS SECTION - 83" FROM FACE
6" 12" 12" 12" 12" 6" MAXIMUM =260 FPM

6"
MINIMUM =240 FPM
260 260 260 260 255
1 o II
RANGE =20 FPM
255 260 250 260 260
1 "D II
±Z
MEAN =254 FPM
240 260 240 250 240
12"
S.D. = 7.2 FPM
250 255 250 250 260
6"
C.V. = 2.8 %
SUMMARY DATA
MAXIMUM = 270 FPM RANGE = 40 FPM S.D. = 10.3 FPM

MINIMUM = 230 FPM MEAN = 249 FPM C.V. = 4.1 %
TOTAL FLOW RATE = 6225 CFM VENTURI PRESSURE DROP = 2.40"
45

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

6" MINIMUM =130 FPM
150 140 145 140 130
12" RANGE = 30 FPM
140 150 160 150 135
12" MEAN =142 FPM
140 140 140 145 135
12" S.D. = 7.7 FPM
135 145 155 135 135
6" C.V. = 5.4 %

1 1
6" 12 12 12 1
12 1 6" MAXIMUM =150 FPM
6" MINIMUM =125 FPM

135 145 145 150 130
12" RANGE =25 FPM
140 150 150 150 130
12"
MEAN =143 FPM
125 150 150 150 130
12 •» S.D. = 8.7 FPM
135 145 150 145 150
6" C.V. = 6.1 %
6" 12" 12" 12" 12" 6" MAXIMUM = 155 FPM

6"
MINIMUM = 130 FPM
145 150 145 155 135
12"
RANGE =25 FPM
140 155 150 155 135
12"
MEAN =145 FPM
130 145 145 150 135
12"
S.D. = 7.3 FPM
150 150 140 145 150
6"
C.V. = 5.1 %
SUMMARY DATA

46


1
6 12" 12" 12" 12" 6" MAXIMUM 50 FPM
6" MINIMUM 30 FPM

40 45 50 45 40
12" RANGE = 20 FPM
40 45 40 43 38
12" MEAN = 42 FPM
40 50 48 40 30
12" S.D. = 5. 2 FPM
45 40 50 35 40
6" C.V. = .L2. 3 %

1
6 12" 12" 12" 12" 6" MAXIMUM 55 FPM
6" MINIMUM 30 FPM

50 45 45 50 45
12 •• RANGE = 25 FPM
30 45 55 53 45
12" MEAN = 47 FPM
50 45 50 52 42
12" S.D. = 5.6 FPM
40 45 50 55 45
6" C.V. = 12. 3 %
6" 12" 12" 12" 12" 6" MAXIMUM = 55 FPM

6"
MINIMUM = 30 FPM
45 50 50 50 50
12"
RANGE =25 FPM
40 50 55 40 40
12"
MEAN = 48 FPM
30 50 50 55 45
12"
S.D. = 6.3 FPM
45 55 48 53 50
6"
C.V. = 13.0 %
SUMMARY DATA

4T
riGURE 13
VENTURI CALIBRATION CHECK

FLOW RATE BASED ON ANEMOMETER READING
a: o
0'-'
SQUARE ROOT OF PRESSURE DROP (in. H20)

a DATA POINTS + REGRESSION POINTS
48
the tunnel cross section. This increases the velocity in

proportion to the amount of tunnel area blocked by the
object. Therefore, a "blockage ratio" must be calculated.
A blockage ratio is basically a factor by which the
unblocked velocity must be multiplied to get the true tunnel
velocity. To determine the blockage ratio, the amount of
tunnel cross section blocked by the object must be
estimated. The following formula can then be applied:
(18)
Blockage = Tunnel Cross Section
Ratio Tunnel Cross section - Object Cross Section
(19)
Corrected = Measured X Blockage
Velocity Velocity Ratio
Notice from table (3) that by appling the blockage
ratio the corrected velocities are 265 fpm, 152 fpm, and 49
fpm for the mannequin and 292, 167, and 54 fpm for the
circular cylinder.
The Re for air flow around the objects at these

velocities is also given in table (3). These velocities
were selected because the Re are in the same range as those
for air flowing around an industrial worker in a uniform
flow such as a spray paint booth.
TEST OBJECT DESCRIPTION
The circular cylinder used in this project was

constructed of sheet metal and was 48 inches tall and 12
inches in diameter. Two holes were drilled in the cylinder,
one in the front about 15 inches from the top and another in
the 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: 46 143 249

CORRECTED VELOCITIES, CYLINDER: 54 167 292
REYNOLDS NUMBERS, CYLINDER: 5616 17,368 30,368
CORRECTED VELOCITIES, MANNEQUIN: 49 152 265
REYNOLDS NUMBERS, MANNEQUIN: 3414 10,539 18,465
*The mannequin diameter is not actually a diameter because

the mannequin cross section is more elliptical than circular
in shape. The value reported here as the diameter is the
breadth of the mannequin chest as measured just under the
armpits and at the same distance from the floor (27") as the
source 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" of

the 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 feet
from 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 in
all 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 the
mixing 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
SMOKE
OUTLET
T <J1
VACUUM PUMP
TITANIUM
TETRACHLORWE W
w
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
SV6 PIFFUSER
/ r\
ROTAMETER
MI RAW
DATALOGGER
SF6
JNCIAMEV
TAWK
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 #1
CYLINDER 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
EXPERIMENT #2
CYLINDER MANNEQUIN
(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

12" CYUNDER IN WIND TUNNEL
E
a
a
\^
z
O
I
u
u
z
8
Ifl
u.
m
DISTANCE OF SOURCE FROM CYLINDER (IN.)

D 54 fpm
5»
FIGURE 17

E
a
a
Z
o
t
o
z
8
10
k.
DISTANCE OF SOURCE FROM CYUNDER (IN.)

D 167 FPM
59
FIGURE n

12" CYLINDER IN WIND TUNNEL
E
a.
a.
\^
X
o
{=
u
O
z
8
10
b.
H

a 292 FPM
60
FIGURE 19

E
a
a
z
O
o
z
01
0 2

a 292 FPM +167 FPM O 54 FPM
61
FIGURE 20

MANNEQUIN IN WIND TUNNEL
E
a
a
8
h.
10

O 49 FPM
62
FIGURE 21

z
o
i
o
z
8
ID
ii.

D 152 FPM
63
FIGURE 11

7 -
6 -
E
a
a
5 -
Z
o
4 -
u
O
z
3 -
8
u>
u.
V)
1 -
1------1------r 1------1------1------1------r 1-----1-----1-----1-----1-----1-----1-----1-----r

2 6 8 10 12 14 16 18 20

D 265 FPM
63
FIGURE 22

E
a
a.
Z
o
B
8
ID
k.
M

D 265 FPM
64
FIGURE 23

40
35
30 -
a
a.
s^
25 -
Z
o
20 -
u
o
z
15 -
8
10 -
5 -

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

CYLINDER IN WIND TUNNEL
E
a.
a.
>w
S
I
I-
8
ID
la.
01

D 54 FPM
67
FIGURE 25

CYUNDER IN WIND TUNNEL
E
a.
Z
o
I-
z
u
u
z
8
10
Ik

D 167 FPM
6i
FIGURE 26

E
a
a
O
z
8
10
b.
n

D 292 FPM
69
FIGURE 27

E
a
1
Ui
O
Z
8
U)
in

a 292 FPM + 167 FPM O 54 FPM
10
FIGURE 2g

E
a
a.
I
i
u
O
Z

a +9 FPM
71
FIGURE 29

a
o.
ͣw
Z
o
i
o
z
ID

a 152 FPM
72
FIGURE 30

E
a
a.
I
I
r
u
z
8
10
in

a 265 FPM
73
FIGURE 31

E
a
I
u
O
Z
8
10

265 FPM +152 FPM O 49 FPM
74
DETERMINATION OF ZONE DEPTH: Cbz = 0.5Co
m 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 obtained
for 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 OF

CONCENTRATION VERSUS DISTANCE
EXPERIMENT ONE - CYLINDER
VELOCITY De USING MAXIMUM De USING INITIAL

(FPMl CONCENTRATION AS Co CONCENTRATION AS Co
* 54 8.3" 7.8"
167 3.9" 4.3"
292 2.1" 2.1"
EXPERIMENT ONE - MANNEQUIN

fFPM) CONCENTRATION AS Co CONCENTRATION AS Co
49 4.4" 4.6"
152 4.2" 4.3"
265 5.8" 4.2"
EXPERIMENT TWO - CYLINDER

(FPM) CONCENTRATION AS Co CONCENTRATION AS Co
54 5.2" 6.9"
167 1.9" 1.9"
292 1.9" 1.9"
EXPERIMENT TWO - MANNEQUIN

rFPM) CONCENTRATION AS Co CONCENTRATION AS Co
49 5.0" 10.1"
152 4.8" 4.9"
265 4.7" 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 two
values for Co mentioned above were used in this calculation.
Table (8) gives the calculated values of De from the
predictive model.
SUMMARY
Table (9) summarizes the results of the model

evaluation experiment by comparing the depth of the zone as
determined by;
1) the observation of test smoke,

2) the concentration versus distance curves, and
3) the theoretical equation.
77
TABLE 8
ZONE DEPTH AS DETERMINED FROM THEORETICAL EQUATION

54 13.8" 15.4"
167 9.6" 10.5"
292 10.2" 10.2"

49 9.6" 12.6"
152 9.3" 12.9"
265 9.4" 10.6"

54 12.1" 13.8"
167 9.0" 9.0"
292 7.3" 7.3"

(FPM) CONCENTRATION AS Co CONCENTRATION AS Co
49 13.3" 17.6"
152 10.7" 12.9"
265 9.0" 10.6"
78
TABLE 9
SUMMARY TABLE OF ZONE DEPTH BY VARIOUS METHODS
[ALL MEASUREMENTS FOR ZONE DEPTH GIVEN IN INCHES]

VELOCITY SMOKE LOG-LOG THEORETICAL

fFPM^ VISUALIZATION PLOTS EQUATION
54 14.0 7.8/8.3 **15.4/13.8

167 16.0 4.3/3.9 10.5/ 9.6
292 22.0 2.1/2.1 10.2/10.2

fFPM) VISUALIZATION PLOTS EQUATION
49 10.0 4.6/4.4 12.6/ 9.6

152 9.0 4.3/4.2 12.9/ 9.3
265 13.0 4.2/5.8 10.6/ 9.4

54 14.0 6.9/5.2 13.8/12.1

167 16.0 1.9/1.9 9.0/ 9.0
292 22.0 1.9/1.9 7.3/ 7.3

49 10.0 10.1/5.0 17.6/13.3

167 9.0 4.9/4.8 12.9/10.7
292 13.0 4.3/4.7 10.6/ 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 calculated

from 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 layer

separation on breathing zone concentration is to monitor
concentration in the breathing zone under conditions in
which the boundary layer will interact with the contaminant
source and compare the result to a situation in which it
does not. As has been mentioned, in the typical orientation
of a worker with respect to LEV, the separated boundary
layer can possibly extend downstream far enough to cause the
contaminant 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 be
less likely to reach out and interact with the contaminant
source but will rather extend downstream.
To evaluate the effect of a 90 degree orientation with

respect to ventilation on breathing zone concentration, a
method similar to that previously described was used. The
mannequin was again placed in the wind tunnel. However,
this time it was oriented at 90 degrees from the flow. That
is, the wind was flowing from its side rather than around
its back. A set of SF6 concentration versus distance curves
was obtained in exactly the same manner as before for 2 65
fpm, 152 fpm, and 49 fpm.
The curves obtained with the mannequin at 90 degrees

are graphically compared to the curves obtained in the
previous experiment in figures (33), (34), and (35). A
JS^SS--:-
S1
>
WPICAL ORIENTATION
->
FIGURE 32
"^^^
>
> o
9(? VEGREES TO FLOW

B2
FIGURE 33

MANNEQUIN - 49 FPM
E
a
a
\
O
z
8
If

MANNEQUIN/90 + MANNEQUIN
S3
FIGURE 34

MANNEQUIN - 1 52 FPM
E
a
a.
^^
Z
o
u
O
z
ID
U.
in

MANNEQUIN/90 + MANNEQUIN
S4
FIGURE 35

MANNEQUIN - 265 FPM
a
a.
I
t
O
z
8
10
Ik

a MANNEQUIN/90 i- MANNEQUIN
85
table of these values is provided as table (10)

86
TABLE 10
MANNEQUIN/9 0 MANNEQUIN
(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 since

they were based on the blockage ratio at the maximum
mannequin cross section. However, for comparitive purposes
they 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 in
table (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 concentration

once 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 concentration
appears to fall off exponentially with increasing distance
throughout the zone, dropping to essentially zero by the
time the edge of the zone (as determined from test smoke) is
reached. However, at 167 fpm (figure (17)), experiment 1
data shows that a little more mixing is occurring, giving
concentrations that are relatively uniform up to 4 to 6
inches before falling off. The best mixing within the zone
seems to occur at 54 fpm (figures (16) and (24)). Here, a
somewhat 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 to
about 6 to 8 inches for all velocities (figures (20), (21),
(22), (28), (29), (30)). Note that the mannequin provides a
more realistic view of the effect of the turbulent zone
phenomenon on human exposure because of the hands and arms

which 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 the
mannequin at lower flows. Perhaps some smaller scale
turbulent 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 at

laboratory 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 original
value (see figure (8)).
In table (9), the zone depth as visualized by the test

smoke 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 the
test 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 when
Co is defined as the breathing zone concentration at a
source distance of 0.5 inches. This zone depth determined
from 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 points
obtained 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 this

velocity for the first mannequin experiment and both
cylinder experiments. Therefore, it is concluded that the
close agreement at this single point is not significant.
On the basis of these experiments, it appears that the
fundamental assumption of a well mixed turbulent zone may
not be accurate. The data seems to indicate that even
within the zone a definite variation in concentration with
distance is observed. Perhaps the periodic shedding of
vortices from this zone is not the only contaminant removing
process involved. Other phenomenon, such as turbulent
diffusion, may also affect concentration, giving rise to
gradients within the zone.
DISCUSSION OF MANNEQUIN VERSUS MANNEQUIN AT 90 DEGREES
Perhaps the most dramatic indication of the
significance of the reverse flow phenomenon on worker
exposure can be seen in the comparison of breathing zone
concentration between the mannequin at typical orientation
with respect to LEV and the mannequin oriented at right
angles to the flow. From figures (33) through (35) it is
obvious that standing 90 degrees to the direction of air
flow significantly reduces the concentration of contaminant
in the breathing zone. Concentrations measured with airflow
coming from behind the mannequin ranged from 6 to 43 times
higher than those measured at corresponding distances with
the 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 of
vortices.
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 the
zone.
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/Cbz

riNCHES) @ 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 smoke

observarion (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

fINCHES) 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.)

riNCHES) 0 292 FPM Co (ppm) (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

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

(INCHES) @ 152 FPM Co (DDm) (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 0.56
3.0 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 1.29
10.0 1.9 4.0 2.11
12.0 1.4 4.0 2.86
fIGURE 36
MEASURED VS. PREDICTED CONCENTRATIONS

FROM ORIGINAL THEORETICAL EQUATION
£
a
a
Ui
z
O
P
UJ
O
z
o
o
Q
U
a:
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

99
Q r e? s s i o n CJ u t i::) u t if
-v„ (j;.:::/4!.j
Std Err ot Y iEst 0.. 286667
R Squared
H o., a -f O ta s e r vat i o n s :i. o
D e c:j 1-ͣ e? e B !jf F r e e ci o fn 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

CYLINDER — 54 FPM
5£
i
o
z
X
2
Z/D (DIMENSIONLESS DISTANCE)

D DATA POINTS -I- REGRESSION POINTS
FIGURE 37
100
C a n s t a n t O „ i.56> 3 O i
Ejtd Err of Y EZst 0„ 303773
R S q I ..I a r • e ci ('!;.. 5 3 Ei 9 2 4
No. of Observations IS
Degrees of Freedom 13
Std E"'T Q-f Coef a C>„214437

MANNEQUIN----49 FPM & 152 FPM
I
§
Z/D (DIMENSIONL£SS DISTANCE)

n DATA POINTS + REGRESSION POINTS
FIGURE 38
««f
702
Keg re :put"
0. 2-.:i.6i:,:i:l.
R 5Qi..iare?d ("1, Q "/ 3 Q 9 9
N a,. i;:i t 0 !::> b e r vat :i. a n;;
Degrees of Frͣ eeci om
A I.;, ("i e T 'f-11:: i e n c i s j .;.„.!. •.::! / *..ͣ' a ͣ.

MANNEQUIN-----265 FPM
I
o
z
X
-0.7
Z/D (DIMENSIONLESS DISTANCE)

D DATA POINTS + REGRESSION POINTS
FIGURE 4(;
103
FIGURE 41
DIMENSIONLESS De AS A FUNCTION OF Re
ZONE DEPTH DOWNSTREAM OF OBJECT
1.9
>
Q
U 18
(Thousands)
REYNOLDS NUMBER
DATA POINTS + REGRESSION POINTS
R e g r e s s i o n 0 i.i t p:) i..( t. n

Constant 1„029587
Std Err of Y Est 0„144758
R Squared 0«7S7622
M o, o f 0 b s e r V a t :i. g p. s 6
X Coe-f -f i c i en t < s) 0, 000025
S 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) 1
K
De = D[(0.000025)(Re) + 1.03]
MIXING FACTOR (K):
CYLINDER
- Re < 10,000; K = exp[(1.04)(Z/D) - 0.03]

- Re > 10,000! K = exp[(1.37)(Z/D) - 1.10]
MANNEQUIN
- Re < 10,000: K = exp[(0.93)(Z/D) + 0.16]

- Re > 10,000: 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;

FROM CYUNDER MODEL (Re > 10,COO)
a -
X
7
6 -
_
y^
bJ
O
5 -
D
J! y^
i 4 -
o
a y^
i
'O.y^
I)
z Vi y^
1 -
r'T —
r 1 1 1 i 1 1 1
MEASURED cONCEfJTRATION fpptT,1

Q VELOCITY - 292 FPM
107
FIGURE 43

FROM CYUNDER MODEL (Re > 10,000)
E
a
a.
ͣ^-•
_i
Ui
£i
O
i
0
MEASURED cQNcENTRATiQN fppt-n)

D '/ELOCrrf - 16? FPM
TOS
FIGURE 44

FROM OOJNDER MOCCL (Re < 10,000)
a
Q.
_l
u
Q
O
2
I
u
MEASURQJI CuNCENTftATION (ppm)

D VELOCITY" - 54 FPM
'09
FIGURE 45
MEASURED VS, PREDICTED CONCENTRATIONS

FROM MANNEQUIN MODEL (R« > 10,0)0)
8:
-I
u
d
0
2
z
O
I
i
u
O
z
o
o
MEASURED cuNcENTRATiON f'ppm^

D VELOCITY" - 266 FPM
10
FIGURE 46
MEASURED VS. PREDiCTED CONCENTRATIONS

FROM MANNEQUIN MODEL (Re < 10,000)
E
a
a
hi
a
0
2
z
0
e
u
U
z
o
MEASUKkl) CUNCtNTRATluN (ppmi

D VELOCITt' - 152 FPM
msm
FIGURE 47

FROM MANHEOUiN MODEL (Re < 10.1X30)
60 ^
y'^
50 -
i:
a
a.
a
-I
u AQ -
a
O Q ,
d"
30 - y
z ͤ
o
ͤ y'
2D -
a
0
8
10 -^
/'
1 \ 1 1 \
&i 40 60
MEASURED CUNcENTRATItJN (ppmi

a 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 test
smoke. 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 which
a worker is immersed in a uniform flow (such as a paint
booth) with the contaminant source between himself and the
source 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 more
uniform throughout.
Considering the edge of the zone to be the point at

which the initial concentration drops to one half its
original 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 mixed
zone.
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 was

B
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 out

of the mixing zone. Therefore, at these Re, the K factor,
which in effect is accounting for turbulent diffusion, does
not work as well at predicting the concentration (figures
(44) and (47)). On the other hand, at Re greater than
10,000, turbulent diffusion may predominate since the
vortices are becomming more and more turbulent and thus the
model (including the K factor) predicts concentration within
the 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 that

the entrainment flow of the turbulent wake is governed
mainly by the length of the turbulent shear layer as it
crosses the axis of the wake and that this flow increases
with increasing Re. Gerrard terms the thickness of this
shear layer the "diffusion length". This could explain the
increasing zone depth with increasing Re observed with the
test smoke.
To summarize, it appears that the two crucial

assumptions involved in the development of the model are not
completely correct. The distance downstream to which the
turbulent mixing zone extends seems to be greater than the
diameter of a vortex. The turbulent wake appears to effect
contaminant concentrations in the zone at distances beyond
the region of vortex formation. Consequently, the alternate
shedding 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, object
heights, 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 model
is 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 a

significant mechanism involved in the transport of
contaminants 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 data
with a mixing factor. To do this it will be essential to
study the current literature on this subject to gain a more
thorough understanding of this phenomenon. To proceed with
this research without attempting to account for turbulence
would seem to be unreasonable.
121
STUDY OF CONCENTRATION DECREASE AS A FUNCTION OF DISTANCE
Another entirely different approach to predicting

breathing zone concentration from LEV parameters is
suggested by noting the decrease in concentration as the
source moves farther and farther away from the sampling
probe. Figures (48) through (59) illustrate log-log plots
of concentration versus absolute distance for the data
collected in this experiment. [Absolute distance is defined

as the distance from the tip of the sampling probe to the
point source of contaminant.] Note from these graphs that
there appears to be a reasonably strong inverse relationship
between log concentration and log distance. Regression
lines give values for R squared ranging from 0.84 to 0.99.
The 95% confidence interval for the slopes of these
regression lines (table (12)) indicate that there is no
statistically significant difference between the slopes of
the lines at 54 fpm, 167 fpm, and 292 fpm for the cylinder
data. The average value of the slope being approximately -
1.6. Likewise for the mannequin data, slopes at all three
velocities are statistically the same with the exception of
the 49 fpm line from experiment two data. For the mannequin
the average slope is about -2.4. Thus, for a given object
geometry, the slope appears to be independent of velocity.
122
TABLE 13
95% CONFIDENCE INTERVALS FOR

THE SLOPES OF THE REGRESSION LINES
CYLINDER
EXPERIMENT VELOCITY
NUMBER rFPM) SLOPE 95% CONFIDENCE INTERVAL
1 54 -1.27 -1.61----0.92
2 54 -1.48 -1.72----1.25
1 167 -1.71 -1.83----1.59
2 167 -1.42 -1.92----0.92
1 292 -1.71 -1.95----1.45
2 292 -1.83 -2.47----1.19
AVERAGE ---1.57
MANNEQUIN
EXPERIMENT VELOCITY
NUMBER rFPM) SLOPE 95% CONFIDENCE INTERVAL
1 49 -2.35 -2.81----1.90
2 49 -1.72 -2.19----1.25
1 152 -2.74 -3.13 — -2.36
2 152 -2.48 -3.07----1.90
1 265 -2.23 -2.53----2.04
2 265 -2.74 -3.49----2.00
AVERAGE = -2.34
123
FIGURE 48

CYUNDER IN TUNNEL 5+ FPM
3.2
J.I
2.9
2.8
E 2.7
a.
a.
>-•
2.6
Z
o 2.5
p
$
H
2.4
z 2.J
U
^ 2.2
8 2.1
if 2
U)
1,9
z
_l
1.8
1.7
1.6
1.5
1.4
P DATA POINTS
LM ABSOLUTE DISTANCE (IN.)
+ REGRESSION POINTS
R e g rͣ e s s i q n 0 u t p i.i. t .1
Constant. 5n 294948
Std Err at Y Est 0., 250954
t< -Squared 0.853593
No. of Observations 13
Degr'ees o-f Freedom 11
X C;oef 11 c: i en t (s) .....1 „ 2657£B
Std Err o-f Coef „ 0., 158058
124
FIGURE 49

CYLINDER IN TUNNEL 1 67 FPM
2.8
2.6
2.4
2.2
E
a.
2
•5
5 1.8
H
i^ 1.6
2
Ul
o 1.4
z
8 1.2
U)
h.
at 1
z
_i
0.8
0.6
0.4
0.2
a DATA POINTS
LN ABSOLUTE DISTANCE (IN.)
+ REGRESSION POINTS
R e g r e s s i D n 0 n t p u t. s
Constant _, 5,. 650873
Std Err o-f Y Est 0.084051
R Squared 0.. 989556
!n|o. of 0bser^ Vat i ons 13
Deqrees of Freedom 11
X CoG-ft i ci ent (s, -•1 » 70903

Std Elrr of Coef !"i fi 'ͣ! '7 Q '"ͣͣ, "f
?25
FIGURE 50

CYLINDER IN TUNNEL 292 FPM
E
a
a
Z
g
I
z
O
z

Reg r e s s1on 0ut p u t s

Constant 4„660508
Std Err a+ Y Est 0„174467
F^ Squared 0.956482
No., of Observations 13
X L;oef + 1 c 1 en t (s ) — 1 „ 7()fS5S
Std Err of Coef „ 0.109£!84
J 26
FIGURE 5/

MANNEQUIN 1 N TUNNEL 49 FPM
A
D
D
3.5 -
n
n
3 -
E
a
a \^^^
2.5 -
n ^\{n n
2 -
iij
u
^V^^ n
z
a
8 1.5 -
10 D
u.
in
3 1 - "^
D ^
0.5 -
1 1 1 1 1 1 1 1 1 1 1 1
n 11
1,7 1.9 2.1 2.3 2,5 2,7 2.9 3.1

Regression Output.,';
Constant 7» 79156 :L
Gitd Err o-f Y Est . 0,326626
R Squared 0,922.394
No,, of Qtaservati ons 13
Degrees o-f F'resdom 11
X Coe-f-f i cient (s) --2., 35224

Std lErr of Coef „ 0.205718
127
FIGURE 52

MANNEQUIN IN TUNNEL 152 FPM
2,5
a
a.
Z
o
f-
Z
111
u
z
8
H
-0,5 -
LN ABSOLUTE DISTAN(£ (IN.)

Regression Output;:
C o n s t ant 7„429680
Std Err of Y Est 0.277235
R Squared 0.957294
i'^-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 7

Std Eirr of Coef . 0.174610
128
FIGURE 53

2,2
1.8
1.6
E 1.4
a
a
^^
1.2
z
o 1
I 0.8
.Z
U
u 0.6
z
8 0.4
10
h. 0.2
U)
0
-0.2
-0.4
-0,6
-0.8 ------1----------,------ —T----------1----------1----------r

1.7 1.9 2.1 2.3 3.1
lU ABSOLUTE DISTANCE (IN.)

R e g r a B s i o n 0 u. t p u t ;i
Constant 6.221454
Stci Err o+ Y Est 0.179319
R Squared 0.973832
Degrees of Freedom '^ 11
X Coef f i c i en t (s) --2. 2851 1

Std Err of Coef„ 0.112940
J29
FIGURE 54

CYUNDER IN TUNNEL 54 FPM
J.4
D a
3.3
3.2
3.1
E 2.9
a
a
2.8
2.7
2.6
i 2.5
Z
2.4
z
2.3
8 2.2
il.
2.1
1.9
1.8
1.7
1.6
1.5
1.7

Regress.! on Outputs
Constant 5„899558
Std Err of Y Est 0,123820
R Squarsd 0„963876
1^1 o. ot Observations 10
X Uaett :i. ci ent (s ) ͣ-1 . 48232
br.d ot Coef 0.101457
130
FIGURE 55

£
a
a
\^
z
o
5
u
O
z
8
10
h.
V)

'Regression Output.!;
Constant 4., 860895
Std Err o-f Y E^st 0.264082
R Squared 0,843299
No,, of Ubservations 10
Deqrees of F-'reedom 8
X C;oBt t i c: i en t (s) ͣ-1. 419£! 1

Std Birr o+ Coe+ .
/37
FIGURE 56

E
a
a
^-
z
u
o
z
to
z
-I
UN ABSOLUTE DISTANCE (IN.)

DATA POINTS +ͣ REGRESSION POINTS
Regr ess .1 on 0ut p ut:

Constant 5.465901
Std Err of Y Est 0„33£i653
R S q u a r e d 0« 8 4 4 -319
ix!o., o-f Observations 10
Degrees of Freedom ,8
X Coef f i c i en t(s) .....1„82779

Std Err of Coef. 0.277489
j^'<!
J32
FIGURE 57

E
a
o.
ͣ»•
z
o
i
Ul
O
z
8
10
b.
»

Regression Output:
Constant 5.970544
Std Err of Y Est 0,, 249771
F; SquaiTE^d 0.898118
Mcj . of 01:) seI- Vat :i. on s , 10
Degrees o-f Freedom 8
X LJaet t :i. c: i en t (s) -1,. 71 £368

J '
J33
FIGURE 58

E
a.
a.
Z
o
ki
O
z
8
40
ll.
(n
HD,2
IH ABSOUUTE DISTANCE (IN.)

D DATA POINTS + REGFESSION POINTS
Regressi c:)n Output:

Con st an t 6 „ 815163
Std Err of Y Est 0„310511
f< Squared 0.922532
No., of Observations 10
Degrees of F-'reedom 8
X Coe-f t i. c: i en t (s) ͣ•-2.46338
Std Err of Coef, 0,. 2 5 4 4 -3 0
134
FIGURE 59

2.5
2 -
1.5 -
E
a
a
1 -
0.5 -
5
o
z
8
ii. -0.5 -
V)
-1 -
-1.5 -
-2

D DATA POINTS ͣt- REGRESSION POINTS
Regrss?iion Outputs
Constant 7,.1.33664
Std EErr of Y Est " 0„:390552
R Squared 0„901783
N o,. o -f t!) ta s e r v a t :i. o n s 10
Deqrees 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 acceleration
given to the air as it moves around the object would alter
the 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 of
this 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 Dh
on 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 plate
Vf = 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 by
vortex (inches)
139
cylinder, the smoke was pulled back towards the cylinder by

the reverse flow. By moving the source slowly back and
forth,the point at which the smoke first began to be drawn
back toward the cylinder could be estimated. This distance
from the source to the cylinder at which backflow starts to
occur is called s. This was considered to be the depth of
the reverse flow zone.
Appendix V tabulates values of s for various

combinations of one hood diameter (6"), three cylinder
diameters (1.5", 6.0", and 12.0"), and five flow rates (1075
cfm, 995 cfm, 910 cfm, 810 cfm, and 700 cfm). At higher
flow rates, values for s were measured at z distances
(centerline distance from cylinder to hood face) ranging
from 2" to 18". However at lower flow rates, the z distance
ranged from 2" to only 10" because the influence of the hood
flow field does not extend as far away from the hood face
for the lower flow rates.
A complete analysis of the data contained in Appendix V

has not yet been conducted. However, preliminary analysis
suggests that the size of the turbulent mixing zone may vary
as the size of the object relative to the size of the hood
varies. Figures (61) and (62) illustrate s/Do versus z for
flow rates of 1075 cfm and 995 cfm respectively. This
indicates that, relatively speaking, the size of the zone is
much larger when the object is small relative to the hood
diameter and becomes smaller as the object gets larger
relative to the hood. This relationship holds true for the
140
FIGURE 61
RELATIVE ZONE DEPTH VS. DISTANCE

6" HOOD — 995 CFM
0,7
0,6 -
0.5 -
o
a
Z INCHES
a Do-1 ,5" + Do-6" O Da-12"
141
FIGURE 62
RELATIVE ZONE DEPTH VS. DISTANCE

6" HOOD — 1075 CFM
0.7
0.6
0.5 -\
0.4
0
Q
0,3
0.2
0.1
Z INCHES
n Do-1.5" + Do-6" O Do-12"
142
other flow rates as well. In an industrial situation, the

effect of the reverse flow phenomenon on breathing
zoneconcentrations may be more significant for employees
working in front of a large hood as opposed to a smaller
hood.
Perhaps as the hood gets large relative to the object,

the zone formation approaches the uniform flow situation as
discribed previously whereas when the object is much larger
than the hood, most of the flow comes from the side rather
than around the object. However, this explanetion is
speculative. Much empirical and theoretical work is yet to
be done to adequately explain this phenomenon.
143
REFERENCES
1. National Safety Council, Fundamentals of Industrial

Hygiene. 2nd Ed., New York (1979).
2. American Conference of Governmental Industrial
Hygienists, Industrial Ventilation - A Manual of
Jêcommended Practice. 19th Ed., ACGIH, Cincinnati, OH
(1986).
3. McDermott, H.J., Handbook of Ventilation for

Contaminant Control. Ann Arbor Science, Ann Arbor, MI
(1976).
4. National Institute for Occupational Safety and Health,

The Industrial Environment - Its Evaluation and
Control, U.S. Government Printing Office, Washington,
D.C. (1973).
5. Dalla Valle, J.M., and Hatch, T., Studies in the Design

of Local Exhaust Hoods, presented at the 6th Annual
Wood-Industries Meeting, Winston-Salem, NC, Oct. 15-
16, 1931 of the American Society of Mechanical
Engineers.
6. Silverman, L., "Centerline Velocity Characteristics of

Round Openings Under Suction", Journal of Industrial
Hygiene and Toxicology. Vol. 24, No. 9, pp. 257-266.
(Nov. 1942)
7. Silverman, L., "Centerline Velocity Characteristics of

Narrow Exhaust Slots", Journal of Industrial
Hygiene and Toxicology. Vol. 24, No. 9, pp. 267-276.
(Nov. 1942)
8. Fletcher, B., "Centerline Velocity Characteristics of

Rectangular Unflanged Hoods and Slots Under Suction",
Annals
(1977).
of Occupational Hygiene. Vol. 20, pp. 141-146
9. Fletcher, B., "Effects of Flanges on the Velocity in

Front of Exhaust Ventilation Hoods", Annals of
Occupational Hygiene. Vol. 21, pp. 265-269 (1978).
10. Fletcher, B., and Johnson, A.E., "Velocity Profiles
Around Hoods and Slots and the Effects of an Adjacent
Plane", Annals of Occupational Hygiene,. Vol. 25, pp.
365-372 (1982).
11. Garrison, R., "Centerline Velocity Gradients for Plain

and Flanged Local Exhaust Inlets", American Industrial
Hygiene Association Journal^ Vol. 42, No. 10, pp. 739-
746 (1981).
144
12. Garrison, R., "Velocity Calculation for Local Exhaust

Inlets - Empirical Design Equations", American
Industrial Hygiene Association Journal. Vol. 44, No.
12, pp. 937-940 (1983).
13. Garrison, R., "Velocity Calculation for Local Exhaust

Inlets - Graphical Design Concepts", American
Industrial 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 Ventilation
Systems", American Industrial Hygiene Association
Journal. Vol. 44, No. 10, pp. 752-755 (1983).
15. Esmen, N.A., Weyel, D.A., McGuigan, F.P., "Aerodynamic
Properties of Exhaust Hoods", American Industrial
Hycfiene Association Journal. Vol. 47, No. 8, pp. 448-
453 (1986).
16. Flynn, M.R. and Ellenbecker, M.J., "Capture Efficiency

of Flanged Circular Local Exhaust Hoods", Annals of
Occupational Hygiene. Vol. 30, NO. 4, pp. 497-513
(1986).
17. Flynn, M.R. and Ellenbecker, M.J., "The Potential Flow

Solution 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 in
Ventilation", Annals of Occupational Hygiene. Vol. 24,
No. 1, pp. 105-112 (1981).
19. Fletcher, B. and Johnson, A.E., "The Capture Efficiency

of Local Exhaust Ventilation Hoods and the Role of
Capture Velocity", Ventilation '85 - Proceedings of the
1st International Symposium on Ventilation for
Contaminant Control, Toronto, Canada, Oct. 1 - 3, 1985.
20. Heinson, R.J. and Choi, M.S., "Advanced Design Methods
in Industrial Ventilation", Ventilation '85 -
Proceedings of the 1st International Symposium on
Ventilation for Contaminant Controll Toronto, Canada,
Oct. 1-3, 1985. y
21. White, F.M., Fluid Mechanics. 2nd Ed., McGraw - Hill
Book Company, New York (1986). \
22. Flynn, M.R., "The Impact of Separation on Exposure and
Hood Capture", Grant Proposal, National Institute for
Occupational 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 from
Vortex Streets", NACA Rep. 1191 (1954).
25. Fage, A. and Johansen, F.C., "The Structure of Vortex
Streets",
(1928) .
Phil. Magazine. Vol. 5, No. 28, pp. 417-441
26. Bloor, M.S., "The Transition to Turbulence in the Wake

of a Circular Cylinder", Journal of Fluid Mechanics,
Vol. 19, part 2, pp. 290-304 (1964).
27. Bearman, P.W., "On Vortex Shedding from a Circular
Cylinder in the Critical Reynolds Regime", Journal of
Fluid Mechanics. Vol. 37, part 3, pp. 577-585 (1969).
28. Achenbach, E. and Heinke, E., "On Vortex Shedding from
Smooth and Rough Cylinders in the Range of Reynolds
Numbers 6X10'3 to 5X10*6", Journal of Fluid Mechanics.
Vol. 109, pp. 239-251 (1981).
29. Ljungqvist, B., "Some Observations on the Interaction

Between 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 Exhaust

Control 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 Control

Technology Techniques for Welding Fumes", Research
Report, National Institute for Occupational Safety and
Health, Cincinnati, OH (1979).
32. Occupational Safety and Health Administration General
Industry
(1985).
Standards 29 CFR 1910.94c6i, Table G-10,
33. Fitzgerald, M.L., "Comparison of Three Dimensional

Velocity Models for Flanged Rectangular Hoods",
Master's Technical Report # 1087, University of North
Carolina, Chapel Hill, NC
34. Lilley, D.B., "Using A Tracer Gas Technique to Evaluate

Laboratory Hood Effectiveness", Master's Technical
Report
Hill, NC
# 1061, University of North Carolina, Chapel
"x
\
146
35. Humphries, W. and Vincent, J.H., "An Experimental

Investigation of the Detention of Airborne Smoke in the
Wake Bubble Behind a Disk", Journal of Fluid Mechanics.
Vol. 73, part 3, pp. 453-464 (1976).
36. Humphries, W. and Vincent, J.H., "Experiments to
Investigate Transport Processes in the Near Wakes of
Disks In Turbulent Air Flow", Journal of Fluid
Mechanics. Vol. 75, part 4, pp. 737-749 (1976).
37. Humphries, W. and Vincent, J.H., "Near Wake Properties
of Axisymmetric Bluff Body Flows", Applied Scientific
Research. Vol. 32, pp. 649-669 (1976).
38. Vincent, J.H., "Scalar Transport in the Near
Aerodynamic Wakes of Surface-Mounted Cubes",
Atmospheric Environment, Vol. 12, pp. 1319-1322 (1978).
39. Vincent, J.H., "Model Experiments on the Nature of Air
Pollution Transport Near Buildings", Atmospheric
Environmentf Vol. 11, pp. 765-774 (1977).
40. Gerrard, J.H., "The Mechanics of the Formation Region
of Vortices Behind Bluff Bodies", Journal of Fluid
Mechanics. 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 following
concentration 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, twelve
points each were plotted.
For each point a series of three injections of known

concentration of SF6 was made with a gas tight syringe. The
absorbance 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 completely
between 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 calculate

concentrations.
^
MIRAN CALIBRATION CURVE - SF6
ABSORBANCE < 0.1 A.U.
a.
a.
O
z
ABSORBANCE (A.U.)
Ml RAN CALIBRATION CURVE - SF6
ABSORBANCE - 0.1 - 0.3 A.U.
1.1 -
D
1 -
0.9 -
0.8 -
r^
'•^ 1
?
a
0.7 -
a
y^^
0.6 -
^/ n
0.5 -
>/â
g
O
0.4 -
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
ABSORBANCE (A.U.)
a DATA POINTS + REGRESSION POINTS
MIRAN CALIBRATION CURVE - SF6
ABSORBANCE > 0.3 A.U.
E
a
a
o
z
-0.6
ABSORBANCE (A.U.)
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 CURVE
APPARATUS CAUBRATED IN TACT
800
700 -
E 600 -
t SOO
400 -
c
111
300 -
111
-I
oa
o
D
m 200 -
100 -
ROTAMETER READING
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/în, 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 report

the minimum, maximum, and average values.
SAMPLE OUTPUT
When the appropriate software is used, the Datalogger

can "dump" the recorded data into a personal computer for a
subsequent hard copy report. The following report types are
available; overall statistics, time history, amplitude
distribution, raw data, and multiple channel.
For this project, the time history report was the most
useful. This report gives the minimum, maximum, and average
values for each averaging period of the logged data. A
graphical representation is also provided by the computer
which graphs a minus (-) sign for the period's minimum
value, a plus (+) sign for the period's maximiim period, and
an asterisk (*) for the period's average value. The graph's
accuracy and resolution depend on the variation (spread) of
the test data. Please note attachment (1) for an example of
a time history report.
"TEST START DATE: 08/03/1988"
"TEST START TIME: 13:30"
" TEST LOCATION: CHADYRU - PUBI..IC HEALTH"
" EMPLOYEE NAME: DENNIS K. GEORGE"
"EMPLOYEE NUMBER: 419-aB-t-J297"
DEPARTMENT: ESE ~ INDUSTRIAL HYGIENE"
REPEAT OF CYLINDER AT 150"
^j^~iMMENT FIELD 1:
W'MMENT FIELD 2: SF6 FLOW = 15"
" NUMERIC CODES:
"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 POINTS
0. 0000 V — 0.0000
1.0000 V = 1.0000
ALARM VALUES
LOWER: 1.192E-07 ???
UPPER: 1.192E~07 ???
PERIOD LENGTH: 0: O:10

!=• PERIODS COMBINED:
MIN AVG MAX
DATE: 8/ 3/88 TIME: 13:31:21 TAG #:

3334 3374 - *
o 3270 o,
o, 3376 -*+
0 3290 3338
0. 3297 —V
o 3259 3274
0 3230 3240 0 3263
o 3206 o 3248
0 ,3145 n.75 0 3201
:.094 0 3133 —*+
0 3055
liOlO 0 3057 —*+
o 2978
0 , 2886 o 2968
O.2809 ^047 o 2885
2754 0.2780 0 2812
2717 0.2740 0 2765
2711 2726 o <? -y t; Cy
2728 0.2741
2737 O,. 2748
2732 0.2765
2746 2755 0„2766
2736 0.2748 0.2766
0.2712 0.2725 0.2737
O. ?69; 0.2714 0.2729
8/ 3/88 TIME: 13: 34:41 TAG #:

0. 2671 0.2693 0.2712
0. 2654 0.2668 0.2688
0. 2612 0.2631 0,2650
0. 2551 0.2571 0.2601 —h
0. 2461 0.2499 0.2550 -* +
0.2396 0.2426 0.2451 -* +

0.2426 0.2456 O.2487
0.251.3 0.2543 0.2587
@ 0.2621 0.2642
DATE: 8/ -3/88 TIME: 13:50:18 TAG #:

U. 0-2596 u 2615
O. 2578 0.2595 o 2612
0. 2597 0.2617 0 2634
0. "^ 6 3 '^-' 0.2639 o 2646 ---+•
0. 2609 0.2620 0 2627
o. 2574 0.2588 o 2607
0. 0.2551 o 2566
o. 2487 0.2508 0 2534
0. 2438 0 2463 o 2480
o. 2401 0 2418 o 2436
0. 2404 O 2500 o 2579
0.2586 2681 C) 2775 — -*-+ͣ
o, 2800 2886 0 2946 --* -+•
0, 2948 2989 0 3028
0, 3030 3037 0 3053
0 3035 3047 o 3056
0, 3006 3018 0 3029
0 .;. 966 2986 0 3018
o 2919 2949 0 2973
o 2885 2913 o 2934 —#
DATE: 8/ 3/88 TIME: 13: 5:38 TAG

0,2853 0.2870 0-2883
0-2852 0.2862 0-2877
0.2829 0.2850 0.2862 -*
0. 280-3 0.2823 0-2854
0.2797 0.2842 0,2915
0.2928 0-2995 O.3060 -»-+-
0-3077 0-3166 O.3212
0-3215 O-3234 0.3258
O.3252 0.3276 0.3296
@ 0.3283 0.3326 0,3348
DATE: I 8/ 3/88 TIME: 14: 8:56 TAG

0.2669 O.2685 0.2700
0.2611 0-2641 o. 2671 —^ͣ
0.2560 0.2586 o. 2603
--+ͣ
0,2543 0.2556 o. 2569
0-2519 0.2533 0.2544
0-2501 0.2514 0.2530
0-2494 0.2501 0. 514 -*
0-2489 0.2492 u 501 •+
0.2472 0.2488 0.2497
0-2467 0,2478 0,2491
0.2458 0-2474 0,2490
0,2464 0.2477 0.2487
0,2476 0.2485 0.2498
0-2460 0,2473 o, 2494
0.2443 0-2463 O u Z480
0.2476 0. ?531 0.
0.2595 0-2643 0. 2691
O.2693 0-2717 o. 2741
0-2708 0. 2731
0.2695 0,2707 0.272;
DATE: 8/ 3/88 TIME: 14: i: !: 16 TAG #:

0.2646 0.2671 0.2685
0.2610 0,2636 0,2660
0,2577 0,2587 0.2617
0.2569 0,2581 0-2589
APPENDIX IV
CONC VS DISTANCE
REBRESSION DATA

54 FPM INITIAL CONC. AS Co
Regression Output:
Constant 3. 06(: >496
Std Err of Y Est 0. 4i: 909
R Squared 0. 60Z .649
No. of Observations
s 13
X Coefficient(s) -0 .41807
Std Err of Coef. 0. lo;:i42

Regression Output:
Constant 2.577530
R Squared 0,666629
X Coefficient(s) -0.89865
Std Err of Coef, 0.191610
EXPERIMENT ONE ~ CYLINDER

292 FPM MAXIMUM CONC. AS Co
Regression Output:
Constant 1. 810686
Std Err of Y Est 0. 225870
R Squared 0.9: 7061
X Coefficient<s) -0.66066

Regression Outputs
Constant. 3.313068
R Squared 0.766993
Ho. of Observations 10
l<b7 FPM INITIAL CONC. AS Co
Regression Output:
Constant 2,710482
Std Err of Y Est0.354691
R Squared 0.814014
No- of Observations 13
Degrees of Freedom il

Regression Output:
Con st an t 2.201013
R Squared 0,702925
Degrees of F'reedom 11
X Coe-ff icient (s) -0.76251


Regression Output:
Constant 5,060071
R Squared 0.882762

Regression Output:
Constant 2.482728
R Squared 0,899092
No- of Observations 10
X Coefficient's) -0.51403
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 0.927061
No. o+ Observation 13
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 Outputs
Constant 4.736741
R Squared 0.878546
No. of Observations
s 9

Regression Output:
Constant 4.272520
R Squared 0,917495
X Coef fici ent (s) --1.63257

Std Err of Coef. 0,163188

F;egre5sion Output:
Constant 2.410440
Std Err of Y Est , 0.254802
R Squared 0.911868
Regression Outputs
Constant y, .6: 13S6S
Std Err of Y Est 0 . 6B6S.39
R Squared 0 .6?16S36
No. o-f Observations
ns 13
Degrees o-f Freedom
m 11

Regression Output:
Constant 3,072344
R Squared 0,892396
EXPERIMENT ONE ~ MANNEQUIN

Regression Output;
Constant 4. lOr ;816
R Squared 0.957649
Std Err of Coef. 0,116536

Regression Output:
Constant
2.882279
Std Err of Y Est 0 ,520805
R Squared O 557042
152 FF'M INITIAL CONC. AS Co
Regression Output:
Constant 2.337873
R Squared 0„554721

Regression Output:
Constant 2.410440
R Squared 0.911868
X Coefficient(s) -0,66603

Regression Output:
Constant 2.190597
R Squared 0.543977

Regression Output:
Constant 0.. 870794
Std Err of Y Est 0. 320223
R Squared 0. 857881
No. of Observations
s 8
Degrees of Freedom h
X Coefficient(s) -0 .95.:373
Std Err of Coef. 0. 158476
54 FPM MAXIMUM CONG. AS Co
Regression Output:
Constant 3.943893
R Squared 0.960585
No. of Observations S
X Coef f i c i en t < s > -0.77063


Regression Outputs
Constant 3. 808668
Std Err of Y Est 0, 470688
R Squared 0. 854493
No. of Observations
s 8
X Coefficient(s) -1 .38272
Std Err of Coef. 0. 232941

Regression Output:
Constant 2.482728
R Squared 0.899092

Regression Output:
Constant 3.775568
R Squared 0.818931
No, of Observations 8
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 0.438

h 1075 1.5 3 0.563
6 1075 1.5 4 0.375
6 1075 1.5 5 0.625
h 1075 1.5 6 0.5
6 1075 1.5 7 0.625
6 1075 1.5 8 0.625
6 1075 1.5 9 0.689
6 1075 1.5 10 0.875
6 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 0.813
6 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
6 995 6 7 1
h 995 6 8 1
6 995 6 9 1. 063
h 995 6 10 1. 188
6 995 6 12 1.375
6 995 6 14 1. 375
h 995 12 3 0.813
h 995 12 4 1. 125
6 995 12 5 1
6 995 12 6 0.75
6 995 12 7 0.875
6 995 12 3 1, 063
6 995 12 9 1
6 995 12 10 1.25
6 995 12 12 1.375
6 995 12 14 1.5
a 910 1.5 2 0.75
6 910 1.5 •2' 0. 688
h 910 1,5 4 0.75
6 910 1.5 5 0.688
6 910 1.5 6 0. 938
6 910 1.5 7 0.75
6 910 1.5 8 0.75
6 910 1.5 9 0.875
6 910 1.5 10 0.938
6 910 1.5 12 0. 938
6 910 1.5 14 0.875
6 910 6 O' 0.688
6 910 6 4 0.75
6 910 6 5 0.813
6 910 6 0.938
910 h 7 0.875
910 6 8 1. 063
6 910 6 9 1. 125
6 910 6 10 1. 063
6 910 6 12 1. 125
6 910 6 14 1.25
~Z
6 910 12 0.75
910 12 4 0.75
6 910 12 5 0.75
6 910 12 6 0.875
6 910 12 7 0.875
910 12 8 1
6 910 12 9 0.875
6 910 12 10 0.875
910 12 12 1. 125
& 910 12 14 1 . 063
810 1.5 0.813
6 810 1.5 3 0.75
6 910 1.5 4 0. 75
6 810 1.5 5 0.75
6 810 1.5 6 0.75
6 810 1.5 7 0.75
6 810 1.5 8 0.875
6 810 1.5 9 1
6 810 1.5 10 ͣ 0.875
6 810 1.5 12 1. 063
^p^ 6
6
810
810
1.5 14 1. 188
6 y. 0,75
6 810 6 4 0.75
6 810 6 5 0.813
6 810 6 6 0.813
6 810 6 7 0,875
6 810 6 8 0.875
6 810 6 9 0.875
6 810 6 10 0. 938
6 810 6 12 1. 063
6 810 14 1.25
h 810 12 3 0.75
6 810 12 4 0.875
6 810 12 5 0.875
6 810 12 6 0.813
6 810 12 7 0.813
6 810 12 8 1
h 810 12 9 0.875
6 700 1.5 r^
0.813
-^r
6 700 1.5 0.625
6 700 1.5 4 0.813
6 700 1„5 5 0.875
t) 700 1.5 6 1
6 700 1.5 7 0.875
6 700 1 . .J 8 0. 75
6 700 1 .5 9 0.875
6 700 1 .5 10 1
6 700 1 .5 12 0,625
6 700 1 ,5 14 0.688
6 700 6 3 0.875
6 700 6 4 0.875
700 6 5 1. 125
6 700 6 6 1. 125
6 700 6 7 1
6 700 6 8 0. 938
h 700 6 9 1.125
6 700 6 10 1. 063
6 700 6 12 1. 125
6 700 a 14 1
6 700 12 3 0.75
6 700 12 4 0, 75
h 700 12 5 0,75
b 700 12 6 0.938
700 12 7 0.875
6 700 12 a 0.875
6 700 12 9 0.875
6 700 12 10 0,75

Velocity Determination............................... 41 Test Object Description.............................. 48

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Velocity Determination............................... 41 Test Object Description.............................. 48

Uploaded by

Copyright:

Available Formats

CONTENTS

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

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

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

CURRENT DESIGN PROCEDURE VERSUS BREATHING ZONE

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

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

DISCUSSION OF MODEL EVALUATION............................87

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

or neoplasm. Occupational health professionals,

particularly industrial hygienists, constantly seek ways to

eliminate these exposures or at least to control them to

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

GENERALIZED DIAGRAM OF METHODS OF CONTROL

SOURCE AIR PATH RECEIVER

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

ventilation is often employed as a viable engineering

Local exhaust ventilation is a means of inducing air

its source. The basic elements comprising a local exhaust

industrial hygiene literature and are illustrated in figure

The hood is perhaps the most important part of the

process as possible. If complete enclosure is not feasible,

shaped and positioned so as to make use of any

to the outside environment. Many different types of

The fan provides the means of inducing air flow by

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

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

VORTEX SHEDDING_. ._...___..........................25