November
30,
1998
Air
and
Radiation
Docket
and
 [
nfonnationCenter
(
6102)
Attention:
Docket
No.
A­
97­
43
for
the
section
126proposal
and
Docket
No.
A­
98­
12
for
the
FIP
proposal
U.
S.
Environmental
Protection
Agency
401
Nl
Street
SW,
room
M­
1580
Washington,
DC
20460
Dear
Sir/
Madame:

Attached
for
your
consideration
are
the
comments
of
the
Interstate
Natural
Gas
Association
of
America
(
INGAA)
on
EPA s
Federal
implementation
Plan
to
Reduce
the
Regional
Transport
of
Ozone
(
Docket
No.
A­
98­
12).
INGAA
is
pleased
to
provide
comments
and
is
confident
that
EPA
will
carefully
consider
the
many
important
elements
we
have
raised.

We
understand
that
EPA
is
preparing
technical
amendments
to
the
OTAG
SIP
final
rule
and
believe
that
many
of
the
points
raised
in
our
comnnents
here
are
directly
related
to
some
of
those
amendments.
As
such,
INGAA
is
looking
forward
to
meeting
with
EPA
staff
to
discuss
issues
related
to
both
the
OTAG
SIP
CallInventory
and
the
FIP
proposal
in
detail,
preferably
prior
to
issuance
of
the
technical
amendments.

If
you
have
any
questions,
please
feel
free
to
cantact
me
at
202­
2165935

Sincerely,

Lisa
S.
Beal
Director,
Environmental
Affairs
Enclosure/

CC:
D.
Doniger,
EPA
J.
Seitz,
EPA
INGAA
EH&
S
Committee
INTERSTATENATURAL
GAS
ASSOCIATION
OFAMERICA
10
G
STREET,
N.
E.,
SUITE:
700
WASHINGTON,
D.
C.
20002
202/
216­
5900FAX
202/
216­
0877
1
November
30,1998
Air
and
Radiation
Docket
and
Information
Center
(
6102)
Attention:
Docket
No.
A­
97­
43for
the
section
126proposal
and
Docket
No.
A­
98­
12for
the
FIP
proposal
U.
S.
E;
nvironmental
Protection
Agency
401M
Street
SW,
room
M­
1500
Washington,
DC
20460
To
whom
it
may
concern:

The
Interstate
Natural
Gas
Association
of
America
(
INGAA)
is
a
non­
profit
trade
association
representing
virtually
all
interstate
natural
gas
transmission
pipeline
companies
operating
in
the
United
States
and
interprovincial
pipelines
operating
in
Canada,
as
well
as
natural
gas
companies
in
Mexico
and
Europe.
INGMs
U.
S.
members
operate
over
200,000
miles
of
pipeline
and
related
facilities
and
account
for
over
'
90
percent
of
all
natural
gas
transported
and
sold
in
interstate
commerce.

INGM
appreciates
the
opportunity
to
comment
on
the
proposed
Federal
Implementation
Plans
(
FIPS)
which
may
be
needed
if
any
State
fails
to
revise
its
State
Implementation
Plan
(
SIP)
to
comply
with
the
nitrogen
oxides
(
NO)
SIP
call
recently
promulgated
by
E;
PA
(
63FR57356).
In
general,
INGAA's
comments
are
related
to
the
inconsistent
nature
of
the
proposed
FIP
Call
compared
with
that
of
the
finalized
SIP
call,
the
proposal
of
a
control
efficiency
that
is
unachievable,
and
monitoring
requirements.
Specifically,
INGAA
has
the
following
concerns:

0
EPA
should
clearly
establish
a
distinction
between
small
and
large
units
that
is
dependent
on
the
cutoff
level
specified
in
the
fmal
SIP
rule
of
first,

2
INTERSTATE
NATURAL
GAS
ASSOCIATION
OF
AMERICA
10
G
STREET,
N.
E.,
SUITE
700
WASHINGTON,
D.
C.
20002
202/
216­
5900
FAX
2OW216­
0877
250
MMBtu
per
hour,
and
second,
in
the
absence
of
heat
input
data,
greater
than
1
ton
per
day.

EPA
has
failed
to
use
ia
consistent
methodology
to
designate
 
large 
units
in
the
non­
EGU
inventory.

EPA
ignored
information
in
the
ACT
document
that
indicates
SCRis
not
a
reliable
control
technique
to
achieve
90
percent
reduction
of
NOx
fi­
om
lean­
burn
engines
in
variable
load
applications.

EPA s
control
cost
effectiveness
and
economic
impact
analysis
are
incorrect
The
proposed
defmitiom
of
rich
burn­
engineis
inconsistent
with
existing
state
regulations
in
the
22
eastern
states
Grams
Per
Horsepower
are
a
more
appropriate
expression
of
emissions
limitations
than
Parts
Per
Million
The
use
of
CEMS
or
PEMS
is
unnecessary
Specific
Comments
EPA
should
clearly
establish
a
distinction
between
small
and
large
units
lhat
is
dependent
on
the
cutoff
level
specified
in
the
finalSIPrulemaking
of
first,
250
MMBtu
per
hour,
and
in
the
absence
of
heat
input
data,
greater
than
1
ton
per
clay
second.

INGAA
is
very
disturbed
bywhat
appears
to
be
EPA s
inconsistent
interpretation
of
the
Non­
EGU
source
category
size
cutoff
determination
methodology.

 
When
considering
size
cutoffs
for
non­
EGU
units,
the
finalOTAG
recommendations
were
at
result
of
careful
consideration
of
a
number
of
factors
that
finally
led
to
a
designation
criterion
based
upon
horsepower.

In
the
finalSIP
rulemaking,
EPA
recognized
a
need
to
continue
to
make
a
distinction
between
large
and
medium
EGU­
units
when
considering
inclusion
of
these
units
for
emissions
control
and
promulgated
a
250
MMBtu
per
hour
3
and
as
an
appropriate
cutoff
level
for
large
non­
EGU
point
sources.
In
the
fmal
SIP
rulemaking,
EPA
stated
that
it
relied
upon
a
capacity
approach
first
and
where
capacity
data
was
not
available,
a
tons
per
day
approach
second.
Under
this
methodology,
if
specific
heat
input
capacity
data
were
not
available
in
the
updated
inventory,
emissions
reductions
should
have
been
assumed
only
if
the
units
average
summer
day
emissions
were
greater
than
one
ton
per
day.
EPA
clearly
intended
to
give
sources
an
opportunity
to
provided
the
necessary
capacity
data
because
in
the
SIP
call
finalrule,
the
agency
asked
for
additional
comment
that
would
speclfically
address
the
emissions
inventory.

hi
response
to
the
SIP
prciposal,
INGAA
has
already
submitted
a
summary
of
the
draft
results
of
cumulative
information
from
the
natural
gas
industry
that
included
detailed
information
about
reciprocating
engines
and
combustion
turbines
in
the
OTAG
region.
In
response
to
EPA's
most
recent
request,
many
of
INGAA's
individual
members
are
gathering
even
more
detailed
and
unit
specific
information
to
update
the
emissions
inventory
with
specifically,
the
capacity
data
needed
to
niake
the
250MMBtu
determination.

I'NGAA
is
very
concerned
that
EPA's
language
in
the
FIP
call
omits
all
reference
to
what
should
be
the
primary
indication
of
size
cutoff
and
corresponding
emissions
control
consideration
­
250
MMBtu
per
hour.
In
fact,
the
only
place
were
one
can
find
any
reference
to
the
250MMBtu
cutoff
is
for
boilers
included
in
the
trading
program,
not
stationary
combustion
engines
that
have
been
excluded
from
the
trading
program.

EPA
has
clearly
deviated
from
what
was
presented
in
the
finalSIP
call
and
in
the
FIP
proposal,
is
attempting
to
establish
a
significantly
more
stringent
and
restrictive
FIP.
INGAA
strongly
protest
to
this
action
and
urges
EPA
to
maintain
the
regulatory
structure
it
established
in
the
final
SIP
rule.

4
EPA
has
failed
to
use
a
consistent
methodology
to
designate
"
large"
units
in
the
non­
EGU
inventory
In
the
FIP
proposal,
EPA
claims
to
have
used
the
same
methodology
as
that
used
in
the
SIP
call
to
designate
"
large"
non­
EGU
units.
The
SIP
call,
63
FR
57356
(
Tuesday,
October
27,
19981,
at
57416,
explicitly
set
forth
the
five­
step
process
to
be
followed
to
designate
Large
sources
from
Small
sources.
Under
the
EtPA'salgorithm,
heat
capacity
is
first
detennined
either
by
inventory
data
(
step
1)
or
a
default
procedure
(
step
2).
Sources,
withheat
inputs
of
greater
than
250
MMBtu/
hr
are
desibwated
as
Large
sources
and
have
assumed
emission
reductions
(
step
3).
Emission
reductions
based
upon
tonnage
per
day
are
assumed
onlv
if
heat
capacitv
data
are
unavailable
bv
inventorv
or
default
­
data
(
step
4).
All
other
sources
are
considered
Small
(
step
5).
Under
this
algorithm,
all
sources
with
kniown
heat
inputs
of
250
MMBtu/
hr
or
less
(
either
by
inventory
or
default
data)
are
to
be
considered
small.
EPA
now
indicates
that
lhis
explicit
algorithm
is
not
be
followed.
Now
in
the
FIP
EPA
says
that
dividing
line
between
Large
and
Small
non­
EGU
sources
is
to
be
2400
horsepower.
But
the
published
inventory
is
not
consistent
with
this
interpretation
nor
the
250
MMBtu/
hour
or
even
the
one
tonlday.
In
the
final
inventory,
1,887units
in
the
inon­
EGU
inventory
with
no
capacity
information
are
designated
as
"
large"
even.
though
the
unit's
average
summer
day
emissions
are
less
than
one
ton
per
day.

The
proposed
2400
horsepower
cutoff
for
reciprocating
internal
combustion
engines
is
inconsistent
with
OTAG
recommended
horsepower
cutoffs
and
inconsistent
with
the
size
cutoffs
used
by
EPA
in
calculation
of
the
SIP
Call
budgets.
For
non­
utility
point
sources,
OTAG
recommended
that
"
Large"
sources
include
reciprocating
internal
combustion
engines
equal
to
or
larger
than
8,000horsepower.
In
the
final
SIP
Call
inventory,
EPA
indicated
that
the
Agency
chose
to
use
capacity
indicators
primarily
to
set
cutoff
levels.
If
the
source
was
greater
than
250
IMMBtu,
additional
emission
reductions
were
included
to
calculate
the
state
NOx
budgets.
Also,
the
proposed
2400
horsepower
cutoff
far
reciprocating
internal
combustion
engines
is
inconsistent
with
the
OTAG
recommended
emissions
cutoff
for
"
Large"
sources.
For
non5
1
utility
point
sources,
OTAG
recommended
that
"
Large"
sources
include
sources
with
2
tons
NOx
emissions
per
average
summer
day
or
more.
The
2400
horsepower
cutoff
is
based
on
E;
PA's
assumption
that
sources
.
with
NOx
emissions
of
greater
than
or
equal
to
one
ton
per
day
should
be
controlled.

While
ECPA
indicated
that
an
applicability
threshold
of
2400
hp
was
selected
because
engines
of
that
size
have
the
potential
to
emit
at
least
1
tone
per
day
of
NOx,
the
applicability
criterion
(
doesnot
recognize
the
fact
that
not
all
engines
2
2400
hp
emit
at
least
1
ton
per
day
of
NOx.
Many
2400
hp
engines
emit
less
than
1
ton
per
day
of
NOx.
The
applicability
threshold
of
2400
hp
was
selected
based
on
EPA's
use
of
emission
factors
for
lean­
bum
and
rich­
bum
engines
found
in
the
ACT
document,
16.8
and
15.8g/
hp­
hr
respectively.
If
EPA
intends
to
regulate
only
"
Large"
reciprocating
internal
combustion
engines,
EPA
should
revise
the
applicability
criterion
to
be
consistent
with
the
state
budgets
and
EPA
actions
in
the
SIP
Call
rulemaloing.

The
applicabilitycriterion
does
not
recognize
units
that
have
already
been
controlled
forNOx
The
proposed
2400
horsepower
cutoffwould
cause
additional
NOx
controls
for
many
more
reciprocating
engines
than
those
designated
"
Large"
in
the
SIP
Call
budgets.
As
proposed,
the
FIP
would
impose
emission
limitations
and
emissions
monitoring
requirements
on
allengines
equal
to
or
greater
than
2400
hp
within
the
FIP
area,
regardless
of
the
unit's
emissions.
This
would
result
in
the
imposiition
of
emission
limitations
and
emissions
monitoring
requirements
on
engines
that
were
designated
"
Small"
in
the
state
budgets
and
designated
"
Small"
using
the
methodology
described
by
EPA
in
the
SIP
Call
rulemaking.

The
purpose
of
the
FIP
should
be
to
implement
a
program
of
control
that
would
meet
the
promulgated
NOx
budgets,
not
to
expand
the
program
of
control
to
other
sources.

EPA
must
apply
a
consistent
methodology
for
determining
the
size
of
a
unit.

INGAA
recommends
that
EPA
consistently
apply
the
size
designation
methodology
used
in
the
SIP
cadl
final
rule.

6
EPA
states
that
the
FIP
proposal
is
intended
to
achieve
the
NOp
emissions
reductiions
required
by
the
NOx
SIP
call
rulemaking
and
that
rulemaking
docket
cOnt0.
s
information
and
analyses
that
are
relied
upon
in
the
NOx
FIP
proposal.
gaough
EPA
has
included
by
reference
the
entire
NOx
SIP
call
docket,
the
proposed
rule
States
that
the
Ody
portions
that
form
the
basis
for
the
FIP
mlemalking
are
those
that
address
feasibility
and
cost
effectiveness
of
control
measures
and
the
projection
of
emissions
reductions
that
various
control
measures
would
achieve.
Nevertheless,
EPA s
proposed
FIP
rule
goes
beyond
these
meas
because
it
does
not
apply
the
same
methodology
described
in
the
SIP
rulemaking
to
designating
 
large 
non­
EGU
Units
in
the
final
inventory.

EPAignored
information
in
the
ACT
document
that
indicates
SCR
is
not
a
reliable:
control
technique
to
achieve
90
percent
reduction
of
NOx
from 
lean­
burn
engines
in
variable
load
applications.

AccoFding
to
the
Technical
Suppiort
Document
(
TSD)
for
Stationary
Internal
Combustion
Engines
(
VI­
B13
IEPA
adopted
SCR
and
90
percent
reduction
as
the
control
technique
for
lean­
bum
engines
based
on
information
provided
in
Table
2­
5of
the
Alternative
Conlxol
Techniques
Document
­­
NOx
Emissions
from
Stationan
Reciprocating
Internal
Combustion
Engines
(
EPA­
453/
R­

93032
(
ACTdocument):

The
control
level
for
spark
ignited
lean­
bum
engines
that
meets
the
$
2,00O/
ton
criteria
above,
is
a
limit
of
125
ppmv
NOx
at
15%
02.
This
represents
selective
catalytic
reduction
(
SCR)
control.
SCR
provides
the
greatest
NOx
reduction
of
all
technologies
considered
in
the
ACT
document
for
lean­
bum
engines
and
is
capable
of
providing
a
90
percent
reduction
in
NOx
emissions.
This
emission
limitation
corresponds
to
the
 
Expected
controlled
NOx
emission
levels 
(
SCR)
from
Table
2­
5
of
the
ACT
document.

Table
2­
5
does
indicate
a
90
percent
reduction
in
NOx
emissions
for
SCR
and
a
footnote
indicates
this
percent
reduction
is
the
 
guaranteed
NOx
reduction
available
from
most
catalyst
vendors. 
Section
5.2.4.4
of
the
ACT
states
that
­­
I
emlssion
reductions
of
80
to
90+
percent
are
possible
using
SCR
on
lean­
bum
engines
that
are
used
in
baseload
applications:

Based
on
the
available
information
and
the
emission
test
data
presented
in
'
Tables
5­
8
and
A­
5,
it
is
estimated
that
the
achievable
NOx
emission
reduction
for
SCR
in
gas­
fired
applications
is
80
to
90+
percent
for
baseload
applications,
with
an
NH,
slip
level
of
10
ppmv
or
less.

However,
Section
5.2.4.4
of
,
theACT
states
that
inadequate
information
was
available
to
determine
achievable
continuous
NOx
reductions
and
ammonia
slip
levels
for
SCR
in
variable
load
applications:

The
available
data
are
not
sufficient
to
assess
the
achievable
continuous
NOx
reductions
and
ammonia
slip
levels
for
SCR
used
invariable
load
applications.

Section
2.2.2.4
of
the
ACT
notes
that
"
there
is
also
little
experience
with
SCR
in
variable
load
applications
due
to
ammonia
injection
control
limitations."
In
addition,
Section
5.2.4.4
states
that
variable
load
applications
may
pose
problems
for
the
SCR
system:

The
duty
cycle
of
the
engine
should
also
be
considered
in
determining
the
applicability
of
SCR.
Exhaust
temperature
and
NO,
emission
levels
depend
lipon
engine
power
output,
and
variable
load
applications
may
cause
exhaust
temperature
and
NO,
concentration
swings
that
pose
problems
for
the
SCR
system.
The
lower
exhaust
temperature
at
reduced
power
output
may
result
in
a
reduced
NO,
reduction
efficiency
fkom
the
catalyst.

When
the
additional
informattion
presented
in
the
ACT
document
is
considered,
along
with'infomation
presented
in
the
comments
provided
below,
it
is
clear
that.
EPA
should
not
rely
on
SCR
as
the
NOx
control
technique
for
lean­
burn
engines
in
variable
load
appllications.
SCR
is
not
feasible
for
variable
load
8
applications
and
a
90
percent
reduction
of
NOx
is
not
achievable
using
SCR
on
engines
in
variable
load
applications.

SCRis
not
feasible
for
variable
load
applications,
such
as
natural
gas
compression.

The
natural
gas
transmission
industry
has
evaluated
NOx
emissions
controls
for
hundreds
of
engines
under
NOx
RACT
and
found
that
SCR
is
not
a
reliable
control
technique
for
engines
in
natural
gas
compression
service
(
avariable
load
application).
State
regulators
have
agreed
­­
in
no
case
is
SCR
required
as
NOlx
MCT
for
engines
in
trimsmission
service.
Instead,
NQx
emissions
reductions
have
been
achieved
using
low
emissions
combustion
(
LEC),

cornbustion
modifications
or
parametric
controls
(
PC).
These
technologies
are
proven
for
natural
gas­
fired
lean­
burn
engines,
while
SCR
is
not.

Reciprocating
internal
combustion
engines
in
natural
gas
compression
service
are
used
to
compress
and
move
natural
gas
along
the
pipeline.
An
engine
must
respond
to
pipeline
conditions,
supply,
and
demand
for
natural
gas.
As
a
result,
engines
are
frequently
required
to
change
load
and
speed
conditions.
This
variable
load
operation
complicates
the
successful
operation
of
an
SCR
system
given:
1)
variations
in
NOx
emissions,
2)
variations
in
exhaust
gas
flow
andl
temperature,
and
3)
thermal
cycling.

.1.
Variations
in
NQx
Emissions.
In
order
for
SCRto
work
properly,
with
the
appropriate
control
of
NOx
emissions
and
without
excessive
ammonia
emissions,
it
is
necessary
that
the
proportions
of
NOx
and
ammonia
be
correct.
Changes
in
load
and
speed
produce
changes
in
the
NOx
emissions
from
the
engines.
For
example,
a
load
change
of
only
five
percent
can
significantly
change
the
NOx
emitted
fi­
om
an
engine.
When
these
changes
occur,
the
ammonia
feed
system
must
be
adjusted
to
maintain
the
proper
ratio
of
ammonia
to
NOx.
In
Section
5.2.4.2,
the
ACT
notes
that
variable
load
can
cause
NOx
concentration
swings
that
pose
problems
for
the
SCR
system.
SCR
systems
traditionally
use
a
feedback
system
that
relies
on
a
NOx
or
ammonia
analyzer
to
adjust
the
9
ammonia
feed
system.
It
takes
approximately
five
minutes
time
to
extract
a
sample
and
report
NOx
emission
levels
using
a
NOx
CEMs.
A
recent
installation
of
SCR
at
the
Buckeye
Pipeline
station
relies
on
a
predictive
emissions
model
(
PEMs)
as
the
feed­
forward
control
for
ammonia
feed.
The
engines
are
not
equipped
with
CEMs.
INGAA
understands
that
EPA
and
the
South
Coast
Air
Management
District
are
reviewing
this
installation
to
determine
if
it
is
cost­
effective.
At
the
Buckeye
station,
the
engines
were
purchased
new
and
the
manufacturer
designed
the
PEMs
to
control
the
ammonia
feed
system.
In
addition,
the
engines
are
used
to
drive
liquid
pumps
and
the
variations
in
load
and
speed
for
the
engines
are
limited
by
the
range
of
operation
of
the
pumps.
In
contrast,
engines
on
a
natural
gas
pipeline
must
respond
to
the
conditions
of
the
natural
gas
within
the
pipeline
­­
a
much
wider
range
of
operation
than
is
required
for
liquid
pumps
like
those
at
the
Buckeye
station.
There
are
no
known
applications
where
PEMs
have
been
successfully
used
as
feed­
forward
controls
for
existing
reciprocating
internal
combustion
engines.

­
Variations
in
Exhaust
Gas
Flow
and
Temperature
2.
Variable
load
conditions
also
result
in
variations
in
exhaust
gas
flow
and
temperature
conditions.
The
ammonia
control
system
would
need
to
be
adjusted
to
respond
to
'
the
changes
in
the
exhaust
mass
flow
rate.
Variable
exhaust
gas
flow
rates
would
also
affect
mixing
in
the
catalyst.
The
wider
the
range
in
exhaust
gas
flow
rates,
the
more
difficult
it
is
to
design
a
catalyst
that
can
reliably
achieve
high
emission
reductions
(
e.
g.,

90
percent).
Reduced
engine
load
can
also
decrease
the
exhaust
gas
temperature
below
the
range
for
optimal
NOx
conversion,
which
would
not
allow
the
catalyst
to1
reliably
achieve
high
emission
reductions.

Section
5.2.4.2
of
the
ACT
states
that
"
lower
exhaust
temperature
at
reduced
power
output
may
result
in
a
reduced
NOx
reduction
efficiency
from
the
catalyst."

10
­
Thermal
Cvcling
3.

'
Variable
load
conditions
also
result
in
thermal
cycling
of
the
catalyst
­

i2S
the
engine
exhaust
temperature
changes.
This
thermal
cycling
leads
to
reduced
catalyst
life.

The
complications
introduced
Qy
variable
load
operation
have
not
been
sufficiently
resolved
to
ensure
reliable
operation
of
SCR
on
engines
in
variable
load
applications,
such
as
natural
gas
compression.
EPA
concluded
that
SCR
was
a
feasible
control
technology
(
thatachieves
90
percent
reduction
of
NOx)

based
am
a
limited
review
of
infarmation
presented
in
the
ACT
document.

SCRwill
not
reliably
achieve
90
percent
reduction
of
NOx
for
variable
load
applications.

The
TSI)
indicates
that
EPA
adopted
SCR
and
90
percent
reduction
based
on
vendor
quotes
presented
in
Table
2­
5
of
the
ACT
document.
However,
as
presented
in
the
comments
above,
EPA
ignored
other
information
in
the
ACT
document
that
indicates
SCR
is
not
a
reliable
control
technology
to
achieve
90
percent
reduction
of
NOx
for
variable
load
applications.

The
available
data
are
not
sufficient
to
assess
the
achievable
continuous
NOx
reductions
anid
ammonia
slip
levels
for
SCR
used
in
variable:
load
applications.

Variable
load
applications
present
unique
difficulties
for
SCR
that
have
not
yet
been
adequately
addressed
to
allow
widespread
application
of
SCR
for
engines
in
load­
following
applications,
such
as
natural
gas
compression..

In
the
ACT
document,
emission
test
data
for
six
engines
with
SCR
are
presented.
Results
for
over
30
emission
tests
for
the
six
small
engines
are
provided
in
the
ACT
document.
All
the
engines
tested
are
smaller
than
the
proposed
size
cutoff
of
2400
horsepower
(
hp)
proposed
by
EPA
in
the
FIP.
The
11
emissions
data
suggest
that
the
percent
NOx
reduction
achieved
with
SCR
varies.
Although
the
variability
is
not
discussed
in
the
ACT
document,
it
is
likelly
that
the
variability
is
dlue
to
changes
in
the
engine's
operation.
The
range
of
effectiveness
for
the
six
engines
is
presented
in
the
table
below.

Engin
Manufacturer
Model
Horsepower
Number
of
Range
of
NOx
Control
e
No.
Tests
Reported
(
YO)
45
Clark
HRA­
6
660
5
84­
91
47
Clark
HRA­
6
660
5
82­
88
139
Cooper
GMV
660
1*
50
Bessemer
248
Cooper
GMV­
8
800
9*
87­
93
Bessemer
309
Clark
HRA­
32
350
12
65­
84
357
Tecogen
CM­
200
29
1
2
95­
97
The
only
engine
that
achieved
levels
greater
than
90
percent
for
alltests
conducted
was
a
small
engine
designed
for
cogeneration
(
the
Tecogen).
The
Tecogen
would
be
installed
in
a
baseload
application
­­
not
a
load­
following
application.
Two
other
engines
reported
levels
at
or
above
90
percent
for
at
least
one
test,
but
levels
less
than
90
percent
for
other
tests.
The
other
engines
did
riot
include
any
test
results
at
or
above
90
percent
NOx
reduction.

EPA
adopted
SCR
and
90
percent
reduction
based
on
limited
information
presented
in
Table
2­
5
of
the
ACT.
However,
information
priesented
in
Section
5.2.4.4
of
the
ACT
indicates
that
EPA
was
not
able
to
determine
an
achievable
NOx
reduction
level
for
the
use
of
SCR
in
load­
following
applications.
Therefore,
EPA
should
not
use
SCR
and
90
percent
reduction
for
lean­
burn
engines
in
the
FIP
or
in
the
state
budgets
for
the
SIP
Call.

12
SCRis
not
a
demonstrated
NOx
control
technology
for
reciprocating
internal
combustion
engines
in
variable
load
applications.

The
ACT
document
states
that
"
there
is
also
little
experience
with
SCRin
variable
load
applications
due
to
ammonia
injection
control
limitations."
As
indicated
above,
the
natural
g,
astransmission
industry
has
evaluated
NOx
emissions
controls
for
hundrelds
of
engines
under
NOx
RACTand
found
that
SCR
is
not
a
reliable
control
technique
for
engines
in
natural
gas
compression
service
(
avariable
load
applicattion).

According
to
a
review
of
SCR
conducted
by
the
Gas
Research
Institute
(
GRI)
(
attadhed
to
these
comments
ais
Appendix
A),
SCRhas
been
kstalled
on
reciprocating
engines
in
very
few
instances.
No
natural
gas­
fced
lean­
bum
engines
were
identifled
by
GRI
in
the
EPA
RACT/
BACT/
LAERclearinghouse
with
SCR
controls
(
based
on
a
query
of
post­
1991
determinations
for
natural
gas­
fu­
ed
internal
combustion
­­
15.004).
RACT/
BACT/
LAER.
deterrninations
for
lean­
burn
engines
relied
on
low
emissions
combustion
(
LEC)
technology,
other
combustion
modification.(
such
as
high
energy
ignition
systems),
or
parametric
controls
(
such
as
retarded
timing
and
airto
fuel
ratio
adjustment).

A
total
of
18
diesel
engines
with
SCR
controls
(
PA­
0096
and
PA­
0097)
were
identitied
in
the
clearinghouse.
The
SCR
controls
for
the
diesel
engines
reduce
NOx
emissions
by
80
percent.

The
GRI
study
also
cites
the
fact
that
results
from
a
1996
survey
of
North
American
interstate
natural
gas
transmission
companies
indicate
that
of
the
599
lean­
burn
engines
with
NOx
controls,
only
two
engines
inthe
United
States
have
SCRinstalled.
Those
twa
engines
since
have
been
mothballed
due
to
difficulties
operating
the
SCRunits
in
a
load­
following
application.
NOx
emissions
from
lean­
bum
engines
in
pipeline
service
have
been
controlled
using
combustion
modifications
or
ptmametriccontrols
because
of
the
problems
associated
with
using
SCRin
the
variable
load
application.

The
information
presented
above
indicates
that
SCRis
not
demonstrated
in
practice
for
load­
following
applications.
The
information
presented
in
the
ACT
13
document
is
not
a
suMcient
basis
for
EPA
to
conclude
that
SCR
is
feasible
(
with
90
percent
reduction)
for
all
existing
engines
that
would
be
subject
to
control
requirements
under
the
FIP
proposal
or
the
SIP
Call.
EPA
should
not
base
its
FIP
proposal
or
state
budgets
for
the
FIP
and
SIP
Call
on
the
use
of
SCR
for
lea&
burn
engines.

EPA's
cost­
effectivenessanalysis
for
the
use
of
SCR
relies
on
90
percent
reduction
of
NOx,
which
is
not
feasible
for
load­
following
applications.

The
TSDindicates
that
EPA
relied
on
Figure
2­
6
of
the
ACT
to
evaluate
the
cost­
effectiveness
of
SCR:

As
illustrated
in
Figure
2­
6
of
the
ACT
.
..
.
The
cost­
effectiveness
is
about
$
800/
ton
for
a
2200
hp
engine
operated
8,000
hours
per
year.
Therefore,

SCR
meets
the
criteria
of'
less
than
$
2,00O/
ton
of
NOx
rteduction.

The
cost­
effectivenessresults
presented
in
the
ACT
document
rely
on
90
percent
reduction
of
NOx.
As
indicated
in
the
comments
above,
SCR
has
not
been
shown
to
reliably
reduce
NOx
emissions
by
90
percent
in
load­
following
appllications.
The
ACT
document
states
that
although
90
percent
is
used
for
the
cost­
effectiveness
calculations
in
that
document,
EPA
had
insufficient
infoinnation
to
determine
"
achievable
continuous
NOx
reductions
and
ammonia
slip
levels
for
SCR
used
in
variable
load
applications."
Variable
load
operation
pres,
entsunique
difficulties,
including
variations
in
NOx
emissions,
variations
in
exhaust
gas
flow
and
temperature,
and
thermal
cycling,
that
have
not
been
adequately
addressed
in
the
EPA
cost
analysis.

Therefore,
EPA
should
not
conclude
that
SCR
is
cost­
effective
for
all
engines
that
would
be
subject
to
the
FIP
or
SIP
Call,
since
clearly
EPA
has
not
established
the
effectiveness
or
feasibility
of
SCR
for
load­
following
applications.
EPA
should
work
with
industry
stakeholders
to
assess
the
cost­
effectiveness
of
NOx
control
techniques
for
reciprocating
internal
combustion
engines.

14
Projection
of
Emission
Reductions
that
Various
Control
Measures
Would
Achieve.

EP.
A
calculated
the
state
budgets
for
the
proposed
FIP
and
the
SIP
Call
based
on
the
Agency's
assessment
that
SCR
can
achieve
90
percent
NOx
reduction
from
lean­
burn
reciprocating
internal
combustion
engines.
As
indicated
in
the
comments
above,
EPA
concluded
in
the
ACT
document
that,
while
SCR
could
achieve
80
to
90
percent
NOx
reductions
for
engines
in
baseload
applications,
there
was
insufficient
information
to
assess
achievable
NOx
reductions
for
SCR
on
engines
in
variable
load
applications.
90
percent
reduction
of
NOx
using
SCR
is
not
proven
in
practice
for
lean­
burn
reciprocating
engines
in
load­

following
applications.

EPA
should
reevaluate
the
emission
reductions
for
lean­
burn
engines
based
on
NOx
controls
that
are
proven
and
are
shown
to
be
cost­
effective
for
lean­
burn
reciprocating
internal
combustion
engines.

The
emission
levels
for
lean­
bum
engines
cannot
be
achieved
at
a
cost
of
$
2,000
per
ton
of
NOx.
Based
on
the
ACT
document,
EPA
incorrectly
concluded
that
SCR
for
lean­
burn
engines
could
be
implemented
on
a
2200
hp
lean­
bum
engine
for
about
$
800per
ton
of
NOx.
Industry
data
suggests
that
these
costs
are
not
correct.

While
INGAA
supports
the
30­
day
rolling
average,
EPA
should
specify
that
emission
limits
are
at
100%
speed
and
100940torque.

The
attached
GRI
(
Appendix:
A)
report
on
the
effectiveness
of
SCR
applied
to
pipeline
compressor
engines
also
concludes
that
SCR
is
infeasible
for
pipeline
compressor
engines.
The
report
documents
that
research
data
collected
on
internal
combustion
engines
utilizing
SCR
is
biased
towards
small
engines
in
baseload
applicationsrather
than
the
type
of
units.
The
industry
has
only
identified
two
compressor
unitsthat
have
been
tested
with
SCR
These
compressor
units
have
been
subsequently
been
removed
&
om
service
due
to
poor
pelrformance.
The
report
further
concludes,
as
did
the
original
ACT
document
15
that
SCR
is
not
designed
for
load
following
applications.
Finally
the
test
data
from
the
ACT
document
shows
that
the
average
performance
of
the
tested
engines
was
sigrniscantly
below
90%
emission
reduction.

Control
Cost
Effectiveness
and
Economic
Impact
Analysis
INGAA
has
examined
the
Regulatory
Impact
Analysis
for
the
NOx
SIP
Call,
FIP,
and
Section
126
Petitions,
tlie
Non­
Electricity
Generating
Unit
Economic
Impact
Analysis
for
the
NOx
SIP
Call
and
Ozone
Transport
Rulemaking
­
Non­
Electricity
Generating
Unit
Cost
Analysis.
These
documents
contain
cost
analysis
for
control
options
jor
Large
Stationary
IC
engines.
INGAA
strongly
believes
that
errors
in
the
basic
underling
assumptions
that
support
these
analysis
has
severely
distorted
the
conclusions
reach
regarding
the
control
cost
effectiveness
of
Large
Stationary
IC
engine
and
the
resultant
economic
impact
on
the
natural
gas
transmission
industry.

First,
the
emission
inventory
for
IC
engines
category
is
significantly
flawed
regarding
the
number
sources
and
their
corresponding
level
of
emissions.
This
directlv
affects
all
subsequent
economic
analvsis
using
this
inventory.

Second,
EPA
cost
analysis
far
all
control
options
depend
on
tons
of
NOx
removed
as
a
percentage
of
uncontrolled
emissions.
If
uncontrolled
emissions
are
significantly
overstated,
­
whichwe
believe
they
are
in
the
EPA
analysis,
then
cost
effectivenessis
likewise
equally
overstated.
As
stated
above
the
ACT
uses
16.8
g/
hp­
hr
while
AP­
42
is
1lg/
hp­
hr.
This
difference
alone
could
lead
to
und.
er
estimating
cost
by
50
percent.

Third,
cost
effectiveness
is
also
affected
by
percentage
of
source
utilization
during
the
ozone
season.
INGAA
members
operate
a
number
of
IC
engines
at
natural
gas
storage
fields.
Allthough
source
utilization
varies
widely
average
summer
time
values
for
IC
engines
with
potential
to
exceed
one
ton
a
day
are
typically
in
the
range
50
Yo
or
less.
Thus
cost
effectiveness
calculations
for
this
sector
could
be
underestimated
by
a
factor
of
two
from
those
used
in
EPA's
cost
analysis
which
was
91%.
This
fact
point
is
confirmed
by
EPA's
contractor
16
Pechan­
Avanti
Group
on
page
57
in
the
September
17,
1998report
entitled
Ozone
Transport
Rulemaking
Non­
Electricity
Generating
Unit
Cost
Analysis:
There
are
uncertainties
in
the
cost
per
ton
values
for
IC
engines
because
we
do
not
know
which
engines
are
lean
bum
versus
rich
bum
(
lean
bum
is
assumed),

nor
operating
practices
(
hours
of
operation
and
load).
AJl
of
these
factors
affect
control
cost.
Following
on
page
61:"
For
the
non­
trading
source
analysis,
the
cost
effectivenessof
controls
applied
to
IC
engines
are
a
key
factor
in
the
cost
analysis.

Finally
the
cost
for
annual
emissions
monitoring
are
also
significantly
underestimated.
In
a
study
recently
completed
for
the
natural
gas
industry,
the
annual
cost
for
CEM
on
a
single
IC
engines
is
estimated
as
$
107,000per
year.

Thisis
about
two
and
one
half
times
the
$
43,353
per
year
reported
in
the
Regulatory
Impact
Analysis
Report.

In
summary,
after
the
corrections
noted
above
are
made,
EPA's
conclusion
that
SCR
and
LEC
NOx
emission
controls
for
IC
engines
are
highly
cost
effective,
i.
e.
iunder
$
2000
per
ton,
is
in
error.

The
proposed
definition
ofrich­
burn
engine
is
inconsistent
with
existing
state
regulations
in
the
22
eastern
states.

Theproposed
FIP
defines
rich­
bum
engine
as
any
engine
with
a
lambda
of
1.1
or
less,
based
on
the
manufacturer's
original
air
to
fuel
ratio
compared
to
the
stoichiometric
air
to
fuel
ratio.
This
definition
has
been
implemented
only
in
a
few
air
pollution
control
districts
in
California.
The
Technical
Support
Document
(
VI­
B­
13)
indicates
that
the
California
FIP
rule
was
reviewed
by
EPA
to
develop
the
definitions
arid
control
requirements
for
reciprocating
internal
combustion
engines.
However,
the
California
FIP
has
no
regulatory
standing
because
the
rule
was
withdrawn
by
EPA
(
60
FR
43468­
43469),
including
the
regulations
issued
on
Febniary
14,
1995,
but
not
published
in
the
Federal
Register.
For
the
22
eastem
states,
a
more
appropriate
definition
of
rich­
burn
engine
is
'
I
a
four­
stroke,
spark­
ignited
engine
where
the
oxygen
content
in
the
ediaust
stream
before
any
dilution
is
1%
or
less
measured
on
a
dry
basis."
Definitions
based
on
1%
oxygen
are
currently
used
in
a
number
of
the
22
17
eastern
states.
None
of
the
22
eastern
states
use
the
California
defintion
of
rich­
burn
engine.

The
proposed
defmition
of
rich­
burn
engine
does
not
recognize
that
many
engines
that
were
originally
considered
"
rich­
burn"
have
been
modified
so
that
the
design
and
operating
characteristics
of
the
unit
are
akinto
"
lean­
burn"

engines.

These
changes
were
made
largely
to
comply
with
NOx
emission
reduction
requirements.
Under
the
proposed
FIP,
these
engines
would
wrongly
be
considered
"
rich­
burn"
since
the
original
manufacturer's
airto
fuel
ratio
would
be
used.

NSCR
would
not
be
achievalde
for
all
engines
in
the
rich­
burn
subcategory
if
this
definition
were
adopted.

The
proposed
definition
of
rich­
burn
engine
includes
engines
with
exhaust
oxygen
concentrations
greater
than
1
percent.
Although
some
engines
with
exhaust
oxygen
content
greater
than
1
percent
may
be
adjusted
to
run
at
levels
compatible
with
NSCR,
there
are
some
engines
that
cannot
be
adjusted
to
run
at
such
fuel­
rich
conditions.
Therefore,
the
proposed
definition
incorporates
some
engines
for
which
the
control
technology
will
be
infeasible
for
NOx
control.

A
lambda
of
1.1
corresponds
to
approximately
2
percent
oxygen
in
the
exhaust.

An
exhaust
concentration
of
2
percent
would
not
be
compatible
with
the
use
of
NSCR
to
control
NOx.
In
order
to
use
NSCR
as
intended
for
NOx
control,
it
would
be
necessary
to
adjust;
the
engine
to
run
at
1
percent
oxygen
or
less.

In
order
to
use
NSCR
as
intended
for
NOx
control,
it
would
be
necessary
to
adjust
the
engine
to
run
at
1percent
oxygen
or
less.
However,
for
some
older
models
of
engines,
commercjially
available
air­
to­
fuel
ratio
controllers
cannot
ensure
that
the
engines
will
operate
with
exhaust
concentrations
of
1
percent
oxygen
or
less,
at
all
load
conditions,
including
low
loads.

18
The
definition
of
rich­
bum
engine
should
not
rely
on
the
manufacturer's
recommended
air­
to­
fuel
ratio.

This
approach
does
not
reflect
the
fact
that
existing
engines
have
been
rem
mufactured
or
re­
constructed
to
convert
an
engine
from
a
rich
bum
to
a
lean
bum.
These
changes
result
in
different
air­
to­
fuel
ratios
khan
those
specified
by
the
design
of
the
engine
manufacturer.

The
definition
of
rich­
bum
engine
should
rely
on
exhaust
oxygen
content.

Lambda
is
more
difficult
to,
measure/
calculate
than
exhaust
oxygen
levels.
Two
air­
to­
fuel
ratios
are
necessary
to
determine
lambda:
the
stoichiometric
air­
tofuel
ratio
and
the
manufacturer's
recommended
air­
to­
fuel
ratio.

A
definition.
of
rich­
bum
engine
that
relies
on
1
percent
exhaust
oxygen
concentration
would
be
appropriate
for
the
FIP.

The
oxygen
content
of
the
exhaust
is
easy
to
measure
on­
site
and
determine
whether
an
engine
meets
the
criteria
of
1
percent
or
less.
The
definition
is
colnsistent
with
the
use
of
NSCR
for
NOx
control,
which
requires
that
oxygen
be
1
percent
or
less.
With
this
definition,
the
use
of
NSCR
for
NOx
control
would
be
achievable
for
all
"
rich­
burn"
engines.
The
definition
reflects
the
operating
conditions
of
the
engine,
not
simply
the
conditions
specitied
by
the
manufacturer.
Therefore,
the
definition
accommodates
diverse
operating
conditions
of
existing
engines,
which
may
result
in
an
exhaust
gas
oxygen
content
different
than
that
specified
by
the
manufacturer.
The
definition
also
takes
into
account
the
possible
re­
manufacture
or
re­
construction
of
existing
engines,
which
may
result
in
an
exhaust
gas
oxygen
content
different
than
that
specified
by
the
manufacturer.

Parts
Per
Million
versus
Grams
Per
Horsepower
Hour
IIVGAA
believes
that
the
expression
of
the
emission
limitations
in
terms
of
parts
per
million
(
ppm)
is
unnecessary
to
assure
compliance
with
seasonal
tons
of
emissions
and
is
inconsistent
with
the
seasonal
tonnage
budgets.
Tonnage
limits
would
be
more
appropriate,
especially
for
engines
and
turbines.
If
tonnage
is
not
19
acceptable,
then
grams
per
horsepower
hour
is
more
meaningful
in
the
case
of
internal
combustion
engines.

Emissions
Trading
Program
The:
Use
of
CEMS
or
PEMS
is
Unnecessary
While,
INGAA
supports
the
use
of
Parametric
Emissions
Monitoring
Systems
(
PEIMS)
for
sources
to
opt
into
an
emissions
trading
program,
the
proposed
use
of
CFR
Part
60
requirements
of
continuous
emissions
monitoring
(
CEMs)
or
PEMs
under
seasonal
tonnage
limits
is
overly
burdensome
to
assure
compliance
for
IC
engines.
Part
60
requires
relative
accuracy
(
RA)
of
20
percent.

At
the
proposed
control
levell
of
125
ppm,
this
would
mean
a
RA
of
25
ppm.
Tesiting
of
IC
engines
has
established
that
this
20
percent
FWactually
represents
7
percent
error
in
actual
accuracy
or
9
ppm.
Contrast
this
to
utility
boilers
where
NOx
concentrations
are
in
the
thousands
of
ppm.
For
these
sources
the
actual
errors
allowed
would
be
greater
than
100
ppm.

Also,
the
use
of
CEMs
or
PEMs
is
inappropriate
for
compliance
with
a
seasonal
tonnage
limit.
The
IC
source
category
accounts
for
only
a
small
ii­
actionof
the
total
SIP
Call
region
emissions.
Requiring
Part
60
level
of
accuracy
is
not
justified
to
meet
the
overall
budget
goals.

EPA
should
allow
the
use
of
the
same
monitoring
on
IC
engines
as
it
uses
in
its
Title
V
operating
permit
program
In
the
proposed
FIP,
EPA
indicated
that
the
emission
limitations
will
be
considered
applicable
requirements
and
must
be
incorporated
into
Title
V
operating
permits.
If
only
"
Large"
sources
are
regulated
by
the
FIP,
then
those
sources
would
be
considered
major
sources
under
Title
V
and
required
to
obtain
a
Title
V
permit.
Therefore,
appropriate
monitoring
may
be
determined
as
a
part
of
the
Title
V
permitting
process.
The
requirement
to
use
CEMs
in
the
proposed
FIP
is
inconsistent
with
the
flexibility
that
EPA
has
incorporated
in
the
Agency's
guidance
on
periodic
monitoring
and
compliance
assurance
monitoring
(
CAM).
The
FIP
should
preserve
the
flexibility
to
allow
sources
and
20
I'

Title
V
permitting
authorities
tc,
determine
the
most
appropriate
monitoring
to
ensure
compliance.

The
FIP
should
not
dictate
cont.
inuous
emissions
monitoring
for
reciprocating
internal
combustion
engines
that
do
not
participate
in
the
emissions
trading
program.
The
FIP
should
require
that
sources
and
Title
V
permitting
authorities
determine
the
most
appropriate
monitoring
to
ensure
compliance
when
the
emission
limitations
are
incorporated
into
the
source's
Title
V
permit.

In
the
final
SIP
Call
rulemaking,
EPA
did
not
require
continuous
emissions
monitoring
for
non­
EGU
units
ithat
would
not
participate
in
the
NOx
trading
program.
Rather,
51.121(
i)
only
required
continuous
emissions
monitoring
for
units
serving
electric
generators
with
a
nameplate
capacity
greater
than
25
MWe
or
boilers,
combustion
turbines,
or
combined
cycle
units
with
a
maximum
design
heat
input
greater
than
250
mmBtu/
hr.
The
FIP
should
include
the
same
flexibility
with
regards
to
monitoring
as
the
requirements
for
the
SIP
Call.

For
noln­
trading
sources,
the
piurpose
of
monitoring
is
to
demonstrate
compliance
with
a
NOx
emissicin
limitation.
EPA
has
issued
guidance
for
CAM
and
periodic
monitoring
that
alllow
sources
and
permitting
authorities
discretion
to
determine
the
most
appropriate
monitoring
requirements
to
ensure
compliancewith
applicable
requirements.
However,
EPA's
approach
under
the
FIP
is
inconsistent
with
the
Agency's
policy
that
compliance
can
be
determined
through
monitoring
other
than
continuous
emissions
monitoring.

EPA
n.
eedsto
clarifs
the
differences
between
98.4
compliance
determination
and
98.5(
b)
reporting
requirements.

In
98.
a4
of
the
proposed
FIP,
EPA
indicates
that
reciprocating
internal
combustion
engines
must
dem.
onstratecompliance
through
the
use
of
continuous
ehssions
monitors,
pursuant
to
40
CFR
60.
In
98.5(
b),
however,
EPA
includes
monitoring
requirements
that
indicate
a
source
must
operate
a
CEMs
which
meets
the
applicaible
requirements
of
40
CFR
part
60,
subpart
A
or
alternate
calculational
and
recordkeeping
procedures
based
upon
actual
21
emissions
testing
and
correlations
with
operating
parameters.
It
is
unclear
why
the
requirements
of
98.4
and
98.5
are
not
the
same.
In
the
find
rule,
EPA
should
incorporate
the
flexibility
included
in
98.5
into
the
compliance
determination
requirements
under
98.4.
In
addition,
EPA
should
incorporate
add.
itional
flexibility
that
would
be
consistent
with
the
principles
for
periodic
monitoring
and
CAM.
EPA
should
not
impose
additional
monitoring
requirements
under
98.5
beyond
those
requirements
included
in
98.4
to
determine
compliance.

Allocation
ofAllowances
INGAA
has
been
a
long
and
ardent
supporter
of
an
output
based
allocation
of
dl6wances
in
the
NOx
trading
system
for
large
stationary
sources
under
the
FIF.
INGAA
is
encouraged
by
EPA's
apparent
recognition
of
the
value
an
output­
based
system
may
bring
to
the
proposed
trading
program.
INGAA
is
an
active
member
of
the
Coalition
for
Gas
Based
Environmental
Solutions
which
is
submitting
comments
on
reLgarding
this
subject.
INGAA
fiilly
supports
and
concurs
with
the
comments
of
the
Coalition
which
have
been
attached
at
Appendix
B
INGAA
appreciates
the
opportunity
to
comment
on
this
proposal
and
looks
forward
to
working
with
EPA
in
the
future.
If
you
have
any
questions
or
need
clarification,
please
feel
free
to
contact
me
at
202
216­
5935.

Sincerely,

Lisa
S.
Bed
Director,
Environmental
Afbirs
22
Appendix
A
Gats
Research
Institute
Technical
Issues
Related
to
the
Potential
Use
of
Selective
Catalytic
Reduction
(
SCR)
to
Reduce
NOx
Emissions
from
Natural
Gas­
FiredLean
Bum
Engines
23
Technical
IssuesRelated
to
the
Potential
Use
of
Selective
Catalytic
Reduction
(
SCR)
to
Reduce
NOx
Emissions
frolm
Natural
Gas­
Fired
Lean
Burn
Engines
Prepared
by:
Coerr
Emironmental
Corporation
and
Radian
International,
LLC
Prepared
for:
Gas
Research
Institute
November
9,
1998
24
Table
ofContents
I.
Introduction
..........................................................................
1
I1.
Overview
of
SCR
Technologg..................................................
2
III.
Basis
for
EPA's
Concllusion
that
SCR
is
Effective
....................
3
Iv.
SCR
Installations
on
Lean­
BurnEngines
................................
7
Difficulties
Operating
SCR
Systems
on
Load­
FollowhgApplications9
Negative
Environmental
Impacts
from
SCR
................................
12
VII.
Conclusions
.........................................................................
13
25
I.
INTRODUCTION
On
October
27,
1998
(
63FR
57356),
EPA
promulgated
the
Agency's
rulemaking
to
call
for
22
eastern
states
and
the
District
of
Columbia
to
revise
their
State
Implementation
Plans
(
SIP)
to
address
the
regional
transport
of
ozone
(
the
SIP
Call).
EPA
also
proposed
a
Federal
Implementation
Plan
(
FIP)
that
would
be
implemented
if
any
state
failed
to
subm.
it
an
adequate
SIP
revision
in
response
to
the
EPA
SIP
Call
(
63FR
56394).

In
a
technical
support
document
for
both
rulemakings
(
A­
96­
56,
VI­
B­
13),

EPA
stated
that
emissions
of
nitrogen
oxides
(
NOx)
from
large
lean­
burn
reciprocating
internal
combustion
engines
can
be
reduced
by
90
percent,

cost­
effectively,
using
Selective
Catalytic
Reduction
(
SCR):

The
control
level
for
spark
ignited
lean
burn
engines
that
meets
the
$
2,00O/
ton
criteria
.
.
.
is
a
limit
of
125
ppmv
NOx
at
15%
0,.
This
represents
selective:
catalytic
reduction
(
SCR)
control
.
.
.
providing
a
90
percent
reduction
in
NOx
emissions.

The
Agency
stated
in
the
SIP
Call
rulemaking
that
90
percent
control
with
SCR
was
adopted
based
on
information
presented
in
the
Alternative
Control
Techiaues
Document
­­
NOx
Emissions
from
Station­
Reciprocating
Internal
Combustion
Engines
(
EPA­
453/
R­
93­
0321
(
ACTdocument):

For
[
reciprocating
internal
combustion
engines],
EPA
determined
based
on
the
relevant
ACT
document,
that
post­
combustion
controls
are
available
that
would
achieve
a
90
percent
reduction
from
uncontrolled
levels
.
.
.
.
(
63FR
57418)

EPA.
used
90
percent
control
for
large
reciprocating
internal
combustion
engines
to
calculate
the
state
NOx
budgets
for
the
SIP
Call
and
the
FIP.
In
the
FIP,
IEPA
also
proposed
an
emission
limitation
of
125
ppmv
for
natural
gas­
fired
lean­
burn
engines
based
on
90
percent
control
of
NOx
using
SCR.

The
Gas
Research
Institute
(
GFLI)
presents
in
this
white
paper
an
assessment
of
the
technical
issues
related
to
the
potential
use
of
SCR
to
reduce
NOx
emissions
from
natural
gas­
fired
lean­
burn
engines
and
the
reasonableness
of
the
EPAs
estimate
of
a
90
percent
reduction
of
NOx
emissions
from
lean­

burn
engines
using
SCR.
GRI
intends
to
continue
investigating
NOx
control
techniqu.
es
for
natural
gas
lean­
bum
engines
and
NOx
emissions
monitoring
techniques.
This
white
paper
includes
information
on
SCR
that
could
be
assembled
prior
to
the
November
30
deadline
for
comments
on
the
proposed
FIP.

Overview
ofSCR
Technology
SCR
was
developed
to
reduce
NOx
emissions
from
fossil
fuel­
fired
boilers.
In
the
early­
to­
mid
1980s,
SCR
was
applied
to
cogeneration
gas
turbine
engines
and
reciprocating
internal
com bustionengines.

NOx
reduction
with
SCR
is
based
on
the
reaction
of
ammonia
and
NOx
to
form
nitrogen
and
water.
SCR
uses
a
catalyst
to
promote
the
ammonia
and
NOx
reaction.
In
the
absence
of
a
catalyst,
the
reducing
reactions
occur
in
the
range
of
1000
°
C.
As
the
temperature
is
increased
above
the
1000
°
C
range,
the
ammonia
oxidizes
to
NO,
thereby
increasing
the
NOx
emissions
and
as
the
temperature
is
decreased
below
the
optimum
range,
unreacted
ammonia
is
emitted
from
the
system.
When
a
catalyst
is
added,
the
ammonia
and
NOx
react
at
temperatures
significantlylower
than
those
obtained
without
the
catalyst.

The
essential
components
of
the
SCR
system
are:
SCR
catalyst
Exhaust
ducting
and
SCR
housing
Ammonia
system
2
Control
system
and
coiitinuous
emissions
monitors
(
CEMs)

Traditional
SCR
catalysts
are
base
metal
catalysts
composed
of
vanadium
pentoxjde,
titanium
dioxide,
or
tungsten
trioxide.
The
temperature
range
for
the
base
metal
SCR
catalysts
for
NOx
conversions
greater
than
80
percent
is
about
300
to
425
°
C.
During
the
1980szeolite
SCR
catalysts
were
developed.

These
catalysts
have
a
wider
temperature
operating
range
md
offer
more
resistance
to
sulfur
poisoning,
which
has
been
a
problem
with
some
of
the
other
SCR
catalysts.
The
SCRprocess
requires
about
two
to
three
times
the
catalyst
size
of
that
required
for
oxidation
exhaust
catalysts
(
such
as
used
for
CO
control),
since
SCR
space
velocities
are
low
compared
to
those
required
for
other
post­
combustion
catalybc
controls.
Typically,
2
to
3
catalyst
stages
are
required
in
the
SCR
system.

SCR
requires
on­
site
storage
of
ammonia
and
an
ammonia
delivery
system.

The
ammonia
system
includes
the
ammonia
storage
tank,
the
ammonia
flow
regulation
system,
and
the
ammonia
injectors.
Heaters
are
required
to
vaporize
the
ammonia.
The
type
of
vaporization
system
used
depends
on
the
type
off
ammonia
used
­­
anhydrous
or
aqueous
­­
and
on
the
ambient
temperature
range
experienced
at
the
site.
In
order
for
SCR
to
work,
the
ammonia
must
be
fully
vaporized
and
evenly
dispersed
across
the
face
of
the
catalyst.

The
arnmonia
control
system
generally
includes
NOx
CEMs
upstream
and
downstream
of
the
SCR
catallyst.
The
ammonia
flow
rate
required
is
based
on
a
preset
ammonia
to
NOx
ratio
and
determined
from
a
measurement
of
the
inlet
SCR
NOx
concentration
and
fi­
om
determination
of
the
exhaust
flow
rate
of
the
engine.
System
performance
is
checked
by
measuring
the
outlet
SCR
NOx
concentration.

The
exhaust
gas
samples
are
conditioned
(
e.
g.,
dried
and
filtered)
and
are
pumped
to
the
NOx
analyzer.
Typically,
the
NOx
analyzer
is
cycled
between
the
inlet
=
andoutlet
of
the
SCR
unit.
Generally
this
type
of
system
requires
a
response
time
of
approximately
five
minutes
to
obtain
an
accurate,
stable
reading
of
the
NOx
concentrations.

Basis
for
EPA's
Conclusion
that
SCR
is
Effective
EPA
used
information
found
in
the
ACT
document
as
the
basis
for
use
of
90
percent
NOx
reduction
in
the
SIP
Call
and
FIP
rulemakings.
In
the
technical
support
document
for
stationaqr
internal
combustion
engines
(
A­
96­
56,
VI­

B­
13),
EPA
stated
that
90
perceint
effectiveness
for
the
use
of
SCR
on
natural
gas­
fired
'
lean
bum
engines
was
adopted
based
on
the
"
achievableNOx
reduction"
presented
in
Table
2­
5
of
the
ACT
document.
Table
2­
5
is
reproduced
below.

4
TABLE
2­
5.
EXPECTED
RANGE
OF
NO,
EMISSION
REDUCTIONS
AND
CONTROLLED
EMISSION
LEVELS
FOR
CONTROL
TECHNIQUES
APPLIED
TO
LEAN­
BURN
SI
ENGINES
[
NATURAL
GAS
FUEL)

Control
Average
uncontrolled
NO,
emission
level"
Achievable
Expected
controlled
NOx
emission
levels
technique
NO,
reduction,
YO
­
g/
hp­
hr
Ppmv
g/
hp­
hr
ppmv
AF­
16.8
1,230
5
­
30
11.8
­
16.0
860
­
1,170
IR­
16.8
1,230
0
­
20
13.4
­
16.8
980
­
1,260
AF
+
IR­
16.8
1,230
20
­
40
10.1
­
13.4
740
­
980
SCR­
16.8
1,230
90"
1.7
125
­
LE
I
16.8
I
1,230
87
2.0'
150
_
I
The
uncontrolled
emission
rate
shown
is
a
representative
average
for
lean­
bum
SI
engines.
The
actual
uncontrolled
emission
rate
will
vary
from
engine
to
engine.
bGuaranteedNO,
reduction
available
from
most
catalyst
vendors.
'
Guaranteed
controlled
NO,
emission
level
available
from
engine
manufacturers.

Footnote
"
b"
to
Table
2­
5
indicates
that
the
basis
for
the
90
percent
reduction
for
SCR
is
based
om
the
"
guaranteed
NOx
reduction
available
from
most
catalyst
vendors."
Section
2
the
ACT
does
not
include
a
detailed
discussion
of
the
information
provided
by
the
vendors
to
support
the
claims
for
90
percent
reduction
or
the
long­
term
performance
for
SCR
Section
5.2.4.4
of
the
ACT
provides
further
discussion
of
the
achievable
emission
reduction
using
SCR.
In
that
section,
the
ACT
states
that
emission
reductions
of
80
to
90+
percent
is
possible
using
SCR
on
engines
in
baseload
applications:
Based
on
the
available
information
and
the
emission
test
data
presented
in
Tables
5­
8
and
A­
5,
it
is
estimated
that
the
achievable
NOx
emission
reduction
for
SCR
in
gas­
fired
applications
is
80
to
90+
percent
for
baseload
applications,
with
an
NH,
slip
level
of
10
ppmv
or
less.

The:
ACT
states
that
90
percent
NOx
reduction
was
used
in
Chapter
6
to
calculate
controlled
NOx
emission
levels
and
cost
effectiveness.
However,
the
ACT
also
states
that
inadequate
information
was
available
to
determine
achdevable
continuous
NCbx
reductions
and
ammonia
slip
levels
for
SCR
in
variable
load
applications:

The
available
data
are
not
sufficient
to
assess
the
achievable
continuous
NOx
reductions
and
ammonia
slip
levels
for
SCR
used
in
variable
load
applications.

As
:
indicated
in
Section
V
of
this
paper,
variable
load
applications
present
unique
difficulties
for
SCR
that
have
not
yet
been
adequately
addressed
to
allow
widespread
application
of
SCR
for
engines
in
load­
following
applications,
such
as
natural
gas
compression.

In
the
ACT
there
is
emissions
test
information
for
6
reciprocating
internal
combustion
engines,
3
Clark
engines,
2
Cooper
Bessemer
engines,
and
1
.
Tecogen
engine.
The
engines
range
in
size
from
291
to
800
horsepower.
All
the
engines
are
smaller
thian
the
engine
horsepower
EPA
adopted
in
the
FIP
(
2400hp).
A
total
of
34
eimission
test
results
are
provided
in
the
ACT
that
report
NOx
emission
reductions.
Two
emission
test
results
report
zero
NOx
emission
reductions.
The
emission
results
presented
were
drawn
from
emission
tests
conducted
in
California
from
1986
­
1992.
Two
Clark
engines
and
the
Tecogen
engine
were
stillin
service
at
the
time
the
ACT
document
was
published.
The
other
three
SCR
units
had
been
removed
from
service
or
the
engines
had
been
replaced
with
electric
units.
The
ACT
document
stated
tha.
t
emissions
test
data
fior
existing
installations
with
SCR
ranged
from
65
to
95
percent
reduction.
According
to
the
technical
support
document,
EPA
calculated
the
expected
control1e:
d
NOx
emission
level
ais
1.7
g/
hp­
hr
(
125ppmv)
based
on
an
average
uncontrolled
NOx
emission
level
of
16.8
g/
hp­
hr
(
1230
ppmv)
and
90
percent
reduction
of
NOx
using
SCR.
However,
the
emissions
data
in
the
ACT
document
suggests
that
the
percent
NOx
reduction
achieved
by
engines
varies.
No
explanation
is
provided
in
the
ACT
document
for
the
variability
of
the
SCR
effectiveness.
No
infoimation
is
provided
on
the
operating
conditioins
of
the
engines
when
tested.
Also,
no
information
is
provided
on
ammonia
emissions
(
i.
e.,
ammonia
slip)
from
the
units.
The
range
of
effectiveness
reported
for
the
6
engines
is
presented
in
the
table
below.

Engin
Manufacturer­
Model
I
Horsepowe
I
Number
of
I
Range
of
NOx
­
e
No.
r
Tests
ControlReported
(%)
45
Clark
HRA­
6
660
5
84­
91
­­­__
__
47
Clark
HRA­
6
660
5
82­
88
139
Cooper
GMV
660
1*
50
Bessemer
~
~
~~~

248
Cooper
GMV­
8
800
9*
87­
93
Bessemer
309
Clark
HRA­
350
12
65­
84
32
357
Tecogen
CM­
29
1
2
95­
97
I
I200
I
I
a
)
ne
additional
test
was
conducted,
where
zero
emission
reduction
was
reported.
No
explanation
is
provided
in
the
report
for
those
results.

Only
one
engine,
the
Tecogen,
ireported
levels
greater
than
90
percent
for
all
tests
conducted.
This
small
engine
is
designed
to
be
used
for
cogeneration.

Two
other
engines
(
Nos.
45
and
248)
reported
levels
at
or
above
90
percent
for
at
least
one
test,
but
levels
less
than
90
percent
for
other
tests.
The
other
engines
did
not
include
any
test
results
at
or
above
90
percent
NOx
reduction.

The
emissions
data
included
ini
the
ACT
document
for
natural
gas­
fired
lean­
bum
engines
demonstrate
that
the
effectiveness
of
SCR
systems
are
variable.

7
The
ACT
document
does
not
discuss
or
explain
the
variable
emissions
results
and
does
not
identlfy
the
factors
that
would
contribute
to
this
variability.
Only
the
small
cogeneration
engine
consistently
reported
emission
reductions
greater
than
90
percent.
The
other
engines
did
not
Consistently
achieve
emission
reductions
at
or
above
90
percent.
As
indicated
in
the
ACT
document,
the
90
percent
reduction
presented
by
EPA
in
the
SIP
and
FIP
was
based
on
information
presented
in
Table
2­
5.
However,
information
presented1
in
Section
5.2.4.4
indicates
that
EPA
was
not
able
to
determine
an
achievable
NOx
reduction
level
for
the
use
of
SCR
in
load­
following
applications.

SCR
Installations
on
Lean­
Burn
Engines
There
is
limited
operating
experience
to
date
with
the
use
of
SCR
on
reciproca.
tinginternal
combustion
engines.
At
the
time
the
ACT
document
was
published,
EPA
identified
a.
total
of
23
engines
(
including
3
engines
using
digester
gas)
with
lean­
burn
engines
in
the
United
States
based
on
information
provided
by
catalyst
vendors
and
approximately
40
installations
of
SCR
overseas.
The
ACT
document
also
noted
that
there
is
little
experience
with
SCE
in
variable
load
applications
due
to
ammonia
injection
control
limitations.

At
present,
no
natural
gas­
fired
lean
burn
engines
were
identified
in
the
EPA
RACT/
BACT/
LAER
clearinghouse
with
SCR
controls
(
Queryof
15.004,
post
1991).
HACT/
BACT/
LAER
determinations
for
lean
burn
engines
have
relied
on
low
emissions
combustion
(
LEC)
technology,
other
combustion
modification
(
such
as
high
energy
ignition
systems),
or
parametric
controls
(
such
as
retarded
timing
and
alir
to
fuel
ratio
adjustment).
A
total
of
18
diesel
engines
with
SCR
controls
(
PA­
0096
and
PA­
0097)
were
identified
in
the
clearinghouse.
The
SCR
colntrols
for
the
diesel
engines
reduce
NOx
emissions
by
80
percent.

8
For
the
natural
gas
transmission
industry,
there
are
no
hown
SCR
installations
on
lean
burn
engines
in
pipeline
service.
Two
lean
bum
engines
at
a
natural
gas
storage
facility
in
California
have
SCR.
These
engines
have
been
mothballed
due
to
the
operations
and
maintenance
prob1e:
msrelated
to
the
use
of
SCR
with
the
variable
loads
experienced
in
the
naturatl
gas
storage
operation.
The
engines
are
not
used
presently.

Therefore,
there
are
no
engines
in
pipeline
service
within
the
United
States
at
this
tirne
that
have
SCR
installed.
The
two
engines
in
California
where
SCR
was
installed
have
been
mothballed
and
are
not
currently
in
use.

In
1996,
the
natural
gas
transmission
industry
conducted
a
survey
to
gather
information
on
the
reciprocating
engines
and
turbines
in
natural
gas
transmission
service.
The
transmission
companies
provided
information
on
over
5,000
reciprocating
inteirnal
combustion
engines.
As
a
part
of
the
survey,
information
was
provided
on
over
800
reciprocating
internal
combustion
engines
with
some
form
of
NOx
emission
controls
in
place.
The
survey
results
indicate
that
tlhe
natural
gas
transmission
industry
has
installed
technology
that
relies
on
combustion
modification
or
parametric
controls
to
reduce
NOx
emisstionsfrom
lean­
bum
engines,
rather
than
post­

combustion
controls,
such
as
SCR.
Only
2
of
the
599
lean­
bum
engines
with
NOx
controls
have
SCR
listalled
(
the
California
units).
For
natural
gas
transmission,
SCR
is
the
NOrr
control
technique
that
has
been
used
least
to
reduce
emissions
from
lean­
burn
engines.
Low
emissions
combustion
(
LEC)

or
other
combustion
or
parametric
control
techniques
have
been
used
most
often
to
reduce
NOx
emissions
from
lean­
bum
engines.
In
contrast,
for
rich­

burn
engines,
post­
combustion
catalytic
control
(
non­
selectivecataly­
hc
reduction
­­
NSCR)
is
the
NOx
control
technique
that
has
been
used
most
often
to
reduce
NOx
emissions.
The
transmission
industry
survey
indicates
that
100
of
201
rich­
bum
engines
with
NOx
controls
have
NSCR
installed.
The
results
of
the
industry
survey
are
presented
below.
ban
Burn
Engines
qpe
of
Control
In
SIP
Call
Not
in
SIP
Call
Total
Units
States
States
Low
Emissions
Combustion
High
Energy
Ignition
Other
NOx
Control
Parametric
Controls
SCR
Total
245
139
384
87
21
108
49
13
62
39
4
43
0
2*
2*
371
179
599
*
Possible
conversion
to
lean­
bum
conversion.
Data
for
these
engines
are
under
review.

The
survey
results
include
NOx
controls
that
were
installed
by
natural
gas
transmission
companies
to
comply
with
NOx
RACT
and
BACT
requirements.

No
natural
gas
transmission
company
was
required
to
install
SCR
as
NOx
RACT
or
BACT.
Although
some
companies
did
consider
SCR
as
a
potential
control
option,
SCR
was
found
to
be
unacceptable
due
to
the
limits
of
the
ammonia
delivery
system
and
the
need
for
load­
following
horsepower
to
respond
to
pipeline
conditions.
(
More
discussion
of
this
issue
is
included
in
the
following
section
of
this
white
paper.)

Therefore,
while
NSCR
is
widely
applied
for
natural
gas­
fired
rich­
burn
engines,
there
are
very
few
instances
where
SCR
has
been
applied
to
natural
gas­
fired
lean­
burn
engines.
State
air
regulatory
agencies
did
not
require
10
SCR
to
implement
NOx
RACT
or
BACT
requirements.
At
present,
there
are
no
engines
in
the
natural
gas
transmission
industry
that
are
operating
with
SCR.
Rather,
combustion
modification
and
parametric
controls
have
been
the
techniques
applied
to
reduce
NOx
emissions
from
existing
lean­
burn
engines.

Difficulties
Operating
SCR
Systems
The
following
issues
must
be
considered
in
the
context
of
NOx
control
with
SCR
Variable
load
operation
Ammonia
handling
and
storage
Catalyst
maintenance
arid
disposal
Engine
maintenance
Continuous
emissions
monitoring
Variable
load
operation
complicates
the
successful
operation
of
an
SCR
system
given:

1)
variations
in
NOx
emissions,
2)
variations
in
exhaust
gas
flow
and
temperature,
and
3)
thermal
cycling.

In
order
for
SCR
to
work
properly,
with
the
appropriate
control
of
NOx
emissions
and
without
excessive
ammonia
emissions,
it
is
necessary
that
the
proportions
of
NOx
and
ammonia
be
correct.
Reciprocating
internal
combustion
engines
in
natural
gas
transmission
service
are
used
to
compress
and
move
natural
gas
along
the
pipeline.
The
load
and
speed
of
the
engines
must
respond
to
pipeline
conditions,
supply,
and
demand
for
natural
gas.
As
a
result,
engines
are
frequently
required
to
change
load
and
speed
conditions.
These
changes
produce
changes
in
the
NOx
emissions
from
the
engines.
For
examplie,
a
load
change
of
only
5
percent
can
significantly
change
the
NOx
emitted
from
an
engine
­­
for
some
engines
NOx
can
increase
as
much
as
50
piercent.
When
these
changes
occur,
the
11
ammonia
feed
system
must
be
adjusted
to
maintain
the
proper
ratio
of
ammonia
to
NOx.
In
Section
5.2.4.2,
the
ACT
notes
that
variable
load
can
cause
NOx
concentration
swings
that
pose
problems
for
the
SCR
system.

SCR
systems
traditionally
use
a
feedback
system
that
relies
on
a
NOx
or
ammonia
analyzer
to
adjust
the
ammonia
feed
system.
It
takes
approximately
5
minutes
time
to
extract
a
sample
and
report
NOx
emission
levels
using
a
NOx
CEMs.
Therefore,
for
load­
 allowing
applications,
a
feedback
system
is
not
sufficient
alone
to
allow
the
ammonia
feed
system
to
be
adequately
responsive
to
maintain
NOx
reductions.
Some
load­
following
engines
may
be
able
to
use
SCR
if
the
systems
incorporate
feed­
forward
controls.
The
feed­
forward
controls
incorporate
sophisticated
predictive
emissions
models
(
PEMs),
which
estimate
NOx
and
adjust
the
ammonia
feed
system
accordingly.
A
recent
installation
of
SCR
at
the
Buckeye
Pipeline
station
uses
PEMs
as
the
feed­
forward
controls
for
an
SCR
with
new
Waukesha
engines.
For
existing
engines,
no
applications
are
known
where
PEMs
have
been
successfully
used
as
feed­
forward
controls
for
SCR.
One
reason
that
the
feed­
fomard
system
may
not
have
been
applied
to
existing
engines
is
the
need
to
develop
a
sophisticated
predictive
emissions
model
to
estimate
NOx.
For
new
engines,
engine
manufacturers
have
information
readily
available
to
develop
the:
PEMs.
However,
for
existing
engines,
the
owner
or
operator
would
have
to
independently
develop
a
PEMs
that
would
be
sophisticated
enough
to
reliably
control
the
ammonia
feed
system.
The
engines
at
the
Buckeye
Pipeline
station
also
represent
a
load­
following
application
with
a
limited
range
of
operating
conditions
since
the
engines
are
used
to
drive
liquid
pumps.
For
natural
gas
transmission,
the
engines
are
used
to
compress
gas
along
the
pipeline
and
the
pipeline
conditions
create
a
wider
ratnge
of
operating
conditions
for
the
load­
following
engines.
The
Buckeye
Pipeline
station
experienced
difficulties
with
catalyst
masking
when
lube­
oil
residuals
were
deposited
on
the
catalysts
during
frequent
start­
up
tests.
For
natural
gas
transrnlssion
service,
engines
often
experience
12
frequent
start­
ups,
which
may
lead
to
catalyst
masking,
as
experienced
at
Buckeye.

Variable
load
conditions
also
result
in
variations
in
exhaust
gas
flow
and
temperature
conditions.
The
imonia
control
system
would
need
to
be
adjusted
to
respond
to
the
changes
in
the
exhaust
mass
flow
rate
(
especially
dual
shaft
engines).
Variable
exhaust
gas
flow
rates
would
also
affect
mixing
in
the
catalyst.
The
wider
the
range
in
exhaust
gas
flow
rates
velocities,
the
more
difficult
it
is
to
design
a
catalyst
that
can
reliably
achieve
high
emission
reductions
(
e.
g.,
90
percent).
Reduced
engine
load
can
also
decrease
the
exhaust
gas
temperature
below
the
range
for
optimal
NOx
conversion,
which
would
not
allow
the
catalyst
to
reliably
achieve
high
emission
reductions.

Section
5.2.4.2
of
the
ACT
states
that
 
lower
exhaust
temperature
at
reduced
power
output
may
result
in
a
reduced
NOx
reduction
efficiency
from
the
catalyst. 
Variable
load
conditions
also
result
in
thermal
cycling
of
the
catalyst,
which
leads
to
reduced
catalyst
life.
These
variable
load
conditions
are
cornmon
for
engines
in
natural
gas
transmission
service.

As
indilcated
in
the
EPA
ACT
document,
there
is
little
experience
using
SCR
systems
on
engines
in
load­
following
applications,
such
as
natural
gas
transmission.
Although
catalyst
vendors
suggest
that
SCR
can
be
designed
to
work
in
load­
following
applications,
installations
of
SCR
systems
on
load­

following
applications
have
been
limited.
The
results
of
the
1996
survey
of
natural
gas
transmission
companies
were
queried
to
identify
SCR
installations.
Based
on
that
search,
only
two
engines
have
SCR
installed
and
those
engines
have
been
mothballed.

The
continuous
and
reliable
operation
of
an
SCR
system
requires
close
monitoring
of
several
subsystems.
This
includes
ammonia
storage,
control,
metering,
injection,
and
leak
detection
systems.
Problems
can
occur
with
the
clogging
of
the
ammonia
inozzles
and
these
have
to
be
closely
monitored
and
corrective
action
taken
immediately
to
ensure
proper
injection
rates
and
even
distribution
of
ammonia
in
the
flue
gas
for
proper
mixing.
In
addition,

most
SCRsystems
rely
on
CEMs
as
the
feedback
control
for
the
ammonia
feed
system.
In
order
to
ensure
reliable
performance,
NOx
CEMs,
including
sample
conditioning,
need
to
be
properly
calibrated
and
maintained,
which
requires
specially
trained
personnel.
Finally,
instrumentation
for
ammonia
injection
rate
needs
to
be
calibrated
frequently
to
ensure
proper
control.

These
activities
become
difficult
if
the
station
is
unmanned,
which
is
often
the
case
in
pipeline
operation.

Maintenance
of
the
SCR
system
requires
periodic
catalyst
cleaning
in
order
to
restore
its
reactivity.
Cleimbg
is
accomplished
by
washing
of
the
catalyst
with
water
and
typically
requires
several
days
downtime
for
removal
of
the
catalyst,
the
washing
and
the
re­
installation
of
the
catalyst.
Disposal
of
the
solution
remaining
from
this
procedure
can
be
an
issue.
Contaminants
from
lube
oil
and
engine
wear,
such
as
metals
or
silicon
oxides,
will
affect
the
frequency
of
catalyst
cleaning
and
can
reduce
catalyst
effectiveness
and
catalyst
life.
To
minimize
contamination,
specially
formulated
lubricating
oils
must
be
used.
For
four­
stroke
engines,
the
use
of
low­
contaminant
lube
oil
may
result
in
increased
valve
wear.
The
vendor
generally
handles
disposal
of
SCRcatalysts.

Most
SCR
systems
rely
on
continuous
emissions
monitors
(
CEMs)
as
the
feedback
contxol
for
the
ammonia
feed
system.
The
installation,
operation
and
:
maintenance
of
a
NOx
CEMs
are
complicated
and
require
specially
trainled
personnel.

14
Negative
Environmental
Impacts
from
SCR
I
SCR
relies
on
the
use
of
ammonia
injected
in
the
exhaust
stream
in
the
presence
of
a
catalyst
to
control
NOx
emissions.
Ammonia
is
a
colorless
gas
wiith
a
pungent
odor.
Ammonia
is
poisonous
if
inhaled
in
great
quantities
and
is
irritating
to
the
eyes,
nose,
and
throat
in
lesser
amounts.
Ammonia
is
explosive
when
mixed
with
air
in
certain
proportions
(
approximately
one
volume
of
ammonia
to
two
volumes
of
air).

Unlike
technology
like
combustion
modifications
or
parametric
controls
(
which
is
used
on
engines
in
natural
gas
transmission
service),
SCR
is
not
a
pollution
prevention
technique
and
uses
additional
resources,

which
in
turn
create
safety
hazards
and
hazardous
waste
disposal
prob1e:
ms.
For
example,
periodic
cleaning
of
the
catalyst
is
required
to
maintain
catalyst
effectiveness.
After
cleaning,
the
cleaning
solution
is
considered
a
hazardous
waste
and
creates
disposal
problems.

Additionally,
catalyst
disposall
itself
can
create
environmental
problems
because
the
catalyst
may
contain
heavy
metals
and
other
toxic
substances.

SCRsystems
usually
operate:
with
an
ammonia/
NOx
molar
ratio
of
about
1.0.
This
is
close
to
the
theoretical
limit
required
for
complete
NOx
reduction.
However,
due
to
kinetic
limitations
and
improper
mixing
between
the
ammonia
and
the
flue
gas,
complete
reduction
is
not
achieved,
leading
to
unreacted
ammonia
escaping
into
the
atmosphere.

Increasing
the
ammonia
injection
rate
to
overcome
deleterious
kinetic
and
mixing
effects
increases
the
probability
of
NOx
reduction,
but
also
increases
ammonia
slip
emissions.
To
achieve
high
NOx
reduction
efficiencies
consistently,
SCF:
systems
typically
operate
with
a
relative
excess
of
ammonia,
resulting
in
ammonia
slip
emissions.
This
issue
creates
a
potential
for
an
increase
in
ammonia
emissions
into
the
15
atmosphere,
which
is
further
exacerbated
by
ammonia
flow
difficulties
associated
with
variable
load
operation.

There
are
safety
concerns
associated
with
accidental
spills
of
ammonia.

At
low
concentrations,
ammonia
can
cause
health
effects
and
can
be
a
nuisance
due
to
its
objectionable
odor.
At
high
concentrations,
it
is
a
toxic
compound
and
a
fire
hazard.
Safety
hazards
can
occur
if
the
ammonia
is
spilled
or
there
are
leaks
from
ammonia
storage
vessels.

Safety
hazards
can
be
reduced
if
aqueous
ammonia
or
urea
is
used.

However,
because
water
becomes
saturated
at
about
25
percent
ammonia
by
weight,
aqueous
ammonia
tanks
must
be
four
times
larger
than
anhydrous
ammonia
tanks
for
the
same
application.
Consequently,

the
cost
of
storage
tanks
and
transportation
costs
for
aqueous
ammonia
will
be
greater
than
if
anhydrous
ammonia
were
used.
If
a
concentrated
aqueous
solution
of
urea
is
used,
the
urea
tank
must
be
heated
to
avoid
re­
crystallization
of
the
urea.
The
corrosive
nature
of
ammonia
leads
to
failure
of
piping
and
other
components.
Additionally
suspended
rust
particlles
often
clog
filters
and
ammonia
injection
nozzles
requiring
more
fi­
equent
cleaning
or
replacement
of
equipment.

Storing
and
using
ammonia
in
quantities
above
the
thresholds
specified
in
the
RMP
rule
may
create
situations
where
the
facility
has
to
comply
with
the
provisions
of
40
CFlR
part
68
as
well.

ConcXusions
As
presented
in
this
paper,
90
percent
reduction
of
NOx
emissions
from
lean­
burn
reciprocating
internal
combustion
engines
using
SCR
is
not
demoinstrated
in
practice.
Although
catalyst
vendors
indicated
to
EPA
that
90
percent
reduction
typically
would
be
achievable
for
gas­
fired
applications,
the
emissions
data
presented
in
the
ACT
document
suggests
that
NOx
reduction
varies
for
engines
with
SCR.
Also,
the
ACT
states
that
insufficient
data
were
available
to
estimate
achievable
NOx
16
reduction
levels
for
the
use
of
SCR
in
variable
load
applications,
such
as
natural
gas
transmission.
Since
engine
operation
can
significantly
affect
partmeters
that
affect
SCR
performance
(
such
as
NOx
emission
levels,

exhaust
gas
flow
and
exhaust
gas
temperature),
SCR
systems
would
rieeld
to
be
evaluated
over
the
full
range
of
engine
operations
to
determine
achievable
NOx
control
levels.
Load­
following
applications,
such
as
natural
gas
compression,
present
unique
difficulties
for
SCR
systems,

which
were
not
fully
addressed
by
EPA
in
determining
the
applicability
of
SCIZ
or
the
90
percent
effectiveness.
Also,
EPAs
assessment
of
the
cost­

effectiveness
of
SCR
did
not
include
costs
that
would
be
incurred
for
variable
load
applications,
including
costs
to
develop
a
forward­
feed
PEM
and
costs
that
result
from
vendors
"
over­
engineering"
the
SCR
system
to
try
to
offset
reduced
NOx
control
efficiencies.

SCR
has
been
installed
on
reciprocating
internal
combustion
engines
in
very
few
instances.
Of
the
599
lean­
bum
engines
with
NOx
controls
that.

were
included
in
a
survey
of
natural
gas
transmission
companies,
only
two
engines
in
the
United
States
have
SCR
installed
and
those
engines
have
been
mothballed
due:
to
difficulties
operating
the
SCR
units
in
a
load­
following
application.,
Instead,
because
of
the
problems
associated
witln
using
SCR
in
variable
load
applications,
NOx
emissions
from
lean­

burn
engines
in
pipeline
service
have
been
controlled
using
combustion
modification
or
parametric
controls.

17
Appendix
B
COMMENTS
OF
THE
COALITION
FOR
GAS­
BASIED
ENVIRONMENTAL
SOLUTIONS
ON
THE
FEDERAL
IMPLEMENTATION
PLAN
TO
REDUCE
THE
EtEGIOMAL
TRANSPORT
OF
OZONE
18
Joel
Bluestein­
Director
1655
North
Fort
Myer
Drive
Suite
600,
Arlington,
VA
22209
703­
528­
1900
­
FAX
703­
528­
5106
COALITIONFOR
6AS­
BASED
COMMENTS
OF
THE.
COALITION
FOR
GAS­
BASED
ENVIRONMENTAL
SOLUTIONS
ON
THE
FEDERAL
IMPLEMENTATION
PLAN
TO
REDUCE
THE
:
REGIONALTRANSPORT
OF
OZONE
The
Coalition
for
Gas­
Based
Environmental
Solutions
is
a
group
of
natural
gas
producers,
pipelines
and
distribution
companies
formed
to
prom.
ote
environmental
regulation
that
recognizes
and
encourages
the
use
of
low­
emitting,
high­
efificiency
fuels
and
technologies.
The
Coalition
is
pleased
to
have
this
opportunity
to
comment
on
the
 
Federal
Implementation
Plan
to
Redluce
the
Regional
Transport
of
Ozone. 
The
Coalition s
comments
focus
on
the
allocation
of
allowances
in
the
NO,
trading
system
that
EPA
would
implement
for
large
stationary
sources
under
the
FIP.
The
Coalitioln
strongly
supports
output­
based
allocation
as
part
of
the
trading
program
and
urges
the
EPA
to
adopt
options
2
or
3
presented
in
the
FIP
proposal.
In
addition,
this
letter
presents
some
specific
recommendations
for
improving
the
proposed
output­
based
allocation
system,
in
particular
more
equitable
and
appropriate
treatment
of
cogeneration
facilities.

The
allocation
of
allowances
in
the
trading
program
represents
the
annutal
distribution
of
a
large
amount
of
money.
If
NO,
allowances
are
worUi
$
3000/
ton,
the
allocation
will
distribute
over
$
1.5
billion
worth
of
allowances
to
affected
sources
each
year.
In
a
competitive
electric
market,
this
value
will
affect
the
utilization
of
generating
assets.
If
allowances
are
distributed
lbased
on
historical
emissions
or
heat
input,
they
will
subsidize
and
incentivize
higher
emissions
and
higher
energy
consumption.
Only
an
output­
based
allocation
provides
the
correct
signal
by
rewarding
sources
that
produce
the
greatest
output
with
the
lowest
energy
consumption
and
emissions.

The
IEPA
has
begun
to
suggest
that
it
understands
the
value
of
output­
based
allocation.
In
addition
to
providing
the
output­
based
options
2
and
3in
the
FIP,
EPA
recently
promulgated
output­
based
New
Source
Perfctrmance
Standards
for
NO,
emissions
from
large
boilers.
In
the
preamble
to
that
rule
as
in
the
SIP
call,
EPA
discussed
the
advantages
of
output­
based
regulation.
The
Coalition
applauds
the
EPA s
support
for
output­
based
allocation.

19
The
greatest
barrier
to
output­
based
regulation
in
the
FIP
at
this
time
seems
to
be
a
perceived
lack
of
data
to
use
in
making
the
allocations.
The
EPA
has
committed
to
a
process
to
develop
monitoring
programs
thait
will
clearly
provide
the
data
to
support
output­
based
allocations.
These
data
are
expected
to
be
available
within
a
few
years,
so
the
real
issue
at
this
time
is
how
to
do
the
allocations
for
2003
and
2004
with
the
available
data.
After
that
time,
detailed
data
should
be
available
to
do
the
allocations.

If
we
expect
to
be
moving
to
output­
based
allocation
within
a
few
years,
it
seems
a
waste
of
effort
to
develop
an
input­
based
program
for
2003
and
2004,
only
to
immediately
begin
developing
an
output­
based
program.
This
is
even
more
true
whien
we
consider
that
the
data
required
and
avatilable
for
the
first
years
of
an
input­
based
program
are
not
very
different
from
those
for
an
output­
based
program.
The
Coalition
therefore
urges
the
EPA
to
adopt
an
output­
based
allocation
methodology
such
as
option
2
or
3
for
the
final
FIP.
Some
specific
recommendations
on
the
methodology
follovr.

These
recommendations
ffocus
on
the
electric
generating
sector.
Although
the
Coalition
supports
the
use
of
output­
based
allocation
for
all
sources,
the
value
of
output­
based
allocation
for
industrial
boilers
is
too
small
given
the
type
of
data
available
at
this
time
and
the
effort
involved.
We
suggest
that
the
EPA
focus
on
electric
generators
initially
and
bring
in
the
industrial
boilers
in
a
few
years
as
the
improved
monitoring
data
become
available.

._
Output­
based
allocation
to
utility
boilers
The
EPA
approach
to
calculation
of
output
for
these
units
is
to
use
heat
input
data
from
the
acid
rain
CEMs
and
apply
a
heat
rate
to
calculate
output.
This
is
not
unreasonable
as
a
short
term
solution.
There
will
be
some
difficulty
matching
boilers
to
CEMs
to
generators,
but
that
can
be
dealt
with.
The
Coalition
does
recommend
the
use
of
ELA
Form
767
as
a
source
of
heat
rate
data
rather
than
Form
860
as
proposed
by
EPA.
Form
767
tracks
heat
input
and
power
generation
on
a
monthly
basis
and
is
a
more
accurate,
complete
and
reliable
source
of
heat
rate
than
Form
860.
For
that
matter,
Fonn
767
would
be
a
more
reliable
source
of
output
data
than
the
approach
suggested
by
EPA
since
the
measurement
of
output
(
and
for
that
matter
input)
as
specified
by
Form
767
is
more
accurate
than
the
estimation
method
used
to
calculate
heat
input
for
acid
rain
CEM
purposes.
However,
the
approach
suggested
by
EPA
is
adequate
and
may
be
more
attractive
to
EPA
since
involves
less
reliance
on
outside
data
sources.
Finally,
many
states
may
require
sources
to
provide
heat
iriput
data
which
the
EF A
could
use
in
making
the
allocations.

20
Output­
based
allocation
to
non­
utilitv
electric
generators
This
is
a
more
difficult
case
since
EPA
has
no
direct
access
to
data
on
either
input
or
output
for
NUGs
at
this
time.
The
proposed
approach
is
again
to
multiply
heat
input
times
a
heat
rate.
Although
the
FIP
states
that
heat
input
data
from
EIA
Form
867
were
used,
this
is
apparently
an
error
since
lhese
data
are
confidential
at
this
time.
Instead,
it
appears
that
EPA
used
heat
input
data
frcbm
the
2007
IPM
forecast,
backcast
to
1995.
For
he:
atrates,
the
EPA
suggests
using
generic
values
for
different
equipment
types
since
specific
data
are
not
available.
The
Coalition
suggests
several
possible
improvements:
IPM
has
specific
heat
rate
values
for
many
non­
utility
generators.
These
could
be
used
instead
of
the
generic
values.

Better
still,
if
EPA
is
going
to
use
IPM
data,
simply
use
the
IPM
generation
forecast
and
backcast
to
1995.
This
is
probably
more
accurate
than
the
proposed
approach.

Again,
many
states
require
non­
utility
generators
to
provide
data
on
heat
input
which
the
EPA
could
use
instead
of
the
IPM
data.
It
could
then
use
either
specific
heat
rate
data
or
the
generic
values
to
calculate
generation.

Cogeneration
Cogeneration,
also
known
as
combined
heat
and
power,
­

has
been
recognized
as
a
high
efficiency
generation
technique
that
can
make
a
significant
contribution
to
national
environmental
and
energy
efficiencygoals.
President
Clinton
is
on
record
supporting
expanded
use
of
CHI?
as
an
economically
beneficial
strategy
to
address
energy,
environmental,
and
global
climate
change
challenges.
In
response,
the
U.
S.
Department
of
Energy
has
initiated
a
CHI?
Challenge
program
that
seeks
to
address
barriers
to
expanded
deployment
of
CHP
in
the
marketplace.
EPA s
Atmosplheric
Pollution
Prevention
Division
(
APPD)
has
supported
DOE SCHP
Challenge
program,
particularly
in
the
area
of
identifymg
and
addressing
regulatory
impacts
on
cogeneration.

A
Cogeneration
facility
provides
both
electric
and
thermal
energy
at
a
combined
efficiency
higher
dhan
normal
separate
generation
of
them
output.
Because
it
is
providing
both
thermal
and
electric
output,
however,
it
consumes
more
energy
per
kwh
thana
standard
electric
generator.
The
basic
approatch
to
output­
based
allocation
for
cogeneration
facilities
should
be
to
allocate
them
allowances
equivalent
to
those
allocated
for
conventional
separate
generation
of
the
same
output.
In
the
hlly­
developed
output­
based
allocation
system,
this
is
straightforward:
The
thermal
and
electric
output
are
measured
and
the
facility
receives
its
proportional
share
of
the
allowance
pool
for
electric
and
thermal
generation
respectively.

The
proposed
output­
based
allocation
in
the
FIP
does
not
recognize
cogeneration
and
in
fact
perializes
cogeneration
facilities
because
it
gives
21
them
the
same
credit
as
simple
electric
generators,
even
though
they
must
consume
more
energy
in
order
to
provide
useful
thermal
energy.
The
EPA
should
remedy
this
oversight
by
allocating
allowances
to
cogeneration
facilities
for
thermal
and
electric
output.
This
should
be
done
by
using
generic
techniology­
specific
factors
similar
to
those
proposed
for
heat
rate.
Specifically,
the
EPA
should
calculate
the
thermal
output
of
cogeneration
facilities
by
using
the
following
generic
power­
to­
heat
ratios.

Recommended
Power­
to­
HeatRatios
for
Cogeneration
Facilities
Power­
to­
Heat
System
mpe
Steam
Turbine
Simple
Cycle
Combustion
.

Turbine
RUtio
0.2
0.5
1IO
1.75
Combined
Cycle
I.
C.
Engine
The
electric
output
should
'
be
divided
by
the
power­
to­
heat
ratio
and
converted
from
kwh
to
MMI3tu
thermal
output.
This
can
be
used
in
one
of
two
ways
depending
on
how
industrial
sources
are
being
treated:
If
industrial
sources
are
being
treated
on
an
output­
basis,
the
thermal
output
can
be
used
directly
to
calculate
the
share
of
allowances
to
be
allocated
to
the
facility.

If
industrial
sources
are
being
treated
on
an
input
basis,
the
thermal
output
should
be
divided
by
a
typical
boiler
efficiency
such
as
SO%,
to
calculate
the
equivalent
heat
input
that
would
be
required
to
provide
that
amount
of
thermal
energy.
The
equivalent
heat
input
would
be
multiplied
by
0.17
lbMME3tu
to
calculate
the
allocation.

As
noted
above,
some
states
may
have
more
specific
information
which
the
EPA
could
use
to
calculate
these
allocations.

Finally,
all
of
these
approaches
should
be
applied
in
the
Section
126
language
and
should
be
provided
to
the
states
for
use
in
making
allocations
for
2003..

Pending
the
development
of
more
accurate
monitoring
data,
the
EPA
has
very
little
data
on
which
base
an
allocation
­
either
input
or
output
­
which
can
truly
be
characterized
as
highly
accurate,
verifiable
and
quality
assured.
For
electric
generators
affected
by
the
acid
rain
program,
the
EPA
may
have
better
data
than
for
the
others.
However,
it
has
not
been
shown
that
the
CEM
heat
input
data
are
more
reliable
than
generation
data
provided
to
EIA
and
there
are
engineeringreasons
to
believe
that
it
might
be
less
accurate.
For
the
remainder
of
the
sources,
EPA
has
no
reliable
data
for
either
input
or
output.

22
3
If
the
ultimate
goal
is
an
output­
based
system
­
and
it
should
be.

at
If
the
EPA
is
planning
to
provide
reliable
output
data
within
the
next
few
years
­
and
it
is.

Then
we
should
not
quibble
over
which
set
of
inaccurate
data
is
less
bad.
We
should
proceed
to
set
up
the
regulatory
structures
that
we
will
need
and
use
in
the
future
using
the
best
available
data
and
apply
better
data
in
the
next
few
years
as
they
become
available.
We
have
the
ability
to
do
output­
based
allocation
for
2003
with
as
much
accuracy
as
any
other
approach.
Data
and
methodological
shortcomings
should
not
be
used
as
an
excuse
to
delay
the
development
of
regulatory
systems
that
reduce
costs
and
pollution
and
which
have
been
identified
is
our
goal.

For
clarifications
or
hrther­
information
on
these
comments,
please
contact
Joel
Bluestein
at
7103­
528­
1900
or
jbluesteint3eea­
inc.
com.

23
FAX
TKANSMITTA
DATE
June
5,2002
NUMBER
OF
PAGES
(
incfudlngthis
cwver
Sh88t):

TO:
Name
Bill
Neuffer
I
Pmtectlog
Texas
by
pmlng
end
Pmentrng
~
allullan
Organization
US
EPA
OAQPS
FAX
Number
(
919)
641­
0824
FROM:
TEXAS
NATURAL
RESOURCE
CONSERVATION
COMNllSSlON
Name
Randy
Hamilton
­

DivisiodRegion
Air
Permits
Division
Telephone
Number
(
512)
2394512
FAX
Number
(
512)
239­
1300
NOTES:

The
DukeRexas
Eiastern
test
results
for
the
SIXrich
bum
YVG103s
converted
to
lean
burn
in
1999with
Enginuity's
precombustlonchamber
engineering.
We
seem
to
have
misplaced
this
test
report,
but
the
company
sent
acopy
today.

TO
'
d
20
'
d
LT:
8T
2002
S
UnC
2
7
L
,
la
L
t
5
330
3.1
97.1
650
914
70
n
111
2dO
16.0
10149
0'
010
10113546
ti653
992
1730
m5
lo42
1743
1.48
973
802
330
3.
I
99.0
663
,
911
­
78
136
.
24.0
16.0
LW8a
*
.
10002.422
S653
992
17.20
9349
1006
172.7
173.7
sa7
28.1
32­
0
80.7
PO'd
8T:
8I
ZOOZ
S
unlC
82
1
1
11fIf99
11/
1/
93
1&
1S
le31
90
'
d
6T:
8T.
ZOOZ
S
unf
ao.
1
s95
7­
12
la55
143.1
142.0
138
8S7
854
37.7
333
363
/
b&
6
a3
67.8
9
713­
OM­
8820
U2U6
1
1lMM
m40
2I:
M
1R
33#
*

1.1
67.0
a4
8.79
74
I23
add
13.0
9959
adlo
9w.
993
11653
992
17s
n44
11%

dm30.36
43
40
0.
W
sl
1964
193.0
135
9f.
l
oa.
4
*
1AmS
$
B
333
1134
0.80
0.9029
o.
ooia
­
034
020
10
