1
dwcgi­
7150­
1077822954­
148160000.
wpd
MEMORANDUM
TO:
Docket
No.
OAR­
2002­
0058
FROM:
Jim
Eddinger,
U.
S.
Environmental
Protection
Agency,
OAQPS
(
C439­
01)

DATE:
February,
2004
SUBJECT:
Revised
MACT
Floor
Analysis
for
the
Industrial,
Commercial,
and
Institutional
Boilers
and
Process
Heaters
National
Emission
Standards
for
Hazardous
Air
Pollutants
Based
on
Public
Comments
1.0
INTRODUCTION
This
memorandum
describes
the
development
of
the
Maximum
Achievable
Control
Technology
(
MACT)
floor
and
is
a
revision
of
the
memorandum
previously
prepared
for
the
proposed
rulemaking
for
the
industrial,
commercial,
and
institutional
boilers
and
process
heaters
National
Emission
Standard
for
Hazardous
Air
Pollutants
(
NESHAP).
The
methodology
used
to
develop
the
MACT
floor,
the
assumptions
used
for
the
analysis,
the
data
sources,
and
the
resulting
MACT
floor
for
new
and
existing
sources
are
presented.
The
memorandum
includes
the
following
sections:

Section
2.0
Background
Information
Section
3.0
Data
Sources
Section
4.0
Affected
Source
and
Subcategories
Section
5.0
General
Methodology
for
the
MACT
Floor
Analysis
Section
6.0
Determination
of
Best
Performing
Controls
Section
7.0
Analysis
of
Good
Combustion
Practices
Section
8.0
Determination
of
MACT
Floor
Emission
Limits
Section
9.0
Analysis
for
Process
Heaters
Section
10.0
Determination
of
Health­
Based
Alternative
TSM
Limit
Section
11.0
References
2
dwcgi­
7150­
1077822954­
148160000.
wpd
Appendices
2.0
BACKGROUND
INFORMATION
Industrial
boilers,
commercial
and
institutional
boilers,
and
process
heaters
were
identified
as
source
categories
of
HAP
under
section
112(
c)
of
the
Clean
Air
Act
(
the
Act),
to
be
regulated
by
a
NESHAP
under
section
112(
d)
of
the
Act.
Indirect­
fired
process
heaters
are
similar
to
boilers
in
fuel
use,
emissions,
and
applicable
controls,
and,
consequently
are
combined
with
industrial,
commercial
and
institutional
boilers
for
purposes
of
developing
emission
standards.

Direct­
fired
units
are
covered
in
other
MACT
standards
or
rulemakings
pertaining
to
industrial
process
operations.
For
example,
lime
kilns
are
covered
by
the
Pulp
and
Paper
NESHAP
(
40
CFR
Part
63,
subpart
S).
The
source
category
also
does
not
include
combustion
units
regulated
in
other
standards,
including
municipal
waste
combustion
units,
industrial/
commercial
waste
incinerators,
medical
waste
incinerators,
hazardous
waste
boilers,
or
pulp
and
paper
recovery
boilers.

The
Act
specifically
requires
that
fossil
fuel­
fired
steam
generating
units
of
more
than
25
megawatts
that
produce
electricity
for
sale
(
i.
e.,
utility
boilers)
be
reviewed
separately
by
EPA.

Consequently,
fossil
fuel­
fired
utility
boilers
greater
than
25
megawatts
are
not
examined
in
this
source
category,
but
fossil
fuel­
fired
units
less
than
25
megawatts
and
all
nonfossil
fuel­
fired
utility
boilers
are
included
in
this
source
category.
Emissions
from
combustion
units
with
waste
heat
boilers
are
also
not
included
in
the
source
category.
Emissions
from
any
commercial
or
industrial
solid
waste
incinerator
(
CISWI)
or
other
incinerator
unit
that
has
a
waste
heat
boiler
will
be
covered
by
regulations
promulgated
under
section
129
of
the
CAA.

Many
industrial
facilities
have
office
buildings
located
onsite
which
use
hot
water
heaters.

Such
hot
water
heaters,
by
their
design
and
operation,
could
be
considered
boilers.
However,

since
hot
water
heaters
generally
are
small
and
use
natural
gas
as
fuel,
their
emissions
are
negligible
compared
to
the
emissions
from
the
industrial
operations
that
make
such
facilities
major
sources,
and
compared
to
boilers
that
are
used
for
industrial,
commercial,
or
institutional
purposes.
Moreover,
such
hot
water
heaters
are
more
appropriately
described
as
residential­
type
boilers,
not
industrial,
commercial
or
institutional
boilers.
Therefore,
residential
type
hot
water
heaters
are
not
included
in
this
source
category.
3
dwcgi­
7150­
1077822954­
148160000.
wpd
Section
112(
d)
of
the
Act
directs
EPA
to
develop
standards
that
require
the
maximum
degree
of
reduction
in
emissions
of
HAP
that
is
achievable,
which
are
commonly
referred
to
as
MACT
standards.
For
existing
major
sources,
the
Act
requires
MACT
to
be
no
less
stringent
than
the
average
emission
limitation
achieved
by
the
best­
performing
12
percent
of
existing
sources
among
the
data
available
to
the
Administrator.
For
new
major
sources,
the
Act
requires
MACT
to
be
no
less
stringent
than
the
emission
control
that
is
achieved
in
practice
by
the
bestcontrolled
similar
source.
These
minimum
stringency
levels
are
often
referred
to
as
the
"
MACT
floor."

The
term
"
average",
as
it
pertains
to
MACT
floor
determinations
for
existing
sources,

described
in
section
112(
d)(
3)
of
the
Act,
is
not
defined
in
the
statute.
In
a
Federal
Register
notice
published
on
June
6,
1994
(
59
FR
29196),
the
EPA
announced
its
conclusion
that
Congress
intended
"
average"
as
used
in
section
112(
d)(
3)
to
mean
a
measure
of
mean,
median,

mode,
or
some
other
measure
of
central
tendency.
The
EPA
concluded
that
it
retains
substantial
discretion
within
the
statutory
framework
to
set
MACT
floors
at
appropriate
levels,
and
that
it
construes
the
word
"
average"
(
as
used
in
section
112(
d)(
3))
to
authorize
the
EPA
to
use
any
reasonable
method,
in
a
particular
factual
context,
of
determining
the
central
tendency
of
a
data
set.

3.0
DATA
SOURCES
Various
sources
of
data
were
used
in
the
MACT
floor
analysis
for
boilers
and
process
heaters.
The
boiler
and
process
heater
population
database
was
used
to
characterize
the
number
and
types
of
existing
units,
the
types
of
fuels
burned,
the
capacity
of
the
units,
the
types
of
existing
add­
on
control
technologies,
and
the
locations
of
these
units.
This
database
includes
information
on
approximately
42,000
boilers
and
15,000
process
heaters.
The
development
of
this
database
is
discussed
in
the
memorandum
"
Development
of
the
Population
Database
for
the
Industrial,
Commercial,
and
Institutional
Boiler
and
Process
Heater
National
Emission
Standard
for
Hazardous
Air
Pollutants
(
NESHAP)".
1
The
boiler
emissions
test
database
was
used
in
correlation
with
the
population
database
to
characterize
the
type
and
magnitude
of
hazardous
air
pollutants
(
HAP)
that
are
emitted
from
various
types
of
combustion
units
that
burn
different
fuel
combinations
and
have
different
levels
and
types
of
existing
add­
on
control
technologies.
The
development
of
the
emissions
test
4
dwcgi­
7150­
1077822954­
148160000.
wpd
database
is
discussed
in
detail
in
the
memorandum
"
Development
of
the
Emissions
Test
Database
for
the
Industrial,
Commercial,
and
Institutional
Boiler
National
Emission
Standard
for
Hazardous
Air
Pollutants
(
NESHAP)".
2
In
addition,
emission
data
submitted
during
the
public
comment
period
were
included
in
the
final
analysis,
where
appropriate.

Other
sources
of
data
were
reviewed
to
assess
the
performance
of
various
types
of
add­
on
control
devices.
The
sources
reviewed
and
the
conclusions
drawn
from
this
review
regarding
the
performance
and
applicability
of
add­
on
control
techniques
to
the
combustion
units
included
in
this
source
category
are
discussed
in
the
memorandum
"
Methodology
for
Estimating
Cost
and
Emissions
Impacts
for
Industrial,
Commercial,
Institutional
Boilers
and
Process
Heaters
National
Emission
Standards
for
Hazardous
Air
Pollutants".
3
Another
data
source
used
during
the
MACT
floor
analysis
was
regulations
that
pertain
to
boilers
and
process
heaters
from
various
state
air
pollution
control
agencies.
Regulations
pertaining
to
these
sources
were
reviewed
for
all
states
that
had
rules
that
apply
to
combustion
sources.

4.0
AFFECTED
SOURCE
AND
SUBCATEGORIES
4.1
Description
of
Affected
source
This
MACT
includes
the
industrial
boilers,
institutional
and
commercial
boilers,
and
process
heaters
source
categories.
The
definition
of
affected
source
has
been
revised
based
on
public
comments
to
be
the
collection
of
existing
industrial,
commercial,
or
institutional
boilers
and
process
heaters
located
at
a
major
source
facility.
Process
heaters
are
defined
as
units
in
which
the
combustion
gases
do
not
directly
come
into
contact
with
process
gases
in
the
combustion
chamber
(
e.
g.
indirect
fired).
Boiler
means
an
enclosed
device
using
controlled
flame
combustion
and
having
the
primary
purpose
of
recovering
thermal
energy
in
the
form
of
steam
or
hot
water.

Because
facilities
could
have
multiple
boilers
and
process
heaters
on­
site
that
burn
different
types
of
fuels
and
have
different
levels
of
add­
on
controls,
the
MACT
floor
is
determined
by
evaluating
emissions
and
feasability
of
controls
separately
for
particular
subcategories
of
units
within
the
affected
source.
A
major
source
of
HAP
emissions
is
any
stationary
source
or
group
of
stationary
sources
located
within
a
contiguous
area
and
under
common
control
that
emits
or
has
the
potential
to
emit
any
single
HAP
at
a
rate
of
10
tons
or
more
per
year
or
any
combination
of
HAP
5
dwcgi­
7150­
1077822954­
148160000.
wpd
at
a
rate
of
25
tons
or
more
a
year.
The
affected
source
does
not
include
those
units
in
Section
2.0
that
are
excluded
from
the
source
category.

A
wide
variety
of
pollutants
may
be
emitted
from
boilers
and
process
heaters,
including
HAP's,
VOC's,
and
criteria
pollutants.
The
HAP's
emitted
from
boilers
and
process
heaters
can
be
categorized
as
either
inorganic
HAP
(
primarily
acid
gases
such
as
hydrogen
chloride
or
hydrogen
fluoride),
organic
HAP's
(
such
as
benzene
or
PAH's),
and
metallic
HAP
(
such
as
mercury
or
lead).
Due
to
its
health
affects
and
different
emission
characteristics,
mercury
is
often
analyzed
separately
from
non­
mercury
metallic
HAPs.
The
types
and
amounts
of
pollutants
emitted
from
these
sources
depends
greatly
on
the
type
of
fuel
being
burned
in
the
combustion
device.

4.2
Subcategories
The
Act
allows
source
categories
to
be
divided
into
subcategories
when
differences
between
given
types
of
units
lead
to
corresponding
differences
in
the
nature
of
emissions
and
the
technical
feasibility
of
applying
emission
control
techniques.
The
design,
operating,
and
emissions
information
that
EPA
has
reviewed
indicate
the
need
to
subcategorize
boilers
and
process
heaters
based
on
the
physical
state
of
the
fuel
burned,
i.
e.,
solid,
liquid,
or
gas.
Data
indicate
that
there
are
significant
design
and
operational
differences
between
units
that
burn
solid,
liquid
and
gaseous
fuels.

Boiler
systems
are
designed
for
specific
fuel
types
and
will
encounter
problems
if
a
fuel
with
characteristics
other
than
those
originally
specified
is
fired.
While
many
boilers
in
the
population
database
are
indicated
to
co­
fire
liquids
or
gases
with
solid
fuels,
in
actuality
most
of
these
commonly
use
fuel
oil
or
natural
gas
as
a
startup
fuel
only.
Other
co­
fired
units
are
specifically
designed
to
fire
combinations
of
solids,
liquids,
and
gases.
Changes
to
the
fuel
type
(
solid,
liquid,
or
gas)
would
require
extensive
changes
to
the
fuel
handling
and
feeding
system
(
e.
g.,
a
stoker
using
wood
as
fuel
would
need
to
be
redesigned
to
handle
fuel
oil
or
gaseous
fuel).

Additionally,
the
burners
and
combustion
chamber
would
need
to
be
redesigned
and
modified
to
handle
different
fuel
types
and
account
for
increases
or
decreases
in
the
fuel
volume
and
shape.
In
some
cases,
the
changes
may
reduce
the
capacity
and
efficiency
of
the
boiler
or
process
heater.

An
additional
effect
of
these
changes
would
be
extensive
retrofit
costs.

Emissions
from
boilers
and
process
heaters
burning
solids,
liquids,
and
gaseous
fuels
will
also
differ.
Boilers
and
process
heaters
emit
a
number
of
different
types
of
HAP
emissions.
In
6
dwcgi­
7150­
1077822954­
148160000.
wpd
general,
their
formation
is
dependent
upon
the
composition
of
the
fuel.
The
combustion
quality
and
temperature
may
also
play
an
important
role.
The
fuel
dependent
HAP
emissions
from
boilers
and
process
heaters
are
metals,
including
mercury,
and
acid
gases.
These
fuel
dependent
HAP
emissions
generally
can
be
controlled
by
either
changing
the
fuel
property
before
combustion
or
by
removing
the
HAP
from
the
flue
gas
after
combustion.
Organic
HAP,
on
the
other
hand,
are
formed
from
incomplete
combustion
and
are
much
less
influenced
by
the
characteristics
of
the
fuel
being
burned.
The
degree
of
combustion
may
be
greatly
influenced
by
three
general
factors:
time,

turbulence,
and
temperature.
These
factors
are
a
function
of
the
design
of
the
boiler
or
process
heater
which
is
dependent
in
part
on
the
type
of
fuel
being
burned.
The
different
emission
characteristics
will
affect
the
type
of
air
pollution
controls
that
may
be
used.
Accordingly,
the
source
category
was
divided
into
three
subcategories
to
consider
these
differences:
solid
fuel­
fired
units,
liquid
fuel­
fired
units,
and
gaseous
fuel­
fired
units.
The
solid
subcategory
includes
units
that
burn
any
amount
of
solid
fuel.
The
gaseous
subcategory
includes
units
that
only
burn
gaseous
fuel,
except
during
periods
of
natural
gas
curtailment.
The
liquid
subcategory
includes
the
remaining
units.

Another
factor
that
affects
emissions
from
boilers
and
process
heaters
is
the
combustor
design.
The
combustor
design
influences
the
completeness
of
the
combustion
process
and
the
formation
of
organic
compounds.
Boilers
with
capacities
less
than
10
MMBtu/
hr
use
combustor
designs
(
e.
g.,
firetube
or
cast­
iron)
which
are
not
common
in
units
above
10
MMBtu/
hr.
Large
boilers
generally
are
field­
erected
using
watertube
combustor
design
with
capacities
above
10
MMBtu/
hr.
The
vast
majority
of
these
small
units
use
natural
gas
as
fuel.
Additionally,
most
existing
State
and
Federal
regulations
for
boilers
and
process
heaters
do
not
regulate
units
with
a
heat
input
capacity
of
less
than
10
MMBtu/
hr,
due
to
their
low
emissions.
Accordingly,
the
three
subcategories
were
further
divided
into
large
units
(
watertube
boilers
and
process
heaters
>
10
MMBtu/
hr
capacity)
and
small
units
(
all
firetube
boilers
and
process
heaters

10
MMBtu/
hr
capacity)
to
differentiate
the
combustor
designs
typically
found
in
these
size
ranges.

A
third
subcategory
classification
was
also
considered
to
distinguish
units
that
are
operated
infrequently,
such
as
back­
up
or
emergency
units.
Back­
up
or
emergency
units
only
operate
if
another
boiler
that
is
the
regular
source
of
energy
or
steam
is
not
operating
(
for
example
due
to
a
shutdown
for
maintenance
and
repair).
Peaking
units
operate
only
during
peak
energy
use
periods,
typically
in
the
summer
months.
The
boiler
database
indicates
that
these
7
dwcgi­
7150­
1077822954­
148160000.
wpd
infrequently
operated
units
typically
operate
10
percent
of
the
year
or
less.
These
limited
use
boilers,
when
called
upon
to
operate,
must
respond
without
failure
and
without
lengthy
periods
of
startup.
This
subcategorization
was
made
because
the
limited
use
units,
those
with
capacity
utilizations
less
than
10
percent,
have
a
specialized
use
and
operation
that
are
different
from
typical
industrial,
commercial,
and
institutional
units.

Thus,
a
total
of
nine
subcategories
were
developed
for
this
source
category:
(
1)
large
solid
fuel­
fired
boilers
and
process
heaters,
(
2)
large
liquid
fuel­
fired
boilers
and
process
heaters,

(
3)
large
gaseous
fuel­
fired
boilers
and
process
heaters,
(
4)
limited
use
solid
fuel­
fired
boilers
and
process
heaters,
(
5)
limited
use
liquid
fuel­
fired
boilers
and
process
heaters,
(
6)
limited
use
gaseous
fuel­
fired
boilers
and
process
heaters,
(
7)
small
solid
fuel­
fired
boilers
and
process
heaters,
(
8)
small
liquid
fuel­
fired
boilers
and
process
heaters,
and
(
9)
small
gaseous
fuel­
fired
boilers
and
process
heaters.
Because
these
subcategories
were
defined
based
on
fundamental
differences
in
the
types
of
emissions,
all
MACT
floor
analyses
were
done
separately
for
each
individual
subcategory.

5.0
GENERAL
METHODOLOGY
FOR
MACT
FLOOR
ANALYSIS
Many
approaches
were
considered
for
determining
the
MACT
floor
including
use
of
emissions
data
only,
use
of
state
regulations
and
permits,
review
of
possible
process
changes,
and
review
of
add­
on
controls.
The
limitations
of
the
data
available
resulted
in
some
of
these
approaches
not
being
appropriate
options
for
developing
the
MACT
floor.
Consequently,
the
most
appropriate
approach
for
determining
MACT
floors
for
boilers
and
process
heaters
is
to
look
at
the
control
options
used
by
the
units
within
each
subcategory
in
order
to
identify
the
best
performing
units.
The
methodology
used
consisted
of
using
information
on
controls
from
the
population
database,
emissions
from
the
emissions
database
and
public
comments,
and
State
regulations.
The
consideration
of
the
approaches
that
were
not
used
is
discussed
below.
The
consideration
of
process
changes
or
work
practices
is
discussed
in
Section
7.0.

The
first
step
in
the
methodology
was
to
identify
the
control
technologies
used
by
the
best­
controlled
sources
in
each
subcategory
for
controlling
four
classes
of
pollutants:

nonmercury
metallic
HAP,
mercury,
inorganic
HAP,
and
organic
HAP.
The
population
database
was
used
to
determine
the
existing
numbers
and
types
of
boilers
and
process
heaters
with
the
best
technologies
used
to
control
these
HAP
emissions.
The
database
contains
specific
information
on
8
dwcgi­
7150­
1077822954­
148160000.
wpd
the
types
of
control
devices
that
are
present
for
most
of
the
boilers
and
process
heaters.

However,
it
does
not
include
any
unit­
specific
data
on
the
control
device
design
or
the
actual
emissions
from
the
individual
combustion
units
in
the
population
database.
The
emission
limits
for
each
class
of
pollutant
associated
with
the
best
control
technologies
were
then
determined
using
information
in
the
emissions
database
and
any
additional
emission
data
obtained
during
the
public
comment
period.
Limits
were
identified
in
units
of
pound
of
pollutant
per
million
Btu
(
lb/
MMBtu)
of
heat
input
to
be
consistent
with
the
format
of
the
New
Source
Performance
Standard
(
NSPS)
for
industrial
boilers
as
well
as
other
existing
boiler
regulations.

5.1
Consideration
of
Emission
Test
Data
Only
Under
one
approach,
the
MACT
floor
for
a
category
of
sources
could
be
calculated
by
ranking
the
emission
test
results
from
units
within
the
category
from
lowest
to
highest,
and
then
taking
the
numerical
average
of
the
test
results
from
the
best
performing
(
lowest
emitting)
12
percent
of
sources.

However,
review
of
the
available
HAP
emission
test
data
indicated
several
problems
with
using
this
MACT
floor
approach
to
establish
emission
limits
for
boilers
and
process
heaters.
First,

the
emissions
database
is
very
limited
for
HAP
emissions
from
industrial
boilers.
Prior
to
proposal
and
during
the
Industrial
Combustion
Coordinated
Rulemaking
(
ICCR)
process,
EPA
conducted
a
thorough
search
for
HAP
emission
test
reports.
This
search
was
supported
by
industry,
trade
groups,
and
States.
For
criteria
pollutants,
such
as
PM,
substantial
emission
information
was
available
and
gathered.
For
HAP,
this
was
not
the
case.
Industrial
boilers
have
not
generally
been
required
to
test
for
HAP
emissions.
In
the
proposed
rule,
we
requested
commenters
to
provide
additional
emissions
information.
However,
only
one
source
provided
any
additional
emissions
data
(
mercury
test
results
from
three
additional
coal­
fired
industrial
boilers).

The
main
problem
with
using
only
the
HAP
emissions
data
is
that,
based
on
the
test
data
alone,

uncontrolled
units
(
or
units
with
low
efficiency
add­
on
controls)
were
frequently
identified
as
being
among
the
best
performing
12
percent
of
sources
in
a
subcategory,
while
many
units
with
high
efficiency
controls
were
not.
However,
these
uncontrolled
or
poorly
controlled
units
are
not
truly
among
the
best
controlled
units
in
the
category.
Rather,
the
emissions
from
these
units
are
relatively
low
because
of
particular
characteristics
of
the
fuel
that
they
burn,
that
can
not
reasonably
be
replicated
by
other
units
in
the
category
or
subcategory.
This
kind
of
variability
in
9
dwcgi­
7150­
1077822954­
148160000.
wpd
emission
rates
is
expected
given
the
variety
of
fuel
types
included
within
each
subcategory
of
boilers
and
process
heaters.

A
review
of
fuel
analyses
indicate
that
the
concentration
of
HAP
(
metals,
HCl,
mercury)

can
vary
greatly,
not
only
between
fuel
types,
but
also
within
each
fuel
type.
Some
fuels
even
have
pollutant
concentration
levels
below
the
detection
limit
of
the
applicable
analytical
test
method.
Therefore,
a
unit
without
any
add­
on
controls,
but
burning
a
fuel
containing
lower
amounts
of
HAP,
can
have
emission
levels
that
are
lower
than
the
emissions
from
a
unit
with
the
best
available
add­
on
controls.
If
only
the
available
HAP
emissions
data
are
used,
the
resulting
MACT
floor
levels
would,
in
most
cases,
be
unachievable
for
many,
if
not
most,
existing
units,

even
those
that
employ
the
most
effective
available
emission
control
technology.
For
example,
an
uncontrolled
boiler
burning
wood
may
have
lower
emissions
of
mercury
than
a
well
controlled
boiler
burning
coal.
This
would
result
in
some
coal
burning
boilers
never
being
able
to
achieve
the
mercury
HAP
level
of
the
wood­
fired
unit,
no
matter
what
add­
on
controls
are
used.
In
this
instance,
establishing
a
MACT
standard
based
on
emission
data
alone
would
force
the
coal
units
to
switch
to
different
fuels
to
achieve
the
MACT
limits.

Another
problem
with
using
only
emissions
data
is
that
there
is
no
HAP
emissions
information
for
some
subcategories.
This
is
consistent
with
the
fact
that
units
in
these
source
categories
have
not
historically
been
required
to
test
for
HAP
emissions.

5.2
Consideration
of
State
Regulations
and
Permits
HAP
emission
limits
contained
in
State
regulations
and
permits
were
also
reviewed
as
a
surrogate
for
actual
emission
data
in
order
to
identify
the
emissions
levels
from
the
best
performing
units
in
the
category
for
purposes
of
establishing
MACT
standards.
However,
no
State
regulations
or
State
permits
were
found
which
specifically
limit
HAP
emissions
from
these
sources.

5.3
Consideration
of
Fuel
Switching
Fuel
switching
was
examined
as
an
appropriate
control
option
for
sources
in
each
subcategory.
The
feasibility
of
both
fuel
switching
to
other
fuels
used
in
the
subcategory
and
to
fuels
from
other
subcategories
were
considered.
This
consideration
included
determining
whether
switching
fuels
would
achieve
lower
HAP
emissions.
A
second
consideration
was
whether
fuel
switching
could
be
technically
achieved
by
boilers
and
process
heaters
in
the
subcategory
10
dwcgi­
7150­
1077822954­
148160000.
wpd
considering
the
existing
design
of
boilers
and
process
heaters.
The
availability
of
various
types
of
fuel
was
also
reviewed.

After
considering
these
factors,
fuel
switching
was
determined
to
not
be
an
appropriate
control
technology
for
purposes
of
determining
the
MACT
floor
level
of
control
for
any
subcategory.
This
decision
was
based
on
the
overall
effect
of
fuel
switching
on
HAP
emissions,

technical
and
design
considerations
discussed
earlier,
and
concerns
about
fuel
availability.

Data
available
in
the
emissions
database
indicates
that
while
fuel
switching
from
solid
fuels
to
gaseous
or
liquid
fuels
would
decrease
PM
and
some
metals
emissions,
emissions
of
some
organic
HAP
would
increase,
resulting
in
uncertain
benefits.
This
determination
is
discussed
in
the
memorandum
"
Development
of
Fuel
Switching
Costs
and
Emission
Reductions
for
Industrial/
Commercial/
Institutional
Boilers
and
Process
Heaters
National
Emission
Standards
for
Hazardous
Air
Pollutants".
4
In
order
to
adopt
such
a
strategy,
the
relative
risk
associated
which
each
HAP
emitted
would
need
to
be
analyzed,
as
well
as
whether
requiring
the
control
in
question
would
result
in
overall
lower
risk.

A
similar
determination
was
made
when
considering
fuel
switching
to
cleaner
fuels
within
a
subcategory.
For
example,
the
term
"
clean
coal"
refers
to
coal
that
is
lower
in
sulfur
content
and
not
necessarily
lower
in
HAP
content.
Data
gathered
also
indicates
that
within
specific
coal
types
HAP
content
can
vary
significantly.
Switching
to
a
low
sulfur
coal
may
actually
increase
emissions
of
some
HAP.
Therefore,
fuel
switching
to
a
low
sulfur
coal
as
part
of
the
MACT
standards
for
boilers
and
process
heaters
could
not
be
included
in
the
analysis.
Fuel
switching
from
coal
to
biomass
would
result
in
similar
impacts
on
HAP
emissions.
While
this
would
reduce
metallic
HAP
emissions,
it
would
likely
increase
emissions
of
organics
based
on
information
in
the
emissions
database.

Another
factor
considered
was
the
availability
of
alternative
fuel
types.
Natural
gas
pipelines
are
not
available
in
all
regions
of
the
U.
S.,
and
natural
gas
is
simply
not
available
as
a
fuel
for
many
industrial,
commercial,
and
institutional
boilers
and
process
heaters.
Moreover,

even
where
pipelines
provide
access
to
natural
gas,
supplies
of
natural
gas
may
not
be
adequate.

For
example,
it
is
common
practice
in
cities
during
winter
months
(
or
periods
of
peak
demand)
to
prioritize
natural
gas
usage
for
residential
areas
before
industrial
usage.
Consequently,
even
where
pipelines
exist
some
units
would
not
be
able
to
run
at
normal
or
full
capacity
during
these
11
dwcgi­
7150­
1077822954­
148160000.
wpd
times
if
shortages
were
to
occur.
Therefore,
under
any
circumstances,
there
would
be
some
units
that
could
not
comply
with
a
requirement
to
switch
to
natural
gas.

Similar
problems
for
fuel
switching
to
biomass
could
arise.
Existing
sources
burning
biomass
generally
are
combusting
a
recovered
material
from
the
manufacturing
or
agriculture
process.
Industrial,
commercial,
and
institutional
facilities
that
are
not
associated
with
the
wood
products
industry
or
agriculture
may
not
have
access
to
a
sufficient
supply
of
biomass
materials
to
replace
their
fossil
fuel.

There
is
also
a
significant
concern
that
switching
fuels
would
be
infeasible
for
sources
designed
and
operated
to
burn
specific
fuel
types.
Changes
in
the
type
of
fuel
burned
by
a
boiler
or
process
heater
(
solid,
liquid,
or
gas)
may
require
extensive
changes
to
the
fuel
handling
and
feeding
system
(
e.
g.,
a
stoker
using
wood
as
fuel
would
need
to
be
redesigned
to
handle
fuel
oil
or
gaseous
fuel).
Additionally,
burners
and
combustion
chamber
designs
are
generally
not
capable
of
handling
different
fuel
types,
and
generally
cannot
accommodate
increases
or
decreases
in
the
fuel
volume
and
shape.
Design
changes
to
allow
different
fuel
use,
in
some
cases,
may
reduce
the
capacity
and
efficiency
of
the
boiler
or
process
heater.
Reduced
efficiency
may
result
in
less
complete
combustion
and,
thus,
an
increase
in
organic
HAP
emissions.

6.0
DETERMINATION
OF
THE
MACT
FLOOR
BASED
ON
CONTROL
TECHNIQUES
6.1
Identification
of
Typical
Add­
on
Control
Devices
in
Population
Database
The
initial
step
for
the
MACT
floor
analysis
based
on
control
technologies
was
to
identify
the
typical
types
of
add­
on
control
technologies
used
on
existing
boilers
and
process
heaters
in
the
population
database.
The
population
database
sometimes
includes
specific
descriptions
regarding
the
types
of
add­
on
devices
that
are
on
the
combustion
units.
These
specific
control
devices
in
the
population
database
were
grouped
into
more
general
control
device
categories
in
order
to
simplify
the
analysis.
For
instance,
high
temperature
and
low
temperature
fabric
filters
were
grouped
into
a
general
fabric
filter
category.
Also,
control
techniques
listed
in
the
population
database
that
were
assumed
to
have
no
effect
on
HAP
emissions,
such
as
low
NOx
burners
or
fuel
air
recirculation,
were
not
considered
in
these
control
device
groupings.
Because
many
of
the
specific
control
devices
listed
in
the
population
database
were
assumed
to
achieve
similar
control
12
dwcgi­
7150­
1077822954­
148160000.
wpd
efficiencies,
this
grouping
process
did
not
result
in
a
less
accurate
MACT
floor
analysis.
The
control
device
groupings
are
presented
in
Appendix
A­
1.

6.2
Control
Technology
Assessments
Once
the
types
of
existing
add­
on
control
devices
were
determined
and
grouped
into
more
general
control
categories,
the
technologies
were
ranked
in
terms
of
their
relative
performance.

The
rankings
for
each
control
device
category
were
based
on
the
typical
control
efficiency
each
was
expected
to
achieve.
The
memorandum
"
Methodology
for
Estimating
Cost
and
Emissions
Impacts
for
Industrial,
Commercial,
Institutional
Boilers
and
Process
Heaters
National
Emission
Standards
for
Hazardous
Air
Pollutants"
3
discusses
typical
efficiencies
assigned
to
the
control
devices.
The
rankings
were
assigned
as
follows:
ranking
of
"
1"
means
the
control
device
can
achieve
greater
than
99%
control
efficiency,
ranking
of
"
2"
means
greater
than
98%
control
efficiency,
ranking
of
"
3"
means
greater
than
90%
control
efficiency,
ranking
of
"
4"
means
greater
than
75%
control
efficiency,
ranking
of
"
5"
means
greater
than
50%
control
efficiency,
ranking
of
"
6"
means
greater
than
30%
control
efficiency,
ranking
of
"
7"
means
less
than
30%
control
efficiency,
and
a
ranking
of
"
8"
means
that
the
control
device
achieves
no
control.
The
control
devices
were
ranked
in
this
manner
by
relative
control
efficiencies
individually
for
each
of
the
pollutant
categories
(
inorganic
HAP,
organic
HAP,
non­
mercury
metallic
HAP,
and
mercury)

because
the
most
effective
control
devices
for
each
of
these
pollutant
categories
are
sometimes
different.
For
example,
ESP's
are
effective
in
controlling
metallic
HAP
emissions,
but
are
ineffective
in
controlling
organic
HAP
or
inorganic
HAP
emissions.

6.3
Determination
of
the
Best­
performing
Sources
based
on
Control
Technologies
The
boilers
and
process
heaters
in
the
population
database
in
each
subcategory
were
ranked
based
on
their
controls
in
order
of
decreasing
control
effectiveness
for
each
of
the
pollutant
categories.
That
is,
the
boilers
and
process
heaters
in
each
subcategory
were
ranked
separately
for
each
of
these
pollutant
categories
according
to
the
units
that
have
the
bestperforming
controls
for
each
specific
type
of
pollutant.
The
best­
performing
12
percent
of
sources
for
existing
sources
or
best­
performing
"
similar
source"
for
new
sources
was
identified
for
each
of
these
pollutant
categories
separately.

Once
the
control
device
categories
were
ranked
along
with
the
number
of
units
in
each
category,
the
percentage
of
units
with
the
best­
performing
control
devices
was
determined
for
13
dwcgi­
7150­
1077822954­
148160000.
wpd
each
pollutant
category.
This
calculation
was
done
by
dividing
the
number
of
units
with
a
ranked
control
device
by
the
total
number
of
units.
However,
the
percentage
of
units
with
each
type
of
control
device
was
based
only
on
the
population
of
units
for
which
control
device
information
was
available.
The
population
database,
on
which
this
analysis
is
based,
does
not
have
control
information
available
for
every
boiler
and
process
heater.
Often
the
control
device
database
field
specified
a
particular
control
device
or
combination
of
control
devices
on
a
unit,
sometimes
the
field
specified
that
a
unit
had
no
control
devices,
but
sometimes
the
database
field
was
blank.

These
units
with
blank
control
information
data
fields
were
excluded
from
the
MACT
floor
analysis
because
using
them
would
have
required
that
broad
assumptions
be
made
about
the
types
of
controls
that
might
be
on
these
units.
Therefore,
the
MACT
floor
analysis
is
actually
done
using
a
subset
of
the
population
database
that
is
assumed
to
be
representative
of
the
entire
population.
The
summary
tables
in
Appendix
A
show
the
number
of
units
with
"
no
information"

for
each
of
the
subcategories.

For
new
sources,
the
best­
performing
control
devices
in
each
subcategory
are
those
ranked
with
the
highest
removal
efficiency
for
each
pollutant.
For
existing
sources,
the
bestperforming
12
percent
of
sources
needed
to
be
identified.
Once
the
control
device
categories
were
ranked
from
best­
performing
to
worst­
performing
for
each
subcategory
and
pollutant
category
by
the
control
rankings,
and
the
percentages
of
units
using
each
control
were
calculated,

the
cumulative
percentage
of
units
represented
was
reviewed
to
determine
the
best­
performing
12
percent
of
units.
The
median
unit
in
the
best­
performing
12
percent
of
units
(
i.
e.,
the
boiler
or
process
heater
unit
representing
the
94th
percentile)
was
used
to
represent
the
technology
associated
with
the
MACT
floor
level
of
control
for
each
subcategory.
Because
the
control
device
rankings
were
done
using
a
scale
from
1
to
8
based
on
control
efficiencies,
different
control
device
categories
might
have
the
same
efficiency
ranking
for
a
pollutant
category.

Because
there
is
no
distinction
in
performance
between
control
devices
with
the
same
efficiency
ranking,
if
the
six
percent
level
occurred
in
the
middle
of
a
control
device
category
ranking,
then
all
sources
that
had
existing
controls
ranked
at
that
level
or
better
were
included
in
the
group
of
units
that
were
considered
to
be
the
best­
performing
12
percent
of
sources.

The
summary
tables
in
Appendix
A
indicate
which
units
and
control
device
categories
are
included
in
the
best­
performing
12
percent
of
sources
for
each
subcategory
and
pollutant
14
dwcgi­
7150­
1077822954­
148160000.
wpd
category.
Table
6­
1
summarizes
the
results
of
the
MACT
floor
control
technologies
analysis.
A
discussion
of
the
results
for
each
subcategory
is
presented
in
the
following
sections.

6.4
Best
Performing
Control
Technologies
for
Existing
Sources
6.4.1
Existing
Solid
Fuel
Boilers
and
Process
Heaters
Large
Units
­
Heat
Inputs
Greater
than
10
MMBtu/
hr.
The
most
effective
control
technologies
identified
for
removing
non­
mercury
metallic
HAP
are
fabric
filters.
About
14
percent
of
solid
fuel­
fired
boilers
and
process
heater
use
fabric
filters.
The
most
effective
control
technologies
identified
for
removing
inorganic
HAP
that
are
acid
gases,
such
as
HCl,
are
wet
scrubbers
and
packed
bed
scrubbers.
These
technologies
are
used
by
about
13
percent
of
the
boilers
and
process
heaters
in
the
large
solid
fuel
subcategory.
About
12
percent
of
solid
fuelfired
boilers
and
process
heaters
use
wet
or
dry
scrubbers,
and
approximately
1
percent
use
packed
bed
scrubbers.
Based
on
test
information
on
utility
boilers,
fabric
filters
are
determined
to
be
the
most
effective
technology
for
controlling
mercury
emissions.
3
As
discussed
previously,

approximately
14
percent
of
sources
in
the
subcategory
use
fabric
filters.
No
add­
on
control
technologies
that
would
reduce
organic
HAP
emissions
were
identified
as
being
used.

Therefore,
the
combination
of
fabric
filter
and
wet
scrubber
control
technologies
forms
the
basis
for
the
MACT
floor
level
of
control
for
existing
large
solid
fuel
boilers
or
process
heaters.

This
analysis
is
shown
in
Appendix
A­
2.

Small
Units
­
Heat
Inputs
Less
than
or
Equal
to
10
MMBtu/
hr.
For
each
pollutant
group
(
non­
mercury
metallic
HAP,
mercury,
inorganic
HAP/
HCl,
and
organic
HAP),
less
than
6
percent
of
the
units
in
this
subcategory
used
control
techniques
that
limit
emissions.
This
analysis
is
shown
in
Appendix
A­
3.

Limited
Use
Units
­
Capacity
Utilizations
Less
than
or
Equal
to
10
Percent.
The
most
effective
control
technologies
identified
for
removing
non­
mercury
metallic
HAP
are
ESP
and
fabric
filters.
Less
than
2
percent
of
limited
use
solid
fuel­
fired
boilers
and
process
heater
use
fabric
filters,
and
14
percent
use
ESP.

Similar
control
technology
analyses
were
done
for
the
boilers
and
process
heaters
in
this
subcategory
for
inorganic
HAP,
organic
HAP
and
mercury.
For
each
of
these
pollutant
groups,

less
than
6
percent
of
the
units
in
this
subcategory
used
control
techniques
that
limit
emissions.

Consequently,
ESP
and
fabric
filters,
which
achieve
non­
mercury
metallic
HAP
control,
form
the
15
dwcgi­
7150­
1077822954­
148160000.
wpd
basis
for
the
MACT
floor
level
of
control
for
existing
solid
fuel
boilers
and
process
heaters
in
this
subcategory.
This
analysis
is
shown
in
Appendix
A­
4.

6.4.2
Existing
Liquid
Fuel
Boilers
and
Process
Heaters
Less
than
6
percent
of
the
units
in
each
of
the
liquid
subcategories
used
control
techniques
that
would
reduce
non­
mercury
metallic
HAP,
mercury,
organic
HAP,
or
acid
gases,
(
such
as
HCl).
This
analysis
is
shown
in
Appendices
A­
5
through
A­
7.

6.4.3
Existing
Gaseous
Fuel
Boilers
and
Process
Heaters
No
existing
units
in
the
gaseous
fuel­
fired
subcategories
were
using
control
technologies
that
achieve
consistently
lower
emission
rates
than
uncontrolled
sources
for
any
of
the
pollutant
groups
of
interest.
This
analysis
is
shown
in
Appendices
A­
8
through
A­
10.

6.5
Best
Performing
Control
Technologies
for
New
Sources
6.5.1
New
Solid
Fuel­
fired
Units
Large
Units
­
Heat
Inputs
Greater
than
10
MMBtu/
hr.
The
most
effective
control
technology
identified
for
removing
non­
mercury
metallic
HAP
are
fabric
filters.
The
most
effective
control
technologies
identified
for
removing
inorganic
HAP
including
acid
gases,
such
as
HCl,
are
wet
or
dry
scrubbers.
Wet
scrubbers
is
a
generic
term
that
is
most
often
used
to
describe
venturi
scrubbers,
but
can
include
packed
bed
scrubbers,
impingement
scrubbers,
etc.
One
percent
of
boilers
and
process
heaters
in
this
subcategory
reported
using
a
packed
bed
scrubber.

Emission
test
data
from
other
industries
suggests
that
packed
bed
scrubbers
achieve
consistently
lower
emission
levels
than
other
types
of
wet
scrubbers.

For
mercury
control,
one
technology,
carbon
injection,
that
has
demonstrated
mercury
reductions
in
other
source
categories
(
i.
e.,
municipal
waste
combustors),
was
identified
as
being
used
at
one
existing
industrial
boiler
facility.
However,
test
data
on
this
carbon
injection
system
indicated
that
this
unit
was
not
achieving
mercury
emissions
reductions.
Therefore,
carbon
injection
was
not
considered
to
be
a
MACT
floor
control
technology
for
industrial,
commercial,

and
institutional
boilers
and
process
heaters.
Data
from
electric
utility
boilers
indicate
that
fabric
filters
are
the
most
effective
technology
for
controlling
mercury
emissions.
No
add­
on
control
technologies
that
would
reduce
organic
HAP
emissions
were
identified
as
being
used
on
units
in
this
subcategory.
16
dwcgi­
7150­
1077822954­
148160000.
wpd
The
combination
of
a
fabric
filter
and
a
packed
bed
scrubber
forms
the
technology
basis
for
the
MACT
floor
level
of
control
for
new
solid
fuel
boilers
and
process
heaters
in
this
subcategory.
See
Appendix
A­
2.

Small
Units
­
Heat
Inputs
Less
than
or
Equal
to
10
MMBtu/
hr.
The
most
effective
control
technology
identified
for
removing
nonmercury
metallic
HAP
are
fabric
filters.
The
most
effective
control
technology
identified
for
units
in
this
subcategory
for
removing
acid
gases,
such
as
HCl,
are
wet
scrubbers.
The
most
effective
control
technology
identified
for
removing
mercury
is
fabric
filters.
No
add­
on
control
technologies
that
would
reduce
organic
HAP
emissions
were
identified
as
being
used
on
units
in
this
subcategory.

The
combination
of
a
fabric
filter
and
a
wet
scrubber
forms
the
technology
basis
for
the
MACT
floor
level
of
control
for
new
solid
fuel
boilers
and
process
heaters
in
this
subcategory.

See
Appendix
A­
3.

Limited
Use
Units
­
Capacity
Utilizations
Less
than
or
Equal
to
10
Percent.
The
most
effective
control
technology
identified
for
removing
non­
mercury
metallic
HAP
and
mercury
are
fabric
filters.
The
most
effective
control
technology
identified
for
units
in
this
subcategory
for
removing
acid
gases,
such
as
HCl,
are
wet
scrubbers.
No
add­
on
control
technologies
that
would
reduce
organic
HAP
emissions
were
identified
as
being
used
on
units
in
this
subcategory.

The
combination
of
a
fabric
filter
and
a
wet
scrubber
forms
the
technology
basis
for
the
MACT
floor
level
of
control
for
new
solid
fuel
boilers
and
process
heaters
in
this
subcategory.

See
Appendix
A­
4.

6.5.2
New
Liquid
Fuel­
fired
Units
Large
Units
­
Heat
Inputs
Greater
than
10
MMBtu/
hr.
The
most
effective
control
technology
identified
for
removing
non­
mercury
metallic
HAP
are
ESPs.
The
most
effective
control
technology
identified
for
removing
inorganic
HAP
that
are
acid
gases,
such
as
HCl,
are
packed
bed
scrubbers.
Information
in
the
emissions
database
or
from
other
source
categories
does
not
show
that
control
technologies,
such
as
fabric
filters,
ESP,
or
wet
scrubbers,
achieve
reductions
in
mercury
emissions
from
liquid
fuel­
fired
industrial,
commercial,
and
institutional
boilers
and
process
heaters.
No
add­
on
control
technology
being
used
in
the
existing
population
of
boilers
and
process
heaters
in
these
subcategories
that
consistently
achieved
lower
emission
rates
than
uncontrolled
levels,
such
that
a
best
controlled
similar
source
for
organic
HAP
could
be
identified.
17
dwcgi­
7150­
1077822954­
148160000.
wpd
The
combination
of
an
ESP
and
a
packed
bed
scrubber
forms
the
technology
basis
for
the
MACT
floor
level
of
control
for
new
liquid
fuel
boilers
and
process
heaters
in
this
subcategory.

See
Appendix
A­
5.

Small
Units
­
Heat
Inputs
Less
than
or
Equal
to
10
MMBtu/
hr.
The
most
effective
control
technology
identified
for
removing
non­
mercury
metallic
HAP
used
by
units
in
this
subcategory
are
ESPs.
The
most
effective
control
technology
identified
for
units
in
this
subcategory
for
removing
acid
gases,
such
as
HCl,
are
wet
scrubbers.

Information
in
the
emissions
database
or
from
other
source
categories
does
not
show
that
control
technologies,
such
as
fabric
filters,
ESP,
or
wet
scrubbers,
achieve
reductions
in
mercury
emissions
from
liquid
fuel­
fired
industrial,
commercial,
and
institutional
boilers
and
process
heaters.
No
add­
on
control
technology
being
used
in
the
existing
population
of
boilers
and
process
heaters
that
consistently
achieved
lower
emission
rates
than
uncontrolled
levels,
such
that
a
best
controlled
similar
source
for
mercury
or
organic
HAP
could
be
identified.

The
combination
of
a
fabric
filter
and
a
wet
scrubber
forms
the
technology
basis
for
the
MACT
floor
level
of
control
for
new
liquid
fuel
boilers
and
process
heaters
in
this
subcategory.

See
Appendix
A­
6.

Limited
Use
Units
­
Capacity
Utilizations
Less
than
or
Equal
to
10
Percent.
The
most
effective
control
technology
identified
for
removing
non­
mercury
metallic
HAP
used
by
units
in
this
subcategory
are
ESPs.
The
most
effective
control
technology
identified
for
units
in
this
subcategory
for
removing
acid
gases,
such
as
HCl,
are
wet
scrubbers.

Information
in
the
emissions
database
or
from
other
source
categories
does
not
show
that
other
control
technologies,
such
as
fabric
filters,
ESP,
or
wet
scrubbers,
achieve
reductions
in
mercury
emissions
from
liquid
fuel­
fired
industrial,
commercial,
and
institutional
boilers
and
process
heaters.
No
add­
on
control
technology
being
used
in
the
existing
population
of
boilers
and
process
heaters
that
consistently
achieved
lower
emission
rates
than
uncontrolled
levels,
such
that
a
best
controlled
similar
source
for
mercury
or
organic
HAP
could
be
identified.
See
Appendix
A­
7.

Gaseous
Fuel
Subcategories.
No
existing
units
were
using
control
technologies
that
achieve
consistently
lower
emission
rates
than
uncontrolled
sources
for
any
of
the
pollutant
groups
of
interest.
See
Appendices
A­
8
through
A­
10.
18
dwcgi­
7150­
1077822954­
148160000.
wpd
7.0
ANALYSIS
OF
WORK
PRACTICES
AND
PROCESS
CHANGES
Upon
review
of
the
emissions
test
data,
it
was
determined
that
no
control
technology
consistently
achieved
organic
HAP
emission
levels
any
lower
than
those
from
uncontrolled
boilers.
Therefore,
there
is
no
achievable
MACT
floor
emissions
level
that
can
be
established
for
the
organic
HAP
pollutant
category.

The
HAP
emissions
from
boilers
and
process
heaters
are
primarily
dependent
upon
the
composition
of
the
fuel.
Fuel
dependent
HAP
are
metals,
including
mercury,
and
acid
gases.
Fuel
dependent
HAP
are
typically
controlled
by
removing
them
from
the
flue
gas
after
combustion.

Therefore,
they
are
not
affected
by
the
operation
of
the
boiler
or
process
heater.
Consequently,

process
changes
would
be
ineffective
in
reducing
these
fuel­
related
HAP
emissions.

Organic
HAP
can
be
formed
from
incomplete
combustion
of
the
fuel.
Combustion
is
defined
as
the
rapid
chemical
combination
of
oxygen
with
the
combustible
elements
of
a
fuel.
The
objective
of
good
combustion
is
to
release
all
the
energy
in
the
fuel
while
minimizing
losses
from
combustion
imperfections
and
excess
air.
The
combination
of
the
fuel
with
the
oxygen
requires
temperature
(
high
enough
to
ignite
the
fuel
constituents),
mixing
or
turbulence
(
to
provide
intimate
oxygen­
fuel
contact),
and
sufficient
time
(
to
complete
the
process),
sometimes
referred
to
the
three
Ts
of
combustion.
Good
combustion
practice
(
GCP),
in
terms
of
boilers
and
process
heaters,
could
be
defined
as
the
system
design
and
work
practices
expected
to
minimize
organic
HAP
emissions.
The
GCP
control
strategy
could
include
a
number
of
combustion
conditions
and
work
practices
which
are
applied
collectively
to
achieve
this
goal.

While
few
sources
specifically
reported
using
good
combustion
practices,
boilers
and
process
heaters
within
each
subcategory
might
use
any
of
a
wide
variety
of
different
work
practices,
depending
on
the
characteristics
of
the
individual
unit.
The
lack
of
information,
and
lack
of
a
uniform
approach
to
assuring
combustion
efficiency,
is
not
surprising
given
the
extreme
diversity
of
boilers
and
process
heaters,
and
given
the
fact
that
no
applicable
Federal
standards,

and
most
applicable
State
standards,
do
not
include
work
practice
requirements
for
boilers
and
process
heaters.
Even
those
States
that
do
have
such
requirements
do
not
require
the
same
work
practices.

Consequently,
any
uniform
requirements
or
set
of
work
practices
that
would
meaningfully
reflect
the
use
of
good
combustion
practices,
or
that
could
be
meaningfully
implemented
across
any
subcategory
of
boilers
and
process
heaters
could
not
be
identified.
19
dwcgi­
7150­
1077822954­
148160000.
wpd
Additionally,
few
of
the
GCP's
have
been
documented
to
reduce
organic
HAP
emissions,

and
they
could
not
be
considered
in
the
MACT
analysis.
One
GCP
that
may
effect
organic
HAP
emissions
is
maintaining
CO
emission
levels.
CO
is
generally
an
indicator
of
incomplete
combustion
because
CO
will
burn
to
carbon
dioxide
if
adequate
oxygen
is
available.
Controlling
CO
emissions
is
a
mechanism
for
ensuring
combustion
efficiency,
and
therefore
may
be
viewed
as
a
kind
of
GCP.
5
As
discussed
in
section
8.0,
CO
is
also
considered
a
surrogate
for
organic
HAP.

To
determine
if
CO
monitoring
would
be
the
basis
of
the
existing
and
new
source
MACT
floor
for
organic
HAP
emissions
control,
available
information
was
examined.
The
population
database
does
not
contain
information
on
existing
units
monitoring
CO
emissions.
State
regulations
applicable
to
boilers
and
process
heaters
that
required
CO
monitoring
to
maintain
a
specific
CO
limit
were
then
reviewed.
Many
of
the
state
regulations
identified
were
applicable
to
units
of
only
certain
capacities,
heat
inputs,
or
fuel
types.
The
applicability
of
these
state
requirements
were
matched
to
the
units
in
the
population
database
to
determine
which
units
were
subject
to
a
particular
requirement
and
which
were
not.
First,
the
units
that
were
located
in
states
with
CO
requirements
were
identified
using
the
state
codes
in
the
population
database.
Then
the
corresponding
unit
capacities
and
fuel
types
were
reviewed
to
determine
if
the
CO
requirement
applied.
In
some
cases,
the
applicability
requirements
were
too
specific
to
be
able
to
identify
whether
a
unit
would
be
subject
to
the
requirement
or
the
population
database
would
not
have
enough
information
regarding
a
specific
unit
(
such
as
unit
capacity)
to
determine
if
the
requirement
would
apply.
In
the
cases
where
the
applicability
of
a
requirement
could
not
be
determined,
the
associated
units
were
not
included
in
the
MACT
floor
analysis
because
too
many
assumptions
would
have
to
be
made
regarding
whether
requirements
applied.
Instead,
as
with
the
add­
on
control
technology
analysis,
the
MACT
floor
analysis
based
on
CO
requirements
was
done
using
a
subset
of
the
population
in
each
subcategory
for
which
the
applicability
could
be
determined.
This
subset
was
assumed
to
be
representative
of
the
entire
subcategory.
The
results
showed
that
less
than
6
percent
of
the
existing
units
in
any
subcategory
were
subject
to
CO
monitoring
requirements
or
emission
limits.
Therefore,
it
did
not
constitute
a
MACT
floor
level
of
control.
This
analysis
is
presented
in
Appendix
B.

For
new
sources,
the
analysis
of
State
regulations
indicated
that
at
least
one
of
the
boilers
and
process
heaters
in
the
large
and
limited
use
subcategories
for
solid
fuel,
liquid
fuel,
and
gaseous
fuel
were
required
to
meet
a
CO
emissions
limit.
The
State
with
the
most
stringent
CO
20
dwcgi­
7150­
1077822954­
148160000.
wpd
emission
limit
that
applies
to
all
units
within
a
subcategory
is
California,
which
requires
monitoring
and
maintaining
a
CO
limit
of
400
ppm.
Another
state,
Massachusetts,
has
a
limit
of
200
ppm.
However,
the
limit
does
not
necessarily
apply
to
all
boilers
in
a
subcategory,
(
i.
e.,
it
would
apply
to
large
solid
fuel
boilers
but
would
not
be
applicable
to
wood­
fired
units
or
units
in
lower
size
ranges).
Consequently,
the
200
ppm
limit
would
not
be
appropriate
for
the
entire
subcategory.
Therefore,
the
new
source
MACT
floor
includes
a
CO
emission
limit
of
400
ppm
to
reflect
the
MACT
floor
level
of
control
for
emissions
of
organic
HAP
from
the
large
and
limited
use
solid,
liquid,
and
gaseous
subcategories.
(
The
California
State
regulations
reviewed
are
included
in
the
boiler
and
process
heater
docket
as
items
II­
I­
83
through
II­
I­
86)

8.0
MACT
FLOOR
EMISSION
LIMIT
METHODOLOGY
The
available
emissions
data
for
boilers
and
process
heaters
controlled
by
the
bestperforming
technologies
in
each
subcategory
were
reviewed
to
determine
the
emissions
levels
associated
with
the
MACT
floor
control
technology.
Using
the
technology­
basis
for
the
MACT
floor
for
each
subcategory,
the
corresponding
emission
limitations
were
determined
for
each
pollutant
category.

An
outlet
emission
rate
format
was
used
for
the
MACT
floor
analysis
because
outlet
data
are
available
for
boilers
and
process
heaters
that
use
the
control
techniques
that
provide
the
greatest
reduction
in
HAP
emissions.
The
individual
limits
reflect
the
achievable
performance
of
boilers
and
process
heaters
using
the
appropriate
controls
for
each
type
of
emissions.

The
most
typical
units
for
the
limits
are
pounds
of
pollutant
emitted
per
million
British
thermal
units
(
Btu)
of
heat
input.
The
mass
per
heat
input
units
are
consistent
with
other
Federal
and
many
State
boiler
regulations
and
allows
easy
comparison
between
such
requirements.

8.1
Surrogates
for
Pollutant
Categories
The
MACT
floor
based
on
control
technology
was
conducted
for
each
subcategory
and
for
four
pollutant
categories:
non­
mercury
metallic
HAP,
mercury,
inorganic
HAP,
and
organic
HAP.
These
categories,
which
cover
all
the
HAP
emitted,
include
a
large
number
of
compounds,

making
it
infeasible
to
develop
emission
limits
for
each
one.
Consequently,
surrogate
pollutants
were
identified
to
represent
the
pollutants
in
each
category.

8.1.1
Non­
Mercury
Metallic
HAP
21
dwcgi­
7150­
1077822954­
148160000.
wpd
There
are
many
different
non­
mercury
metallic
HAP
that
could
be
emitted
from
boilers
and
process
heaters
including
arsenic,
beryllium,
cadmium,
chromium,
lead,
manganese,
and
nickel.
Most,
if
not
all,
non­
mercury
metallic
HAP
emitted
from
combustion
sources
will
appear
on
the
flue
gas
fly­
ash.
Therefore,
the
same
control
techniques
that
would
be
used
to
control
the
fly­
ash
PM
will
control
non­
mercury
metallic
HAP.
Also,
all
fuels
do
not
emit
the
same
type
and
amount
of
metallic
HAP
but
most
generally
emit
PM
that
includes
some
amount
and
combination
of
metallic
HAP.
Therefore,
the
MACT
floor
emission
level
associated
with
the
best­
performing
12
percent
of
sources
for
the
non­
mercury
metallic
HAP
category
was
set
using
particulate
matter
as
a
surrogate.

However,
there
are
some
sources
in
the
solid
fuel­
fired
categories
that
burn
a
fuel
containing
very
little
metals,
but
with
sufficient
PM
emissions
to
require
control.
In
such
cases,

PM
would
not
be
an
appropriate
surrogate
for
metallic
HAP.
Therefore,
an
alternative
metals
emission
limit
was
also
developed
for
solid
fuel­
fired
sources.
The
metals
emission
limit
is
for
the
sum
of
emissions
of
eight
selected
metals:
arsenic,
beryllium,
cadmium,
chromium,
lead,

manganese,
nickel,
and
selenium.
These
eight
pollutants
represent
the
most
common
and
the
largest
emitted
metallic
HAP
from
boilers
and
process
heaters.

8.1.2
Inorganic
HAP
As
with
non­
mercury
metallic
HAP,
there
are
several
pollutants
which
fall
into
the
inorganic
HAP
pollutant
category
including
hydrogen
chloride
and
hydrogen
fluoride.
The
emissions
test
information
available
to
EPA
indicate
that
the
primary
inorganic
HAP
emitted
from
boilers
and
process
heaters
are
acid
gases,
with
HCl
present
in
the
largest
amounts.
Other
inorganic
compounds
emitted
are
found
in
much
smaller
quantities.
Also,
control
technologies
that
would
reduce
HCl
would
also
control
other
inorganic
compounds
that
are
acid
gases.

Therefore,
HCl
is
considered
a
good
surrogate
for
inorganic
HAP
and
controlling
HCl
will
result
in
a
corresponding
control
of
other
inorganic
HAP
emissions.

8.1.3
Mercury
A
MACT
floor
emission
limit
was
determined
specifically
for
mercury
and
not
for
a
surrogate
compound.
All
the
mercury
emissions
data
were
reviewed
to
determine
the
associated
emission
level
that
corresponds
to
the
levels
from
the
units
determined
to
be
the
technology
basis
of
the
MACT
floor.

8.1.4
Organic
HAP
22
dwcgi­
7150­
1077822954­
148160000.
wpd
For
organic
HAP,
carbon
monoxide
(
CO)
was
chosen
as
a
surrogate
to
represent
the
variety
of
organic
compounds,
including
dioxins,
emitted
from
the
various
fuels
burned
in
boilers
and
process
heaters.
CO
is
a
good
indicator
of
incomplete
combustion
and,
thus,
the
formation
of
organic
HAP
emissions.
Therefore,
using
CO
as
a
surrogate
for
organic
HAP
is
a
reasonable
approach
because
minimizing
CO
emissions
will
result
in
minimizing
organic
HAP
emissions.

8.2
Methodology
for
establishing
MACT
floor
emission
levels
After
the
MACT
floor
based
on
control
techniques
was
identified
for
each
subcategory
and
pollutant
group,
the
emissions
database
was
reviewed
to
identify
all
emission
tests
for
the
pollutant
groups
that
also
had
the
MACT
floor
control
technology.
Then,
the
emission
levels,
in
units
of
pound
pollutant
per
MMBtu
heat
input,
were
reviewed
for
each
pollutant
group
surrogate
in
order
to
determine
an
emission
level
associated
with
the
MACT
floor
level
of
control.

First
the
data
and
associated
emission
test
reports
for
all
the
higher
emission
points
were
reviewed
to
identify
any
outliers
and
determine
if
there
was
something
about
the
test
conditions
or
control
device
operation
that
made
it
unrepresentative
of
the
MACT
floor
level
of
control
or
the
entire
subcategory
population.
Several
data
points
were
removed
from
the
analysis
because
their
unrepresentativeness.

The
summary
tables
in
Appendix
C
indicate
which
test
data
were
used
in
the
calculation
of
emission
limits
for
each
subcategory
and
pollutant
group.
Table
8­
1
summarizes
the
results
of
the
MACT
floor
control
technologies
analysis.
A
discussion
of
the
results
for
each
subcategory
is
presented
in
following
sections.

8.2.1
Existing
Source
MACT
Floor
Emission
Levels
For
existing
sources,
the
calculation
of
numerical
emission
limits
was
a
two­
step
analysis.

The
first
step
involved
calculating
a
numerical
average
of
an
appropriate
subset
of
the
emission
test
data
from
units
using
the
same
technology,
or
technologies,
as
the
units
in
the
top
12
percent.

Based
on
the
initial
ranking,
the
proportion
of
the
units
using
a
particular
technology
that
were
among
the
top
12
percent
of
units
in
the
subcategory
were
identified.
Then,
a
corresponding
proportion
of
the
emission
test
data
from
units
using
that
type
of
control
technology
were
reviewed,
and
an
overall
average
measured
performance
level
was
calculated.
For
example,
in
the
large
solid­
fuel
subcategory,
approximately
14
percent
of
units
used
the
best
performing
control
technology
for
PM/
metallic
HAP
(
baghouses).
In
order
to
rank
the
units
using
the
best
technology
for
which
there
were
emission
test
data,
unit
by
unit
measured
performance
levels
23
dwcgi­
7150­
1077822954­
148160000.
wpd
were
calculated
by
averaging
the
multiple
tests
from
each
individual
unit
(
if
multiple
tests
were
available).
The
best
12/
14
of
the
units
for
which
we
generated
such
individual
averages
were
identified,
and
the
unit
by
unit
averages
from
all
of
these
units
was
averaged.
This
resulted
in
an
overall
average
measured
emissions
performance
level
for
units
representative
of
the
top
12
percent
of
units
in
the
subcategory.

The
second
step
in
this
part
of
the
process
involved
generating
and
applying
an
appropriate
variability
factor
to
account
for
unavoidable
variations
in
emissions
due
primarily
to
uncontrollable
differences
in
fuel
characteristics
and
ordinary
operational
variability.
All
the
units
for
which
we
had
emission
test
data
using
the
same
technology,
or
technologies,
were
identified
as
units
in
the
top
12
percent.
Then,
for
each
such
unit
with
multiple
emission
tests,
the
variability
in
the
measured
emissions
was
calculated
from
that
unit
by
dividing
the
highest
three­
run
test
result
by
the
lowest
three­
run
test
result.
Finally,
the
overall
variability
in
the
measured
emissions
from
these
units
was
calculated
by
averaging
all
the
individual
unit
variability
factors.
This
overall
variability
factor
was
multiplied
by
the
overall
average
measured
emissions
performance
level
(
as
described
above)
to
derive
a
emission
limit
representative
of
the
average
emission
limitation
achieved
by
the
top
12
percent
of
units.

This
approach
reasonably
ensures
that
the
emission
limit
selected
as
the
MACT
floor
adequately
represents
the
average
level
of
control
actually
achieved
by
units
in
the
top
12
percent,

considering
ordinary
operational
variability.

During
the
public
comment
period,
commenters
requested
that
EPA
account
for
variability
in
fuel
composition
as
MACT
floors
are
established
and
to
provide
adequate
allowances
for
inherent
fuel
supply
variability.
Commenters
contended
that
EPA's
calculation
of
variability
was
statistically
unsound
and
recommended
that
EPA
estimate
statistically
the
variance
in
the
distribution
of
control
technology
efficiency
rather
than
calculate
a
variability
factor.
Based
on
comments,
we
did
conduct
a
statistical
analysis
of
the
data
to
identify
the
95th
and
99th
percent
confidence
limits.
This
analysis
provided
similar
results
to
the
variability
analysis
approach
conducted
for
the
proposed
rule.
Consequently,
we
decided
not
to
change
the
variability
methodology.

Some
boilers
and
process
heaters
within
each
subcategory
may
be
able
to
meet
the
floor
emission
levels
without
using
the
air
pollution
control
technology
that
is
used
by
the
top
12
percent
of
units
in
the
subcategory.
This
is
to
be
expected,
given
the
variety
of
fuel
types,
fuel
24
dwcgi­
7150­
1077822954­
148160000.
wpd
input
rates,
and
boiler
designs
included
within
each
subcategory
and
the
resulting
variability
in
emission
rates.
Thus,
for
instance,
boilers
or
process
heaters
within
the
large
unit
solid
fuel
subcategory
that
burn
lower
percentages
of
solid
fuels
may
be
able
to
achieve
the
emission
levels
for
the
large
unit
solid
fuel
subcategory
without
the
need
for
additional
control
devices.

Furthermore,
solid
fuels,
especially
coal,
are
very
heterogeneous
and
can
vary
in
composition
by
location.
Coal
analysis
data
obtained
from
the
electric
utility
industry
in
another
rulemaking
contained
information
on
the
mercury,
chlorine,
and
ash
content
of
various
coals.
A
preliminary
review
of
this
data
indicate
that
the
composition
can
vary
greatly
from
location
to
location,
and
also
within
a
particular
location.
Based
on
the
range
of
variation
of
mercury,

chlorine,
and
ash
content
in
coal,
it
is
possible
for
a
unit
with
a
lower
performing
control
system
to
have
emission
levels
lower
than
a
unit
considered
to
be
included
in
the
best
performing
12
percent
of
the
units.

This
situation
is
reflected
in
the
emissions
information
used
to
set
the
MACT
floor
emission
limits.
In
some
instances
there
are
boilers
with
ESP
or
other
controls
that
achieve
similar,
or
lower,
outlet
emission
levels
of
non­
mercury
metallic
HAP,
PM,
or
mercury
than
fabric
filters.
In
most
cases,
this
is
due
to
concentrations
entering
these
other
control
devices
being
lower,
even
though
the
percent
reduction
achieved
is
lower
than
fabric
filters.

Additionally,
the
design
of
some
control
devices
may
have
a
substantial
effect
on
the
their
emission
reduction
capability.
For
example,
fabric
filters
are
largely
insensitive
to
the
physical
characteristics
of
the
inlet
gas
stream.
Thus,
their
design
does
not
vary
widely,
and
emissions
reductions
are
expected
to
be
similar
(
e.
g.
99
percent
reduction
of
PM).
However,
ESP
design
can
vary
significantly.
Some
ESP
are
2
fields,
others
may
have
3
or
4.
The
more
fields
the
larger
the
emission
reduction
for
PM.
Similarly,
other
devices
can
be
designed
to
achieve
higher
emission
reductions.
This
level
of
detail
was
not
available
for
the
information
used
to
develop
the
MACT
floor
emission
limits.

For
existing
unit
subcategories
where
less
than
12
percent
of
units
in
the
subcategory
use
any
type
of
control
technology,
the
same
approach
could
not
be
used
to
identify
the
average
level
of
control
achieved
by
the
top
12
percent.
Therefore,
the
central
tendency
of
the
best
controlled
units
was
estimated
by
looking
at
the
median
unit
of
the
top
12
percent
(
the
unit
at
the
94th
percentile).
If
the
median
unit
of
the
top
12
percent
is
using
some
control
technology,
the
measured
emission
performance
of
that
individual
unit
was
used
as
the
basis
for
estimating
an
25
dwcgi­
7150­
1077822954­
148160000.
wpd
appropriate
average
level
of
control
of
the
top
12
percent.
For
subcategories
where
even
the
median
unit
is
using
no
control
technology,
the
average
control
of
the
top
12
percent
of
units
is
no
emission
reductions.

Large
Solid
Fuel
Units
­
Heat
Inputs
Greater
than
10
MMBtu/
hr.
As
described
earlier,
a
PM
level
is
set
as
a
surrogate
for
non­
mercury
metallic
HAP,
and
fabric
filters
are
the
MACT
floor
control
technology.
Using
the
two­
step
methodology
described
in
Section
8.2,
the
proportion
of
the
units
using
fabric
filters
in
the
population
database
that
were
among
the
top
12
percent
of
units
in
the
subcategory
were
identified
and
a
corresponding
proportion
of
the
emission
test
data
from
units
using
fabric
filters
were
reviewed,
and
an
overall
average
measured
performance
level
was
calculated.
Approximately
14
percent
of
the
boilers
in
the
population
database
used
fabric
filters.
Including
the
emission
data
obtained
during
the
public
comment
period,
the
emissions
database
contains
PM
information
on
11
different
boilers
using
fabric
filters.

(
Emission
data
from
one
unit,
Energy
Products
of
Idaho,
was
removed
from
the
analysis
since
this
unit
is
used
for
research
and
development
and,
thus,
is
not
in
this
subcategory.)
In
order
to
rank
the
units
using
the
best
technology
for
which
there
were
emission
test
data,
unit
by
unit
measured
performance
levels
were
calculated
by
averaging
the
multiple
tests
from
each
individual
test
(
if
multiple
tests
were
available).
The
best
12/
14
of
the
units
in
the
emissions
database
were
identified
(
i.
e.,
the
best
9
boilers).
The
average
PM
emission
limit
from
the
best
9
boilers
is
0.015
lb/
MMBtu,
and
the
average
variability
level
is
4.55.
Incorporating
the
variability,
the
MACT
floor
emission
level
for
PM
is
0.07
lb/
MMBtu.
This
analysis
is
shown
in
Appendix
C­
1.

An
alternative
metals
limit
was
also
calculated
and
can
be
used
to
show
compliance
in
cases
where
metal
HAP
emissions
are
low
in
proportion
to
PM
emissions.
This
is
because,

according
to
the
emissions
database,
some
biomass
units
have
low
metals
content
but
high
PM
emissions.
The
available
emission
test
data
for
solid
fuel
boilers
with
either
an
ESP
or
a
fabric
filter
control
were
identified.
These
tests
were
further
screened
for
only
those
tests
that
included
emission
results
for
all
of
the
eight
total
selected
metals
(
arsenic,
beryllium,
cadmium,
chromium,

lead,
manganese,
nickel,
and
selenium)
and
corresponding
PM
results.
The
sum
of
the
emissions
of
these
eight
metals,
in
terms
of
lb/
MMBtu,
were
then
ranked
from
highest
to
lowest
emissions.

Then,
beginning
with
the
highest
tests,
those
tests
that
also
included
corresponding
PM
data
were
identified.
For
existing
sources,
the
highest
test
results
for
metals
having
corresponding
PM
data
available
indicated
a
PM
emission
level
of
0.0232
lb/
MMBtu
which
is
below
the
MACT
floor
PM
26
dwcgi­
7150­
1077822954­
148160000.
wpd
emission
level
for
existing
solid
fuel
sources.
Because
this
source
is
meeting
the
MACT
floor
PM
emission
level,
the
corresponding
alternative
metallic
HAP
emissions
level
for
existing
sources
was
set
based
on
this
source.
However,
since
the
corresponding
PM
emission
limit
(
0.023lb/
MMBtu)
is
well
under
the
MACT
floor
PM
emission
level
(
0.7lb/
MMBtu),
the
alternative
total
selected
metals
(
TSM)
MACT
floor
emissions
level
for
existing
sources
was
determined
by
extrapolating
the
TSM
emission
level
from
this
source
based
on
the
difference
in
the
PM
emission
levels.
Therefore,
the
TSM
MACT
floor
emission
level
is
0.001
lb/
MMBtu
[
0.000416
x
(
0.07/
0.023)].
This
analysis
is
shown
in
Appendix
C­
2.

At
proposal,
the
TSM
MACT
floor
level
for
existing
sources
was
determined
to
be
0.001
lb/
MMBtu.
However,
this
level
did
not
incorporate
variability.
Based
on
comments,
we
reexamined
our
analysis
for
determining
the
TSM
MACT
floor
level.
First,
it
was
determined
that
the
metals
test
results
from
the
unit
selected,
at
proposal,
as
the
basis
had
a
questionable
test
run
result
for
manganese
that
should
not
have
been
included
in
calculating
the
test
results.
(
The
high
manganese
concentration
in
one
test
run
was
attributable
to
transfer
from
the
potassium
manganese
impingers.)
Eliminating
this
test
run
resulted
in
a
metals
emission
level
of
0.000167
lb/
MMBtu
instead
of
0.00084
lb/
MMBtu.
Thus,
this
unit
is
no
longer
the
basis.
The
new
basis
is
a
unit
with
a
metal
level
of
0.000416
lb/
MMBtu
and
a
corresponding
PM
level
of
0.0232
lb/
MMBtu.
Since
we
have
no
multiple
metals
test
results
from
any
unit
to
determine
variability
and
PM
is
a
surrogate
for
metallic
HAP,
the
appropriate
approach
was
deemed
to
be
the
use
of
extrapolating
based
the
difference
between
the
source
PM
level
and
the
MACT
floor
PM
level.
.

The
MACT
floor
emission
level
for
inorganic
HAP
is
based
on
HCl
emissions
test
information
from
units
using
wet
or
dry
scrubbers
or
packed
bed
scrubbers.
Approximately
13
percent
of
the
boilers
in
the
population
database
used
scrubbers.
The
emissions
database
contains
HCl
information
on
9
different
boilers
using
scrubbers.
In
order
to
rank
the
units
using
the
best
technology
for
which
there
were
emission
test
data,
unit
by
unit
measured
performance
levels
were
calculated
by
averaging
the
multiple
tests
from
each
individual
test
(
if
multiple
tests
were
available).
The
best
12/
13
of
the
units
in
the
emissions
database
were
identified
(
i.
e.,
the
best
8
boilers).
The
average
HCl
emission
limit
from
the
best
8
boilers
is
0.00962
lb/
MMBtu,
and
the
average
variability
level
is
9.08.
Incorporating
the
variability,
the
MACT
floor
emission
level
for
HCl
is
0.09
lb/
MMBtu.
This
analysis
is
shown
in
Appendix
C­
3.
27
dwcgi­
7150­
1077822954­
148160000.
wpd
The
MACT
floor
emission
level
for
mercury
is
based
on
emissions
test
information
from
units
using
fabric
filters.
Approximately
14
percent
of
the
boilers
in
the
population
database
used
scrubbers.
Including
the
mercury
emission
data
obtained
during
ther
public
comment
period,
the
emissions
database
contains
mercury
information
on
10
different
boilers
using
fabric
filters.
In
order
to
rank
the
units
using
the
best
technology
for
which
there
were
emission
test
data,
unit
by
unit
measured
performance
levels
were
calculated
by
averaging
the
multiple
tests
from
each
individual
test
(
if
multiple
tests
were
available).
The
best
12/
14
of
the
units
in
the
emissions
database
were
identified
(
i.
e.,
the
best
9
boilers).
The
average
mercury
emission
limit
from
the
best
9
boilers
is
0.00000302
lb/
MMBtu,
and
the
average
variability
level
is
2.98.
Incorporating
the
variability,
the
MACT
floor
emission
level
for
mercury
is
0.000009
lb/
MMBtu.
This
analysis
is
shown
in
Appendix
C­
4.

Some
boilers
and
process
heaters
that
use
technologies
other
than
those
used
as
the
basis
of
the
MACT
floor
may
be
able
to
achieve
the
MACT
floor
emission
levels.
For
example,

emission
test
data
show
that
many
boilers
with
well
designed
and
operated
ESP
can
meet
the
MACT
floor
emission
levels
for
nonmercury
metallic
HAP
and
PM,
even
though
the
floor
emission
level
for
these
pollutants
is
based
on
units
using
a
fabric
filters
(
however,
we
would
not
expect
that
all
units
using
ESP
would
be
able
to
meet
the
emission
limits
in
the
proposed
rule).

Small
Solid
Fuel
Units
­
Heat
Inputs
Less
than
or
Equal
to
10
MMBtu/
hr.

Because
less
than
6
percent
of
the
units
in
this
subcategory
used
control
techniques
that
limit
emissions
from
any
of
the
pollutant
groups,
the
MACT
floor
emission
level
for
existing
units
for
each
of
the
pollutant
categories
in
this
subcategory
is
no
emissions
reductions.

Limited
Use
Solid
Fuel
Units
­
Capacity
Utilizations
Less
than
or
Equal
to
10
Percent.

A
PM
limit
was
established
as
a
surrogate
for
non­
mercury
metallic
HAP
control,

reflecting
the
emission
test
data
from
units
using
ESP
and
fabric
filters
that
were
representative
of
the
top
12
percent
of
units
in
the
subcategory.

The
emissions
test
database
did
not
contain
test
data
for
limited
use
boilers
and
process
heaters.
In
order
to
develop
emission
levels
for
this
subcategory,
information
from
units
in
the
large
solid
fuel
subcategory
was
used.
This
was
considered
to
be
an
appropriate
methodology
because
although
the
units
in
this
subcategory
are
different
enough
to
warrant
their
own
subcategory
(
i.
e.,
different
purposes
and
operation),
emissions
of
the
specific
types
of
HAP
for
which
limits
are
being
proposed
(
nonmercury
metals)
are
expected
to
be
related
more
to
the
type
28
dwcgi­
7150­
1077822954­
148160000.
wpd
of
fuel
burned
and
the
type
of
control
used,
than
to
unit
operation.
Consequently,
the
emissions
information
from
the
large
solid
fuel
subcategory
that
is
most
representative
of
the
units
in
this
subcategory
was
used
to
establish
MACT
floor
levels
for
this
subcategory
because
the
fuels
and
controls
are
similar.

Appendix
A­
4
shows
that
of
the
top
12
percent
of
units
in
this
subcategory,
5.8
percent
use
fabric
filters
and
6.2
percent
use
ESPs.
In
order
to
account
for
both
controls,
the
emissions
database
was
reviewed
for
information
on
fabric
filters
and
ESPs
from
solid
fuel
fired
units.
The
emissions
database
contains
significantly
more
information
on
units
with
ESPS
than
units
with
fabric
filters.
Less
than
5.8
percent
of
the
units
in
the
emissions
database
have
fabric
filters.

Therefore,
the
analysis
used
all
the
information
from
fabric
filters
and
the
remaining
information
from
6.2
percent
of
the
ESPs
to
calculate
the
MACT
floor
limit.
The
ESP
information
was
first
divided
into
units
burning
coal
and
those
burning
biomass.
The
population
database
indicates
that
the
majority
of
boilers
in
this
subcategory
burn
coal.
However,
more
emissions
information
is
available
for
units
burning
biomass.
In
order
to
reflect
the
population
database,
all
the
units
burning
coal
were
incorporated
into
the
analysis.
The
remaining
ESPs
burning
biomass
were
ranked
from
lowest
to
highest
emissions
and
the
units
with
the
lowest
emissions
were
included
in
analysis
of
MACT
floor
emission
limits.
The
average
emission
limit
from
this
population
of
units
(
i.
e.,
units
with
fabric
filters,
units
with
ESPs
firing
coal,
and
lowest
emitting
units
with
ESPS
firing
biomass)
was
calculated
to
be
0.0273
lb/
MMBtu,
and
the
operational
variability
was
calculated
to
be
8.11.
The
MACT
floor
emission
level
based
on
this
test
data,
considering
operational
variability,
is
0.21
lb
PM/
MMBtu.
An
alternative
metals
limit
of
0.004
lb
metals/
MMBtu
was
also
calculated
so
that
sources
could
show
compliance
in
cases
where
metal
HAP
emissions
are
low
in
proportion
to
PM
emissions.
The
emissions
database
indicates
that
some
biomass
units
have
low
metals
content
but
high
PM
emissions.
The
emission
level
for
metals
was
selected
from
metals
test
data
associated
with
PM
emission
tests
that
met
the
MACT
floor
PM
emission
level.
The
same
methodology
used
to
calculate
the
alternative
metals
limit
for
large
solid
units
was
used
for
limited
use
units.
Appendix
C­
5
and
C­
6
presents
the
calculation
of
PM
emission
limits
for
this
subcategory.

Because
fewer
than
6
percent
of
units
used
controls
that
would
reduce
emissions
of
organic
HAP,
inorganic
HAP,
and
mercury,
the
median
unit
for
these
HAP
grouping
reflects
no
29
dwcgi­
7150­
1077822954­
148160000.
wpd
emission
reductions.
Therefore,
the
MACT
floor
for
inorganic
HAP,
organic
HAP
and
mercury
in
this
subcategory
is
no
emission
reductions.

Existing
Liquid
Fuel
Boilers
and
Process
Heaters.
Less
than
6
percent
of
the
units
in
each
of
the
liquid
subcategories
used
control
techniques
that
would
reduce
nonmercury
metallic
HAP
and
PM,
mercury,
organic
HAP,
or
acid
gases,
(
such
as
HCl).
Therefore,
for
each
subcategory
of
liquid
fueled
boilers
and
process
heaters,
the
MACT
floor
is
no
emission
reductions
for
nonmercury
metallic
HAP,
mercury,
inorganic
HAP,
and
organic
HAP.

Existing
Gaseous
Fuel
Boilers
and
Process
Heaters.
No
existing
units
in
the
gaseous
fuel­
fired
subcategories
were
using
control
technologies
that
achieve
consistently
lower
emission
rates
than
uncontrolled
sources
for
any
of
the
pollutant
groups
of
interest.
Therefore,
the
MACT
floor
for
existing
sources
in
this
subcategory
is
no
emissions
reductions
for
nonmercury
metallic
HAP,
mercury,
inorganic
HAP,
and
organic
HAP.

8.2.2
New
Source
MACT
Floor
Emission
Levels
For
each
pollutant
type
in
each
subcategory,
the
available
emission
test
data
from
units
using
the
best
control
technology
was
used
to
identify
the
single
unit
with
the
best
average
measured
performance.
An
emission
limit,
based
on
the
measured
performance
of
this
single
unit
was
calculated
by
applying
an
appropriate
variability
factor
to
account
for
unavoidable
variations
in
emissions
due
to
uncontrollable
variations
in
fuel
characteristics
and
control
device
performance.

The
approach
use
to
calculate
the
MACT
floors
for
new
sources
is
somewhat
different
from
the
approach
used
to
calculate
the
MACT
floors
for
existing
sources.
While
the
MACT
floors
for
existing
units
are
intended
to
reflect
the
average
performance
achieved
by
a
representative
group
of
sources,
the
MACT
floors
for
new
units
are
meant
to
reflect
the
"
emission
control
that
is
achieved
in
practice"
by
the
best
controlled
similar
source.
Thus,
for
existing
units,

the
central
tendency
of
a
set
of
multiple
units
is
the
focus,
while
for
new
units,
the
level
of
control
that
is
representative
of
that
achieved
by
a
single
"
best
controlled"
similar
source
is
calculated.
As
with
the
analysis
for
existing
sources
the
new
unit
analysis
must
account
for
variability.
To
accomplish
this
for
new
sources,
for
the
fuel
dependant
HAP
emissions,
what
the
"
best
controlled"
similar
source
can
achieve
in
light
of
the
inherent
and
unavoidable
variations
in
the
HAP
content
of
the
fuel
that
such
unit
might
potentially
use
was
necessary
to
be
determined.
For
30
dwcgi­
7150­
1077822954­
148160000.
wpd
non­
fuel
dependent
HAP
emissions,
on
the
other
hand,
the
inherent
variability
of
the
control
technology
used
by
sources
in
the
category
was
analyzed.

Thus,
for
new
units,
after
identifying
the
best
control
technology
for
each
pollutant
group
within
each
subcategory
(
based
on
the
control
technology
rankings),
the
emissions
data
available
for
boilers
and
process
heaters
controlled
by
these
technologies
was
examined
to
determine
achievable
emission
levels
for
PM
(
as
a
surrogate
for
nonmercury
metallic
HAP),
total
selected
nonmercury
metallic
HAP,
mercury,
HCl
(
as
a
surrogate
for
inorganic
HAP),
and
CO
(
as
a
surrogate
for
organic
HAP).
First,
the
units
using
the
best
control
technology
for
which
we
had
emissions
data
were
determined.
Then,
the
emission
data
for
any
unit
with
multiple
test
results
was
average,
and
the
units
were
ranked
based
on
the
unit
by
unit
average
measured
emissions
performance.
Then,
the
unit
with
the
best
average
measured
emissions
performance
was
identified.
Finally,
to
estimate
the
emission
control
achievable
by
this
unit,
a
variability
factor
was
applied
to
the
average
measured
emissions
performance
of
the
best
unit.
For
fuel
dependant
HAP
emissions
(
mercury
and
HCl),
the
variability
factor
was
calculated
by
looking
at
data
on
HAP
variability
in
coal,
from
an
analysis
of
coal
properties
obtained
through
a
utility­
related
information
collection
request.
The
fuel
dependant
variability
factor
was
derived
by
dividing
the
highest
observed
HAP
concentration
by
the
lowest
observed
HAP
concentration
from
the
utility
coal
analysis.
This
was
done
because
coal
available
to
utilities
and
industrial
boilers
and
process
heaters
are
expected
to
be
similar,
and
coal
is
the
solid
fuel
that
is
routinely
used
in
such
units
that
has
generally
the
greatest
degree
of
HAP
variability.
Once
the
fuel
dependant
variability
factors
were
calculated,
they
were
applied
to
the
average
measured
emissions
performance
of
the
unit
with
the
best
data
to
derive
the
MACT
floor
level
of
control.
This
approach
reasonably
estimates
the
best
source's
level
of
control,
adjusted
for
unavoidable
variation
in
fuel
characteristics
which
have
a
direct
impact
on
emissions.

For
non­
fuel
dependant
HAP
emissions
(
PM),
the
appropriate
variability
factor
was
calculated
in
the
same
general
manner
as
for
existing
units.
A
variability
factor
for
each
unit
using
the
same
control
technology
as
the
unit
with
the
best
emissions
data
was
calculated,
and
then
the
overall
variability
in
the
measured
emissions
from
units
was
determined
using
this
technology
by
averaging
all
the
individual
unit
variability
factors.
Finally,
this
overall
variability
factor
was
applied
to
the
average
measured
emissions
performance
of
the
unit
with
the
best
emissions
data.
31
dwcgi­
7150­
1077822954­
148160000.
wpd
For
new
unit
subcategories
where
no
units
in
the
subcategory
employed
any
type
of
control
technology,
data
could
not
be
identified
to
represent
the
level
of
control
of
the
best
controlled
similar
unit.
Accordingly,
the
MACT
floor
level
of
control
for
such
subcategories
is
no
emission
reductions.

Large
Solid
Fuel
Units
­
Heat
Inputs
Greater
than
10
MMBtu/
hr.
As
described
earlier,
a
PM
level
is
set
as
a
surrogate
for
non­
mercury
metallic
HAP
and
the
MACT
floor
level
of
control
is
a
fabric
filter.
The
best
performing
boiler
in
the
emissions
database
with
a
fabric
filter
has
an
average
emission
limit
of
0.0054.
See
Appendix
C­
1.
Incorporating
the
average
variability
for
all
the
units
with
fabric
filters,
4.55,
results
in
the
MACT
floor
PM
emission
limit
of
0.025
lb/
MMBtu.

An
alternative
metals
limit
was
also
calculated
and
can
be
used
to
show
compliance
in
cases
where
metals
HAP
emissions
are
low
in
proportion
to
PM
emissions.
This
is
because,

according
to
the
emissions
database,
some
biomass
units
have
low
metals
content
but
high
PM
emissions.
The
available
emission
test
data
for
solid
fuel
boilers
with
a
fabric
filter
control
were
identified.
These
tests
were
further
screened
for
only
those
tests
that
included
emission
results
for
all
of
the
eight
total
selected
metals
(
arsenic,
beryllium,
cadmium,
chromium,
lead,
manganese,

nickel,
and
selenium)
and
corresponding
PM
levels.
The
sum
of
the
emissions
of
these
eight
metals,
in
terms
of
lb/
MMBtu,
were
then
ranked
from
highest
to
lowest
emissions.
For
new
sources,
the
highest
test
results
for
metals
from
a
fabric
filter
having
corresponding
PM
data
available
indicated
a
PM
emission
level
of
0.0025
lb/
MMBtu
which
is
below
the
new
source
MACT
floor
PM
emission
level.
Because
this
source
is
meeting
the
MACT
floor
PM
emission
level,
the
corresponding
alternative
metallic
HAP
emissions
level
for
new
sources
was
set
based
on
this
source.
Incorporating
variability,
the
alternative
metallic
HAP
emissions
level
for
new
sources
is
0.0003
lb/
MMBtu.
See
Appendix
C­
2.
As
discussed
earlier,
at
proposal,
variability
was
not
incorporated.
Incorporating
variability
resulted
in
a
revising
the
proposed
metals
limit
(
0.0001
lb/
MMBtu)
to
0.0003
lb/
MMBtu.

Hydrogen
chloride
emissions
are
dependent
on
the
quantity
of
chlorine
in
the
fuel
burned.

To
estimate
the
emission
control
achievable
by
this
unit,
a
variability
factor
was
applied
to
the
average
measured
emissions
performance
of
the
best
unit.
The
variability
factor
was
calculated
by
looking
at
data
on
HAP
variability
in
coal,
from
an
analysis
of
coal
properties
obtained
through
a
utility­
related
information
collection
request.
6
The
fuel
dependant
variability
factor
was
derived
32
dwcgi­
7150­
1077822954­
148160000.
wpd
by
dividing
the
highest
observed
HAP
concentration
by
the
lowest
observed
HAP
concentration
from
the
utility
coal
analysis.
This
was
done
because
coal
available
to
utilities
and
industrial
boilers
and
process
heaters
are
expected
to
be
similar,
and
coal
is
the
solid
fuel
that
is
routinely
used
in
such
units
that
has
generally
the
greatest
degree
of
HAP
variability.
Once
the
fuel
dependant
variability
factors
were
calculated,
they
were
applied
to
the
highest
test
result
of
the
unit
with
the
lowest
average
emission
level
to
derive
the
MACT
floor
level
of
control.
This
unit
had
two
multiple
test
results.
Using
the
highest
of
the
two
test
results
was
deemed
more
appropriate
than
using
the
average
of
the
two
because
a
difference
fuel
mixture
was
combusted
during
the
two
tests.
Fuel
analysis
information
shows
that
chlorine
content
can
vary
from
20
ppm
to
3620
ppm
for
solid
fired
units.
A
variability
factory
calculated
by
dividing
the
highest
value
by
the
lowest
value
results
in
a
value
of
181.
See
appendix
C­
7.
The
variability
factor
was
multiplied
by
the
highest
test
run
average
of
the
best
performing
unit,
0.0000996
lb/
MMBtu
(
indicated
in
Appendix
C­
3),
resulting
a
MACT
floor
HCl
emission
level
of
0.02
lb/
MMBtu.

Mercury
emissions
are
dependent
on
the
amount
of
mercury
in
the
fuel
burned.
Similar
to
the
HCl
analysis,
a
variability
factor
for
mercury
was
derived
from
the
mercury
content
of
coal.

The
fuel
dependant
variability
factor
was
derived
by
dividing
the
highest
observed
HAP
concentration
by
the
lowest
observed
HAP
concentration
from
the
utility
coal
analysis.
This
was
done
because
coal
available
to
utilities
and
industrial
boilers
and
process
heaters
are
expected
to
be
similar,
and
coal
is
the
solid
fuel
that
is
routinely
used
in
such
units
that
has
generally
the
greatest
degree
of
HAP
variability.
Once
the
fuel
dependant
variability
factors
were
calculated,

they
were
applied
to
the
average
emission
level
from
the
"
best­
controlled
similar
unit"
(
lowest
emitting)
to
derive
the
MACT
floor
level
of
control.
Available
fuel
analysis
information
shows
that
mercury
content
of
coal
boilers
varies
from
0.0254
ppm
to
0.3186
ppm.
See
Appendix
C­
7.

A
variability
factor
calculated
by
dividing
the
highest
value
by
the
lowest
results
in
a
value
of
12.54.
The
best
performing
unit
in
the
emissions
database
has
a
mercury
emission
level
of
0.00000023
lb/
MMBtu,
as
indicated
in
Appendix
C­
4.
Incorporating
the
variability
with
the
lowest
emission
level
results
in
the
MACT
floor
mercury
emission
level
of
0.000003
lb
mercury/
MMBtu.

Small
Solid
Fuel
Units
­
Heat
Inputs
Less
than
or
Equal
to
10
MMBtu/
hr.
The
emissions
database
did
not
contain
test
data
for
boilers
and
process
heaters
less
than
10
MMBtu/
hr
heat
input.
In
order
to
develop
emission
levels
for
this
subcategory,
test
data
were
data
from
units
in
33
dwcgi­
7150­
1077822954­
148160000.
wpd
the
large
solid
subcategory
were
used.
This
is
considered
an
appropriate
methodology
because
although
the
units
in
this
subcategory
are
different
enough
to
warrant
their
own
subcategory
(
i.
e.,

different
designs
and
emissions),
emissions
of
the
specific
HAP
for
which
limits
are
being
proposed
(
HCl,
mercury,
PM
and
metals)
are
expected
to
be
related
more
to
the
type
of
fuel
burned
and
the
type
of
control
used
than
to
the
unit
design.
Consequently,
emissions
test
data
from
units
greater
than
10
MMBtu/
hr
heat
input
were
used
to
establish
the
MACT
floor
levels
for
this
subcategory
for
HCl,
PM,
nonmercury
metallic
HAP
(
using
PM
as
a
surrogate),
and
mercury
because
the
fuels
and
controls
are
similar.

Because
the
same
emissions
data
for
large
units
are
used
for
the
small
subcategory,
the
MACT
floor
emission
levels
are
also
the
same.
The
MACT
floor
emission
levels
based
on
emissions
data
from
the
unit
representing
the
best
controlled
similar
source,
and
incorporating
operational
variability,
are
0.025
lb
PM/
MMBtu
or
0.0003
lb
selected
nonmercury
metals/
MMBtu,
0.000003
lb
mercury/
MMBtu,
and
0.02
lb
HCl/
MMBtu.

Limited
Use
Solid
Fuel
Units
­
Capacity
Utilizations
Less
than
or
Equal
to
10
Percent.

The
emissions
test
database
did
not
contain
test
data
for
limited
use
boilers
and
process
heaters.

In
order
to
develop
emission
levels
for
this
subcategory,
test
data
from
units
in
the
large
solid
fuel
subcategory
were
used.
This
was
considered
to
be
an
appropriate
methodology
because
although
the
units
in
this
subcategory
are
different
enough
to
warrant
their
own
subcategory
(
i.
e.,
different
purposes
and
operation),
emissions
of
the
specific
types
of
HAP
for
which
limits
are
being
proposed
(
HCl,
mercury,
and
metals)
are
expected
to
be
related
more
to
the
type
of
fuel
burned
and
the
type
of
control
used,
than
to
unit
operation.
Consequently,
emissions
information
from
the
large
solid
fuel
subcategory
could
be
used
to
establish
MACT
floor
levels
for
this
subcategory
because
the
fuels
and
controls
are
similar.

Because
the
same
emissions
data
are
used
for
limited
use
and
large
units,
the
MACT
floor
emission
levels
are
also
the
same.
The
MACT
floor
emission
levels
based
on
test
data
from
unit
representing
the
best
controlled
similar
source,
and
incorporating
operational
variability,
are
0.025
lb
PM/
MMBtu
or
0.0003
lb
metals/
MMBtu,
0.000003
lb
mercury/
MMBtu,
and
0.02
lb
HCl/
MMBtu.

Large
Liquid
Units
­
Heat
Inputs
Greater
than
10
MMBtu/
hr.
As
discussed
earlier,
a
PM
level
is
set
as
a
surrogate
for
nonmercury
metallic
HAP.
The
emissions
database
did
not
contain
test
data
for
boilers
and
process
heaters
with
ESP.
In
order
to
develop
a
PM
emission
level
for
34
dwcgi­
7150­
1077822954­
148160000.
wpd
this
subcategory,
test
data
from
oil­
fired
utility
boilers
controlled
with
ESP
were
used.
Although
the
units
in
this
subcategory
are
generally
smaller
than
utility
boilers,
emissions
of
the
specific
HAP
for
which
limits
are
being
proposed
(
PM
as
a
surrogate
for
metals)
are
expected
to
be
related
more
to
the
type
of
fuel
burned
and
the
type
of
control
used
than
to
the
size
of
the
unit.

Consequently,
emissions
test
data
from
oil­
fired
utility
boilers
could
be
used
to
establish
the
MACT
floor
levels
for
this
subcategory
for
non­
mercury
metallic
HAP
(
using
PM
as
a
surrogate)

because
the
fuels
and
controls
are
similar.
7
However,
none
of
the
utility
boilers
with
ESP's
conducted
multiple
tests.
Consequently,
a
variability
factor
could
not
be
calculated
in
the
manner
described
for
solid
units.
In
order
to
incorporate
variability
and
also
incorporate
the
best
performing
ESP,
the
highest
uncontrolled
PM
emission
level
(
0.414
lb/
MMBtu)
reported
was
multiplied
by
the
emission
reduction
achieve
be
the
best
performing
ESP
(
92
percent
reduction
in
PM).
See
Appendix
D­
1.
The
resulting
emission
limit,
0.03
lb/
MMBtu
was
used
as
the
MACT
floor
emission
level
for
PM.
Unlike
for
solid
fuel
subcategories,
liquid
fuels
that
are
low
in
metals
are
not
high
in
PM
emissions.

Therefore,
an
alternative
metals
standard
for
the
liquid
subcategories
was
not
calculated.

There
was
no
available
emissions
test
data
for
HCl
from
liquid
fuel­
fired
boilers.

Therefore,
the
available
fuel
analysis
chlorine
data
for
residual
oil
and
distillate
oil
was
identified
for
the
purpose
of
determining
a
hydrogen
chloride
emission
limit
for
new
sources
in
the
liquid
subcategory.
There
was
one
chlorine
data
point
for
distillate
oil
and
six
chlorine
data
points
available
for
residual
oil.
The
MACT
floor
emission
limit
calculations
for
HCl
were
done
using
the
highest
residual
oil
data
point
of
160
mg
chlorine/
L.
See
Appendix
D­
2.
Assuming
that
all
chlorine
in
the
fuel
would
be
emitted
as
HCl,
the
chlorine
content
value
was
converted
to
an
uncontrolled
emission
factor
of
0.009
lb
HCl/
MMBtu.

For
new
sources
in
the
large
liquid
fuel
subcategory,
the
emission
limit
is
based
on
the
performance
of
a
packed
scrubber
which
is
assumed
to
achieve
at
least
95%
reduction
of
hydrogen
chloride
(
although
some
can
achieve
up
to
99
percent
reduction).
Applying
a
95%

reduction
to
the
calculated
uncontrolled
residual
oil
emission
factor
results
in
an
HCl
limit
of
0.0005
lb/
MMBtu.

Small
Liquid
Units
­
Heat
Inputs
Less
than
or
Equal
to
10
MMBtu/
hr.
The
emissions
test
database
did
not
contain
test
data
for
liquid
fuel
boilers
and
process
heaters
less
than
10
MMBtu/
hr
heat
input
capacity.
In
order
to
develop
emission
levels
for
this
subcategory,
35
dwcgi­
7150­
1077822954­
148160000.
wpd
information
from
units
in
the
large
liquid
fuel
subcategory
was
used.
Although
the
units
in
this
subcategory
are
different
enough
to
warrant
their
own
subcategory
(
i.
e.,
different
designs
and
emissions),
emissions
of
the
specific
types
of
HAP
for
which
limits
are
being
proposed
(
HCl
and
metals)
are
expected
to
be
more
related
to
the
type
of
fuel
burned
and
the
type
of
control
than
to
unit
design.
Consequently,
emissions
information
from
units
greater
than
10
MMBtu/
hr
heat
input
capacity
could
be
used
to
establish
MACT
floor
levels
for
this
subcategory
because
the
fuels
and
controls
are
similar.
The
MACT
floor
emission
level
based
on
PM
test
data
from
a
liquid
fuel
unit
with
an
ESP
representing
the
best
controlled
similar
unit,
and
incorporating
operational
variability,
is
0.03
lb
PM/
MMBtu,
i.
e.,
the
same
as
for
large
units
because
the
same
information
is
used.
For
new
sources
in
the
small
liquid
fuel
subcategory,
the
same
methodology
described
for
large
units
was
used.
However,
the
emission
limit
is
based
on
the
performance
of
a
wet
scrubber
which
is
assumed
to
achieve
at
least
90
percent
reduction
of
hydrogen
chloride.
Applying
a
90
percent
reduction
to
the
calculated
uncontrolled
residual
oil
emission
factor
results
in
an
HCl
limit
of
0.0009
lb/
MMBtu.
The
MACT
floor
for
new
sources
in
this
subcategory
is
no
emissions
reductions
for
mercury
or
organic
HAP.

Limited
Use
Liquid
Units
­
Capacity
Utilizations
Less
than
or
Equal
to
10
Percent.
The
emissions
test
database
did
not
contain
test
data
for
limited
use
liquid
fuel
boilers
and
process
heaters.
In
order
to
develop
emission
levels
for
this
subcategory,
information
from
units
in
the
large
liquid
fuel
subcategory
was
used.
Although
the
units
in
this
subcategory
are
different
enough
to
warrant
their
own
subcategory
(
i.
e.,
different
purposes
and
operation),
emissions
of
the
specific
HAP
for
which
limits
are
being
proposed
(
HCl
and
metals)
are
more
related
to
the
type
of
fuel
burned
and
the
type
of
control
used
than
to
unit
operation.
Consequently,
emissions
information
from
units
greater
than
10
MMBtu/
hr
heat
input
capacity
could
be
used
to
establish
MACT
floor
levels
for
this
subcategory
because
the
fuels
and
controls
are
similar.
The
MACT
floor
emission
level
based
on
PM
test
data
from
a
liquid
fuel
unit
with
an
ESP
representing
the
best
controlled
similar
unit,
and
incorporating
operational
variability,
is
0.03
lb
PM/
MMBtu,
i.
e.,

the
same
as
for
large
units
because
the
same
data
is
used.
For
new
sources
in
the
limited
use
liquid
fuel
subcategory,
the
same
methodology
described
for
large
units
was
used.
However,
the
emission
limit
is
based
on
the
performance
of
a
wet
scrubber
which
is
assumed
to
achieve
at
least
90
percent
reduction
of
hydrogen
chloride.
Applying
a
90
percent
reduction
to
the
calculated
uncontrolled
residual
oil
emission
factor
results
in
an
HCl
limit
of
0.0009
lb/
MMBtu.
36
dwcgi­
7150­
1077822954­
148160000.
wpd
Gaseous
Fuel
Units.
No
existing
units
were
using
control
technologies
that
achieve
consistently
lower
emission
rates
than
uncontrolled
sources
for
any
of
the
pollutant
groups
of
interest.
Therefore,
no
limits
were
determined.

9.0
ANALYSIS
FOR
INCLUSION
OF
PROCESS
HEATERS
The
process
heaters
in
the
population
database
were
reviewed
to
determine
what
types
of
add­
on
controls
existed.
Many
of
these
units
were
either
gaseous
fuel­
fired
or
they
did
not
have
control
information
available,
both
of
which
had
no
effect
on
the
outcome
of
the
MACT
floor
analysis.
The
few
solid
fuel­
fired
process
heaters
that
did
have
control
information
used
similar
control
devices
to
those
represented
in
the
boiler
MACT
floor
analysis,
so
that
combining
these
units
into
the
overall
MACT
floor
analysis
had
no
effect
on
the
results.

Also,
there
was
very
little
emissions
test
data
available
for
process
heaters
and
even
less
available
for
process
heaters
with
the
MACT
floor
level
of
control.
An
analysis
conducted
for
the
ICCR
process
heaters
workgroup
indicates
that
available
data
show
that
boiler
emissions
are
an
adequate
surrogate
for
heater
data,
and
no
significant
differences
were
identified
in
heater
and
boiler
emissions.
8
Therefore,
the
emissions
data
used
from
boilers
to
determine
the
MACT
floor
emission
levels
was
assumed
to
be
representative
of
process
heater
emissions.

10.0
DETERMINATION
OF
HEALTH­
BASED
ALTERNATIVE
TSM
LIMIT
In
anticipation
of
the
possibility
of
including
in
the
final
rule
a
health­
based
TSM
compliance
alternative,
the
available
emission
test
data
for
solid
fuel
units
that
included
emission
results
for
all
of
the
eight
total
selected
metals
(
arsenic,
beryllium,
cadmium,
chromium,
lead,

manganese,
nickel,
and
selenium)
were
reexamined
based
on
removing
manganese
from
the
summation.
The
sum
of
the
emissions
of
the
remaining
seven
metals,
in
terms
of
lb/
MMBtu,
were
then
ranked
from
highest
to
lowest
emissions.
Then,
beginning
with
the
highest
tests,
those
tests
that
also
included
corresponding
PM
data
were
identified.
For
existing
sources,
the
highest
test
results
(
0.0003965
lb/
MMBtu)
for
metals
(
without
manganese)
having
corresponding
PM
data
available
indicated
a
PM
emission
level
of
0.0232
lb/
MMBtu.
It
is
the
same
units
that
has
the
highest
TSM
test
results
with
or
without
manganese
included.
Because
this
source
is
meeting
the
MACT
floor
PM
emission
level,
the
corresponding
health­
based
TSM
HAP
emissions
level
for
37
dwcgi­
7150­
1077822954­
148160000.
wpd
existing
sources
was
set
based
also
on
this
source.
Again,
since
the
corresponding
PM
emission
limit
(
0.023lb/
MMBtu)
is
well
under
the
MACT
floor
PM
emission
level
(
0.7lb/
MMBtu),
the
health­
based
alternative
TSM
MACT
floor
emissions
level
for
existing
sources
was
determined
by
extrapolating
the
TSM
(
excluding
manganese)
emission
level
from
this
source
based
on
the
difference
in
the
PM
emission
levels.
Therefore,
the
health­
based
TSM
(
excluding
manganese)

MACT
floor
emission
level
is
0.001
lb/
MMBtu
[
0.000397
x
(
0.07/
0.023)].
This
analysis
is
shown
in
Appendix
C­
2.

Table
6­
1.
Summary
of
MACT
Floor
Control
Technologies
Source
Subcategory
Non­
mercury
metallic
HAP
Mercury
Inorganic
HAP
Organic
HAP
Existing
Solid
Large
Fabric
Filter
Fabric
Filter
Scrubber
None
Solid
Small
None
None
None
None
Solid
Limited
ESP
or
Fabric
Filter
None
None
None
Liquid
Large
None
None
None
None
Liquid
Small
None
None
None
None
Liquid
Limited
None
None
None
None
Gas
Large
None
None
None
None
Gas
Small
None
None
None
None
Gas
Limited
None
None
None
None
New
Solid
Large
Fabric
Filter
Fabric
Filter
Packed
Bed
Scrubber
CO
limit
Solid
Small
Fabric
Filter
Fabric
Filter
Wet
Scrubber
None
Solid
Limited
Fabric
Filter
Fabric
Filter
Wet
Scrubber
CO
limit
Liquid
Large
ESP
None
Packed
Bed
Scrubber
CO
limit
Liquid
Small
ESP
None
Wet
Scrubber
None
Liquid
Limited
ESP
None
Wet
Scrubber
CO
limit
Gas
Large
None
None
None
CO
limit
Gas
Small
None
None
None
None
Source
Subcategory
Non­
mercury
metallic
HAP
Mercury
Inorganic
HAP
Organic
HAP
38
dwcgi­
7150­
1077822954­
148160000.
wpd
Gas
Limited
None
None
None
CO
limit
39
dwcgi­
7150­
1077822954­
148160000.
wpd
Table
8­
1.
Summary
of
MACT
Floor
Emission
Limits
(
lb/
MMBtu)

Source
Subcategory
Non­
mercury
metallic
HAP
Mercury
Inorganic
HAP
Organic
HAP
Existing
Solid
Large
0.07
for
PM
0.001
for
metals
0.000009
0.09
for
HCl
None
Solid
Small
None
None
None
None
Solid
Limited
0.21
for
PM
0.004
for
metals
None
None
None
Liquid
Large
None
None
None
None
Liquid
Small
None
None
None
None
Liquid
Limited
None
None
None
None
Gas
Large
None
None
None
None
Gas
Small
None
None
None
None
Gas
Limited
None
None
None
None
New
Solid
Large
0.025
for
PM
0.0003
for
metals
0.000003
0.02
for
HCl
400
ppm
CO
limit
Solid
Small
0.025
for
PM
0.0003
for
metals
0.000003
0.02
for
HCl
None
Solid
Limited
0.025
for
PM
0.0003
for
metals
0.000003
0.02
for
HCl
400
ppm
CO
limit
Liquid
Large
0.03
for
PM
None
0.0005
for
HCl
400
ppm
CO
limit
Liquid
Small
0.03
for
PM
None
0.0009
for
HCl
None
Liquid
Limited
0.03
for
PM
None
0.0009
for
HCl
400
ppm
CO
limit
Gas
Large
None
None
None
400
ppm
CO
limit
Gas
Small
None
None
None
None
Gas
Limited
None
None
None
400
ppm
CO
limit
None
40
dwcgi­
7150­
1077822954­
148160000.
wpd
10.0
References
1.
Jeanette
Alvis
and
Christy
Burlew,
ERG.
Memorandum
to
Jim
Eddinger,
U.
S.
Environmental
Protection
Agency,
OAQPS.
Development
of
the
Population
Database
for
the
Industrial/
Commercial/
Institutional
Boiler
and
Process
Heaters
National
Emission
Standards
for
Hazardous
Air
Pollutants.
October,
2002
2.
Jeanette
Alvis
and
Christy
Burlew,
ERG.
Memorandum
to
Jim
Eddinger,
U.
S.
Environmental
Protection
Agency,
OAQPS.
Development
of
the
Emissions
Test
Database
for
the
Industrial/
Commercial/
Institutional
Boiler
and
Process
Heaters
National
Emission
Standards
for
Hazardous
Air
Pollutants.
October,
2002.

3.
Roy
Oommen,
ERG.
Memorandum
to
Jim
Eddinger,
U.
S.
Environmental
Protection
Agency,
OAQPS.
Methodology
for
Estimating
Cost
and
Emissions
Impacts
for
Industrial,
Commercial,
Institutional
Boilers
and
Process
Heaters
National
Emission
Standards
for
Hazardous
Air
Pollutants.
October,
2002.

4.
Chad
Leatherwood,
ERG.
Memorandum
to
Jim
Eddinger,
U.
S.
Environmental
Protection
Agency,
OAQPS.
Development
of
Fuel
Switching
Costs
and
Emission
Reductions
for
Industrial/
Commercial/
Institutional
Boilers
and
Process
Heaters
National
Emission
Standards
for
Hazardous
Air
Pollutants.
October
2002.

5.
Petroleum
Environmental
Research
Forum.
Project
92­
19.
The
Origin
and
Fate
of
Toxic
Combustion
Byproducts
in
Refinery
Heaters
and
Boilers.

6.
Coal
mercury
data.
Working
Group
distribution
materials
on
EPA
website
for
the
Utility
MACT:
"
www.
epa.
gov/
ttn/
atw/
combust/
utiltox/
utoxpg.
html#
DA2".
January
2002.

7.
Working
Group
distribution
materials
on
EPA
website
for
the
Utility
MACT:
"
www.
epa.
gov/
ttn/
atw/
combust/
utiltox/
utoxpg.
html#
DA2".
January
2002.

8.
Jason
Huckaby,
Eastern
Research
Group.
Memorandum
to
Bill
Maxwell,
U.
S.
Environmental
Protection
Agency.
Boiler
and
Heater
Emissions
Comparison.
November
6,
1998.
41
dwcgi­
7150­
1077822954­
148160000.
wpd
APPENDIX
A
MACT
Floor
Control
Technology
Analysis
Tables
(
See
Excel
Spreadsheet
"
MACTfloorappA­
D.
xls")
42
dwcgi­
7150­
1077822954­
148160000.
wpd
APPENDIX
B
Summary
of
CO
Monitoring
Information
(
See
Excel
Spreadsheet
"
MACTfloorappA­
D.
xls")
43
dwcgi­
7150­
1077822954­
148160000.
wpd
APPENDIX
C
Emission
Limit
Analysis
Tables
for
Solid
Subcategories
(
See
Excel
Spreadsheet
"
MACTfloorappA­
D.
xls")
44
dwcgi­
7150­
1077822954­
148160000.
wpd
APPENDIX
D
Emission
Limit
Analysis
Tables
for
Liquid
Subcategories
(
See
Excel
Spreadsheet
"
MACTfloorappA­
D.
xls")
