This article was downloaded by: [190.238.138.

34]
On: 20 September 2014, At: 18:12
Publisher: Taylor & Francis
Informa Ltd Registered in England and Wales Registered Number: 1072954 Registered office: Mortimer
House, 37-41 Mortimer Street, London W1T 3JH, UK
Journal of the Air Pollution Control Association
Publication details, including instructions for authors and subscription information:
http://www.tandfonline.com/loi/uawm16
A Plume Rise Model Compared with Observations
Gary A. Briggs
a
a
NASA Trainee , The Pennsylvania State University , University Park , Pennsylvania ,
USA
Published online: 16 Mar 2012.
To cite this article: Gary A. Briggs (1965) A Plume Rise Model Compared with Observations, Journal of the Air Pollution
Control Association, 15:9, 433-438, DOI: 10.1080/00022470.1965.10468404
To link to this article: http://dx.doi.org/10.1080/00022470.1965.10468404
PLEASE SCROLL DOWN FOR ARTICLE
Taylor & Francis makes every effort to ensure the accuracy of all the information (the Content) contained
in the publications on our platform. However, Taylor & Francis, our agents, and our licensors make no
representations or warranties whatsoever as to the accuracy, completeness, or suitability for any purpose
of the Content. Any opinions and views expressed in this publication are the opinions and views of the
authors, and are not the views of or endorsed by Taylor & Francis. The accuracy of the Content should not
be relied upon and should be independently verified with primary sources of information. Taylor and Francis
shall not be liable for any losses, actions, claims, proceedings, demands, costs, expenses, damages, and
other liabilities whatsoever or howsoever caused arising directly or indirectly in connection with, in relation
to or arising out of the use of the Content.
This article may be used for research, teaching, and private study purposes. Any substantial or systematic
reproduction, redistribution, reselling, loan, sub-licensing, systematic supply, or distribution in any
form to anyone is expressly forbidden. Terms & Conditions of access and use can be found at http://
www.tandfonline.com/page/terms-and-conditions
GARY A. BRIGGS, NASA Trainee,
The Pennsylvania State University,
University Park, Pennsylvania
A Plume Rise Model
Compared with Observations
Dimensional arguments are used to predict plume rise for buoyant plumes in both stable
and neutral air, for both calm and windy conditions. Dominant terms are assumed to be
windpseed u, "buoyancy flux" F {proportional to heat efflux), and a stability parameter s
{proportional to potential temperature gradient). Observations presented support the
dimensional analysis predictions, except that for final rise in a neutral atmosphere they
are adeauate only for a conservative estimate of rise. The method is extended to predict
maximum ground concentration of effluent gases in the worst situations {windy neutral
and fumigation) for open country, valleys, and "canyons." These predictions are com-
pared ivith limited observations.
l l ui ne rise is closed related to
air pollution, since the concentration of
an effluent at the ground is reduced when
the effective source height is increased.
There are many empirical and theo-
retical plume rise formulas in the litera-
ture, but none of them are supported by
a quantity of good observations. In
recent years a number of observations
have been made, most of which will be
used here. Still lacking is a thorough
field study in which all chimney param-
eters are known, and temperature pro-
file, wind profile, and plume trajectory as
far downwind as the point of maximum
ground concentration are measured over
the same period.
Dimensional analysis is a relatively
straightforward, but often overlooked,
theoretical approach to physical prob-
lems such as plume rise. Dimensional
arguments were first applied to plume
behavior by Batchelor
1
in 1954 and later
by Scorer,
2
,
3
,
4
whose work will be ex-
tended here to include the effect of
stability.
In certain simplified cases it is possible
to extend the dimensional analysis to
calculating maximum ground concentra-
tions directly from stack and meteoro-
logical parameters. This will be done
here for the windy neutral case and the
case of fumigation with no wind, gen-
erally the two worst possible situations.
There is a greater lack of data here, but
comparison with some observations will
be made.
Plume Rise
Description
A smoke plume takes on the hori-
zontal wind speed quickly after leaving
the chimney, but continues to rise with
its original vertical momentum plus
momentum added by buoyancy, deter-
mined by the heat, average molecular
weight, and weight of suspended parti-
cles of the plume. The rise is critically
affected by entrainment of air, due at
first to the plume's own relative motion;
as this dies out, atmospheric turbulence
is chiefly responsible for mixing. In
stratified air the buoyancy of the plume
and entrained air is altered by the
change in relative density of the am-
bient air with height. In stable air this
acts as a restoring force on the plume,
while in unstable air the plume may be
accelerated to unlimited heights, some-
times even initiating cumulus clouds.
Essential Stack Parameters
The most important step in dimen-
sional analysis is to choose the ] param-
eters which are most significant. Much
simplification is achieved by approxi-
mating the chimney as a point source,
specified entirely by its "momentum
flux" and "buoyancy flux."
Momentum flux is just the efflux
velocity w times the mass efflux at the
chimney, but for our purposes we can
divide it by n and the air density p,
since this is the only way to eliminate
the "mass" unit. We can then define a
term proportional to momentum flux by
r
M
*
T
2 2
where M, T, and r are average molecular
weight, absolute temperature, and stack
radius, respectively. Subscript "s" de-
signates the stack gases, as opposed to
the ambient air.
Similarly a term proportional to
buoyancy flux can be defined by dividing
the rate at which buoyant force is added
to the plume (after mixing with a large
volume of ambient air) by IT and p.
This is given by
F =
T C
pP
T
= 4 . 4 x 1 0 -
cal-sec
2
Qll
where g, C
p
, and Q
a
are gravitational
acceleration, specific heat at constant
pressure, and heat efflux at the chimney,
respectively. When the stack gases
have an average molecular weight and
heat capacity similar to air
;
F is approxi-
mated by
AT
F g wr
1
where AT is the temperature elevation
of the stack gases, T
s
T. Note that F
has the units [IH~
Z
].
When the plume becomes bent over in
the wind, these quantities are diluted in
the axial direction in proportion to the
average windspeed at stack level, u. In
this case the essential stack parameters
become F
m
/u and F/u.
For most hot sources, momentum rise
is negligible compared to the stack
height in a high wind, and will not be
considered here.
Transitional Rise
When the plume first leaves the chim-
ney it is not strongly affected by stabil-
ity, so for a bent-over plume, rise de-
pends chiefly on F/u and t. Dimen-
sional anal} sis then gives the height of
rise b}"
Mi ex (F/uy/*i*/>
Using the relationship x = u t, where x is
the distance downwind of the stack, and
estimating the coefficient of propor-
tionality from Fig. 1, the above equation
becomes
Ah = 2.0 F
x
/*u~
l
x
2/l
In Fig. 1 plume rises from seven
widely different sources are plotted
against downwind distance. Both Ah
and x are made non-dimensional in terms
of L, defined b}'
September 1 965 / Volume 1 5, No. 9
433
D
o
w
n
l
o
a
d
e
d

b
y

[
1
9
0
.
2
3
8
.
1
3
8
.
3
4
]

a
t

1
8
:
1
2

2
0

S
e
p
t
e
m
b
e
r

2
0
1
4

The stack plume trajectories plotted are
from Bosanquet, et al.,
r>
Csanady,
6
and
Stewart, et al.
7
at Harwell. Also plot-
ted is rise data from lard-pail type oil
burners of Ball
8
and of exhaust clouds
from horizontally fired rocket motors of
Van Vleck and Boone.
9
Because the wind
was measured much lower than plume
level for the Harwell and rocket exhaust
data, wind speeds were adjusted by esti-
mating roughness and assuming the
logarithmic wind profile generally found
in neutral conditions. All the data fit
the above formula reasonably well,
especially considering that the source
sizes vary by three orders of magnitude.
Source data are shown in Table I.
Rise in Stable Air
The measure of hydrostatic stability
of the atmosphere is the lapse rate. The
most logical temperature to use is the
potential temperature 9, defined as the
temperature air would have if it were
slowly brought adiabatically to a stand-
ard pressure of 1000 mb. Thus poten-
tial temperature allows for the cooling of
rising air due to expansion (since pres-
sure falls with height). Air displaced
vertically by the distance dz then be-
comes cooler than the surrounding air by
the amount (bO/bz)dz. The restoring
acceleration per unit displacement then
becomes
_ g_ be
S
~ Tbz
In the atmosphere be/bz is relate 1 to the
temperature gradient bT/bz bj
r
be
dz
bT
+ 5.4F/1000 ft
Note that s has the units [t~
2
]; physi-
cally it is the square of the angular
frequency of the oscillation of a balloon
or air parcel about its mean altitude in
stable air. If we use s as a stability
parameter for predicting plume rise, it
turns out that the time required for the
plume to level off is proportional to s~
1//2
.
In calm air, the dimensional analysis
solution for final height is readily found
to be
Ah
Experiments in an ice rink by Crawford
and Leonard
10
indicate a coefficient of
proportionality of about 5.25, while the
experiments of Morton, Taj'lor, and
Turner
11
give a constant of 4.25. There
are no observations of vertical rise in the
atmosphere available. As a compro-
mise, we may take
Ah = 4. 7/ <
7l
/ 4
S
-
3
/s
In windy, stable air the dimensional
prediction is
Ah = 2.6 I
The constant has been estimated from
Fig. 2, where heights from two sources
and corresponding stability parameters
were made non-dimensional and plotted.
The potential temperature gradients
were averaged through the layer of
plume rise. The data from the Colbert
plant of TV A (Gartrell, Thomas, and
Carpenter)
12
fit the above power law
well, especially at higher wind speeds.
The most stable runs from the Moses
and Strom data
13
show more scatter;
here the source was three orders of mag-
nitude weaker. Leveling is very evi-
dent in the Colbert data, with the plume
maintaining nearly a constant height
from half a mile to as far as nine miles
downwind.
Rise in Neutral Air
In a neutral atmosphere the rise is
probably never definitely terminated,
but it is impeded by mixing due to
atmospheric turbulence. The rise t hat
has the most meaning is usually t hat at
the point of maximum ground concen-
tration.
In a high wind, plumes of small and
moderate sized plants are confined to the
surface layer of t he atmosphere, char-
acterized by constant stress with height.
The stress between "slices" or "layers"
of the moving air is not due to wind, but
rather to the combination of turbulent
eddies which transport momentum and
wind shear. If we assume stress r a
function only of air density, wind shear
bu/bz, and height z, by dimensional
analysis we must have
bu
bz~
p \Z
define-4/ = u*
p
U KZ h
bz bmz
where the constant of proportionality k
is called the von Karman constant,
found empirically to be about 0.4, and
where u* is called the friction velocity.
Since in the surface layer both r and p
vary little with height, u* can be con-
sidered constant and the wind profile can
be found by integrating the above equa-
tion :
u*
u = - In z + constant
u = In -
k ZQ
The constant of integration is defined in
terms of z
0
, called the roughness length.
It can be found empirically by measuring
the average wind at two heights, and is
usually of the order of Vio the height of
surface features such as trees and build-
ings. The above logarithmic profile is
well supported by observations in neu-
tral conditions.
In the surface, layer of a neutral
atmosphere, the statistics of the turbu-
lence are adequately specified by u* and
the height, except perhaps very close to
the ground. Then the plume rise at the
point of maximum ground concentration
should be a function just of F/u, u*, and
stack height h
s
. Dimensional analysis
yields
F
M
*- * I vv i<v i t s
u u*
2
h
where / is a function of the nondimen-
sionalized chimney height. The inten-
sity of the small turbulent eddies respon-
sible for mixing the plume does not
change rapidly with height, so/ is a weak
function. Since we usually are most
interested in rise when it is somewhat
comparable to stack height, in this
limited range we can approximate / by a
constant and say
F
Ah
Unfortunately plume rise observers in
the past have measured wind at only one
height, so it is not possible to determine
z
0
and u*. However u is not strongly
dependent on height and roughness, and
is proportional to u*, so a crude approxi-
mation of our original function becomes
F
Ah cc
u
s
If we assume the time t max it takes
the plume to reach the point of maxi-
mum ground concentration to be a func-
tion of the total height h and u*, then
t
X
max
max
X
u
max
u
- h oc
oc
In
h
u~*
h
where u is the average wind at plume
level and x max is the distance down-
wind at which maximum ground concen-
tration occurs, observed to be in the
order of 10 h. Because plume observa-
tions are seldom carried out as far as x
max and no ground concentrations are
measured concurrently, there are no
data from which to properly determine
the constants of proportionality in the
above equations. It can be said that
the rise is of the order of 10Wu~
s
, and by
inspection of Fig. 1 a conservative value
of plume rise might be
Ah >400
'F
Strictly speaking the above reasoning-
applies only to low plumes, since the
surface layer is generally only 10 to 100
m high. However in a high wind
plumes even from large plants do not
rise so high as to be in a greatly different
type of turbulent regime. Stress and
consequently u* fall gradually off above
434 Journal of the Air Pollution Control Association
D
o
w
n
l
o
a
d
e
d

b
y

[
1
9
0
.
2
3
8
.
1
3
8
.
3
4
]

a
t

1
8
:
1
2

2
0

S
e
p
t
e
m
b
e
r

2
0
1
4

I 0
3
- -
Ah +
L
10'
-tI 1II I I 1 I
-iII I I I 11
iI I I I I I
10
Tig. 1. Observed plume rises vs Ah = 2.OF
x
J
z
v
x
x
2
^
3
(dashed line).
the surface layer, so the above equations
should still be valid as approximations.
Maximum Ground Concentration
Neutral, Windy Day
At a plant with a low stack the highest
ground concentrations usually occur on
a windy day, since the plume is brought
down to a low height. Since the wind
causes much mechanically induced tur-
bulence near the ground, the air becomes
well mixed and nearly neutral, and the
plume is efficiently diffused toward the
ground.
At some point downwind of the stack
maximum ground concentration is
reached. Assuming that at this point
the cross section of effluent concentra-
tion always has the same shape, the
ground concentration x should be a func-
tion of the mass efflux of pollutant Q, u,
and h. Both dimensionally and by
physical reasoning it can be shown that
at this point
Q
ulr
At a low wind the plume rise is high so
ground concentration is negligible. At
a very high wind the total plume height
h approaches the stack height A, but the
I Mi l l H 1 1 1 1 I I I I | H 1I I I I I I -i H Hh
++
Ah --
IO 1 Mi l l
O GARTRELL, THOMAS AND CARPENTER
A MOSES AND STROM
Fig. 2. Observed plume rises (stable with wind) vs Ah = 2.6 (F/J)
1
^ s "
1
/
3
(solid line).
September 1965 / Volume 15, No. 9
10-z
435
D
o
w
n
l
o
a
d
e
d

b
y

[
1
9
0
.
2
3
8
.
1
3
8
.
3
4
]

a
t

1
8
:
1
2

2
0

S
e
p
t
e
m
b
e
r

2
0
1
4

pollutant is diluted in proportion to the
wind. It is at some intermediate wind
speed u max at which maximum ground
concentration is observed. If we as-
sume the rise to be given by
h = h
s
+ 400 -
u max is found by straightforward
differentiation to occur at
u max =
6
U )
The maximum ground concentration
at this critical wind speed then becomes
given by
Trace
No.
Table IDafa for Sources Shown in Fig. 1
Source
, ( i o. =0 F (
ft
-,)
\ sec/ \sec
3
/
1
2
3
4
5
a
D
7
8
9
10
11
12
13

Bosanquet No. 1
Bosanquet No. 2
Bosanquet No. 3
Harwell
Van Vleck
Csanady
Ball
153
7
158
110
720
273
1
6,900
326
7,130
4,830
31,600
12,000
42
14
33
26
19
10
1 O
lo
20
21
32
46
7-9
16-19
15-23
3-13
Table IIMaximum Ground Concentrations of A
1
The constant of proportionality is esti-
mated from data in Tables II and III to
be about 0.007 when ground concentra-
tion is averaged over a half hour or more.
For a 15 minute average, peak concen-
tration may be several times this
amount, as large thermal eddies some-
times persist this long, causing varia-
tions in plume rise. In unstable air,
while the average gfound concentration
is less than that for neutral air, momen-
tarily very high concentrations may be
recorded as thermal eddies occasionally
carry the plume to the ground near the
chimney.
Fumigation with No Wind
For most plants maximum ground
concentrations of effluent are likely to
occur at fumigation after a c m night
with a steep ground inversion. By
"calm" we mean that the plume rises
essentially vertically, although it may
bend over at stratification level. Such a
plume reaches a limiting height as given
by the formula for calm, stable air, and
then spreads out more or less evenly in a
thin layer. The rate of spreading
should also be a function of F and s.
Fumigation occurs after sunrise when
convection due to ground heating builds
up to the level of the stratified smoke,
mixing it evenly down to the ground.
The concentration should then be in-
versely proportional to the height of
plume rise through the nighttime inver-
sion. Applying dimensional analysis we
find
X max
Q
The constant of proportionality is
conservatively estimated to be of the
order 0.05 from very limited data given
below
r
; this equation should be tested
with more complete data before being
relied upon. The area of spread corre-
sponding to the above constant would be
where At is the time during which the
smoke accumulates in the stratified
Plant
Oak Ridge
Harwell
Type of Average
peak 15-min
"average" 15-min
"peak"
peak 40-min
Wind, Height
19 ft/sec at 154
ft
19 ft/sec at 154
ft
12 m/sec at 35 in
5 m/sec at 1 m
u max
33 ft/sec
11 m/sec
X max F
l
/s
Q
0.019
0.012
0.012
0.006
layer. Note that s in the above equa-
tions should be determined from the
average potential temperature gradient
through the layer of plume rise during
the inversion, not at the time of fumiga-
tion.
The above formulas are valid only for
open country, since no topographical
interference has been taken into account.
Unfortunately there are no data readily
available for other situations, but for
future reference we can write dimen-
sional analysis formulas for fumigation
in idealized valleys and "canyons." In
any case the maximum ground concen-
tration at fumigation is inversely pro-
portional to the total plume rise in the
inversion.
In a valley with very steep sides or a
canyon with a narrow entrance, the
spread of the smoke may be inhibited by
the topography, causing higher ground
concentrations. If we assume the
smoke to spread evenly and the maxi-
mum velocity at which the smoke can
spread or "escape" to be a function of F
and s, then the maximum ground con-
centration possible is proportional to
where IF is the effective width of the
valley or can}
r
on entrance at the smoke
level.
For a completely closed in canyon, the
problem becomes quite simple if we as-
sume completely uniform mixing up to
the top of the stratified smoke at fumi-
Table IIIMaximum Ground Concentration of SO. at TVA Plants
Plant Data:
Neutral:
Fumigation:
r, ft
w, ft/sec
7',, F
F, ft
4
/sec
3
h
lt
ft
period of data
# days u > 15 mph
ii at xmax, mph
u max, mph
NQ, tons/day
N
/(A)
X max, ppm
x maxf/s/i//!
/(A')Q
# days u < 5 mph
$, tons/day
/i = fe. + 4.7FV4S-V*
X max, ppm
x max F
1
/
i
s~h
l
^
i
Q
7
48
290
23,000
250
3/56-10/56
13
18-26
39
558
8
5. 0
1.0
0.006
1
84
1250
0. 4
0.05
8. 2
47
290
31,000
300
3/55-4/58
11
19
41
469
4
3. 3
0. 8
0.007
9
135
1400
0. 6
0.06
12.5
44
290
67,000
500
12/58-1/61
3
27
44
787
2
1.9
0. 4
0.005
12
333
1800
0. 6
0.04
436 Journal of the Air Pollution Control Association
D
o
w
n
l
o
a
d
e
d

b
y

[
1
9
0
.
2
3
8
.
1
3
8
.
3
4
]

a
t

1
8
:
1
2

2
0

S
e
p
t
e
m
b
e
r

2
0
1
4

gation. Such a "box" model gives
Q At
X max =
A (h
s
where At is the duration time of the in-
version and A is the average area of the
canyon.
Concentrations may be less in a valley
if there is always a minimum drainage
wind u diluting the plume downvalley.
For a narrow valley we can use a simple
"trough" model using the height of rise
given by the formula for the stable case
with wind, giving
Q
Y i nnv =
X
' ' u W (hs + 2. 6(/ ' 7M)
1
/ ' S-
1
/ )
Comparison with Observations
Unfortunately maximum ground con-
centration data complete with stack and
meteorological parameters are not abun-
dant, but enough exist that limited
comparisons can be made here.
Table II shows the constant of propor-
tionality in the neutral, windy ground
concentration formula for several differ-
ent types of averages observed at reac-
tors in Harwell, England
7
and in Oak
Ridge, Tennessee.
15
Table III shows the constants of
proportionality corresponding to maxi-
mum ground concentrations observed at
three TV A plants, for both the neutral,
windy and the fumigation cases. High-
est concentrations tended to occur dur-
ing either low or high winds; u > 15
mph was taken to indicate neutral,
windy conditions while u < 5 mph was
assumed to be a case of fumigation. All
observations represent 30 minute aver-
ages, and only days during which the
concentration surpassed 0.1 ppm are
included. No lapse rate data are given
up to plume level, so AT/Az is assumed
to be 2.7F/1000ft. giving = 5x lO"*-
sec~
2
. Fortunately it is not very criti-
cal, e.g., a temperature gradient of
10F/1000 ft would give about a 30%
lower value for the constant of propor-
tionality for the fumigation case.
The number of stacks N at these
plants is more than one, so that in the
neutral case the plumes combine to assist
each other in over-all rise; as a result the
ground concentration will be less than a
linear function of N, and will be assumed
to be multiplied by f(N) as given in Fig.
7 of "StacksHow High?"
16
. In stable
air the plumes do not co-operate so much
in increasing each other's rise, since the
final rise is most critically affected by
mixing just above the chimney, before
the plumes have combined. If the
plumes rise to different levels, they will
not interfere with each other's spreading
and the concentration at fumigation
could be expected to be a superposition
of the concentration due to each stack
separately, mixed evenly up to the level
of highest rise. If the plumes rise to the
same level the situation is more complex,
since in spreading each plume could
"push" the other aside, making the area
of spread proportional to N but leaving
the concentration unaffected. For the
purpose of estimating the constant of
proportionality in the fumigation form-
ula, the conservative assumption to
make is that the concentration does not
depend on N; the truth undoubtedly lies
somewhere between no dependence and
linear dependence.
Summary
The rise of most hot plumes is caused
almost entirely by buoyancy due to
heat; the most important stack param-
eter for such plumes is the "buoyancy
flux" F, which is proportional to the heat
flux. When the plume is bent over in
the wind, it is diluted along its axis in
proportion to the average wind speed u
at plume level, so the buoyancy param-
eter becomes F/u. In stratified air the
plume's buoyancy is dissipated due to
the stability, which can be characterized
by a stability parameter s proportional
to the potential temperature gradient.
When the atmosphere is neutral the
plume is diffused by atmospheric turbul-
ence, whose intensity is a function of
ground roughness, height, and most
importantly, wind speed.
Retaining only the dominant terms
above for each situation, and applying
dimensional analysis with constants
determined from data, the following
plume rise formulas were found:
transitional Ah = 2.0F
1
^u~
1
x
2//a
stable, calm Ah 4.7.F
1
/
4
s~
3/
'
8
stable, wind Ah = 2.6 1 )
neutral Ah > 400 -,
ir
The transitional rise formula applies to
all-bent over plumes before they ap-
proach their final height. In the stable
formulas, the temperature gradient is
averaged through the layer of plume
rise, i.e., from height h
s
to (h
s
-f Ah).
In the neutral case, no data are available
which go downwind far enough that
definite leveling is evident, so only a
minimum rise can be estimated from
observations.
Utilizing the above plume rise form-
ulas, it is possible in special cases
to extend the dimensional analysis to
maximum ground concentrations, in
particular to the two worst cases (neu-
tral windy, and calm with fumigation).
In the neutral case the highest concen-
tration was found to be
X max = 0.007 jr
/3})
j/
3
which occurs as the wind speed at plume
level approaches 12.6
In the fumigation case it was assumed
that the concentration is inversely pro-
portional to the total plume rise during
the inversion and is a function also of Q,
F, and s, yielding
X max ~ 0.05
Q
where the temperature gradient is
averaged through the layer of plume rise
before the inversion breaks'down. The
coefficient of proportionality was deter-
mined from very limited data, and more
observations are needed to prove the
validity of this model.
Acknowledgments
This work was done with the support
of the Pennsylvania Department of
Health, the Atmospheric Turbulence
and Diffusion Laboratory of the United
States Weather Bureau (Oak Ridge,
Tennessee), and a National Aeronautics
and Space Administration Traineeship.
I would like to thank Dr. Frank Gifford
(Oak Ridge), Dr. Frank Pasquill (Me-
teorological Office, Bracknell, England)
and Dr. Hans Panofsky (Pennsylvania
State U.) for their encouragement and
enlightening discussion, as well as the
Tennessee Valley Authority for permis-
sion to use their SO
2
data.
Appendix: the Dimensional
Analysis Method
Given a physical situation in which a
set of dimensional quantities are related
only to each other, it is possible to infer
immediately something about the form
of the equation describing this relation-
ship, just on the argument that the equa-
tion must be dimensionally correct.
For a simple example, suppose the
speed of sound c in a gas is a function
only of state, i.e., of pressure p, density
p, and temperature T. We first write
the dimensions of these quantities:
[c] = It'
1
[p] = ml-*
[T] = 6
Inspecting these units we see that for
dimension correctness we can only have
V
Temperature drops out in this formula-
tion since the units of c do not include
temperature. Of course it can be
introduced through the equation of
state, which incorporates a dimensional
constant (the universal gas constant) to
balance the units. Note that while di-
mensional analysis can predict a power
law in such a case as this, it can-
September 1 965 / Volume 1 5, No. 9 437
D
o
w
n
l
o
a
d
e
d

b
y

[
1
9
0
.
2
3
8
.
1
3
8
.
3
4
]

a
t

1
8
:
1
2

2
0

S
e
p
t
e
m
b
e
r

2
0
1
4

not predict the dimensionless constant
of proportionality; this must be found
either empirically or through a more
sophisticated theoretical model.
When there are more variables the
problem becomes more complex, but still
oome simplification is possible. The pi
theorem of dimensional analysis states
that if we have n variables which are
functions only of each other, who share
among themselves (/ independent dimen-
sions, there are only (?i d) independ-
ent nondimensioiial ratios. By inde-
pendent, we mean (hat one cannot be
derived from the others.
Consider for instance the drag force D
on a sphere of radius .11 in a fluid of
density p. viscosity //, and mean flow
velocity V. The interdependence of
these quantiries i> expressed by
D = function (p, V, R, n)
Since there are rive quantities and three
basic dimensions, there are only two
independent nondimensional ratios.
We might arbitrarily choose them to be
/)
P
VR
and
P \'-H- fi
so the first equation can be rewritten as
unknown function of only one variable
instead of four:
D =
P
V*R* function (^-
= y
2 P
V'(TTR~)C
Thus we get the familiar drag equation,
where the drag coefficient C
D
is a func-
tion only of Reynold's number (p VR/n),
and can be determined empirically in a
wind tunnel.
While dimensional analysis can often
greatly simplify complex problems, such
as those we find in fluid dynamics, it is
well to warn that one cannot expect the
right answer from the wrong assump-
tions, even if it is dimensionally correct.
The critical step in this method is to
choose the most relevant parameters,
which requires a good intuitional under-
standing of the problem. Combined
with differential equations, which by
themselves may be too difficult to solve,
dimensional analysis can be a powerful
tool.
Detailed treatments of the subject can
be found in references 17 through 19.
REFERENCES
1. G. K. Batchelor, "Heat Convection
and Buoyancy Effects in Fluids,"
Quart J. Roy. Meteorol. Soc, 80: 339-
358(1954).
2. 11. S. Scorer, "Natural Aerodynamics,"
pp. 186-217, Pergamon Press, London
(1958).
3. R. S. Scorer, "The Behavior of Chim-
ney Plumes," Intern. J. Air Poll, 1:
198-220(1959).
4. R. S. Scorer, "The Rise of Bent-Over
Hot Plumes," in "Advances in Geo-
physics," 6, Frenkiel and Sheppard,
eds., pp. 399-411, Academic Press,
New York (1959).
5. C. H. Bosanquet, W. F. Carey, and
E. M. Halton, "Dust Deposition from
Chimney Stacks," Proc Inst, Mech.
Engrs., 162, 355-365 (1950).
6. G. T. Csanady, "Some Observations
on Smoke Plumes," Intern. J. Air
Water Poll, 4: 47-51 (1961).
7. N. G. Stewart, H. J. Gale, and R. N.
Crooks, "The Atmospheric Diffusion
of Gases Discharged from the Chimney
of the Harwell Pile," Intern. J. Air
Poll, 1:87-102(1958).
S. F. K. Ball, "Some Observations of
Bent Plumes," Quart J. Roy. Meteorol.
Soc, 84:61-65(1958).
9. L. D. YSLTL Aleck and F. W. Boone,
"Rocket Exhaust Cloud Rise and Size
Studies Hot Yolume Sources," pre-
sented at the 225th National Meeting
of the American Meteorology Soc,
Jan. 29-31, 1964, Los Angeles, Cali-
fornia.
10. T. V. Crawford and A. S. Leonard,
"Observations of Buoyant Plumes in
Calm Stably Stratified Air," / . Appl.
Meteorol, 1: 251-256 (1962).
11. B. R. Morton, G. I. Taylor, and J. S.
Turner, "Turbulent Gravitational
Convection from Maintained and In-
stantaneous Sources," Proc. Roy. Soc.
London, A, 234, 1-23 (1956).
12. F. E. Gartrell, F. W. Thomas, and S.
B. Carpenter, "Full Scale Dispersion of
Stack Gases: a Summary Report,"
Tennessee Valley Authority, Chat-
tannoga, Tennessee (1964).
13. H. Moses and G. H. Storm, "A Com-
parison of Observed Plume Rises with
Values Obtained from Well-Known
Formulas," / . Air Poll. Control. Assoc,
11:455-466(1961).
15. U. S. Weather Bureau, "A Meteoro-
logical Survey of the Oak Ridge Area,"
USAEC report ORO-99, 554-559
(1953).
16. F. W. Thomas, S. B. Carpenter, and
F. E. Gartrell, "StacksHow High?"
/ . Air Poll. Control. Assoc, 13: 198-
204(1963).
17. P. W. Bridgman, "Dimensional An-
alysis," Yale University Press, New
Haven (1931).
18. D. Ispen, "Units, Dimensions, and Di-
mensionless Numbers," McGraw-Hill,
New York (1960).
19. H. L. Langhaar, "Dimensional An-
alysis and Theory of Models," Wiley,
New York (1951).
(No Reference 14)
CLASSIFIED
POLYTECHNIC INC.
An affiliate of
Walter C. Me Crone Associates
493 E. 31st St. Chicago, III. 60616
Telephone (312) 842-7100
CHEMISTS AND ENGINEERS
CONSU LTATION IN:
AIR POLLU TION ANALYSIS
MICROSCOPIC IDENTIFICATION OF
PARTICU LATES
AIR POLLU TION SU RVEYS
INDU STRIAL ZONING AND PER-
FORMANCE STANDARDS
Marvin A. Salzenstein, P.E.
President
AIR POLLUTION ENGINEERSal-
ary range $7920-110,104, may start
at $8724, depending upon qualifica-
tions. Headquarters Denver. Re-
quires engineering degree and 3 years
experience. Contact Dalton Roberts,
Administrative Officer, Colorado State
Department of Public Health, 4210
East 11th Avenue, Denver, Colorado
80220.
JACKSON & MORELAND
DIVISION OF U NITED
ENGINEERS &. CONSTRU CTORS INC.
Air Pollution Evaluation and Control
Services for Utilities and Industrials
DESIGN, SPECIFICATION, AND EVALU-
ATION OF SYSTEMS AND EQUIPMENT
CONSTRUCTION
AIR SAMPLING SURVEYS
SUPERVISION OF
AND OPERATION
Boston Washington
ROY F. WESTON, INC.
Environmental Science
and Engineering Consultants
WaterSewageRefuse
Industrial Wastes
StreamAir Pollution
Industrial Hygiene
Community Planning
SurveysResearchDevelopment
Process EngineeringPlans and
Specifications Operation Supervision-
AnalysesEvaluations and Reports
NEWT0WN SQU ARE,
PENNSYLVANIA
WI LLI AM T. I NGRAM
Consulting Engineer
Environmental Engineering
PlanningDesignResearch
WaterWa stesRefuse
Air Pollution Control
Industrial Health
Laboratory Services
Offices:
East Coast
7 North Drive
Whitestone, N.Y. 11357
West Coast
6426 Regent St.,
Oakland Calif. 94618
438
Journal of the Air Pollution Control Association
D
o
w
n
l
o
a
d
e
d

b
y

[
1
9
0
.
2
3
8
.
1
3
8
.
3
4
]

a
t

1
8
:
1
2

2
0

S
e
p
t
e
m
b
e
r

2
0
1
4