•
AMERICAN PETROLEUM INSTITUTE
Division of Refining
1801 K Street, N.W.
Washington, D.C. 20006
Nothing contained in any API publication is to be construed as granting any
right, by implication or otherwise, for the manufacture, sale, or use in connection
I
with any method, apparatus, or product covered by letters patent, nor as insuring
anyone against liability for infringement of letters patent.
API publications may be used by anyone desiring to do so, and every effort
has been made by the Institute to assure the accuracy and reliability of the data
contained in them. However, the Institute makes no representation, warranty, or
guarantee in connection with API publications and hereby expressly disclaims
any liability or responsibility for loss or damage resulting from their use; for any
violation of any federal, state, or municipal regulation with which an API publica
tion may conflict; or for the infringement of any patent resulting from the use
of an API publication.
•
FOREWORD
• iii
CONTENTS
• v
',1ANUAL ON DISPOSAL OF REFINERY WASTES
VOLUME ON ATMOSPHERIC EMISSIONS
Typical Regulatory Requirements These publications define the contaminants and describe
their properties and their effects on human health and
1': i,'c!cr~d Air Quality Standards, or the current welfare, animals, and vegetation. The human health
_,::1"011[\, are outlined in the Federal Register. I * category includes toxicology data and effects on the
',,' .1."C standards appear in tabular form in the Air respiratory and nervous systems. Standard test methods
.\ " r \ ('irS of May 3,1971, and are reproduced in are also included.
i ,doic 61 also gives the levels at which the
.. ""'1,111 contaminants affect human health and
I able 61 ~Federal Air Quality Standards and levels at Which Effects Show in Humans
Federal Air Quality Standards Levels at Which Effects Show'
~~~"~ .. ~
Primaryt Secondaryt Human Health Welfare§
fIn);.:uIJIt:\ .
Annu~l~ ~~llJl1t:tric mean, [J.g per ell m 75 60 80 60
'h., ... ·Ilr Cl)nc, :1.g per ell mt 260 150 200 150
. s.uur (h,des:
Annu.d ,lfIth aver. f.Lg per ell m 80 (0.03 ppm) 60 (0.20 ppm) 115 85
'h, ~~hr (llne, ag per ell mil 365 (0.14 ppm) 260 (0.1 ppm) 300 285
\f,l, ;hr cune. tJ.g per ell m!1 1,300 (0.5 ppm)
C.l.t~·n \1t1no\ll .'
,:,,u
\ . ut::.
~':lr ('{me, mg per ell ml!
J\ I'il[ ":unc, mg per ell mil
10 (9 ppm) 10 12
40 (35 ppm) 40 58
"x"'h,hcmical Oxidants'
0: ' .
t'·nr l11ax. :1.g rer ell mi i 160 (0.08 ppm) 160 130 100
If • ..!r".Jrthln~'
"..1.\ '·hr ('!.)·nc 69 am,
. ..:. per cu m 160 (0.24 ppm) 160 100
", ,
'''cn (hides:
,d .trith aver, iJ.g per eu m 100 (0.05 ppm) 100 117 4701J[
• '" 1\ ~l\'l'r, 'J..g per ell m 118
Fnkral criteria.
hv June I, 1975.
~ l'lllit unenforcement.
Ili,,.lnd crops.
1 " h<' ~"<'<'cded more than once a year
'<cd \'1\ d,lIllagC to vegetation unly. .
(iI
62 DRW MANUALATMOSPHERIC EMISSIONS
tUrtluklll ;Ilr motion influences effective stack The atmosphere is either stable or unstable depending
Jill' [lossibility of turbulence should upon its ability either to resist or to augment vertical
C1:Je\.lt.!crl"d : 11 [ill..' "tack design.
motion, Atmospheric stability or instability is directly
related to temperature gradients or lapse rates. If the
existing lapse rate is less than the adiabatic lapse rate,
the air is stable, if it is greater, the air is unstable. There
1:~"" . ,:,' cdll he dclined as the vertical temperature are various degrees of conditional stability and neutral
V ... "'~: ':: ck" tl) the vertical temperature distribu ,t', bility in which the existing lapse rate is greater than
':O.t:r'l'.
::"""piJere, The temperature of the atmos the moist adiabatic lapse rate, but less than the dry
I',tlls, or "Ia pses," with increasing adiabatic lapse rate. Also, if a rising parcel of unstable
'" :",' ,,( the increasing distance from the air enters a new stratum where the existing lapse rate is
"'rface of the earth, The normal or lower than the adiabatic rate, the previously unstable
: Ie' is 3,5 F per 1,000 feet (6.5 C per air will become stabilized at that height.
,:11 that is not in vertical motion. This In a mass of stable air there is little or no vertical
:','rCltllre gradient shows considerable varia activity to sweep it aloft. Smoke, dust, minute water
DRW MANUALATMOSPHERIC EMISSIONS
•
6.2.6 INFLUENCE OF LOCAL TERRAIN
Conditions
Daytime Insolation Nighttime Conditions
~.~.~... ,~... Local terrain can have a pronounced influence on
Surface Thin Over
Wind cast or < 3/8 weather and weather characteristics such as air stability.
Speed Strong Moderate Slight > 4/8 Cloudiness' A wind proceeding over a level area changes direction
(m/sec) Cloudi on reaching hills, high structures, or mountains so that
ness *
2 A AB B the obstruction can usually be cleared. Air that is
2 AB B C E F
4 B BC C D E
forced to rise to pass over such obstructions mayor
56 C CoD D D D may not become unstable depending upon the speed of
6 C D D D D ascent, humidity conditions, and contrast between the
"'The degree of cloudiness is that fraction of the sky above the local apparent
horizon that is covered by clouds. The neutral class (D) should be assumed for cooling rate of the rising air and the lapse rate of the
heavy overcast conditions, day Of night.
surrounding air. In the Los Angeles area the incoming
cool marine air does not have sufficient momentum to
clear the mountains, resulting in inversion and very
6.2.5 INVERSIONS stable air.
As discussed previously, the type of terrainrocky,
Temperature inversions occur when the temperature sandy, or woodedinfluences the magnitude of daily
of the air increases with altitude. A common cause of temperature variations. The smaller the daily tempera
inversion is rapid cooling of the ground at night by ture variation, the greater the tendency toward stable
radiation. The surface air is then cooled by convection air.
so that the air temperature some distance from the Areas of strong topographic relief also experience
ground is higher than the air temperature at or near air currents (breezes). These breezes are usually upward
ground surface. Inversion can also be caused by air during the day when the valley floor is relatively warm,
masses, or fronts, and occasionally by turbulence. and downward at night when the obstruction cools
In Los Angeles it is caused by cool marine air forcing faster than the sheltered valley below.
its way under warmer continental air. Both masses of The major effect of local terrain on weather is the
•
air are then trapped by the adjacent coastal mountains. extent to which it introduces instability into air move
When an inversion occurs, the existing lapse rate is ments or promotes turbulence and eddy currents.
negative, and therefore always less than the adiabatic These in turn directly affect gas dispersion.
DISPERSION OF GASES 65
. ,'I plumes can result from these atmos 4. Fumigation is characteristic of inversion above stack
", 'Ii '. and it is possible to categorize the level. Under these conditions the plume tends to dis
'. I h,)ugh the following descriptions are perse downward but is prevented from upward disper
'I :lrc generally applicable, particularly
sion by the inversion layer.
'.[.lllvely low plumes (e.g., below 1,000
5. Lofting is characteristic of inversion below stack
I.,", ;lrc graphically portrayed in Figure
level. It is the reverse of fumigation and somewhat
similar to fanning, although it occurs at a lower wind
, :, ,",:lracteristic of unstable conditions. The velocity.
1"'"
""'
DISTANCE • DISTANCE •
'''~'iG INSTABILITY ("LOOPING") SURFACE STABILITY ("FANNING")
r
I
'""iii
"
DISTANCE DISTANCE_
'l EAR NEUTRAL ("CONING") INVER310N ALOFTABOVE STACK ("FUMIGATION")
I
        
'""
iii
"
DISTANCE
SURFACE INVERSIONBELOW STACK("LOFTING")
•
meander wildly.
and direction with height, must also be considered.
Wind and associated characteristics affect plume Figure 64Plume Dispersing in a Field of Large Eddies.
dilution, plume diffusion, and plume rise, although
these terms are to some extent interrelated.
OF OBSTRUCTIONS, EDDIES, AND 6.3.5 PLUME RISE AND PLUME RISE FORMULAS
•
"iteicilt protection against downwash. To obtain this
comparisons indicate that all formulas are approxima
Till,':1i stack exit velocity, either a venturi nozzle design
tions at best and no one formula is universally applic
:1 Ihe stacK exit or an inward coning of the stack at the able.
·t.tck exit can improve velocity at the expense of a Briggs 9 found that buoyant plumes follow the "¥3"
'!""11 I'rcssurc drop. Law for a considerable distance downwind of the stack
1
68 DRW MANUALATMOSPHERIC EMISSIONS
I
(but not exceeding ten stack heights). This can be
expressed as:
Where:
t..H = (l.6F)li(x)%
I'
(3)
This formula is on the conservative side, thus pre
senting a slight safety factor. It should be reasonably
accurate for unstable conditions by using a factor of
1.1 or 1.2 and for stable conditions by using a factor
of 0.8 or 0.9. These factors are recommended by
•
Holland. 12 Conditions of extreme stability (inversion)
F = value proportional to the rate of buoyancy
(heat) emission. or extreme instability (turbulence) present the usual
x = horizontal distance downwind of the stack. problems.
I' = wind velocity. The critical wind speed for any given atmospheric
stability condition is that speed which causes a maxi
A regression formula 4 was developed to which the mum ground level concentration of the pollutant.
other formulas were compared. This regression formula
Equation (4) and formulas of other authorities indicate
is generally applicable to the total data available, most
that the plume rise (t..H) is an inverse function of wind
of which involved wind speeds in the range of 9 to
speed, conversely, the greater the wind speed, the greater
22.5 mph (4 to 10 m/sec). This formula is not generally
the effect of dilution. These two forces are counter
applicable to wind velocities outside this range nor to
acting. There is a point of equilibrium between these
a wide variety of stacks where local conditions have a
two counteractive effects where the maximum ground
pronounced or predominate effect.
level concentration occurs. This intermediate wind
Anderson, et al., conclude that "no rational choice
speedthe critical wind speedvaries with conditions
can be made between various proposed empirical
of atmospheric stability. This wind should be used for
plume rise formulas" 10 for the determination of effec
design purposes. A method for determining critical
tive stack height. It is the consensus that this observa
wind speed is indicated in Paragraph 6.6.2.
tion is valid for all formulas, empirical and dimensional,
that have been published to date.
Nevertheless, the quest for accurate formulas con 6.3.6 LIMITATIONS AND RELIABILITY OF
tinues apace so that the engineer can design in com
pliance with applicable specifications. The most precise
dimensional analysis is that presented by the U. S.
Atomic Energy Commission II (these formulas have
been reproduced by Anderson, Hippler. and Robin
son 10 in their work for API). These formulas treat
PREDICTIVE TECHNIQUES
wd [ 1.5 + 2.68
I1H = iJ: (I0'p) (T.' i,T") I]
I ( 4)
Variations of ±50 percent between calculated and
observed values of :lH can occur, especially for adverse
Where: meteorological conditions (e.g., a strong inversion).
,\' stack exit velocity. in meters per second. This does not necessarily signify that the parameters in
d stack diameter. in meters. the equations are in error, although constants and
•
I' wind velocity at stack height. in meters per
second. exponential values do vary. It can indicate inadequate
atmospheric pressure. in millibars. or inaccurate measurement of the parameters or that
atmospheric temperature. in degrees Kelvin. the measurements changed with distance downwind.
stack temperature. in degree, Keh in. Ideally, dimensionalized or empirical formulas with
DISPERSION OF GASES 69
,'onstants should be developed for the area It should be noted that under normal flare operation,
," consideration. sulfur dioxide emission is negligible when compared
". 'c·lop equations with suitable constants, con with emissions from refinery stacks in which sulfur
,I [h~ accuracy of the plume rise equations with containing residual fuel oils have been used. Sulfur
::,'" to wind velocity in neutral or nearneutral dioxide assumes serious proportions only when there
1"115 of >lability. Generally, the equations are is an emergency condition such as the loss of a com
,,'lW[[C in low to moderate wind velocitiesthat pressor or the shutdown of the sulfur pidnt.
,:',1 lllPh (2.3 to 9.1 m/sec). The accuracy tends This discussion is concerned with lighted flare
cI ':11 'e somewhat for lower or higher wind veloci operation under design conditions. If the flare is
!" is not generally important because t1H is a unlighted, dispersion is governed by the criteria dis
"I" Ilii'h figure for low wind velocities, thus cussed in this section and in the subsequent section
'" II" '! !!round concentrations. At high wind (see Paragraphs 6.4. I through 6.4.8).
:', dilution effects preclude high concentrations, The calculations of plume rise and atmospheric
1 "line of the plume is occasionally swept to dispersion from stack effluents are relatively com
,l:lnd. plicated. These calculations are even more complex for
III,' rl'asoning with reference to wind velocities flared gas, in which case the stack effluent gas is burned
IIp[llleS to unstable air except that the factor prior to final dispersion and heat release is external
I ".,jiIV usually produces an increase in average to the flare stack.
'. !'!'I~ilt. This is reflected in equation (4) by use of An empirical approach to the problem of plume rise
:111' ;'1 1,1 or 1.2 applied to t1H. Unfortunately, is given in the following equation:
., ,':I;,rl\, with strong instability, the higher stack H = Hs + t1FL + t1H (5)
'I ""'1' 'lot overcome the effect of looping, which Where:
"'[lei uce undesirable ground concentrations at H effective height of plume rise.
I',ilort distances downwind from the stack. Hs = actual stack height.
t1FL = vertical component of flare length.
It a I reg uent occurrence.
t1H = plume rise after burning is complete.
,',:[I"t deviation from these equations is
", '1',','0 wilen there is an inversion above the stack The following paragraphs show that the accuracy of
, :.ilion) where the most precise results should determination of t1FL and t1H is su bject to question.
""hlC, 1'<0 formula wiII accurately indicate the Practically, the flare length can be assumed to be
"c!1elration or breakthrough of an inversion 120 times the flare stack diameter, as indicated in
.. '.IClllon (4) recognizes this by use of a factor API RP 521. 14 This figure is directly applicable at
I ',lJ, I r inversion conditions are frequent or 0.2 sonic velocity, which is the usual basis of design.
II 'C'. ,'1,' 1IIIersions are occasionally experienced, the The flare length varies with velocity and flame tempera
u," III '.\ hat may be termed an inversion factor is ture, which is related to the material being burned.
tr..:"mmencied. Such a factor would depend on local The flare length also varies with the value of Cp/C,
mctenr,)loglcal conditions and cannot be quantified, the ratio of specific heats. It is apparent that the true
tut IIllulJ involve an increased stack height or increased flare length may be greater than or less than the value
\t~,·~ ,'\il velocity, or both. If these factors are such shown.
that the plume rises above the inversion layer, the layer To determine the vertical component of flare length
",II act II> a protective shield. (t1FL) a correction must be made for the horizontal
component, which is related to the angle of incidence
of the flare to the horizontal. This angle is influenced
63.7 BEHAVIOR OF FLARE PLUMES
by the relation of wind velocity to exit gas velocity.
Two methods are presented in API RP 521,14 and the
I,.: Illalor potential contaminant from a flare is recommended method involves the use of the graph
''':iur dio\idc (SO,), and this contaminant is related to shown in Figure 66, from which ~t1 Y (vertical com
"'t ;'I,)rngen sulfide (H,S) content of the flared gas. ponent) can be obtained.
I'H[I"ulllk matter (smoke), odor, and plume visibility The second determination is that for t1H, the burned
,i,,) contribute to pollution. These pollutants can flare gas plume rise. This is calculated by using
1,1111 he eliminated to a large extent by proper Holland's formula: 12
•
.. """ion techniques and by the use of steam to effect
""'k",e combustion. Infrequently heavier hydro
r "IIlh. ,uch as phenolics, and alkylation residues
t1H = :d [1.5 + 2.68 (IOap ) (T.:';:, T}/J (6)
"l,lllllll~ hydrofluoric acids may also be present. This formula is applicable at the theoretical or imagi
610 DRW MANUALATMOSPHERIC EMISSIONS
1.0
f
0.9 ./
. ~.6.LX
~
0.8 / b,L
b,X  5~;; (it)
X
<J
l 0.7
/
T
b, Y =
[I
b,J.,
+E~~12J 0.5 11
L
W
lr~YIIt<;y
b,X= b,!
[I +E ~~)2J
0:
0
0.6
~Il 0.5
\\ b,1=
n
L
0.5
1"
d'~
L:.X
j
W WHERE:
I E.6.X
0 0.4 \ !lW= LATERAL WIND VELOCITY FLAME GEOMETRY
i= IN STILL AIR
'"
<[
0: !lO= EXIT GAS VELOCITY FROM STACK
AND LATERAL WIND
n = NUMBER OF INCREMENTS
~
0.3
"'.
0.2
0.1
'" ?L
B  ~
•
o
o 0./ 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0 1.1 /.2 1.3
RATIO E !lw
!lo
Figure 66Approximote Flame Distortion Due to lateral Wind on Jet Velocity from Flare Stack.
"lilt concentration) from the The Gaussian curve has been found applicable under
',' "liues. Dispersion is most conditions for the determinations of downwind
",Ill, Whcn diffusion occurs, atmospheric concentrations resulting from a stack
,I Ill\)\'C from a region of plume. The dispersion equations herein assume a
"I' lower concentration. Gaussian distribution in both horizontal and vertical
III"table atmospheric con planesthat is, the equations incorporate deviations
k IIlCtlC energy and thermal on a probability basis, ~!I (horizontal) and ~, (vertical),
i,,~ pluilleall of which con and have been developed from the work of Ogura,
T,
612 DRW MANUALATMOSPHERIC EMISSIONS
x
y
z
Q
concentration at any particular point In
space, in grams.
distance downwind.
distance crosswind from xaxis.
vertical distance above ground.
uniform rate of emission, in grams per
•
of atmospheric conditions and only two dispersion second.
parameters are involved.
cr y = standard deviation of plume diffusion in the
horizontal.
The Gaussian distribution curve as applied to the cr, = standard deviation of plume diffusion in the
distribution of atmospheric contaminants is shown in vertical.
Figure 68, in which y (value of the relative concentra tL = wind velocity, in meters per second.
°
tion) varies from to 100 percent when plotted against
values of x  x/cr. The deviation values or dispersion co
For concentrations calculated at ground level,
equation (8) simplifies to:
efficients, cry and cr" are discussed later in this section.
The value of x represents any point directly downwind
of the plume flow, and x represents the distance in
X(x,y,O;H) = ~L
cry cr, tL
exp [  !
2
(l')'J
cry
~ (~),J
meters from x. Characteristics of the Gaussian dis
tribution curve are described in greater detail by exp [  (9)
Turner. IJ
The Gaussian curves retain the same general bell A still simpler equation can be applied to ground
shape at any distance downwind, but with increasing level concentrations calculated along the centerline of
distances downwind the curves become flatter and the plume (i.e., y = 0):
wider. It should be noted that these curves refer to
distribution of contaminants only, as induced by
X(x,O,O;H) = _(l~ exp [  !2
cry cr, tL
(.l!)']
cr,
(10)
dispersionthat is, the dispersion coefficients.
•
For a groundlevel source with no effective plume
The equations for determining the concentration of
rise, the resulting equation is:
gas or particles are given below. The effective stack
height (H) is determined by estimating plume rise and Q
X(x,O,O;O) = r.   (II)
adding the physical stack height. All equations are for cr 11 cr z [J.
particles equal to or less than 20 microns in diameter.'
The accuracy of the mean value of X so determined is
The equations are for a single source of emission at
proportional to the accuracy of the mean value of the
effective stack height (H)t when there is no particle
wind velocity (tL) and is also directly related to the pre
deposition or reaction at the surface. The equations
cision involved in choosing the diffusion coefficients.
are as follows:
Since diffusion (or dispersion) coefficients are based on
X(x,y,z;H) = _ .Q (Y)']

7:2 cru cr~ (1.
exp [ '
2 cr"
a 10minute sampling time, shorter sampling times
yield somewhat higher values for ground level con
•
01
0.0 Xmax
2Q ) (cr,)
= ( e 7r.~H2 ;.;:
(12)
Where:
Figure 68Gaussian Distribution Curve. e = the base of the natural logarithm, 2.718.
DISPERSION OF GASES 613
l11e value of the effective stack height varies with wind In concluding this section on equations, it should
,pecd and stability conditions. Inversion lids, especially again be emphasized that for practical purposes there
,ct appreciable heights (2,000 feet or over), do not is little difference in the various proposed equations.
",Jrmally present a problem since their effect on plume
",persion is negligible except at appreciable distances 6.4.3 PARAMETERS AND DISPERSION COEFFICIENTS
'rom the source. This is not true in the special case of
,,". crsion breakup fumigation. The parameters in equation (13) are the determinants
I nversion breakup fumigation can occur on sunny of the calculated concentrations at any point in space
"l1omings following cool nights; this produces a mild downwind of the source. The parameters are Q, H,
','.crsion. The radiant heat absorbed by the earth heats IJ., X, y, z, and cr u and cr,; all except cry and cr, have been
.,Ir adjacent to the ground by convection so that the previously defined. It is obvious that the accuracy of
"lire layer under the inversion is mixed vertically. the calculated concentrations depends on the accuracy
,oer these conditions, relatively high concentrations with which these parameters are determined. Param
•:"e found at ground levels for periods ranging up to eters x, y, and z can be accurately determined: Q and
'Ile hour. IJ. should present no serious problem, although H can
be troublesome as previously indicated.
,he ground fumigation concentration at a near
Probably the most difficult parameters to determine
':lXimum value can be approximated by:
accurately, particularly if atmospheric stability con
z,(x.)"O;H) =
V2
Q
IJ.crylh l
exp [  ~ (_L)]
cr"1
(13)
ditions are not precisely known, are the standard
deviations (dispersion coefficients) of plume concentra
: Vlrere: tion distribution, :J y (horizontal) and cr, (vertical).
In addition to varying with atmospheric turbulence,
0'0 = the value of cry for stable conditions plus Ys these factors also vary with wind speed and surface
the effective height of emission.
topography and increase with distance from the source.
ill H + 2:J. = h + !:J.H + 2cr •. The data presented later assume relatively open country
and a height not exceeding several hundred meters
Turner's term :J II! is used in equation (13) rather above the ground. The values plotted are more accurate
Ikm cr" so that estimated calculations for XI will not be for suburban than for urban areas.
",,,htly higher than actual concentrations. Under Several authors relate these deviations to only three
""nigation conditions, additional horizontal spreading sets of turbulent conditions (e.g., low, average, and
.:curs when vertical diffusion is limited by an inversion moderate) while others, including Sutton, recognize
layer. Equation (13) is applicable to conditions down four sets of stability conditions. The six stability or
wind of the stack, but not for areas close to the stack turbulence conditions proposed by Pasquill are
base. Its greatest applicability is for relatively long described in Paragraph 6.2.4. This is a more sophisti
downwind distances, perhaps greater than 5 or 6 cated approach (see also Table 61) and is used by
kilometers. CONCA WE: 8 Anderson, Hippler, and Robinson; 10
Equations (I) through (13) can be programmed for ~urner; 13 and many others. Obviously, the actual
the computer to provide both rapid and accurate stability conditions may be at any value (e.g. halfway
solutions to problems. There are many advantages of between moderately unstable and slightlY unstable).
this modern technology; probably the most important The standard deviations, cr, and Gu , are expressed in
is savings in time particularly, when dealing with meters as functions of the distance downwind for each
multiple sources. In turn, the time saving permits of six different meteorological conditions. Graphs for
better economic evaluation of the processing options these conditions are shown in Figures 69 and 610.
available (e.g., whether a reduction in the number of These graphs represent diffusion caused by horizontal
stacks is economical or enables the refiner more readily forces (e.g., variations in wind direction) and vertical
to meet air quality standards). Complicated economic forces (e.g., atmospheric stability and instability). The
studies of this type are not practical when computer accuracy of assessment of stability conditions and the
programs are not available. validity of the graphical representations determine
For relatively simple problems, Turner 13 has devel the accuracy of the calculated concentrations.
oped a solution for exponents, in which the exponential
function is reduced by table to a simple mathematical 6.4.4 INFLUENCE OF ATMOSPHERIC CONDITIONS
multiplier. This method is recommended when a
' computer program is not available or for occasional
usage of the equations.
Atmospheric conditions and the methods of defining
and estimating them are of paramount importance in
614 DRW MANUALATMOSPHERIC EMISSIONS
~
f!"
z
!!!
10
5
4
2
f!"
z
3xl0 3 r,rTTTnn~~,rTY~,~,rTTrrn
2
•
(.)
iL 103 UJ
LL
UJ
~ 2
LL
0 5 LL
(.)
z ~
(.)
10 2 ~,~~~~~~~~~~
Q
In 2 z
a: o
UJ In
0..
In 10 2 a:
UJ 2
C 0..
5 A· Extremely unstable !!!
...J A  E~trernely unstable
c 101 bL.,.,tC.."c.7C+
~ B • Moderately unstable
...J
B _ Mod.rately unstable
z C  Slightly unstable <t e Slightly unstable
0 2 (.)
N 0 Neutral
t
~ a: E· Slightly stable
:r 101 UJ
>
b>o b N 2
4 X 100 L.11.....L.LLI.l.li.__''JJ..1..U'"_'.JJW.J....L.I.J
102 2
determining actual ground level concentrations and accuracy, the use of six stability conditions provides
in developing the equations used for predicting ground adequate coverage for the existing variations in atmos
level concentrations. This is apparent by the terms pheric stability.
used in the equations. All environmental sites (refineries) are subjected to
The predicted ground level concentration is affected at least two of these conditions (e.g., slightly unstable,
by the wind speed with which it varies inversely, by the C, and neutral, D) for a major period of time. At some
effective stack height (H), and by the dispersion co refineries, three or four of these conditions may be
efficients, cry and ~,: frequently encountered. Atmospheric conditions at a
given location and time can change by one or more
I. Effective stack height, H: The ground level con
classes when moving from ground level to a height of
centration varies inversely as an exponential function
perhaps 2,000 feet. If the condition changes by only
of the stack height. The parameter, H, in turn, is
one class (e.g., slightly unstable, C, to moderately
related to atmospheric stability and instability as
unstable, B), the predicted ground level concentration
discussed previously.
can change by a factor of 2 or more. Accordingly, it is
2. Dispersion coefficients, ~y and cr,: The ground level essential to determine accurately the stability condition
concentration varies inversely with the dispersion co immediately above stack height at time of discharge.
efficients, cry and cr,. These, in turn, are related to the Anderson, Hippler, and Robinson state that even
six atmospheric stability conditions (Figures 69 and though these stability classifications are almost entirely
610) and to the distance downwind from the source. empirical, they represent the best known approach for
predicting the dispersion of pollutants. 10 Practically all
In view of the importance of atmospheric stability,
recent literature on this subject is in agreement.
it might be in order to review the stability factors A to
F, discussed previously (see Table 62). Pasquill has
categorized atmospheric turbulence into six types 6.4.5 DISPERSION OF AEROSOLS AND
•
ranging from A, extremely unstable with high turbulence, PARTICULATE MATTER
to F, stable with practically no turbulence. In view of
the large number of parameters in the dispersion An aerosol may be defined as a suspension of fine
equations and the difficulty in achieving quantitative solid or liquid particles. The Air Quality Criteria
DISPERSION OF GASES 615
•
larger particles will fall out first, followed by particles On the other hand, to determine the concentration
of decreasing size. Medium particles, even if present at a point due east or due west of any of the seven
as a small percentage of the total count in refinery stack stacks, the contributions from all seven must be
emissions, can assume a major role on a weight basis, considered.
and thereby affect a "weight" calculation. The contributions from each source are obtained by
the atmospheric dispersion equations, as previously
6.4.6 MULTIPLE SOURCES indicated. It is apparent that this method can involve a
varying number of calculations, depending on wind J
Multiple sources, as defined in this context, include direction and velocity for each of three or more con
only sources within the refinery. For multiple sources taminants (particulates, sulfur dioxide, and nitrogen iJ
outside the refinery, reference is made to Anderson, oxides). Such calculations, even if programmed on a
et al. 10 and to Turner. 13 computer, can easily be time consuming, particularly
To determine the concentration at any given location, if the refinery is located in a populated area. Depending
a reasonably accurate method involves determining the on conditions, a calculation of this type is probably
individual contribution made by each of the multiple reasonably accurate.
sources. For practical purposes, any source or stack A less accurate method, more suitable for determining
1,250 feet * or more removed at a perpendicular distance ground concentrations further from the refinery (at
from the xaxis (direction of wind flow) can be elimi least I mile downwind), involves only one calculation
nated from consideration since such contribution would for any given wind direction and velocity for each
be negligible. The xaxis can also be defined as the line contaminant. With this method, the multiple sources
joining the major source and the point of calculation. are considered as one source in the atmospheric dis
A typical refinery area with seven stacks (location of persion equation. Again, only those stack sources are
stacks arbitrarily chosen) is shown in Figure 612. The considered that are at a horizontal distance (upwind)
three groups of stacks (S I and S2; S3, S4, and S5; and of 1,250 feet or less from the xaxis. The variable Q
S6 and S7) are more than 1,250 feet apart. It is assumed represents the total contaminant emission for all sources
•
that all stacks emit approximately equal amounts of the involved and H, the effective stack height, is determined
pollutant under consideration. For a north wind, the as a weighted average for all individual effective stack
concentration at a location 5,000 feet due south of heights under consideration. The dispersion coefficients
SI and S2 involves the contributions of SI and S2 only. ax, ail' and a, are determined for this weighted effective
Similarly for a location due south of S3, S4, and S5, stack height, the distance being the average distance of
only these stacks are considered, and the same reasoning all contributing stacks. The concentration so deter
applies for a location due south of S6 and S7.
mined represents an estimate, which may be useful for
"'For heavy contaminant sources emitting more than 25 tons predictive purposes. In the extreme case, where there
per day, this distance should be extended possibly to 3,000 feet
or more. are sufficient sources to consider the uniform area
source, these concentrations are dependent upon cr,
only, thus increasing the accuracy of the estimate.
The foregoing approximations are applicable only to
N stacks discharging above the layer of heavy ground
induced turbulence. Short stacks are unpredictable and
increasing their number increases the magnitude of
potential error.
T
,..:
...
0
OSI
OS3
OS6 6.4.7 WIND TUNNEL AND OTHER STUDIES
0
1
OS2 OS7
OS5
standard deviations of plume spread) provides increased
accuracy of calculated data, particularly where there are
no complications of terrain, however, this technique is
1 5000 FT. I not infallible. Wind tunnel studies are a convenient
Figure 612 Typical Refinery Area. means for obtaining data on dispersion of stack
DISPERSION OF GASES 617
·"T the minimum desired radial extent. When these development of new diffusion models, which are
anticipated to yield more accurate values for dispersion
''''''nln'endations were subsequently effected, the
coefficients.
". ':'dl'd results rroved to be satisfactorily accurate.
Nevertheless, the predicted results obtained by these
, '..'I' ,tudies, such as meteorological studies of the
equations are of a higher order of accuracy than would
""'cd site. may have a more practical impact than have been possible ten or twenty years ago. Jn the
.. I tllnnel studies, although standardization of these
*To some extent, concurrent errors in opposite directions for
'!C" rnav he more difficult. H on the one hand, cry and cr z on the other. tend to cancel out.
618 DRW MANUALATMOSPHERIC EMISSIONS
~
, ,
future even better accuracy can be anticipated as more stitute a visibility hazard and can represent an icing
accurate values are determined for dispersion co~ hazard during the colder periods of the year. Thus a
efficients by means of suitable field experiments. low to moderate wind speed, an ambient air temperature
The use of wind tunnel and other meteorological of 25 F to 60 F and a relative humidity of 70 percent
studies, especially for the purpose of augmenting pre~ or above may cause fogging (and/or icing). Under
dictive techniques, is also advisable and was previously certain conditions, cooling tower plumes may merge
discussed. Wind tunnel studies properly conducted can with stack plumes. This will affect downwind concen
serve as excellent tools for new refineries or for complex, tration of contaminants, possibly in an adverse manner.
existing refineries where new construction is proposed. Mathematical models and actual experience have
In existing refineries, whether or not new construction indicated that plume rise is directly proportional to the
is contemplated, there is no substitute for sampling and diameter of the plume discharge.21 On this basis, a
analysis to determine actual ground level concentrations. single large tower for a given capacity is more effective
This serves to indicate the degree of error, if any, in in reducing visible plume formation than several smaller
previous calculations and to indicate the magnitude of towers, particularly if the latter are of relatively low
any corrections that should be applied to the standard elevation. Also, mechanicaldraft cooling towers dis
dispersion coefficients. Sampling and analysis are valid charging small diameter plumes at low elevations and
only to the extent that the refinery is the sole source. at relatively low velocities are more prone to fogging
Otherwise, it is difficult to predict the refinery con~ conditions than naturaldraft towers.
tribution to a definite ambient air concentration. Well Mathematical models designed to predict plume
planned tracer studies can assist in these cases. behavior and concentrations have been correlated with
actual behavior and ground concentrations with reason
able agreement. The data involved were based on tall
6.5 Cooling Tower Plume Rise naturaldraft towers and on mechanicaldraft cooling
towers. The predictions herein should nevertheless be
applicable to refinery cooling towers because of the
6.5.1 BACKGROUND
'I Induced drafts, because the plume temperature is Figure 613Cooling Tower Plume in Low Wind
,!I'll\'e am bicnt. and because plume diameter is rela Condition.
IlIl'ly large. Thi, is especially true in the case of a
L,,~e plume, which has a smaller surfacetovolume
r:tll,) than a small plume. In the large plume, there is 6.5.2.2 Dispersion Method for High Wind Conditions
"'I1,idcrably less surface mixing, and hence a greater
1"l1dency to mamtain its temperature and reach a The dispersion method is used for moderate or high
c"ll1parallvely higher altitude. wind conditions and involves the PasquillGifford
eq uations. This method generally disregards plume
6,5.2.1 Turbulent Jet Method for Low Wind Conditions buoyancy but assumes a Gaussian distribution of plume
in the vertical and horizontal planes and a conservation
For the low wind condition, Kaylor, et aI., 22 utilize of heat and mass within the plume boundaries. A sketch
t he work of Fan 21 and of Abraham 24 and postulate that of the dispersion model is shown in Figure 614.
1 he cooling tower plume for low or moderate wind The difference between H, the effective plume height,
'clocity a"sumes thc characteristics of a round, buoyant, and Hs, the cooling tower exit height, may be indicated
'1lrbulent jd. as j.h, which is not defined. In petroleum refineries,
The turbulent jet method involves the fol:owing where cooling tower exit heights do not exceed 60 to
basic as,umptions: 80 feet, a fairly accurate estimate of j.h is 15 to 25
percent of Hs, or an average of 20 percent. The variable
I. Plumeinduced turbulence prevails.
j.h does not vary greatly but tends to decrease in pro
2. Round. buoyant, turbulent jet theory applied. * portion to the ratio of cooling tower exit velocity to
3. Mixing and cntrainment mechanism takes place wind velocity.
of dispersion mechanism. In the dispersicn method, equations (7) through (12)
are applicable for calculating travel and diffusion of
4. Gau"ian distribution for heat, moisture, density,
cooling tower plumes. The previous discussion of
and velocity profiles.
parameters and standard deviations and the infl uence
5. Heat. moisture, buoyancy, and momentum con
served.
The paramders shown in Figure 613 are obtained by u
numerical solution of differential equations. The plume
under llIost cond itions (except certain types of inversion)
appears ,ullicj~lltIy buoyant so that it will not reach
the ground, except possibly in very low concentrations,
and will ultimately cvaporate. It is difficult to predict
the clfect of a low inversion (e.g., 500 to \,000 feet),
which IHay present some problems. Fortunately, an
illvcr,ioll at this height is an infrequent occurrence.
>J<L"rl~(.; quantity of heat dispelled from tower increases impor Figure 614Cooling Tower Plume in High Wind
tam:c 01 huoyancy of plume. Condition.
620 DRW MANUALATMOSPHERIC EMISSIONS
of atmospheric conditions is also relevant (see Para it is assumed that the tower is equipped with a drift
graphs 6.4.3 and 6.4.4). eliminator, which substantially eliminates physical
Kaylor, et aI.,22 verified the mathematical models carryoff of entrained water.
by observation of visible plume behavior. These com In the paper by Overcamp and Hoult,26 it is shown
parisons show reasonable correlation which can, on that "the condensation of the water vapor cannot cause
the average, be represented by: Observed plume length rain unless the plume mixes with the aerodynamic
+
= 35 (0.91) (predicted plume length). wash of the tower, bringing the droplets in contact
The actual observations were taken from three sources with the ground."
including a mechanical draft tower with five cells and Basically, rain formation occurs when minute water
a natural draft cooling tower. This tower was a hyper droplets condense. A raindrop reaches the ground
bolic type but, as previously indicated, the general before it evaporates, while a cloud particle does not.
conclusions should also be applicable to low or medium By this definition, a raindrop has a minimum size of
height towers. approximately 200 microns. In the effluent vapor
In an article by Westlin, Brenchley, and Smith, 25 the plume, condensation is the dominant process for water
PasquiIlGiffordTurner method, using Gaussian dis droplets in sizes up to 40 microns. It can be shown
persion coefficients, is also used in developing a plume that the average length of residence in the plume is
length equation. Field test data indicate that an 100 seconds or less, which is insufficient time to form
excellent correlation is obtained with a multiple regres droplets larger than 40 microns. Thus Overcamp and
sion coefficient greater than 0.8. In this article, however, Hoult conclude that the condensation process in a
high wind condition is defined differently and is related vapor plume is not conducive to the formation of an
solely to wind velocitythat is, to a wind velocity appreciable number of the larger raindrops. This con
greater than 1m/sec (7.5 mph). The Kaylor article cl usion was also reached by earlier authorities, Blum
defines high wind condition in terms of a ratio. The (1948), Bront (1952), and Chilton (1958).
Westlin article is based on results from a crossftow The effect of wind velocity (crosswind effect) is
induced cooling tower. important in determining whether interaction occurs
As previously indicated, wind tunnel studies are between the visible plume and the aerodynamic tower
advisable for complicated building structures, for a wake. If interaction occurs, there will be ground f
multiplicity of smoke sources or plume sources, or both, precipitation; if no interaction occurs, the plume will
and for unusual topographic effects which might disrupt rise and ultimately disappear downwind. The critical
the normal flow pattern. Wind tunnel studies usually variables to determine whether interaction occurs were
disregard any buoyant effect and are more effective at found by Overcamp and Hoult to be:
higher wind speeds. Accordingly, wind tunnel models
1. Speed ratio, R.
seem to be ideally suited for predicting cooling tower
plume behavior in high wind conditions. Although the R = U/IL
cost of these studies normally precludes their use, they Where:
should be considered when the possibility of hazardous
U = plume exit velocity.
fogging or icing conditions exists.
IL = wind speed.
6.5.3 PLUME CONDENSATION AND PRECIPITATION 2. Froude number. (This may not be valid for Froude
numbers greater than 3.0.)
Plume condensation is the transference from invisible 3. Reynolds number. (This may also be a factor but
to visible plume and occurs when the saturated vapors can usually be disregarded.)
are cooled below the dewpoint. It is of practical
significance to deduce whether the condensate remains If the speed ratio, R, is below a critical value, inter
in the atmosphere as a vapor plume or cloud, or action occurs. The critical value varies as a straight
whether there is further condensation to form rain. line relationship with the Froude numberthat is, the
The evaporative cooling tower evaporates a small critical val ue decreases as the Froude number decreases.
percentage of the water to be cooled, thus releasing its For a given tower exit velocity, the higher the wind
latent heat of vaporization and cooling the water. The speed the lower the speed ratio, thus the possibility of
rate of evaporation can be quite high, and if only a
small portion of the evaporated water is condensed,
it will precipitate in the form of rain. In this discussion
interaction increases.
The conclusions reached include where the plume
struck the ground (2 to 4 tower heights downwind) and
•
DISPERSION OF GASES 621
. ""c,w,tation, where applicable to tall natural drift eliminators were assumed at 0.2 percent, but
;\"""" iOwers. It is reasonable to assume that this value is probably high).
.:.;.:.1::," d()\\nwind (in terms of tower height) and
,ro, li:c total precipitation would be approxi 3. Increase the height of plume rise to the extent
:.c ,ame for lower height refinery cooling practical without heating effluent air.
'.111,' estimates of the severity of precipita These methods are generally more applicable to new
_:r" n"'lie for a 500 megawatt plant. These are tower construction.
I" I·igure 615.
Mathematical models indicate that aligning cooling
."1 n " relative humidity of 90 percent, an exit
towers with existing wind increases the effective stack
"",.llnrc elf 40 C (104 F), and an ambient tempera
height and, hence, decreases fog concentration at ground
~ (. 125 F), the precipitation would be 0.1
level. A reduction in drift decreases the total amount
of water to be evaporated and therefore minimizes
"ne'L1sions are interesting apart from pre fogforming tendencies. The higher plume rise elevates
",l;sibilities. A noninteracting plume will the plume to an area of more stable wind, usually
'Ie'lind fog hazards will be minimized; an higher velocity and lower humidity, thus promoting
',:1"" plume will result in lowered visibility and plume dissipation .
."':"1I, cOllditions at or near the ground. Superheating the effluent air, thus increasing its
moistureholding capacity, also results in a higher
MINlioAlZING VISIBLE PLUMES effective plume height. These two characteristics
facilitate dispersion before the formation of a visible
plume (i.e., before the air mixture is cooled to the dew
111" towcrs in high humidity conditions have a
. ~'.:,T,(\ I,) rrod LIce fog when the effluent air is cooled point). Various methods can be used to accomplish
", dew point. These visible plumes normally this dispersion. The use of burners represents one
., .[,"bient temperatures of 25 to 60 F and at a possibility, but probably the most practical method
. ,'. ,'i' 70 percent or higher, Temperatures below
involves a heat exchange with the incoming hot water .
, lI,ay cause icing, The return hot cooling tower water is passed through
finned heat exchangers, while the effluent air is passed
.,1 IlIdllOds can be used to minimize or even
around them. This is shown in Figures 6 I 6 and 6 I 7,
.. ·!;Ie Ihe plume, Probably the most positive method
which are taken from an article by Buss. 27
'C!Pcrheating the effluent air. This method is
·"d laler in greater detail. Other methods include
The principle involved is evident from an inspection
: ';\ii()\ving: of the psychrometric charts, in which dew point temper
ature, represented as a curve, is related to dry bulb
\II~I) the length of the cooling tower with the direc temperature and to the maximum moisture content of
(Wil ()f the prevalent winds. dry air. If an anticipated mixture of ambient air and
cooling tower effluent air is assumed, the heating
, Employ drift eliminators to reduce drift to a val uc required to maintain the mixture temperature above
"f n()5 percent or lower. (Values of drift loss with the dew point can be determined. The degree of mixing
depends upon many meteorological factors, and heat
and masstransfer values cannot be readily determined;
0.15 therefore, depending on the degree of fog elimination
cr: RELATIVE HUMIDITY desired, a factor of safety should be employed.
1:
.... 90% Using the incoming hot water as a source of heat has
~
u 010  40% several disadvantages:
z
0
;: I. Increases the capital investment.
<l:
f
a: 0.05 2. Increases the operating expense.
u
:" 3. Increases the pressure drop through the tower.
0::
Q
•
75 F DISCHARGE AIR
CONVENTIONAL FILL
(
AMBIENT
AIR
WATER INLEf 15 F
622 •
FAN'
DECK'"
FILL DECKS
COILS
DRAIN AND
623
.
624. DRW MANUALATMOSPHERIC EMISSIONS
•
6.6 Sample Calculations 6.6.1.3 Problem 2
At stability condition B:
= ~)
6
[1 5 + 268 (10 3)(1013) (561= 293) 3J
.. 561
Zm" = 1.35 X 105 X l~1 = 2.3 X 104 g/m' of SO,
= 6 [1.5 + 2.68 (10 )(1013) (~4)3 J
3
At stability condition C:
\2 ~
~
41
JJ
E
Z"
c
0
.;;;
.~
..
u1
0
~
41
Gi
':'
.!
~
0 or:
'"
·iii
:r
41
~
v
41
It
w
..,c
0
~
">
~
~"
• >
2 ~
1i
0
E ci;
..,;
cT ~
0
c
"
E
~v
c
a "
LL
" 0
,..;i « ~
0
~.
:i.
'"
E
.;"
x
0
..,~c
0
c
.g
0
l:
c
41
v
c:
0
U
E
.;"
x
0
~
~
0
41
v
c:
E~
15
I
co
I 0
I
I ""
~
'"
i.i:
.I
i
1 i.!. •
'., ~ w~ 'xew x
625
1
626 DRW MANUALATMOSPHERIC EMISSIONS
•
148 77 4.9 X 10'
0.25 percent NO" analysis. The primary components of NO" are 3 115 1.0 X 10" 51 5.1 X 10'
NO and NO, and are usually updated as NO" which has a molec 5 89 1.8 X 10" 31 5.6 X 10'
ular weight of 46. To the extent that the average molecular weight 6.7 79 2.3 X 10:' 23 5.3 X 10'
of NO, differs from that of NO, the solutions in Problems I. 2, 8 74 2.7 X 10 5 19 5.1 X 10'
and 3 would require modification. 10 70 2.9 X 10" 15 4.4 X 10'
DISPERSION OF GASES 627
At a wind speed of 15 mph (6.7 m/sec), the maximum I. Stack height is 200 feet (61 meters).
concentration of NO x is 5.6 X 104 g/ml and will be
found at a downwind distance of 0.9 km (Figure 618). 2. Stack temperature is 800 F (427 C) (700 K).
3. Ambient temperature is 68 F (20 C) (293 K).
6.6.2.4 Problem 3
4. Wind speed is 4 m/sec (8.9 mph).
To determine the NO x concentration at a distance of 5. Atmospheric pressure is 10\3 millibars (N.P).
2.0 kilometers (1.25 miles) downwind of the source at
the centerline of the plume and at ground level, the 6. Stack diameter is 12 feet (3.66 meters).
equation is: 7. Gas discharge is 600,000 cubic feet per minute
(actual).
Z(x,O,O;H) = Q
7C'cr y cr z [J.
exp [  2! (H)'J
IJ
z 8. SO, in exit gas is 1,000 parts per million (0.1 per
154 cent by volume).
Z(2000, 0,0 ;H) = (3.1416) (200) (lIS) (6.7)
9. S02 emitted is 0.1827 pound per cubic foot (N.T.P.)
z(2000, 0,0;79) =
154
(4.84)(10=5) exp [I
 2 (0.69) 2J
6.6.3.2 Problem 1
z = (3.2 X 104) (7.88 X 10')
Calculate maximum downwind concentration of
= 2.5 X 104 gicu m
SO, and the point of maximum downwind concentra
This is equivalent to 250 micrograms per cubic meter. tion. Use Holland's formula to obtain the effective
stack height.
6.6.3 CATALYTIC CRACKING UNITS
j,H = v~d [1.5 + 2.68 (10'p) C'j.,Ia) dJ
I
In a catalytic cracking unit, the catalyst regulator is Where:
equipped with two stages of internal cyclones in the v., = stack exit velocity = 60,0.000 X __ L_
regenerator, and the catalytic offgases are then pro 113.1 60 X 3.28
cessed in a carbon monoxide (CO) boiler. In general, = 26.9 m 'sec
for all catalytic cracking units having CO boilers, the
j,H=
=
~6.9~3.66)[1.5 + 2.68 (1.013)C()().ffit93 )3.66J
24.6 [1.5 + 2.71(0.581) (3.66)] = 24.6 (7.26)
j Under the following conditions, assumed for a fluid
i bed unit during a 50,000barrel day, calculate the maxi 178.6 X 1.10 = 196 meters (at stability con
mum ground level concentration of S02 at a wind dition B)
speed of 4 m/sec (8.9 mph) and determine the location. Thus,
(See Problem I.) The day is bright and sunny, stability H = H, + j,H = 61 + 196
category B. = 257 meters (effective stack height)
Under the same conditions, calculate the particulate
concentration at ground level at distances of 0.5 mile Determine Q, quantity of S02 emitted in g/sec.
and 2.0 miles, and displaced horizontally 0.25 mile
(0.25 mile from xaxis). Calculate these concentrations Q = ((>OQ.000)(O·20Il(0.1:~7 Ib/cuftl (4}_~lll)G~~)
for emissions if no electrostatic precipitator is present
(see problem 2) and if an electrostatic precipitator is
600 ~~.95) (0.419 829.5 (0.419) 347 g/sec
connected to the CO boiler (see problem 3).
628 DRW MANUALATMOSPHERIC EMISSIONS
e:
Where:
~
Xm.x = (QX:J (;) = 2.7 X
=
= 2.39 X 10 4 glcu m
2.7 X 10'sqm
10'
7
) x (2000,400,0;257) = 0.049 exp [  1/2 (1.38)'J
•
ticulates behave as gases. This is substantially true
at a finite distance is:
after the particulates pass through a precipitator
because the remaining particles are primarily small
X(x,y,O;H) = TO _Q exp [  I (l')'J
cr"O',1L 2 cry (under lOlL); although some mediumsized particles
(lOlL to 60IL) are present. If no precipitator is present,
exp [  ~ (~YJ more particles are in the medium range, and possibly
some in the large range (over 60IL). The large particles,
For stability condition B, at 0.8 km downwind, cry = and to some extent the medium particles, suffer fallout
130 m and crz = 85 m. For 2 km, cry = 290 m and resulting in the socalled "tilted pI ume." Thus, down
cr, = 235 m. wind readings without a precipitator are higher than the
values shown, especially for X(800,400,0 ;257), which
41,900 [ I (400)2J is now shown as 2.74 X 10 5 g/Cll m
x(800,400; 257) = 7: (130) (85r (4) exp  2 130
•
7:
The total effective height of the plume, H, is determined
by:
exp [ 1 G~~YJ H = Hs + J.FL + J.H
DISPERSION OF GASES 629
i ,0 obtain ;::;~~, the ratio of wind velocity (IL".) to stack 3. Annual average wind velocity is 10 miles per hour.
. il velocity ([.I.,,) must be calculated: NOTE: The sources of loss and the formulas for
tankage loss are described in some detail in the API
~'W = ~ = 0.110 Manual on Evaporation Loss, specifically Bulletins
[.1.0 200 2517 29 and 2518. 30 These bulletins discuss evaporation
,d from Figure 66, loss from floatingroof and fixedroof tanks, respec
tively and also include relatively simple nomographs
'E..1y
L
= 044
. that can be used in lieu of the equations. The method
for determining true vapor pressure (TVP) from Reid
Vapor Pressure (RVP) for any given temperature is
.1FL = 240 (0.44) = 105.6 feet (32.3 meters) given in API Bulletin 2513. 2
,'" obtain .1H from the Holland equation, the
[,)llowing data were assumed for the imaginary stack 6.6.5.2 Problem 1: Evaporation Losses of FixedRoof
height: Tanks
•
=
H 24 feet.
H = 45.7 + 32.2 + 4.3 = 82.2 meters (270 feet) T 16 F.
Fp = 1.15 for white paint in poor condition.
Although some of the preceding assumptions may be C = 1.0.
f
63.0 DRW MANUALATMOSPHERIC EMISSIONS
Then, Thus,
L!I = 0.024 (9\)"'8 (125)173(24)0.51(16)050(1.15)(1.0) W = (0.OOO448)e'0~~5000) 7 bbl/yr t,
= 1495 bbl/yr Total loss of floatingroof tank is:
b. FILLING Loss 200 +7 = 207 bbl/yr
F = 0.OOO3PVK, 6.6.5.4 Conclusions
Where: The fixedroof loss will be 4495 barrels per year.
F filling loss, in barrels per year. The floatingroof loss will be only 207 barrels per year.
P vapor pressure of light at bulk temperature,
in psia. 6.6.6 PRODUCT LOADING
V = volume of liquid pumped into tank, in
barrels per year. Determine the loading loss experienced when loading
K, turnover factor (t = throughput in turnovers a tank truck with 10,000 gallons of 1l.0pound RVP
per year, V/tank capacity). gasoline at temperatures of 40 F and 60 F by splash
Then, loading, subsurface loading, and vapor recovery.
F = (0.0003)(5.0)(2,000,000)(1.0) Assume average saturation of truck vapor space of
= (0.0015)(2,000,000) = 3,000 bbl/yr 30 percent.
Total loss of fixedroof tank is: Volumetric loading losses with relation to gasoline
TVP are shown graphically in Figure 619. 31
1495 + 3000 = 4495 bbl/yr
0.4
a. STANDINGSTORAGE Loss
Where:
L" = K f (D)15 ( 14.7 _
P
P
)O.7( VW)0.7
<If
<J)o
o<{
...10
z...l
ou.
~~
0:
O.J
0.0
0
0.3
0.2
WARNING  LOSS BY ENTRAINMENT
DURING SPLASH LOADING
MAY EXCEED EVAPORATION
LOSS BY TWO OR
THREE FOLD •
::l>
L" = standingstorage loss, in barrels per year.
Kr = tank type factor: "'
0.1
0.045 for welded tanks.
0.13 for riveted tanks with pontoon roof and
single seal.
D tank diameter, in feet (for tanks less than 150
ft in diameter). 0.0
P vapor pressure liquid at bulk temperature,
in psia. 12
V". = average wind velocity, in miles per hour.
Thus, 10
•
(BASED ON TYPICAL SLOPE OF 2.5 AT 10% PT. ON ASTM DIST. CURVE)
V volume of liquid withdrawn from tank. In
barrels per year. Figure 619 Gasoline Correlation for Tank Cars and
D tank diamcter, in feet. Tank Trucks.
III !!;
OUTBREATHING
FOR ALL STOCKS
_I_II ~IE 1I·1i
OVER 100°F
FLASH POINT
i,"
t'1
INBREATHING ALL STOCKS
(j) Wt:t:; t.= AND OUTBREATHING FOR STOCKS
J
l1J UNDER 100°F FLASH POINT
0:
0:
« =
,
CD
>
I
U
ct
«
()
\) ~
z +
«
I
10,000'
J J
1,000
o 20,000 40,000 60,000 80,000 100,000
I
Figure 620_ Thermal Breathing Requirements for Various Tank Capacities.
•
!
I
I 631
I
I
!
I
•
a:
:;:)
~
a:
w
0..
en
...J
w STOCKS UNDER 100°F
a:
a: FLASH POINT
ca
TANK FILLING
STOCKS OVER 100°F
•
FLASH POINT
100
50
o 20,000 40,000 60,000 80,000 100,000 120,000
632 •
DISPERSION OF GASES 633
I 'c following data are taken from Figure 619. All 6.6.7.1 Example
10""' tor splash loading have been arbitrarily in Consider the specific case of a I OO,OOObarrel welded
. : <d h\' 50 percent to allow for losses by entrainment.
cone roof tank flash below 100 F with a maximum
filling rate of 5,000 bbl/hr and a maximum emptying
Splash Loading Subsurface Loading
rate of 2,000 bbl/hr. The following estimates are taken
'.1' "nund')) 11.0 11.0 11.0 11.0
"t"[.Ilu[e F 40 60 40 60 from Figures 620 and 621.
· I' ,;'(lllnds) 3,8 5.6 3.8 5.6
IlflH.'tnc loss a. STEP 1
'"[(i.'nt) 0.100 0.175 0.042 0,062
Pressure Vacuum
i\ \ (d umetric (cu ft/hr) (cu ft/hr)
~ (:'t.'rcent) 0.150 0.263 0.042 0.062
:\1',..,
Thermal requirement, 100,000 bbl 60,000 60,000
.dlt)nq 15.0 26,3 4.2 6.2 Filling requirement 60,000
Emptying requirement 11,200
Total cu ft/hr 120,000 71,200
!"t!Om loading is slightly more efficient than sub
.f I;ICCloading. Vapor recovery systems have efficien b. STEP 2
to 95 percent of splash loading losses, ass um
. ,,!' ')() Assume that the welded cone roof tank (100,000 bbl)
.. ,) entrainment. Therefore, losses approximate has a pressure setting of 0.75 ounces per square inch
,'lit i percent of the total splash loading losses. (OSI) and a vacuum setting of 0.50 OS!. (All vac
uum settings are 0.50 OSI.)
~.6.7 ~OOF VENTS C. STEP 3
For a pressure setting of 0.75 OSI, the maximum
:, 'calculate vent sizing, determine the following: required capacity must be achieved at 0.50 OSI buildup
"!I1acity requirements for: above the pressure setting, or 0.75 + 0.50 = 1.25 OS!.
Similarly, the vacuum setting is 0.50 + 0,50 = 1.00
a. Pressurethermal expansion plus maximum OSI. Therefore, the vent must be sized for:
filling rate. 1. Pressure setting of 0.75 OSI and 120,000 cu ft/hr
b. Vacuumthermal contraction plus maximum at 1.25 OS!.
emptying rate. 2. Vacuum setting of 0.50 OSI and 71,200 cu ft/hr
at 1.00 OS!.
",'"ure and vacuum settings.
Vent ratings for these conditions are available from
1. Size of required vents, the manufacturer's data.
REFERENCES
• "National Primary and Secondary Ambient Air Quality "Meteorology and Atomic Energy, ed D. Slade, U.S. Atomic
Siandards," Federal Register 36 [84] Environmental Protection Energy Commission, Division of Technical Information (1968),
Agency (1971). " J. Z. Holland, "A Meteorological Survey of the Oak Ridge
'API Bu/l, 2513, Evaporation Loss in the Petroleum Indus Area," Atomic Energy Commission Rept ORO·99, Washington,
I'",'C£l£lses and Conlrol (1959). D.C. (1953).
'M. E. Smith, Recommended Guide for the Prediction of '" 0, B. Turner, Workbook of Atmospheric Dispersion Esti
flit, Dispersioll 0/ Airborne Effluents, 1st ed., Am. Soc. Mech. mates, Revision, U.S, Dept. of Health, Education, and Welfare,
Engrs. (1968). Public Health Service (1970),
H API RP 521 Guide for Pressure Relief a"d Depressuring
, W. L. Donn, Meteorology, 3rd ed., McGrawHill Book Co.,
Inc., New York (1964). Systems, Sept. (1969).
I;, The Dispersion of Gases from Elevated Refinery Flares,
· C'. H. Bosanquet, "The Rise of a Hot Waste Gas Plume."
1. I",,/ilute of Fuel 30 [322] (1957). CONCAWE working group, presumably unpublished (1967).
,; R. B. Hawkins and E. A. Nonbebel, "Chimneys and the Dis '" F. Pasquill, Atmospheric Diffusion, D. Van Nostrand Co.,
persal of Smoke," J. Institute of Fuel 28 [530] (1955). New York (1962).
" Air Quality Criteria for Particulate Matter, U.S. Dept. of
, W, S. Lucas, 0, L. Moore, and C. H. Spurr, "The Rise of Health, Education, and Welfare, National Air Pollution Con
Hot PI limes from Chimneys," Institute J. Air and Water Pol/u trol Association, Washington, D.C., Jan, (1969).
Ii"" 7 [473J (1963). '" G, T. Csanady, "Turbulent Diffusion of Heavy Particles in
, The Calculation of Atmospheric Dispersion from a Stack, the Atmosphere," J, Atm, Sci. 20, 20108 (1963).
CONCAWE (1966). "R. L. Duprey, Compilation of Air Pol/utant Emission
•
" G. A. Briggs, "Plume Rise," USAEC, Division of Technical Factors, U.S. Dept. of Health, Education, and Welfare, Environ
II, [ormation, USAEC TIP 25075. mental Health Services (1968) .
, .. G. E. Anderson, R. R. Hippler, and G, D. Robinson, "An ,,, G. R. Lord and H. J, Leutheusser, Wind Tunnel Study of
Evaluation of Dispersion Formulas," Final Rept. prepared by Sarnia Refinery, Imperial Oil Enterprises, Ltd., Sarnia, Canada,
Travelers Research Corporation for Am. Petrol. Inst. (1969). Dec, (1965).
63.4 DRW MANUALATMOSPHERIC EMISSIONS
"D. H. Brown and H. J. Sneck, "Cooling Tower Plume pirical Study of the Length of Cooling Tower Plumes." 65th
Rise," Am. Power Conference, 33rd Annual Meeting, Chicago, Assn. Mtg. APCA, Florida, June (1972).
Illinois, Apr. (1971). "T. J. Overcamp and D. P. Haul!, "Precipitation in the
"F. B. Kaylor, J. D. Kangas, J. L. Petrillo, and Y. J. Tsai, Wake of Cooling Towers," Atmospheric Environment, Vol. 5,
"Prediction and Verification of Visible Plume Behavior Asso 75165, Pergamon Press.
ciated with Wet Plume Discharge," 65th Meeting, Air Pollution "J. R. Buss, "How to Control Fog from Cooling Towers,"
Control Assoc., Miami Beach, Florida, June 1822 (1972). Power 112, 723, June (1968).
"L. N. Fan, "Turbulent Buoyant Jets into Stratified or "Catalytic Cracking EmissionsAn Industry Survey, Am.
Flowing Ambient Fluid," Cal. lnst. Tech., Report No. KH Petrol. lnst., June (1972).
R15 (1967). "API Bull. 2517, Evaporation Loss from FloatingRoof
"G. Abraham, "The Flow of Round Buoyant Jets Issuing Tanks (1962).
Vertically into Ambient Fluid Flowing in a Horizontal Direc '" API Bull. 2518, Evaporation Loss from FixedRoof Tanks
tion," 5th Inter. Water Pollution Research Conf., JulyAugust (1962).
(1970). " API Bull. 2514, Evaporation Loss from Tank Cars, Tank
"P. R. Westlin, D. L. Brenchley, and P. J. Smith, "An Em Trucks. and Marine Vessels (1959).
, .
•
APPENDIX
constant in Lucas, Moore, and Spurr formula.' Tl = temperature of combined inlet gases prior to
stack diameter, in meters. combustion, in degrees Fahrenheit.
Briggs formula, a factor proportional to the I' wind velocity, in feet per second or meters per
rate of buoyancy (heat) emission. second.
/ '/. flare length of flare plumes. I' particle size, in microns.
cJL vertical component of flare length, FL. I' = wind velocity at stack height, in meters per
If effective height of a stack. second.
if = effective height of plume rise. U plume exit velocity (cooling towers).
,II.\' stack height.
V" = stack exit velocity, in feet per minute.
plume rise above stack height.
w stack exit velocity, in meters per second.
!fe heat of combustion.
//r heat losses by radiation. Xmax maximum downwind ground level concentra
i.[ heat emission, in megawatts (Lucas, Moore, tion, in grams per cubic meter.
and Spurr formula). XI') relative maximum concentration.
( Q
standard deviation of plume diffusion in the max
•
atmospheric pressure, in millibars. height, H.
Q heat emission, in megawatts. XI ground fumigation concentration.
Q uniform emission rate of pollutants. x = distance from source at any point directly
f? speed ratio of plume exit velocity to wind downwind of plume flow.
speed (cooling towers).
x distance from x, in meters.
atmospheric temperature, in degrees Kelvin.
stack temperature, in degrees Kelvin. y distance crosswind from xaxis.
imaginary stack exit temperature, in degrees z distance (vertical) above ground.
Fahrenheit. 2 m" plume rise above stack height; similar to :1H.
NOTE: This appendix does not include symbols for evaporation loss equations for fixed and floatingroof
tanks. These are defined separately in the loss calculations in Paragraph 6.6.5 .
• 635
t)
3MMarch 1974