MEMO
FOR
THE
RECORD
TO:
Docket
No.
OAR­
2002­
0076
FROM:
Mark
Evangelista,
Chief
June
14,
2005
Applied
Modeling
Branch
NOAA
Air
Resources
Laboratory
In
Partnership
with
U.
S.
EPA
In
conjunction
with
the
document
"
CALPUFF
Analysis
in
Support
of
the
June
2005
Changes
to
the
Regional
Haze
Rule,"
this
memo
is
in
support
of
the
June
2005
revisions
to
the
Regional
Haze
Rule
and
the
Guidelines
for
Best
Available
Retrofit
Technology
(
BART)
Determinations
under
the
Regional
Haze
Rule.
This
memo
documents
the
development
of
estimated
emission
limits
based
upon
simulations
using
the
CALPUFF
dispersion
model
for
assessing
visibility
impairment
associated
with
prototypical
BART
eligible
EGUs
and
industrial
boilers.
As
described
in
the
CALPUFF
Analysis,
we
conducted
23
CALPUFF
simulations
to
inform
conclusions
and
decisions
contained
in
the
Regional
Haze
Rule
and
the
Guidelines
for
Best
Available
Retrofit
Technology
(
BART)
Determinations
under
the
Regional
Haze
Rule.
In
this
memo,
we
describe
a
method
for
using
the
results
of
some
of
those
23
CALPUFF
simulations
to
evaluate
the
reasonableness
of
establishing
presumptive
emissions
and
distance
factors
to
govern
the
need
for
further
source­
specific
modeling.
Results
from
these
CALPUFF
simulations
were
used
to
derive
tables
of
estimated
emission
rates
to
result
in
an
impact
of
0.5
delta­
deciviews
(
ddv)
at
distances
of
50,
100,
and
200
kilometers.
These
values
are
only
estimates
and
are
relative
indicators
of
the
relationship
between
visibility
impacts
expressed
in
ddv
and
emission
rates
specific
to
each
table.
However,
interpretation
of
these
results,
when
considered
with
the
margin
of
safety
within
both
the
CALPUFF
modeling
and
the
method
for
estimating
emission
limits
described
in
this
memo,
should
allow
reasonable
confidence
that
resultant
emission
limits
can
be
established
and
technically
supported.
The
CALPUFF
model
simulations
conducted
for
the
BART
analysis
are
described
in
detail
in
the
"
CALPUFF
Analysis
in
Support
of
the
June
2005
Changes
to
the
Regional
Haze
Rule."
The
estimated
values
of
emission
limits
determined
from
specific
model
simulations
chosen
from
the
23
CALPUFF
simulations
(
referred
to
as
Run
1,
Run
2,
etc.)
reasonably
represent
the
emission
and
source
scenarios
analyzed
and
described
within
the
Regional
Haze
Rule
and
supporting
documents.
The
simulations
chosen
from
the
23
CALPUFF
simulations
are
listed
in
this
table
with
reasons
for
selection
listed
below
Pollutant
to
Estimate
SO2
NOx
EGU
 
East
Run
3
Run
3
EGU
 
West
Run
4
Run
4
Industrial
Boiler
 
East
Run
16
Run
18
Industrial
Boiler
­
West
Run
17
Run
20
Selection
of
EGU
simulations
for
SO2
and
NOx:
We
selected
Run
3
and
Run
4
to
serve
as
the
basis
for
estimating
the
relationship
between
SO2
emission
rates
(
tpy)
and
visibility
impact
(
ddv),
as
well
as
of
the
relationship
between
NOx
emission
rates
(
tpy)
and
visibility
impact
(
ddv).
These
estimates
will
be
used
to
evaluate
the
reasonableness
of
establishing
presumptive
emissions
and
distance
factors
to
govern
the
need
for
further
source­
specific
modeling
for
the
emission
scenarios
for
which
Run
3
and
Run
4
were
conducted,
namely,
EGU
sources
of
SO2
and
NOx
for
the
Eastern
and
Western
domains.
Each
domain
utilized
separate
meteorological
data,
background
ammonia
concentrations,
and
natural
background
visibility
conditions.
Run
3
and
Run
4
differed
in
NOx
emission
rates,
but
used
identical
emission
rates
for
SO2
and
primary
PM
2.5.
This
difference
in
NOx
is
due
to
the
different
ratios
of
NOx
to
SO2
applied
to
the
Eastern
and
Western
domains
in
the
CAPLUFF
model
simulations.
As
described
in
the
CALPUFF
Analysis,
these
ratios
were
selected
to
best
represent
a
prototypical
relationship
between
primary
pollutant
emissions
in
East
and
West
for
the
EGU
modeling
in
Runs
1­
6.
(
NOx
emissions
were
treated
as
a
function
of
SO2
emissions
for
Runs
1
and
2;
those
NOX
emission
values
remained
unchanged
while
SO2
varied
from
10,000
to
1,000
tpy
in
Runs
3­
6.)
Runs
1,
3,
and
5
varied
SO2
emission
rates
from
10,000
to
1,000
while
holding
the
NOx
emission
rate
to
3,500
tpy
for
all
three
runs
for
the
Eastern
domain.
Similarly
Runs
2,
4,
and
6
varied
SO2
emission
rates
from
10,000
to
1,000
while
holding
the
NOx
emission
rate
to
6,250
tpy
for
all
three
runs
for
the
Western
domain.
Resulting
visibility
impacts
in
ddv
from
the
three
runs
for
each
domain
appeared
linear
when
plotted
together
against
emission
rate
and
distance.
To
determine
the
estimated
emission
limits
for
EGU
scenarios,
we
selected
Run
3
and
Run
4.
They
represented
the
midpoints
of
the
linear
relationship
with
SO2
emission
rates
of
5,000
tpy.
Selection
of
non­
EGU
simulations
for
NOx:
Unlike
EGUs,
non­
EGUs
were
in
a
very
diverse
mix
of
emission
rates;
no
ratio
was
apparent.
In
fact,
many
potential
BART
non­
EGU
sources
only
listed
emissions
of
one
pollutant.
Additionally,
there
were
no
instances
of
more
than
two
model
run
scenarios
similar
enough
to
determine
linear
or
other
relationships
by
plotting
the
results
on
a
common
axis.
In
these
non­
EGU
cases,
we
chose
individual
runs
for
their
individual
emission
information,
with
a
preference
for
CALPUFF
runs
in
which
one
pollutant
was
modeled
in
much
greater
quantities
than
the
other
pollutants.
We
were
fortunate
to
have
Run
18
and
Run
20,
which
simulated
industrial
boilers
with
zero
SO2
emissions
and
1000
tpy
NOx
emissions.
Selection
of
non­
EGU
simulations
for
SO2
:
While
we
desired
runs
in
which
emissions
of
1000
tpy
SO2
and
zero
NOx
were
simulated
,
no
CALPUFF
runs
were
conducted
with
zero
NOx
emissions.
We
therefore
selected
Runs
16
and
17,
which
modeled
900
tpy
SO2
and
300
tpy
NOx,
as
the
closest
approximations
to
1000
tpy
SO2
and
zero
NOx.
We
believe
that
it
is
reasonable
for
the
purposes
of
this
analysis
to
treat
the
combination
of
visibility
impacts
of
such
emissions
of
these
pollutant
as
roughly
equivalent
to
that
of
1000
tpy
of
SO2.
Given
the
total
SO2
and
NOx
emissions
of
1200
tpy,
visibility
impacts
of
Run
16
and
Run
17
were
thus
utilized
in
the
estimation
method
as
having
been
derived
from
1000
tpy
SO2
and
zero
NOx.
Calculation
of
estimated
emission
limits:
Once
we
identified
the
best
runs
to
use
for
the
estimation,
we
recorded
the
5­
year
average
visibility
impact
values
for
each
run,
for
50,
100,
and
200
km..
To
determine
the
estimated
emission
limits,
we
estimated
the
emission
rate
needed
to
result
in
a
visibility
impact
of
.5
ddv
at
50,
100,
and
200
km
by
using
a
simple
ratio:

Target
Emission
Rate
/
0.5
ddv
=
Modeled
Emission
Rate
/
Modeled
Visibility
Impact
This
converts
to 

Target
Emission
Rate
=
Modeled
Emission
Rate
/
(
2
*
Modeled
Visibility
Impact)

The
Modeled
Emission
Rate
is
taken
from
the
input
parameters
for
the
particular
CALPUFF
model
simulation.
We
selected
the
Modeled
Visibility
Impact
as
the
98th
and
99th
percentiles
from
the
results
of
each
CALPUFF
simulation.
These
results
are
tabulated
for
each
of
the
23
CALPUFF
simulations
and
are
included
in
a
spreadsheet
file
in
docket
OAR­
2002­
0076
under
separate
cover,
titled
within
the
spreadsheet
as
"
Ten
Highest
ddv
Values
for
Each
Distance."
The
99th
percentile
was
calculated
as
the
midpoint
value
of
visibility
impact
in
ddv
between
the
3rd
and
4th
highest
values
in
the
table.
The
98th
percentile
was
taken
as
the
7th
highest
value
in
the
same
table.
The
98th
and
99th
percentile
values
were
recorded
for
50,
100,
and
200
km.
The
target
emission
rates
were
estimated
using
the
expression
above
and
rounded
to
the
lower
hundred
value.

For
example,
the
SO2
emission
rate
estimated
to
result
in
a
98th
percentile
0.5
ddv
impact
at
50
km
(
the
target
emission
rate)
for
an
EGU
in
the
Eastern
domain
(
Run
3)
is
calculated 

Target
=
5000
/
(
2
*
0.86
ddv)
Target
=
2900
(
rounded
from
2906)

Again,
with
notice
that
these
values
are
estimates
only,
the
table
of
values
is
attached:

Estimated
Emission
Rates
(
tpy)
to
Result
in
.5
ddv
­
Eastern
Domain
Source
Pollutant
Distance
(
km)
98th
percentile
99th
percentile
EGU
NOx
200
3500
2400
EGU
NOx
100
2500
1800
EGU
NOx
50
2000
1300
EGU
SO2
200
5100
3500
EGU
SO2
100
3500
2500
EGU
SO2
50
2900
1900
Ind.
Boiler
NOx
200
6200
4100
Ind.
Boiler
NOx
100
2600
1900
Ind.
Boiler
NOx
50
1300
900
Ind.
Boiler
SO2
200
6200
3500
Ind.
Boiler
SO2
100
3300
2200
Ind.
Boiler
SO2
50
1700
1200
Estimated
Emission
Rates
(
tpy)
to
Result
in
.5
ddv
­
Western
Domain
Source
Pollutant
Distance
(
km)
98th
percentile
99th
percentile
EGU
NOx
200
4200
2600
EGU
NOx
100
2200
1700
EGU
NOx
50
1400
1100
EGU
SO2
200
3300
2100
EGU
SO2
100
1800
1300
EGU
SO2
50
1100
800
Ind.
Boiler
NOx
200
4100
2700
Ind.
Boiler
NOx
100
1900
1300
Ind.
Boiler
NOx
50
700
600
Ind.
Boiler
SO2
200
6200
4100
Ind.
Boiler
SO2
100
3300
2000
Ind.
Boiler
SO2
50
1300
900
Interpretation
and
Conclusions:
The
emission
rates
in
the
tables
above
may
be
interpreted
as
emission
limits
that
would,
at
those
limits,
result
in
a
visibility
impact
of
0.5
ddv
for
the
emission,
source,
and
CALPUFF
model
parameters
associated
with
the
model
simulations
from
which
the
target
emission
rate,
and
therefore
the
emission
limit,
had
been
estimated.
We
may
reasonably
interpret
any
emission
rate
less
than
a
particular
emission
limit
as
resulting
in
a
visibility
impact
less
than
0.5
ddv
for
the
emission,
source,
and
CALPUFF
model
parameters
associated
with
the
model
simulations
from
which
that
same
target
emission
rate,
and
therefore
the
emission
limit,
had
been
estimated.
However,
we
cannot
as
reasonably
interpret
an
emission
rate
greater
than
the
respective
emission
limit
to
result
in
a
visibility
impact
equal
to
or
greater
than
0.5ddv.
This
discontinuity
is
due
to
the
margin
of
safety
in
the
CALPUFF
simulations
and
the
emission
limit
estimation
method.
The
safety
factors
inherent
in
both
the
CALPUFF
model
simulations
and
the
emission
limit
estimating
method
create
a
margin
of
safety
in
the
calculations
that
gives
us
high
confidence
that
we
may
establish
an
effective
emission
limit
as
long
as
that
limit
is
less
than
the
estimated
emission
limit
determined
here.
Several
factors
add
to
a
cumulative
margin
of
safety
in
the
calculations:
the
chemistry
processing
and
dispersion
calculation
within
CALPUFF
tends
to
overpredict
impacts
in
the
type
of
simulations
we
performed;
the
estimation
of
the
linear
relationship
among
CALPUFF
Runs
1­
6,
which
was
used
for
estimating
emission
limits
for
the
EGU
runs,
is
generalized
and
rounds
to
the
higher
impact;
the
prototypical
input
parameters
for
the
illustrative
CALPUFF
simulations
were
rounded
to
greater
values;
the
use
of
ddv,
which
is
the
product
of
a
non­
linear
calculation,
as
the
operand
in
the
estimation
method
carries
full
weight
of
natural
background
light
extinction,
and
therefore
results
in
slightly
higher
estimates;
and
lastly,
the
results
of
the
estimation
were
rounded
to
the
lower
one
hundreds
value.
Considered
individually,
each
of
these
safety
margins
can
only
result
in
a
marginal
increase
in
confidence,
however,
their
sum
allows
us
a
high
level
of
confidence
that
any
emission
rate
less
than
the
emission
limit
estimated
by
this
method
would
not
result
in
a
visibility
impact
equal
to
or
greater
than
0.5
ddv.
We
have
used
the
information
in
this
memo
to
evaluate
the
reasonableness
of
establishing
presumptive
emissions
and
distance
factors
to
govern
the
need
for
further
source­
specific
modeling.
The
results
of
this
review
support
our
conclusion
that
it
would
be
reasonable
to
exempt
from
further
modeling
sources
with
the
following
combinations
of
emissions
and
distance
to
the
nearest
Class
I
area:
a)
emissions
of
SOx
at
a
rate
less
than
500
tons
per
year,
or
emissions
of
NOx
less
than
500
tons
per
year,
or
combined
emissions
of
both
pollutants
less
than
500
tons
per
year,
and
with
a
distance
from
the
nearest
Class
I
areas
50
km
or
greater;
b)
emissions
of
SOx
at
a
rate
less
than
1000
tons
per
year,
or
emissions
of
NOx
less
than
1000
tons
per
year,
or
combined
emissions
of
both
pollutants
less
than
1000
tons
per
year,
and
with
a
distance
from
the
nearest
Class
I
areas
100
km
or
greater.
For
the
same
given
distances,
the
presumptive
emission
limits
mentioned
in
this
paragraph
are
less
than
any
of
the
emission
limits
estimated
by
the
method
described
here.
Given
this
relationship,
and
given
the
added
confidence
of
the
margin
of
safety
in
our
calculations,
we
have
a
high
level
of
confidence
that
any
modeling
of
sources
with
emission
rates
less
than
the
presumptive
limits
for
the
same
distance
would
result
in
impacts
less
than
0.5ddv,
and
therefore
such
modeling
would
be
unnecessary
to
satisfy
our
certainty
that
no
impact
greater
than
0.5
ddv
would
occur.
Of
course,
our
confidence
remains
high
as
long
as
any
source
considered
against
the
presumptive
limits
is,
in
its
source
parameters,
its
geographic
scenario,
and
its
relationship
to
the
Class
I
areas,
reasonably
similar
to
the
scenarios
and
parameters
utilized
in
the
CALPUFF
simulations
upon
which
the
estimations
of
emission
limits
are
based.
Caution
is
strongly
suggested
in
applying
the
presumption
to
a
source
without
such
similarity,
a
source
located
in
a
unique
terrain
that
would
affect
advection
of
pollutant
to
a
Class
I
area,
for
example.
As
always,
conducting
a
source­
specific
CALPUFF
modeling
simulation
would
eliminate
any
question
in
such
a
case.
