905R80013
                                                       30-53.7
 The Location and Magnitude of Maximum Short-Term
Ground  Level Impacts of Effluents from Tall  Stacks
                Dennis A.  Trout
  U.S.  Environmental Protection Agency, Region V
               Chicago, Illinois

-------
                                                                       80-53.7


Introduction

  The 3-nour national  ambient air quality standard for S&2 reauires that at
any point, the SO? concentration not exceed 0.5 ppm (1300 ug/nr)  more than
once per year,  mis standard poses the need to adequately assess such
maximum short-term impacts.  This paper focuses on assessment of  the short-
term impacts of S0£ sources with tall  stacks (stack heights greater than
1UU m).

  The original intent of building tall  stacks for sources of large S02 emis-
sions potential was to reduce the maximum ground-level ambient air quality
impacts of the stack's effluent, the theory being that the effluent's ambient
concentration (mass/unit volume) would  be reduced by spreading the same amount
of pollutant mass through a greater atmospheric volume.  However, there have
been conflicting opinions on how successful tall  stacks have actually been In
reducing maximum ground-level impacts.   These differences in opinion center
about the expected location and magnitude of maximum short-term ground level
impacts of effluents from tall stacks.

  The purpose of this paper is to review the data relevant to the assessment
of the location and magnitude of maximum short-term impacts of effluents from
tall stacks.  The questions this investigation seeks to answer are: 1) At what
distance from a tall stack is its maximum short-term impact observed?  2) What
atmospheric conditions are associated with the occurrence of maximum short-
term impacts? and 3) Which, of various  dispersion coefficients when used in
standardly available models, best estimate the locations and magnitudes of
expected maximum short-term impacts of  effluents  from tall stacks?

Procedure

  Because theory is born out by observation, answers derived from the pertinent
observations are preferred to those based purely  on theory.  In order to avoid
observational oias, sufficient and necessary observations must be evaluated;
that is, the temporal  and spatial characteristics of the observations must be
considered.  It must also be determined that the  observations are not biased
by the manner in which they are made.

  In order to adequately assess the distance from the stacks to short-term
maximum impacts, it would be desirable  to obtain  quality assured  data from a
network consisting of continuous monitors at numerous distances along each of
numerous radials extending in different directions from the stack.  The moni-
tors should be located so that they adequately cover both near (less than
2 km) and far field distances from the  stack with a spatial density sufficient
to determine the distance to maximum impact.  Ideally the source(s) to be
analyzed would be isolated such that background contributions to  the pollutant
concentrations being monitored are minimized.  Furthermore, it would be desir-
able that the period of record of observations extend over several years so that
year to year variations in meteorological extremes may also be accounted for.
The longer the period of record and the greater the monitoring spatial density,
the greater the probability that occurrences of worst case meteorological
conditions and the constraining emission characteristics of the source will
coincide so that concentrations representative of maximum short-term impacts
of effluents from tall stacks can be observed by  a monitor.

-------
                                                                       80-53.7
  Obviously if the effluent characteristics  of the stack  are dependent  on or
controlled as a function of expected  meteorological  conditions  (i.e., by Sup-
plementary Control Systems (SCS), sometimes  referred to as  dynamic  emission con-
trol  systems, or emissions limit control  systems,  etc.),  the probability of
ODserving short-term maximum impacts  may be  greatly reduced.  Therefore, a
monitoring network might not observe  the maximum possible short-term concen-
tration if tall stack effluents are controlled by  a correctly designed  and
properly operating SCS application or similar meteorology dependent emission
control technique.

  Upon determining the periods of observed maximum short-term impact, the
meteorological conditions associated  with those periods can also  be determined.

  Upon determining the location, magnitude,  and meteorological  conditions associ-
ated with maximum observed short-term impacts, it  is then possible  to compare
model calculated locations ana magnitudes of maximum short-term impacts with
observations.  Such comparisons may be used  to assess which model  results best
compare with observations, but cannot necessarily  be used to conclude that the
transport and dispersion of the effluent mass is accurately modeled with
respect to the actual atmospheric cause-effect relationships.  In  order to
evaluate and compare the adequacy of  a model  with  respect to accurately
accounting for plume rise, transport, and dispersion of the pollutant mass,
detailed and comprehensive observations of such factors would be  required.

  In order to determine if a model is satisfactorily performing from a  regula-
tory perspective, comparisons with observations are made with respect to the
upper ends of the frequency distributions of predictions  and observations.  This
comparison is called for, particularly in light of the requirements of  the short-
term ambient air quality standard which is based on the upper end  of the frequency
distribution of observations.  That is, violations of the ambient  air quality
standards (e.g., the 3-hour SC^ standard) are determined  by the observation of
more tnan one occurrence per year of  concentrations exceeding a set maximum.

Observations

  The first observation to be made is that there is no monitoring  network avail-
able which ideally provides all the observational  desires previously set forth.
However, of the limited data available, there appears to be a set  of observations
from a monitoring network in the vicinity of the tall stacks of the Muskingum
River Power Plant operated by the Ohio Power Company of the American Electric
Power System, which comes closest to  satisfying observational requirements.  The
data from the Muskingum River Power Plant's  SC^ monitoring network  is relied
upon for the determination of the distance from the stacks to maximum short-term
(3-hour) S02 impacts of effluents from tall  stacks.  Once the distance  to
maximum 3-hour SOo impacts is determined, data from monitors which  are  located
at comparable distances to other tall stack  sources can also be evaluated with
respect to the magnitude of maximum 3-hour SC^ impacts.

  The Muskingum River Power Plant is  located in Washington County in south-
eastern Ohio in the valley formed by  the Muskingum River.  The terrain  in the
vicinity of the plant is hilly with elevations as  high as 350 m above mean
sea  level (150 m above the base of the plant stacks).  The total  generation
capacity of the Muskingum River Power Plant  facility is approximately  1,500
megawatts.  The current sulfur content of the coal fired averages  about

-------
                                                                       80-53.7
9 to 10 Ib SOg/lO  Btu.   No flue gas desul furlzatlon  system is  employed  and
the effluent is emitted  through two separate  252  m stacks.   The combined SC^
emission rate from the two stacks when the plant  is operating at full  load  is
approximately 15,500 g/s or 62 tons/hr (assuming  9 Ib S02/10° Btu coal  fired).

                      Distance to Maximum Short-Term  Impacts

  Since January 1978, four continuous S02 monitors have been in location along
a radial extending in a  northeasterly direction from  the Muskingum River Power
Plant.  Table I identifies the Muskingum River Power  Plant  S02  monitoring sites
and their locations.  It can be seen that the closest monitor,  the Center Bend
monitor, is 1.7 km from  the centroid of the two stacks.

Taole I.  Muskingum River Power Plant $62 monitors.

                                     Distance       Azimuth       Height above
                                   from Stacks3   from Stacks3      Stack Base
        Name       Monitor No.         (km)           (°)             (m) _
Center Bend             5              1.7             37              74
Hackney                 2              4.6             41              74
Rich Valley             3              8.5             36   •          105
Cal dwell                 4             20.7             33             132
Beverly                 1              5.2            138              56
Mount 01 i vet            6              6.1             240             108

a  Stacks 1 and 2 are located 645 m apart along a 240° azimuth.   Distances  and
   azimuths indicated are from the centroid of the two stacks.

  Because four of these monitors are located within 4° of a straight  line  (37°)
raaial from the source of emissions, it may be assumed that all  four  monitors
on occasion will  be simultaneously impacted by emissions  from  the plant (i.e.,
when the mean transport winds are from the SW (217°)).  Because  these four
monitors are located in the same downwind direction from  the plant,  it was
assumed that data from this array of monitors were sufficient  to determine
tne distance to highest observed impacts.

  Table II lists the three highest observed 3-hour concentrations at  each
monitor since January 1978 and the dates of those occurrences  at each monitor.
It can be seen that the highest 3-hour concentration ever observed at any  of
these four monitors occurred at the monitor closest to the plant on  May 5,  1979.
It can also be seen that the highest 3-hour concentration observed at the
monitor next closest to the plant also occurred on May 5, 1979.

             Meteorology Associated with Maximum Short-Term Impacts

  Table II also presents the meteorological conditions associated with the
monitored concentrations.  It can be seen that the highest 3-hour S02 impact
occurred in the afternoon under conditions associated with very  unstable
atmospheric conditions; that is, very light wind speeds (less  than 1.5 m/s),
clear skies, and a well developed mixed layer (inversion  base  1300 m  above
ground level).  Thus, the maximum 3-hour S02 concentration measured  by the
Muskingum River Power Plant monitoring network was measured at the monitor
closest to the stacks, and occurred under very unstable atmospheric  conditions.

-------
80-53.7
00
+J •
C '-~

S 3
CU
S- •
3 C
to o
13 4-J
cu en
= c
.f^
CM4~>
O C

~^
j-
o '"o
T -a
ro cu
i.
*J 3
1/1 to
CU 13
J= CU
ai S
IE +->
.^
CT C7)
C •!-
•r- CU
i_ r"
3
T3 01
C
1/1 -r-
3 X
O •"-
•i— c:
ro T3
> C

CU
to S_
j-; CU
O >
O
TJ

^J T»
CU in
= 1
C
•a o
c
••a TJ
cu
to i.
C 3
O (/I
•t-  CU
rO *
s- ~
+J 13
c eu
cu cu
u a.
c: i/i
o
U T3
C
CMf-
0 2
co »— ^
•
>W4
M— 4

cu
pw
r-s
fo
1—
O
O CD 4J
€Z C ^ -~~*
i- T— C>1 S
CU X 'I~ "— ^*
«-)•!— CU
^^ 2C -~"
<^


ooo coo x> Z<~o
r— CM I— r— —





^— ^
CU oo
3 -0 S_ .C
CT 3 CU *J
13 O > C
CX^- O Ol
O O O 4J


OOO Oroo OO
f"" 1 ^— r-"
1 O J;




T3 ^^
-o cu ^=
C CU GL
•r- CL S
3 tn>— •

rO"=fCM rO'^CM «TCO
III III II
CMfO^- CMCOi — • — 1 —



+J
C i- C
4- CU O 1-
o S -i-1 ca
CU •!— O
00 i. C C
S. 3 0 0
ZZ t/3 ^u CJ
O fO
HI U 4-* "4—
^•_ rQ O



UO to <-Q ^o ^" r** CM
P-* f—* r"~ i™* I-" r"" UT5 i™-
III 111 II
ro ^* ^~ ^r CM uo ro co
r^ ^« r^ ^~ ^"* r^* i*^





i.
o
-i-!
CM •-- -5J-
O C
00 >, O
T3 S


OT r— r— CTlr— CM LOrO
<— OO f— Lf) CM OCM
« * • * • * • •


S- O
on s.
= C 0
1 -1— +J
ro S- -i- ro
3 C
•o cz o
CU -~-* ^~


CO ^*^ CO CO ^^ '^^ **? C^
CNJOO cNjcocvj «cr^r
• *• *•• ••


!- = =
3 O S.
tO -r- O. i-
13 4-> •— O
(U !T3 -r->
S S_ -1- CM
4-> C
-i-> c: o
00 CU S


OCM.— OCM.— ^— «?r
T3" CM OO ^ OO CO r— CM
• » • • • • • •


cu u
JZ C
01 o s-
•r- CJ O
3^ +• *
•r- lO
^
o



co ^^ O^ co co co r*^« ^f
!»O ^* CO *«O f™* i— f— • «™—
• •• ••• * •





A
c 3
^ ^.
cu +-> cu
U 00 4-) CJ
C CU 13 =
O -= O
0 .? ^3 " — .
'-(- ~ C 4- i-
o cu o s-
H™ — ^^,
i_ O CU !- >,
O 1. O ro
^_j OO ^J ^J -^
•^ CU 1^ •-- ^
— -w .-a c: o
O ^3 D Cy ^~"
S d 2; 2; . 	 .
^-™ * -— ** >-™s.
s = e


(*** ^o ^n
• • •
r"^ ^J" CO
~O CM ro
co
s- en s_ ^ s_ ai
O Olr^CO O CTiCO^, O CT> rv.
4-> r*** ^*x r*^ -w r^* r^. en -+-J r**» "^^
•r* "^^ CM """^ 'I1™ "^^ "^*^ r—* T- *^^ uO
c tn CM _n — ui o ^->. c «^- t—
o ~^.-^^-, a -^^,^- o -v, ^,
^. -T5 Lf) cn si -O en ' — ^. '-O *&


O OOO
o o o ^o
ro ro LO ^O
r^ r™- r-~







ro ro O O
i t f™™
O O 1
Lf)




rf Lf) CO ro
I III
ro ro LO O






^- Lf) ro o
i 	 i 	 r— CM
i ill
CM ro •— cc
|»"- r— i^— r—







i— i — ro CM
Lf) Lf} CM CM
• • • •





^f ^" O *^*
co oo -cr CM
• • • •






C\J CM *T j—
CO CO CM CO
* • • •






OO CO *3- CO

* • • •





^— x
s
^^

r*«.
•
o
^J
^J.
co
S- 01 r-.
CO 0 CO r~~ -s.

^**x *i— ""^-N, ^) r™»
l^O C -Q »^ "*^,
^*s. O "^x '"N. r—
o*\ ^ .n ^o "—

-------
                                                                       80-53.7


  The maximum impacts ooserved at monitors located at greater distances from
the power plant occurred during periods of higher  wind speeds and/or more
stable atmospheric conditions.  It should also be  noted that  these more
distant monitors, monitors 3 and 4,  are higher in  elevation than  monitors
5 and 2, and therefore closer to plume height.  If the monitors were located
at lower elevations, they would be expected to observe lower  concentrations
than those measured at the higher elevations.

  Because it is observed that maximum short-term concentrations occur close
to a source with tall stacks, other sources with quality assured  monitors
close to their tall stacks also warrant review.   In addition  to the Muskingum
River Power Plant, there are three Tennessee Valley Authority (TVA) power
plants which were found to have quality assured  continuous  SC^ monitors close
to their tall  stacks.  Table III lists these four  power plants with quality
assured continuous SC^ monitors within 2 km of their tall  stacks.  Table IV
lists the stack design and full load operating parameters  and emission rates
for the power plants identified in Table III.

  Although there are no other continuous monitors  along the same  radial  from
the stack for the TVA monitors, and  thus no means  of further  assessing the
distance to maximum impact at these plants, the  meteorological conditions
associated with the occurrences of maximum short-term impacts at  these monitors
can be further evaluated.  It is observed upon review of the  meteorological
data during periods of measured maximum short-term SO? concentrations, that
conditions generally consisted of those associated with very  unstable
atmospheric conditions; i.e., very light and variable winds,  clear skies,
and well developed mixing layers, which occurred during the late  morning and
afternoon perioas.

  Thus, the modeling of maximum 3-hour SO? concentrations  resulting from tall
stack effluents should be approximated by~the  transport and dispersion process-
es associated with very unstable atmospheric conditions.  Although such trans-
port and dispersion processes have not necessarily been adequately observed
and understood, Gaussian air quality models which  account  for the occurrence
of very unstable atmospheric conditions by use of  stability dependent dispersion
coefficients have traditionally been used to estimate air  quality impacts.
Pasquill-Gifford class A stability dispersion  coefficients1 are one of the sets
of coefficients that have been used by regulatory  agencies  and others in model-
ing tall stack effluent impacts under very unstable atmospheric conditions.

  Because a controversy has existed regarding  the  use of the  Pasquill-Gifford
(P-G) curves for class A (very unstable) atmospheric conditions in estimating
ground level pollutant impacts from tall stacks,  the sets  of dispersion
coefficient curves to be considered  herein will  include the P-G class A curves
and the curves derived from field studies for  which tracers were  released at
heights of 100 m or higher.  The P-G class B curves are considered because
several investigators have suggested that replacing the P-G A curves with the
P-G class B curves would yield a better comparison of maximum calculated with
maximum measured impacts.  The contention being  that the P-G  class A curves
when used with standardly available models (such as the U.S.  EPA  CRSTER or
MPTER models)  result in overestimates of ground  level  impacts and that these
impacts are calculated to occur too close to the stack for  sources having tall
stacks.

-------
80-53.7








•
CO
-^
03
4->
CO

f^
r—
03
*->

i.
• r—
-
4->
t)
0

2
_s«i

CM

C
•**•
^
4->
2

CO
&«
O
[J

O
""""
CM
CD
V)
"C
CU
^
—
I/I
03

>5
— >
•p-
r—
03
3
CT
JT*
4->
•»—
2

to
4->
C
03
r^
—
i.
0
3
O
a.


*
^^
1— H
•— H

01
^™
-0
03
' —


S_
O
.tf
c
o
s:







s_
0
4->
• r-
CI
O









!_
O
4->
• r—
C
Q
S




















































^
(J ^~*-
03 E
4-> 4J • 	 •
"en 01
•r- 01 CO
Ol > 03
— O 23
.a
03



CO
^
-^ (J
4jj 03
3 | > *-^
E oo O
• r™ ••— ••
N =
< 0
s_
tl




1/1
•V*
Ol O
U 03
C -*-* ^"™s
03 00 S
.^_J \^
(SI ~ *— '
,r- Q
<4»






C i-
-M O)
•r— jQ
c a
ii






x^-s
01

03
C 4->
o oo
•—
03 >,
!J 4->
O C
-J 3
0
t^J
^^ „










^J
C
^
f«*
a.

s_
a;
2
O
2-




^j* o^ ^"












r-x 0*1 o*\
co c\j ro
CM . —












f*** ^^O ^D
• * •
f-mt r— " /—










un *T T
CM






O

r^7
o

* 2Z i
C H- <
°
31 CO C
ceo
•f- -t- CO
g" ^^ ^^
01 S 0

•^- XX
a; s- 01
OI Ol
~ 'f~ ^
3 > O
en ii
C i/1 CO
• r» 3«
Ji C 0
LO — T3
3 0 -r-
S ~3 3




CM












fs^
en













00
•
f*—










^O












•^^
H-

4>J
i.
'5
2

fj
03
4->
OO


•
>
i-^

OI
r—
r~i
03
1 —
C
O

CO
CO
f=
UJ




.^f
X
LU

^*
^J
03
4-i
00



•f™>
X
LLj
X
O
03
•4_J
o^




^
*xj

oo




-X
u
03
4->
OO










































.*— ^
O) co
4-> ^^
03 O"





Ol
3
4_) --^
03 i^
i-O
cu -^

^
CU
1 —



>,
4-1 — »
o •*«.
0 =
r«> ^— ^
CU




^
0)
4->
2 O1
rg ^_-
•r—
O



2
en—*
•r- £
_o> -~.




j_
v- ^j
CJ o
03 S
4-> 3
oo -z.




s_
4-) Ol
•r— ^
cz S
=3 3
•zz











-t->
C
03

O.

^
01
S
0


o o
0 0

O"^ iQ








o to
OO CM
^* ^r








LO CM
4- 00
OO CM









<£> r~-
P-~ LO






CM CM
tO tO
CM CM







i— CM








^- 1^5
1
t— •








t_
Ol
>

2£

f— ;

CD
C

\/
wo
3
S

O O O
O O O
en en r^*
r— r— r-








CM CM CM
i — i — CM
«3- ^f "fl-








OO OO r^"
CM CM CO
r— r— CM









—








CM ^*  o
O)
OO CO
2
C O
-S -o
O -f-
-3 3

O O
0 0
en i^.
^








en r*^
en ^o
oo oo








r— f"--
OO CM
CM CM









oo »a-
UD r~-






CM CM
LO LO
r*~ f—







CM OO








r~» OO


























o o
0 O
LO LO
o"o"








CM CM
CM CM









en en
CM CM
OO 00









tO to
en en






LO LO
o o
OO OO







•— CM








•— CM















-a
c
03

^
O)

—
3


-------
                                                                       30-53.7
  Three independent sets of field experiments  have been  found  in  the literature
which investigate the dispersion of tracers  released  at  heights of  1QO  m and
higher under very unstable atmospheric conditions. Singer and  Smith,^ Vogt, et
al.,4 and Thomas, et al.,  have derived or and err curves based on such  field
investigations.  Table V identifies the curves of dispersion coefficients
considered herein.  This table indicates the investigator,  experiment site,
tracer release height, and surface roughness length (2Q) for each set of curves
considered.

  Figures 1 and 2 graphically present the curves  of o~  and  a^, respectively,
for each of the tracer studies identified in Table V.

                     Analyses of Periods of  High  Short-Term
                     ""Impacts'at'Monitors Near~Ta11  Stacks

  Table VI indicates the highest 3-hour SOo  concentrations  observed under very
unstable atmospheric conditions at each of the monitors  located within  2 km
of the power plants identified in Table IV.   Table VI  also  compares the observed
values with the concentrations calculated using the alternative sets of
dispersion coefficient curves.  Calculated values are  based on the  use  of:
on-site meteorological data, assumed full load operating conditions (for each
stack in operation at the time), Briggs plume  rise,  and steady state Gaussian
dispersion assuming co-located stacks and an unlimited mixing  height (because
no on-site measurements of mixing layer depth  or  actual  plume  height were avail-
aole).  The calculated 3-hour concentrations are  the  average of three 1-hour
calculations.  It should be noted that experience shows  higher impacts  are
calculated at full load under unstable atmospheric conditions  than  for  reduced
load operating conditions.  Furthermore, background concentrations  were found
to be negligible during these periods based  on upwind  observations.

  It can be seen from Table VI that all the  calculated concentrations result-
ing from the use of each of the alternative  sets  of dispersion coefficients
underpredict the observed 3-hour average values at each  monitoring  site.
Calculations based on the use of the P-G A and Karlsruhe A  curves,  however,
appear to best approximate observations.  It should be noted,  although  not
directly shown, that even if the mixing layer  was assumed to be equal to
plume height, the 3-hour average concentration would  still  be  underpredicted
by use of the P-G B, BNL 82 and Jiilich A curves.   Furthermore, analyses have
also shown that even if it is assumed that the mean wind blew  continually in
the direction of the monitor, the concentrations  calculated by use  of the
P-G B, bNL t$2, and Julich A curves still underpredict  observations.8

  Obviously the comparisons made in Table VI do not necessarily consider the
upper ends of the frequency distributions of predicted and  observed concen-
trations (to be addressed later herein) as recommended in the  "Guideline on
Air Quality Models."y  However, Table VI does  provide  an example  of the com-
parative ranking of the impacts calculated at  select monitoring sites using
the alternative sets of dispersion coefficient curves.

  The relative calculated impacts using the  alternative  sets of dispersion
coefficient curves can also be demonstrated  graphically  and are depicted by
Figures 3 and 4.

-------
                                                                                                                                  80-53.7
 X
 CJ
 CJ
 CJ

 CJ

4-
4-
 CU
 O
 CJ

 c
 o

 CO

 CJ

 cO~

73
 CO
 CJ
CO


T3
r Surface
se Roughness
cu -a
CJ CJ
03 r—
^_ QJ
[— ."V
































W*
4J S
CD- — •
C
4-1
-te *"""*
•r— -"^-
_OJ

ro
O
0
r._
V













OJ
^J
• f—
i/*}

-4-J
c
a
~
•r-
^.
CJ
^_
X
1 1 1





H3
£4
tyi
fT3
i.
-Q
OJ
^^

M
r—
r—
•t—
CJ
«z
—
o




o
o
F—
V













^o
^
CO
03
£_
^
U
2Z

M
r—
,_
• r—
^
"^
.
o




-»-.
o o
o o

A> ^'










>}
3
03
>- =
S S-
cu
-o
73

03 3
r—

CO
^
3

4->
CO
r^


.r—
CO
4*)
C
CU

•r-
i-
CJ
X
cu

"^•^
	 i

2
NM^

>)
i.
o
4_J
03
S_
O
^
03
— 1

^M
03
C
O
• r™
•^
03
>>2T













































s_
o
4.}
03
CT
• r—
4-)
l/l
CJ
>
c
w*


^3
i_
0
4-
4-
•r—
^D
1

^_
•r"
3
cr
CO
03
O-














CJ
>
SB
3
CJ





<

•J
1
a.




^D
W
O
*+-
4-
•f«
0
i

r—
<^>
3
3-
co
03
a.














23

O
I
a.








~
4™J •
'^- i—
= 03
CO
1 4-)
i- CU

C CT
•- O
CO ,c>








J



o =C
-NIC
23 CJ
•f~
_J r—

-2 F-J







•
I—m
03

4-J
CU

CO
03
C
Q
~
1—








2
!^

CJ
-S
^.
CO
P-"
i.
'73
^
o

73
CO
>
•r™
^-
CU
73

C
CO
CJ
^

co
>
03
-C

CO
ai
>
V
3
CJ

1 '"^
°
s_
o

b"

0
^^


ra

^
CJ
>
03
J^
J^
O
O
t_
i23

OJ
J^*
4->

4-J
tt3

73
CU
>
i_
CO
CO
^
O

co
c:
o
•r-
_i
Y^
^
O
CJ



ind direction traces
table than f^). Curves
3 co
C
C 3
O
CJ
73 S-
Oi O
CO =
03
_O CO
25 -r"
^^
•?Q
C «
03 •
CU
M •
) S-

•f— ^^
,_
••- 73
o C
O3 03
4->
CO CO
,_•
CJ -O
4-J 03
03 4->
O CO
•^ w
73 3
C
•r- 4-1
CO
.= O
^J —
•r-
= CO
OO 4J
c
73 CJ
^ CJ
i-
1. ZL
CJ CJ
CT i.

^ <
OO ^— ••



























o
•
G

-^
^
n3

•*
^J

A
,__t
CQ

M
CM
ca

co
cu
•r—
i.
O
wl
CU

03
O

s_
c
4-

>,

C
0

73
CJ
>
i-
CJ
-^
CJ

CJ
•^»



o
cu
"3
s_
CO
03
03
CO
CT
CJ
?~

•

^*
o
4-

73
CJ
^
O
I—
cu
>
CU
73

CJ
o

O
CO
,_
03

CU
>
(^
^J^

CO
cu
>
S_
3
CJ

1 N
b

73
C
03

£

4—
O
co
4->
CJ
V)


.a
4-J
4-
4-
OJ
0
03
4-J
CO
03
O
CO
.E
CT
OJ
J=



0
4-J

4J
CO
CU
CO
O
»•_
U
•^.^

co
^j
f~
CT
• r—
CJ
^~

CJ
CO
03
CO

CU
SH

i_
CJ

03
i-
4_)

4.^
CO
CJ

"CT
•P«
_^^ •
c
CU -i-
JJ= CO
4-J i-
CO
S_ JZ
0
4- 73
CU
•*/> 3
cu co
> •»"•
1. >
3 CO
CJ 1-
CU CJ
J= L.
4-> O3



                                                                                                CD  O  —

                                                                                                o  w  ~"
                                                                                                ^-  01  CM
                                                                                                    CJ  C

                                                                                               13 r—  O
                                                                                                03  03
                                                                                                CO  >  73
                                                                                                       CU
                                                                                                O 73  CO
                                                                                               4-J  CJ  CJ
                                                                                                   4J  X
                                                                                                i_  03  CJ

                                                                                               230
                                                                                               •r-  CJ  +J

                                                                                                O  O3  -t-J
                                                                                                =  cj  O
                                                                                                CJ  03

                                                                                                  CU
     cj CJ

 03  O 03
     CJ
 co     73
 C   CNJOJ
 O O f^
 •r- CO >-,
                                                                                               4-1   3  .03

                                                                                                CJ   !- 73
                                                                                                CJ   CT O
                                                                                                C  ^ -.-
                                                                                                O   cj  S_
                                                                                                CJ   03  CJ

                                                                                                CM—  ~
                                                                                               O   CJ —
                                                                                               CV1  73  CJ

                                                                                               73  r—  CU
                                                                                                CJ   cj

                                                                                                03  -^  O
 CJ  C CO
i—  O C
 03 -T- O
 CJ 4-> ••-
     03 4-)

 C W S-
 03  C 4J
     CJ C
73  CJ CJ
 CJ  C CJ
 >  o c
 1-  U O
 CU     CJ
 co 73
.a  cj 73
 o  > c

4-  CU O
 O  co S-
    -Q CT
 C O -*
 O — CJ
 co     03
•t-     23

 03 4-J
                                                                                               O "-  O
                                                                                               OJ
-a
 .-a
                           CJ

                          3
                           s_
                           co

                           i_
                           03
                           CJ
        en i— i— to
        "3- — r— O
                                                                                                              — ^

                                                                                                              C
                                                                                                              C  03
 CO  CO 73
 3J  C  CJ
 co  O 4J
 C -i-  03
-— 4J  S      10 0
                                                                                                      •r-     S_ r—
                                                                                                              CJ
                                                                                                              CJ
                                                                                                              o
                                                                                                              CJ
                                                                                                              O
                                                                                                              C/1
                                                                                                O3 73  O3     3
               _O

               T
cc
o
  i
c.
                          'O

                          a.
                          73
                          CJ
                          CJ
                          CO
 i_

 O
                                 i—< CM  CO C
                                 o —  o o

                                  VI
                                                                                                                                   o o
 o o o  o
 o o o  o
       P~» CM O LO
       *r CM CM o

        V!
                                 OO  CO CO r-«
                                 *O  ^" CM CM
LO I


  I
     I    I    t
     O  CM LO
 O3
a
                                                                                                                         03
;_
cu
a
           r-.
    r*^  CO r^*
cn f^^  [^^ ^^»
r-. -v.  •— CM
^-,^0  O CM
LO CM  CM	

1^5 c^o  cn ^^
                                                                                                                                s_
                                                                                                                                CU
     CJ  CU 73
 = -i-  5-  =
 3  >  O  03
 CT  CJ     —

•Z-     3  CJ

 co ^  "C3  —.
 3  O  —  3

-------
                                 o   _
                                 co   N "o
                                 m
                                'd
            s
                                 p
                                 CO
                                  j*
                               ,  co
                               "d
    •§
                                        §
                                     !C v*. co
                                     8 ° >
   b
                                     Sid
                                     111
                                     S "° o
                                     8 *. w

                                     § Sa
                                     .2 o *-
                                     f — g
                                         o
   .§5-=

r 2 =°g
^S (0 tfl
   c « g

   .§ 11
   5 g tS
   X u T3
                                     §



                                     'I
            8
(UJ) A.

-------
q
d
CN
 I
q
d
q
LO
 I
q
cs
 I
o
                                                                  o
                                                                  o
                                                                  o
                                                                  |) joeduui
                                          o;
                                                               O -C
                                                               03 O5
                                                               fl


                                                               if
                                                               .it
                                                               x *.
                                                               (0 O
                                                               £ c
                                                                 O
                                                                    0)
                                                                    X
                                                                    
-------
                                                                       80-53.7
  Figure 3 indicates the maximum normalized concentration  as a function of
plume height for the alternative sets of dispersion coefficient curves
assuming an unlimited mixing height.   Maximum concentrations would be a
factor of two larger for the worst case assumption of plume trapping, where
the plume height is restricted to a height equal  to the height of the mixing
layer (that is the plume is reflected by the base of an inversion just above
the plume center!ine).  It can also be seen from  Figure 3  that, for plume
heights greater than 200 m, lower maximum normalized concentrations are
calculated using the P-G B, BNL 82, and Jiilich A  curves compared to the P-G A
curves.  The Karlsruhe A values are slightly higher (about 20% higher) than
the P-G A values.

  Figure 4 indicates the distance from the source to the point of maximum
impact as a function of plume height for the alternative sets of dispersion
coefficient curves.  As can be seen from Figure 4, use of  the P-G B, BNL 62,
and Jiilich A curves calculate maximum impacts for plume heights greater than
200 m to occur at  greater distances from the source than calculated by use
of the P-G A curves.  Maximum impacts are calculated to occur closer to the
source by use of the Karlsruhe A curves than by use of the P-G A curves.

                 Analyses of the Upper Ends of the Frequency
                 Distribution of Predictions and  Observations-

  Sulfur Dioxide Emission Limitation (SDEL) Systems (where plant load and
emission rates are reduced when high ground level impacts  are predicted) are
employed at both the TVA Widows Creek and Cumberland power plants.     Because
the emission rates may be reduced at both Widows  Creek and Cumberland during
periods of predicted high impacts, comparison of  the upper ends of the
frequency distribution (based on assumed maximum  emission  rates) would be
inappropriate.  Therefore (lacking readily available hourly operations and
emission data), comparisons of the upper ends of  the frequency distributions
can only be appropriately made for the Muskingum  River and John Sevier power
plants.  In order to compare the upper end of the frequency distribution of
predicted concentrations to maximum observed concentrations, the highest
and second highest concentrations predicted to occur at the Muskingum River
and John Sevier monitoring sites for each year of meteorological data avail-
able are compared in Table VII with the highest and second highest concen-
trations observed during each year of monitoring  data available.  The cal-
culated values indicated in Table VII assume full load operating conditions
as indicated in Table IV.  Calculations were made using the U.S. EPA MPTER
model, a steady state Gaussian model (similar to  the U.S.  EPA CRSTER model
but able to account for multiple sources and separated stacks and locate
receptors using Cartesian coordinates) using the  P-G dispersion coefficients
for class A stability conditions.  The MPTER model was otherwise run with
the same modeling assumptions used in the CRSTER  model.  Hourly sequential
surface meterological data used in the MPTER model were from Huntington, WV,
for Muskingum River and from Bristol, TN, for John Sevier; mixing height
data in both cases were from Huntington, WV.

   It can De seen from Table VII that maximum impacts calculated using P-G A
dispersion coefficients compare well with observations.  The MPTER modeling
results for the Muskingum River monitors also consistently indicated that the
maximum 3-hour impacts were calculated to occur at the monitor closest to the
stacks.

-------
                                                                       80-53.7
Table VII.  Highest and second highest  3-hour
            calculated at the closest monitor
            John Sevier power plants  (ppm).
                                S02
                                site
                       concentrations  observed  and
                        to  the  Muskingum  River  and
Observed
Year
Highest
Second
                                           Year
                                                      Calculated
                                        Hiqhest
                                          Second
Muskingum River
1978
1979
John Sevier

1973
1974
1975
1976
1977
1978
.39
.63
.17
.23
.45
.40
.43
.34
.20
.41
.16
.22
.39
.39
.31
.30
1964
1971
1972
1973
1974
1975
1977
1964
1970
1971
1972
1973
1974
.43
.33
.40
.40
.36
.55
.41
.32
.26
.24
.29
.23
.21
.29
.25
.34
.34
.30
.43
.34
.32
.18
.21
.25
.16
.17
  Review of the meteorological  conditions  for the  periods  of predicted maxi-
mum impacts indicates that very unstable atmospheric  conditions  are  associated
with the maximum calculated values.   These conditions include light  and
variable wind velocities, clear skies, and mixing  depths  of  1000 m or more
which compare well  with the meteorological  conditions associated with the
maximum observed impacts.

  The model calculation methodology  used in deriving  the  concentration esti-
mates shown in Table VII assumes that if the calculated  height of the plume
at final rise by methodology of Briggs'  exceeds  the  height of the mixing
layer that the plume will be completely and permanently  contained in the
stable layer aloft  and therefore that no ground  level  impacts will result.
This modeling methodology assumes that plume rise  will not be influenced
by the height of the mixed layer or  by the temperature structure of  the
atmosphere under unstable conditions as  classified by Turner.11   It  might
be suggested, therefore, that these  modeling assumptions  are not realistic
and that if high concentrations are  observed close to tall stacks that such
concentrations result due to the plume being trapped  within  a mixing layer
having a height much less than  the height  to which the plume would rise if
an unlimited mixing layer existed.  Such a condition  would cause higher
concentrations to be calculated with each  of the sets of  dispersion  coeffi-
cients than if an unlimited mixing height  was assumed.   That is, the cal-
culated maximum concentration would  be greater and occur  closer  to the plant
for lower plume heights, see Figures 3 and 4.

-------
                                                                       30-53.7
  Analyses of the Muskingum River data  assuming  the  effluent  plume  is  trapped
within the mixing layer observed during the  period of  maximum measured
impact, demonstrate that use of the  P-G 3,  BNL B2, and Jiilich A  dispersion
coefficient curves result in underpredictions, as well  as  the predicted
location of maximum impact to occur  further  downwind from  the source than that
observed.   Further analyses of the Huskingum River data demonstrate that if  the
effluent plume is actually trapped within  a  mixing layer sufficiently  shallow
such that a higher concentration is  calculated to occur at the Center  Bend
monitor than at the next closest monitor,  that use of  the  P-G B  or  BNL B2
curves would result in a higner calculated  maximum concentration than  the use
of P-G A curves with the plume trapped  within the actual observed mixing
1ayer.

  Without actual  observations of plume  height, wind  and temperature profiles,
trajectory of tne elevated plume, and the  three  dimensional distribution of
the pollutant mass with time, caution is warranted with respect  to  revision
of tne plume rise, transport and dispersion, and mixing layer trapping
assumptions and associated model algorithms.

Summary ana Recommendations

  Based on existing measurements of  maximum 3-hour  impacts of tall  stack
effluents, it is observed that:

    t The highest measured 3-hour impacts  occur  at the monitor closest to
      stacks (for cases where the directions and elevations of the  different
      monitors relative to the source are  comparable).

    > Maximum impacts measured at monitors  within 2  km of  sources with tall
      stacks occur in the afternoon  with light and variable wind velocities,
      mixing depths of 800 m or more, and  very unstable atmospheric condi-
      tions.

    • Highest impacts measured at monitors  located more distant  from the
      source tend to occur during conditions of  higher wind speeds  and/or
      greater atmospheric stability.

  Based on calculations of maximum impacts  for monitoring  sites  in  the vicinity
of tall stacks, it is observed that:

    t Calculations using on-site meteorological  measurements  and P-G B,  BNL  82,
      or Julich A dispersion coefficient curves  underpredict  observations
      during the periods of maximum  impact observed  at monitors  within 2 km
      of sources with tall stacks.

    t Comparisons of calculations and observations  for the Center Bend monitor
      assuming the effluent plume is trapped within  the mixing  layer observed
      during the period of maximum measured impact demonstrate that the  use
      of the P-G B, BNL B2, and Julich  A dispersion  coefficient  curves result
      in unaerpredictions, as well as the  predicted  location  of  maximum  impact
      to occur further downwind from the source  than that  observed.

-------
                                                                      80-53.7


      a Comparisons  between  calculations  assuming the  effluent  plume  is trapped
        within  a mixing  layer  sufficiently  shallow such that a  higher  impact is
        calculated  to  occur  at Center  Bend  than  at the next closest monitor,
        demonstrate  that use of either the  P-G B or BNL B^ curves would result
        in  higher calculated maximum concentrations than the use of the P-G A
        curves  with  the  plume  trapped  within  the actual observed mixing layer.

      » The highest  and  second highest concentrations  observed  during
        each year of available measurements at monitors within  2 km of
        sources with tall  stacks compare  well with the highest  and second
        highest values calculated by the  U.S. EPA MPTER model for those
        monitoring  sites (considering  each  year  of meteorological input
        data available and assuming full  load operating conditions and
        the use of  the P-G A dispersion coefficient curves for  very unstable
        atmospheric  conditions).

      • Use of  P-G  A and Karlsruhe A curves in models  such as CRSTER  and MPTER
        result  in calculated impacts that better agree with the location and
        magnitude of observed  maxima than use of the P-G B, BNL 82, or Jiilich A
        curves.

  What has  been observed and reported  herein  is  that the magnitudes and distances
to maximum  3-hour concentrations as predicted using model  algorithms  and assump-
tions as contained  in  the U.S. EPA CRSTER and MPTER models compare well (perhaps
fortuitously) with  observations in the vicinity  of tall stacks. Note, it  has
not been suggested,  and  should not be  inferred,  that the P-G A  dispersion
coefficients (or any other set of dispersion  coefficients) do or do not accu-
rately describe the  distribution of pollutant mass of  an elevated plume.   Nor
have any claims been made or inferred  herein  as  to the accuracy of any other
model algorithms or modeling assumptions.

  Due to lack of on-site measurements  of  plume height, wind and temperature
profiles, and trajectory of the elevated  plume,  caution is warranted.  The
potential effects of factors such as increased initial plume growth during
plume rise, enhanced plume rise due to multiple  stacks, transitional  plume
rise effects, changes  in height of the plume  center!ine, and non-Gaussian
conditions  influencing transport and dispersion  have not been applied  in
the calculations or comparisons made herein.

  In order  to better assess  the location  and  magnitude of maximum short-term
impacts of  tall stack  effluents under  unstable atmospheric conditions, more
monitors are required  to be sited near such sources than has been standardly
practiced to date.   Measurement of plume  height, wind  and temperature  profiles,
and distribution of the  effluent mass  as  a  function of downwind distance is
also recommended in  order to better ascertain the transport and dispersion of
tall stack  effluents during  unstable atmospheric conditons.  Investigations
including such  measurements are warranted and necessary for the improvement of
present modeling methodology.

-------
                                                                        80-53.7


 References


 1.  F.A.  Gifford,  Jr.,  "Use of routine  meteorological  observations  for  estimating
    atmospheric dispersion." Nuclear Safety, _2(4):  47  (1961).

 2.  MNL,  Report to the  U.S. EPA of the  Specialists'  Conference  on the EPA  Modeling
    Guideline.  Argonne  National  Laboratory,  Chicago,  IL,  1977,  322  pp.

 3.  I.A.  Singer and M.E.  Smith,  "Atmospheric dispersion at  Brookhaven National
    Laboratory." Int. J.  of Air & Water Pollution,  JjD: 125  (1966).

 4.  K.J.  Vogt,  et  al.,  "New sets of diffusion  parameters  resulting  from tracer
    experiments in 50  and 100 meter release  height."  Proceedings of the NATO/
    CCMS  9th International  Technical Meeting on  Air Pollution Modeling  ana its
    Application, Toronto, August 28-31, 197J-S,  pp. 221-239.

 5.  H.  Kiefer and  W. Koelzer, Janrebericht  1978  der Abteilung Sicherheit.
    Kernforschungszentrum Karlsruhe, 1979,  pp.  191-195.

 6.  M.E.  Smith  (ed), Recommended Guide  for  the Prediction of the Dispersion of
    Airborne Effluents. 2nd ed., Am. Soc. of Mech.  Engineers, 1973,  85  pp.

 7.  G.A.  Briggs  Plume  Rise. U.S. Atomic Energy  Commission, Division of
    Technical  Information,  1969, 81  pp.

 8.  D.A.  Trout, "A review of horizontal and  vertical  dispersion coefficients
    applied in  the calculation of ground level  pollutant  concentrations resulting
    from  emissions from tall stacks under very unstable atmospheric conditions."
    Proceedings of the  NATQ/CCMS 10th International  Technical Meeting on Air
    Pollution Modeling  and  its Application,  Rome, October 23-26, 197§,  10  pp.

 9.  U.S.  EPA, Guideline on  Air Quality  Models.   U.S.  Environmental  Protection
    Agency, Office of  Air Quality Planning  and  Standards, EPA-450/2-78-027, 1970.

10.  J.M.  Leavitt,  et al., "Sulfur dioxide emission  limitation (SDEL) program at
    TVA power plants,"  JAPCA, 26(12): 1133  (1976).

11.  D.B.  Turner, Workbook of Atmospheric Dispersion Estimates.  USDHEW,  PHS Pub.
    No. 995-AP-26, 1970,  84 pp.

12.  D.A.  Trout, Comparisons of predictions  with  observations and analyses  of
    observations of high  short-term pollutant  concentrations monitored  in  the
    vicinity of sources with tall stacks."  Proceedings  of  the  AMS/APCA Second
    Joint Conference on Application of  Air   Pollution  Meteorology,  New  Orleans,
    March 24-27, 1980,  12 pp.
                                                                                    If,

-------