Docket
No.
A­
79­
15
Item
No.
XII­
D­
1
MEMORANDUM
Date:
June
12.1997
TO:
Dennis
Pagano
Dennis
Doll
Tom
&
say
CC:
David
Ailor
From!
Nick
Gudka
Ellzabelh
Anderson
Qubjsct;
Backgrmm~
ihfonnation
used
in
Modeling
of
BSO
Emisslons
from
U.
S.
Steei
Clairton
l'ha
U.
S.
Steel­
Clairton
coke
aven
SSO
emissions
were
modeled
using
a
process
very
similar
to
that
developed
by
Allegheny
County
for
the
PM­
10
SIP,
The
following
clescribes
the
data
usad
to
characterize
the
buoyancy
ofcoke
&
en
emissions,
estimate
plume
rise,
and
model
the
dispersion
of
benzene
salubfe
organic
(
BSO)
Eirnlssions:

Fix
the
PMIO
SIP
rnodellng,
the
Allegheny
County
Health
Department
(
ACHO)
develaped
a
total
buoyancy
flux
far
coke
oven
emissions
that
accounted
for
both
the
buoyancy
associated
with
convective
heat
transfer
from
hot
battery
surfaces
and
the
buoyancy
intrinsic
to
coke
oven
emimians.

.­..,.
..
­..
.
,
..
.
...
I.­­".
Euov­.
with
Convective
Heat
Transfer
ACHD
calculated
the
buoyancy
flux
associated
with
convective
heat
transfer
for
each
battery
line
using
the
same
approach
developed
by
 PA
in
1985
for
the
CIairton
facility.
The
EPA
methodology
is
discussed
in
a
document
authored
by
Layland
and
Met~~
h
(
1985)
(
attached).

ACHD's
resulting
battery­
speciflcbuoyancy
flux
estimates
are
presented
an
the
flnal
page
of
Appendix
d
to
the
Cibedy
Borough
PMlO
SIP
(
see
attachment).

Assodated
with
Q&
QOven
Ernissisns
ACHD
dBVelOped
a
methodology
to
account
for
the
buayancy
intrinsic
to
coke
oven
emissions.
Equations
presented
for
battery
line
1­
2­
3in
the
USX
Glaifbn
Saunse
Parameters
(
see
attachment)
and
Appendix
J
to
the
Liberty
Borough
PMIO
SIP
were
initially
applied
to
develop
avit
flow
rate
estimates
for
the
following
source3
of
emissions
(
charging,
door
leaks,
lid
leaks,
offtake
leaks,
pushing,
travel
car,
and
decarbonizing)
at
each
battery
line.
As
ab
presented
in
Appendix
J,
these
process­
specific
flow
rates
were
then
combined
with
process­
specific
exit
temperatures
and
the
ambient
temperature
to
derive
an
omission­
related
buoyancy
flux
estimate.

The
ACHD
buayancy
flux
wtimates
associated
with
both
convective
heat
transferand
emissions
are
summarized
on
the
last
page
a?
Appendix
J
10
thal'tibertyBorough
PMlO
SIP
ana@
is.
The
approach
is
further
described
in
Modelhg
Cuke
Oven
Fugitives
in
Cmplar
Terrain
(
Graham,
1996)
(
see
attachment),

ACHD
inCorPamted
the
battery­
specifc
total
buoyancy
llox
estimates,
along
with
benery­
specifK
source
configuratran
data
inta
the
Buoyanr
Lfne
and
PUhf
Source
modal
lap)
t0
est'mate
houdy
plme
tiso.
ACHD
provided
Sciences
With
copies
of
the
BLP
input
flies
Ibr
each
battery
Ilne,
along
with
the
meteorological
data
set
from
me
Limoln
monitor
as
used
in
the
PMlo
SIP
madeling
(
PKgO9lM.
EIN).
Input
files
wed
bv
ACHD
for
BLP
modeling
of
PM­
10
SIP
were
obtained
from
ACHD.
The
input
fibs(
BAmRY1.
INP,
BATTERY2.!
NP,
BATTERY8.
INP)
contain
parameters
an
battery
locations,
battery
dimensions,
and
total
buoyancy.
BATl RY
1.
INP
contains
Ihe
characterization
of
parallel
battey
lines
1­
2­
3and
7­
8­
9,
while
BATTERY2.1NP
contains
the
characterization
of
parallel
battery
lines
13­
14­
75
and
19­
20.
BA7TERYQ.
INP
contains
the
characretization
af
battery
6.
In
this
analysis,
the
ACHD
8LP
input
flleswere
used
wilhout
any
modlfication.
Thc
BLP
Input
filesare
attached.

To
model
plume
rise
using
ACHD's
input
data,
Sciences
abtalned
the
BLP
(
version
g0081)
source
codes
from
OAQPS's
llN
BBS.
The
8LP
pragram
was
then
Modified
so
that
program
execution
produces
an
output
file
called
"
BLPPROUT."
that
contains
#
re
hourly
estimates
of
plume
rise.
The
amended
codes
were
added
In
the
subrourlne
"
RISE­(
seeattachment).
The
modified
BLP
FORTRAN
source
des,
named
'
EW
1.
FOR,"
and
the
executabb
pragram
BLPI
.
WE
are
enclosed.

For
each
of
the
three
sgtS
of
Clatrton
coke
oven
batteries.
he
hourly
plume
rise
was
~
characteixadby
running
the
BLPl
.=
E
program
with
input
flee
and
meteordogtcal
)
data
files
provided
by
ACHO.
The
6LPl.
MEprogram
was
executed
with
the
fctlowing
command
\&
ere"
oahyx.
inp"
is
name
of
the
input
file
(
BAllERY1
.
INP,
BAlTERYZ.
INP,
or
13ATERYB.
INP).

'
Moutpul
fila
containing
hourly
pfuma
rise
estimates
fromthe
mddlfled
BLP
program
iscalled
"
BWPRQUT."
Prior
to
using
the
plume
rise
data
in
ISCS3.
it
WEE
necessary
to
append
to
the
end
of
the
ouiput
flle
a
tag
to
indicate
ihe
end
of
plume
rise
dala.
Thb
was
achieved
by
adding
a
fila
called
*
BLmlNY."
to
the
plum
rise
data
file:

copy
blpprout.
+
blptiny.
blpout.

vuhere
"
BLWUT."
is
the
final
name
of
the
plume
rise
data
file
used
lor
ISCST3
nnodeling.
The
BLP
pbrne
rise
output
forthe
three
batteiis
(
81_
QRIG.
OUT.
EIZORIG.
OUT,
6Q­
ORIG.
OUT)
are
enclosed.
Finalfy,
it
should
be
noted
that
the
values
within
these
output
fllsscorrespond
to
the
vatues
prssented
by
ACHD
in
a
ptaper
authored
by
Sulllvan
and
Weaver
(
i995).
Sulhan
and
Weaver
(
1995)
indicate
that
the
buoyancy
flux
and
souma
characterization
data
input
into
the
BLP
modd
produce
plume
rise
estimates
that
correspand
closely
ta
visual
observations
of
plume
rise
that
were
made
a1
the
lacitity
during
extensive
field
investigations.

1SCST3
Dtspersion
Modellng
me
1SCST­
3
program
was
modifled
to
utilize
hourly
plume
rise
data
from
an
external
file
called
 
BLPOU I .
The
ISC
source
codes
(
version
95250)
were
obtained
from
OAQPS
lTN
BBS.
The
SOUI­
CB
codes
were
modifled
in
lour
subroutines:
MAINI.
INC,
HRLOOP,
PUEFF,
and
PHEFFC
(
see
atiachrnent).
The
source
codes
of
the
modified
routines
and
the
executable
program
ISCPRXXE
are
enclosed.

For
ISCST­
3
modeling
the
1991
meteorological
data
tlle
used
by
ACHD
(
PITWQ1M.
61N)
was
converted
from
binaty
to
ASCII
format
using
BINTOASCEXE
pmgrw
lhat
wa~
Qbtainedfrom
OAQPSs
TTN
BBZ.
Tho
final
ASCll
file
used
for
ISCST­
3
was
named
P!
TMET2ASC.
The
meteorologiGd
data
file
and
the
file
format
conversion
program
ar4
enclosed.

In
ISCm­
3,
the
USXGlalrton
batteries
were
characterized
as
llnes
of
point
saurces
beginning
and
ending
at
the
UTM
coordinatea
speoified
by
ACHD
in
the
BLP
 
mut
files.
The
ratio
of
BSQ
emissions
from
aach
of
the
Clalnon
banedes
under
2010
UER
track
limits
was
assumed
In
ISCS73
(
see
memorandum
fmm
Manrin
Bmnscorne,
to
Amanda
Agnew,
 PA
for
battery­
specific
2010
MER
8S0
mimion
estlrnstas
for
the
Clajrton
facility).
The
tatal
ornlssion
rates
from
tfro
three
sets
of
batteries
w0m
scaled
to
a
sum
of
106
gis
in
order
for
the
ISCST­
3
results
to
contah
adequate
significant
dig&.

Battery
1
was
represented
by
60
vlnual
stack
sources
separatsd
by
a
dlstance
equivalent
t6
the
battery
width,
slack
height
of
8.5
meters
as
ACHD
indicated
In
BLP
input
files,
exh
temperature
gf
100
OK,
exit
velocity
of
0.0001
ds,
stadr
diameter
Of
2
mefern,
and
errdsslon
rates
as
shown
in
the
enclosed
1SCST­
3
input
file.
Battery
2
was
mprrsentod
by
56
vlrtual
stack
SGU~
separated
by
a
dlatance
equkalent
to
the
battery
width,
slack
height
of
8.5
meters
as
ACHD
indicafed
in
BLF
input
files,
eltit
temperature
of
100%,
exit
velocity
of
0.0001
mls,
stack
diameter
of
2
metera,
and
emission
rates
a6
ahown
In
the
enclosed
Input
ffle.
Battery
f3was
represented
by
15
VlrtUd
stack
SOurws
separated
by
a
distance
equivalmt
to
the
battery
width,
Stack
height
of
8.6
meters
as
ACHD
indicated
In
BLP
input
files.
exit
temperature
of
I00 
K,
exlt
Velocity
Of
0.0001
ds,
stack
diameter
of
2
maters,
and
omission
rates
as
shown
8
in
the
enclosed
input
file.
In
the
modified
ISC
code,
the
final
plume
rise
esllmates
from
BLP
are
used
Instead
of
the
stack
heights
assigned
within
the
input
flle.
Also,
the
exit
temperature
and
exit
velocity
were
selected
to
ensure
that
no
further
plume
rise
would
be
simulated
in
ISG.

For
this
anaIyGk,
large
and
small
recepror
grlds
were
created
from
DEM
(
1:
2400
scale)
data
for
Glassport
and
McKeesport,
PA.
The
large
grid
has
a
coarser
receptor
density
and
was
u6ed
to
identify
the
area
of
primary
concern.
It
contained
398
nxeptars
and
extended
in
UTM
coordinates
fmm
4466400in
the
North
to
d4565W
in
the
South,
and
fmm
598910
in
the
East
to
590010In
the
West,
Receptor3
of
this
large
grid
were
900meters
apatt.
The
small
grid
was
created
based
upon
the
reslrlts
of
fSC
modeling
with
the
lags
grid
that
identified
the
area
of
maximum
impad.
The
small
grid
contained
483
receptors
and
extended
in
UTM
coordinates
from
4463910
in
the
North
to
4660810
in
the
South,
and
ham
598440
in
the
East
to
595560in
the
Wed.
In
tho
small
grid
the
meptors
were
150
meters
apart.

The
ISCSv­
3
model
was
executed
using
"
RURAL"
mixing
height
ciassiffcation
as
characterizedby
Layland
and
Mew&
(
iBS5).

The
ISCST­
3
input
filesused
for
the
large
grid
{
BI,
lOLG,
INP,
B2­
10LQ.
INP,
E1B­
1
DLGJNP)
and
small
grid
(
Bl­
lOSM.
INP,
82­
1
OSMJNP,
 38­
1OSMJNP)
analysis
air0
enclosed.

The
ISCPREXE
program
was
emutad
for
mcb
of
the
batte'ries
using
the
appropriate
hpul
files
and
plume
rhs
dam
files.
The
program
was
exectit&
uskg
the
Command:

iscpr
bX­
1
OW.
lnp
bX­
1OW.
out
where
"
bX­
1OW.
inp'
and
"
bX­
1DW.
wt'are
the
name
of
the
input
file,
and
Ihe
name
ctf
the
output
file
that
provlde
summary
information­
Another
output
file
from
the
model
nun
is
GSc)
UT­
PLT".
The
GSOUT.
PLT
file
cudaim
air
concentration
estlmares
In
nn9/
m3
at
each
mceptor
Iocat­
m.

lhe
lacation
and
concentretlon
at
the
maximum
impact
polnt
wem
determined
by
surrunlng
concantration
estimates
at
each
receptor
within
the
three
GSWT.
PlT
output
fiiles
that
were
created
by
running
IsCPR.
U<
E
using
the
smll
receptor
grid
For
E!
atteries
1­
2­
m­
8­
9,
13­
14­
15/
19­
20,
and
8.
Finally,
because
the
total
emlsslon
rates
from
the
baneries
were
scaled
to
100
gis,
the
modeled
air
concentrations
are
scaled
to
reflect
aaual
2010
LAEA
emission
rates
(
6.91
Mgyr
or
0.22$
3).
The
maximum
impact
paint
from
This
analysis
is
located
at
4462260
North
and
597090
East
and
the
BSO
concentration
is
estimated
lo
be
0.166
mg/
m'.
While
BSO
rnonltoring
data
are
not
available
to
validate
his
estimate.
ACHD's
PMlO
SIP
modeling
analysis,
which
was
conducted
followingan
approach
very
similar
to
that
of
Sdenceg,
predicted
PMlO
concentratlons
that
correspond
closely
to
monitored
PM10
levels;,
Also.
aa
expected,
the
ACHD
and
Scienter'
studies
predicted
very
similar
maximum
impad
paints,

We
wi8
be
providing
written
docurnontation
early
next
week
of
how
the
concentration
estimates
at
the
offsite
maximum
impact
point
were
combined
with
both
exposure
and
toxicity
factors
to
edirna!
e
n'sk
to
the
rn;
uimally
exposed
individual.
In
the
meantime,
we
twt
that
the
enclosed
infonnatlon
win
plpw
sufficientforyour
review
of
the
technical
approach
used
to
madel
ambient
BSO
conGentrations
associated
wlth
coke
wen
emfsstbnafrom
We
U.
S.
Steel­
Glalnon
facility.
We
look
forward
lo
hearing
your
comments
on
fie
modeling
appmeah
and
the
applicability
of
this
rnethpd
to
Mer
U.
S.
coke
plants.
However,
In
the
meantime,
if
you
should
have
any
questions
concerning
Me
attached
materbl,
please
call
Nick
Gudka
at
(
703)
684­
0123.
