1
August
29,
2003
Establishing
Site­
specific
Fuel
Contaminant
Input
(
FCI)
Values
for
Boiler
MACT
I.
Introduction
AF&
PA's
written
comments
on
EPA's
proposed
Boiler
MACT
rule
expressed
serious
concerns
regarding,
among
other
things,
requirements
for:
1)
setting
fuel
contaminant
input
(
FCI)
operating
limits
for
chlorides
(
Cl­),
mercury
(
Hg)
and,
when
applicable,
Total
Selected
Metals
(
TSM),
2)
identifying
"
worst­
case
fuel"
mixtures
for
conducting
performance
tests,
3)
conducting
performance
tests
when
the
maximum
FCI
for
the
fuel
mixture
will
be
below
the
applicable
emission
standard,
4)
testing
for
FCI
whenever
there
was
a
new
fuel,
change
in
fuel
supplier
or
change
in
fuel
mix,
5)
continuously
demonstrating
that
fuels
used
during
operation
result
in
lower
emissions
than
fuels
burned
during
performance
tests,
and
6)
no
provision
for
adjusting
an
FCI
established
during
the
initial
performance
test
to
an
FCI
commensurate
with
the
emission
limit.

To
address
these
concerns,
AF&
PA
prepared
a
white
paper
(
Establishing
Default
Fuel
Contaminant
Input
Values
for
Fuel
Classes
and
Performing
Compliance
Demonstrations
for
Boiler
MACT)
which
discussed
ways
in
which
default
or
site­
specific
FCI
values
could
be
employed.
Subsequently,
in
response
to
discussions
with
EPA,
AF&
PA
has
pursued
the
development
of
a
method
for
establishing
national
default
FCI
values.
As
a
demonstration
of
the
method,
NCASI
is
collecting
and
analyzing
samples
of
bark
which
is
one
of
the
primary
fuel
classes
burned
in
the
paper
and
forest
products
industry.
A
sampling
methodology
for
this
program
was
submitted
to
the
agency
(
see
A
Proposed
Data
Collection
Effort
to
Establish
Default
Fuel
Contaminant
Input
Values
for
Certain
Solid
Fuel
Classes)
and
the
program
is
now
underway.

AF&
PA
believes
that
national
default
FCI
values
should
be
established
wherever
possible.
However,
development
of
national
FCI
default
values
for
all
fuel
classes
may
not
be
feasible.
The
forest
products
and
other
industries
use
a
large
number
of
miscellaneous
solid
fuels
such
as
board
trim,
culls,
paper
cores,
and
agricultural
and
sludge
residuals
that
would
have
local
FCI
characteristics.
Additionally,
a
facility
may
want
to
establish
a
FCI
for
local
fuels.
It
is
clear
that
an
acceptable
methodology
must
be
found
for
establishing
site­
specific
FCI
values
which
addresses
the
issues
raised
in
AF&
PA
comments
noted
above.
The
purpose
of
this
paper
is
to
propose
such
a
methodology
and
to
describe
the
key
principles
for
establishing
a
site­
specific
FCI
value
that
AF&
PA
believes
are
critical
to
the
successful
implementation
of
the
Boiler
MACT
rule.

II.
Use
of
a
Site­
specific
FCI
Value
In
the
previously
submitted
white
paper
Establishing
Default
Fuel
Contaminant
Input
Values
for
Fuel
Classes
and
Performing
Compliance
Demonstrations
for
Boiler
MACT,
there
are
conceptually
three
different
uses
for
establishing
an
FCI
value.
These
are:
a.
Initial
and
continuous
compliance
demonstration
in
the
case
of
"
no
control"
2
b.
Establishing
a
"
worst
case"
fuel
mixture
for
permitting
and
performance
of
the
initial
performance
test
(
IPT)
c.
Establishing
an
FCI
"
threshold"
value
for
evaluation
of
new
fuel
mixtures
in
the
case
that
"
controls"
are
employed.

In
each
of
these
situations
an
FCI
value
must
be
established
for
each
fuel
class.
A
facility
may
choose
whether
to
conduct
fuel
testing
to
establish
a
site­
specific
FCI
value
for
a
given
fuel
class
or
elect
to
use
an
established
default
value.

III.
Establishing
the
Site­
specific
Value
AF&
PA
recommends
that
EPA
include
language
in
the
rule
identifying
basic
criteria
for
establishing
site­
specific
values.
While
some
details
of
a
particular
site's
sampling
program
may
vary
to
address
regional
and
local
issues,
AF&
PA
believes
that
a
site­
specific
sampling
plan
for
establishing
FCI
values
should
adhere
to
basic
principles
that
would
be
identified
in
the
rule.
Review
of
such
plans,
therefore,
would
be
straight­
forward
and
not
unduly
burdensome
on
the
permitting
authority.
As
such,
AF&
PA
proposes
that
the
following
principles
be
included
in
the
rule.

a.
Analytical
Methods
AF&
PA
recommends
that
the
agency
specify
required
test
methods
in
the
rule.
Analytical
and
sampling
methods
for
bark
are
discussed
in
Appendix
B
of
the
document
A
Proposed
Data
Collection
Effort
to
Establish
Default
Fuel
Contaminant
Input
values
for
Certain
Solid
Fuel
Classes
for
the
development
of
a
national
default
value
for
bark.
These
methods
should
also
apply
for
sampling
programs
conducted
on
a
site­
by­
site
basis
and
be
applicable
to
a
wide
variety
of
fuel
classes.

Standardization
of
site­
specific
sampling
and
analytical
methods
will
also
facilitate
the
consolidation
of
data
for
the
development
of
national
default
FCI
values.

b.
Site­
specific
Sampling
Plan
Criteria
1.
AF&
PA
recommends
that
the
agency
establish
a
minimum
number
of
five
(
5)
samples
for
establishing
an
FCI
on
a
site­
specific
basis.
AF&
PA
examined
various
theoretical
distributions
and
confidence
intervals
for
contaminants
as
a
function
of
the
number
of
samples
collected.
The
analysis
revealed
that
a
point
of
diminishing
returns
was
reached
very
quickly
(
between
five
and
seven
samples).
Distributions
with
low
and
high
standard
deviations
showed
that
little
improvement
was
made
(
i.
e.
the
confidence
interval
about
the
mean
was
not
reduced
significantly)
after
about
five
samples,
but
significant
improvement
(
reduction)
in
the
confidence
bounds
was
achieved
up
to
about
five.
AF&
PA
believes,
therefore,
that
the
minimum
number
of
samples
for
establishing
a
mean
FCI
value
is
five.
3
While
the
type
of
analysis
just
described
demonstrates
that
diminishing
value
is
achieved
by
analyzing
samples
in
excess
of
five,
AF&
PA
believes
that
sites
should
be
allowed
to
sample
more
than
the
minimum
if
they
so
choose.

2.
AF&
PA
believes
that
samples
should
be
collected
at
a
location
that
most
accurately
represents
the
fuel
class,
where
possible,
at
a
point
prior
to
mixing
with
other
dissimilar
fuel
types.
AF&
PA
recommends
that
the
agency
require
that
a
description
of
the
fuel,
sample
location
and,
if
applicable,
contributing
processes
be
included
in
the
site­
specific
sampling
plan
so
that
the
permitting
authority
may
be
confident
that
the
samples
are
representative
of
the
fuel
class.

3.
Analytical
and
sampling
methods
should
be
specified
in
the
site­
specific
sampling
plan
with
the
expected
MDLs.
Analytical
techniques
and
MDLs
should
correspond
to
the
contaminant
levels
expected
in
the
fuel
class.

4.
The
site­
specific
sampling
plan
should
specify
the
nature
and
frequency
of
sampling
for
each
fuel
class
(
e.
g.
grab
samples
daily,
composite
samples
weekly).
The
rationale
for
the
sampling
approach
should
be
described.

5.
When
burning
a
new
fuel
class
is
planned,
the
facility
may
either
establish
a
site­
specific
FCI
for
the
new
fuel
class
or
use
a
national,
regional,
or
fuel
supplier
specific
FCI
value
that
has
already
been
established.
If
there
are
no
FCI
values
established
for
a
fuel
class,
the
facility
would
need
to
establish
a
site­
specific
FCI
for
that
fuel.
For
example,
if
a
coal
contract
specifying
coal
from
Mine
A
changes
to
Mine
B,
the
facility
may
use
a
FCI
established
by
the
vendor,
or
a
national
default
value
for
coal,
or
conduct
the
fuel
sampling
to
establish
a
FCI
for
the
"
new"
fuel
class
from
Mine
B.

c.
Outlier
Test
AF&
PA
strongly
recommends
that
individual
values
within
a
given
data
set
that
are
suspiciously
low
or
high
should
be
subjected
to
an
outlier
test.
It
is
suggested
that
procedures
recommended
in
an
EPA
document
issued
in
2000
(
Guidance
for
Data
Quality
Assessment
­
Practical
Methods
for
Data
Analysis)
be
used
for
these
outlier
tests.
As
the
document
describes,
for
data
sets
with
more
than
25
values,
Rosner's
test
may
be
used,
and
the
Extreme
Value
Test
(
Dixon's
test)
may
be
used
for
data
sets
with
25
or
fewer
samples.
If
the
test
indicates
the
value
is
indeed
an
outlier,
it
should
be
discarded
from
the
data
set
and
supplemental
data
should
be
obtained.

d.
Average
Value
AF&
PA
recommends
that
average
values
should
be
used
for
site­
specific
FCI
determinations.
Since
data
collected
are
representative
of
site
conditions,
more
conservative
values
are
not
necessary
to
assure
long­
term
compliance.
Mean
values
should
be
calculated
in
lb/
106
BTU
(
lb/
1012BTU
for
mercury)
using
individual
sample
lb/
106
BTU
numbers.
4
e.
Verification
AF&
PA
opposes
the
concept
of
periodic
testing
for
updating
FCI
values.
Periodic
updating
of
FCI
values
would
result
in
a
moving
compliance
target
unrelated
to
actual
compliance.
Average
results
of
periodic
testing
are
expected
to
vary
in
accordance
with
statistical
principles.
As
long
as
the
results
are
within
the
expected
variability,
there
is
no
reason
to
believe
that
the
fuel
characteristics
or
the
actual
average
emission
rates
have
changed.
Constant
small
changes
in
FCI
values
would
have
no
real
meaning
or
impact
on
compliance,
and,
as
proposed
in
the
rule,
could
potentially
lead
to
costly
and
unnecessary
performance
testing.
Furthermore,
periodically
changing
FCI
default
values
could
lead
to
unnecessary
questioning
of
retrospective
compliance
status.

Instead,
AF&
PA
recommends
that
once
an
FCI
value
has
been
established
based
on
the
criteria
described
in
Item
b.
above,
that
the
value
not
be
changed
unless
there
is
reasonable
cause
to
believe
that
the
FCI
has
changed.
Reasonable
cause
could
be
an
event
that
would
be
known
to
change
the
fuel
characteristics
(
for
example,
fuel
derived
from
board
trim
where
the
resin
formulation
used
in
manufacture
of
the
bard
is
changed
to
one
containing
a
different
amount
of
salt),
or
if
verification
tests
show
that
the
FCI
has
changed.
We
recognize
that
the
agency
may
wish
to
"
verify"
the
FCI
value,
1
and,
therefore,
AF&
PA
proposes
the
following
verification
approach.

For
an
FCI
value
that
was
established
using
the
minimum
proposed
number
of
samples,
five
(
5),
a
sample
set
of
three
(
3)
should
be
collected
and
analyzed
using
equally
representative
criteria
(
location,
sample
type
(
grab
or
composite),
etc.)
as
for
the
initial
FCI
determination.
A
simplified
student's
t
test
could
be
applied
to
evaluate
whether
any
difference
between
the
two
data
sets
is
statistically
significant.
This
procedure
is
as
follows:
The
difference
between
the
verification
test
average
(
VTA)
and
the
established
fuel
contaminant
input
value
(
EFCI)
is
divided
by
0.73
times
the
standard
deviation
of
the
data
used
to
establish
the
fuel
contaminant
value
(
SEFCI)),
and
the
result
is
compared
to
the
significance
value
of
1.94.

Eq.
1
(
VTA
­
EFCI)
/
(
0.73
SEFCI)
<
1.94
2
1
Note
that
the
FCI
verification
is
not
intended
to
represent
periodic
testing.
Verification,
as
described
here,
may
be
a
one
time
only
verification
conducted
either
one
year
later
or
upon
permit
renewal.
This
verification
approach
would
also
be
applicable
to
enforcement
inspections.
2
This
equation
is
derived
from
a
simplifying
assumption
of
the
student's
t
test
that
the
standard
deviation
of
the
verification
sample
results
will
be
the
same
as
the
standard
deviation
of
the
data
set
used
to
establish
the
default
value.
The
equation
also
assumed
that
initial
data
set
contained
5
samples
and
3
verification
samples.
(
avg
x
­
avg
y)
<
A
ta/
2;
n
where
A
=
[(
nx
­
1)
sx2
+(
ny
­
1)
sy2
/
(
nx
+
ny
­
2)]
1/
2
*
[
1/
nx
+
1/
ny]
1/
2
Assuming
sx
=
sy:
A
=
s
*
[
1/
nx
+
1/
ny]
1/
2
With
nx
=
5
and
ny
=
3then:
A
=
0.73
s
5
Where
the
result
of
this
equation
is
less
than
1.94,
the
established
FCI
is
"
verified"
and
should
remain
the
same.

Where
the
result
of
this
equation
is
greater
than
1.94,
the
facility
should
re­
establish
the
FCI
using
one
of
the
following
methods:
a.
Collect
a
minimum
of
five
samples
(
either
five
new
samples
or
collect
an
additional
two
to
supplement
the
three
VTA
samples)
to
establish
a
new
FCI
for
the
fuel,
or
b.
Use
the
VTA
as
the
new
FCI3.

The
new
FCI
may
be
used
in
the
same
way
a
fuel
class
FCI
would
be
used
as
described
in
the
previously
submitted
white
paper,
Establishing
Default
Fuel
Contaminant
Input
Values
for
Fuel
Classes
and
Performing
Compliance
Demonstrations
for
Boiler
MACT.

For
example,
in
the
case
of
"
no
control"
where
the
FCI
is
used
to
demonstrate
compliance
with
the
standard:
A
new
emission
rate
will
be
calculated
using
the
new
FCI
value.
For
controlled
sources
the
FCI
threshold
approach
may
be
used
to
determine
whether
a
new
IPT
needs
to
be
conducted.
Appendix
A
shows
a
complete
calculation
example
of
how
this
would
be
applied.

f.
Option
for
Substituting
Vendor
Fuel
Certifications
AF&
PA
also
recommends
that
the
Agency
allow
the
use
of
vendor
fuel
certifications
in
lieu
of
site­
specific
sampling
and
analysis.
To
support
the
certification,
the
fuel
vendor
should
be
required
to
perform
sampling
and
analysis
of
the
fuel
using
the
same
sampling
and
analytical
methods
required
by
the
rule.

From
the
t
tables
at
 
/
2
=
0.05
and
n­
2
=
6
t
 
/
2;
n­
2
=
1.94
3
For
units
with
a
very
large
compliance
margin,
there
may
be
no
need
to
do
additional
sampling.
6
Appendix
A
Example
1:
Threshold
Determination
and
FCI
Verification
1.
Boiler
Information:

Rated
Heat
Input:
340
Million
BTU
per
hour
Control
Device:
Wet
Scrubber
­
Metals
and
HCl
No
Credit
for
Controls
­
Hg
Fuels
Burned:
Hog
Fuel
(
HF),
Coal
Permitted
Operation:
Boiler
permitted
to
burn
coal
up
to
75%
of
heat
input
HF
 
Max.
Fuel
Input
 
40
Wet
Tons/
Hour
(
WTPH)
(
100%
rated
capacity)
Site­
specific
Fuel
Cl­
Input
Value
 
0.0047
lb
Cl­/
106BTU
Fuel
Heat
Content
 
8.5x106BTU/
WT
Coal­
Max.
Fuel
Input
 
10
TPH
(
max.
allowed
feed
rate)
Site­
specific
Fuel
Cl­
Input
Value
 
0.20
lb
Cl­/
106BTU
Fuel
Heat
Content
 
26x106BTU/
Ton
2.
Determination
of
Maximum
Fuel
Chloride
Input
The
affected
source
will
comply
with
the
HCl
limit
using
the
wet
scrubber
and
establish
scrubber
operating
parameter
limits
during
the
initial
performance
test.
Coal
has
the
highest
fuel
Cl­
input
rate.
Therefore,
the
maximum
fuel
Cl­
input
will
occur
when
coal
is
fired
at
maximum
permitted
capacity
(
10
TPH
x
26x106BTU/
ton
=
260x106BTU/
hr)
and
the
remaining
rated
heat
input
to
the
boiler
is
from
hog
fuel
(
9.5
WTPH
x
8.5x106
BTU/
WT
=
80x106BTU/
hr).
Sitespecific
FCI
values
have
been
established
for
Coal
and
Hog
Fuel.

Coal
FCI
=
0.20
lb
Cl­/
106BTU
Hog
Fuel
FCI
=
0.0047
lb
Cl­/
106BTU
Coal
­
(
260x106BTU/
hr)(
0.20
lb
Cl­/
106BTU)
=
52.00
lb
Cl­/
hr
HF
­
(
80x106BTU/
hr)(
0.0047
lb
Cl­/
106BTU)=
0.38
lb
Cl­/
hr
Max.
Fuel
Cl­
Input
=
52.38
lb
Cl­/
hr
Max.
Heat
Input
­
340x106BTU/
hr
Max.
Fuel
Cl­
Input
=
52.38
lb
HCl/
hr
=
0.154
lb
Cl­/
106BTU
340x106BTU/
hr
3.
Setting
FCI
Threshold
During
Initial
Performance
Test:

During
the
initial
performance
test
the
boiler
is
operated
at
"
worst
case"
conditions
with
coal
fired
at
the
maximum
permitted
capacity.

Calculated
IPT
Fuel
Chloride
Input
(
see
item
2)
 
0.154
lb
Cl­/
106BTU
Hydrogen
Chloride
Mass
Emissions
During
IPT
 
4.6
lb
HCl/
hr
Heat
Input
During
IPT
 
340x106BTU/
hr
IPT
HCl
Emission
Rate
=
4.6
lb
Cl­/
hr
=
0.014
lb
HCl/
106BTU
(
after
scrubber)
340x106BTU/
hr
7
Scrubber
Efficiency
for
HCl
=
0.154
 
0.014
=
90.9%
0.154
Fuel
Chloride
Input
Threshold
=
IPT
Fuel
Chloride
Input
x
HCl
Emission
Std.
IPT
HCl
Emission
Rate
=
0.154
lb
Cl­/
106BTU
X
0.09
lb
HCl/
106BTU
0.014
lb
HCl/
106BTU
=
0.99
lb
Cl­/
106BTU
This
is
the
approach
referred
to
in
item
II.
c.
above:
the
development
of
a
value
which
considers
the
control
device
efficiency
to
define
the
maximum
allowable
fuel
contaminant
input
without
a
new
performance
test.
The
fuel
chloride
input
threshold
is
the
ratio
of
the
compliance
value
in
the
rule
to
the
observed
emission
rate
times
the
observed
value
during
the
test.

4A.
FCI
Verification
(
Case
1):

HF
Established
Fuel
Chloride
Input
(
EFCI)
=
0.0047
lb
Cl­/
106BTU
Verification
Test
Average
(
VTA)
=
0.007
lb
Cl­/
106BTU
Standard
Deviation
for
EFCI
(
SEFCI)
=
0.003
lb
Cl­/
106BTU
Eq.
1
(
VTA
­
EFCI)
/
(
0.73
SEFCI)
<
1.94
(
0.007
lb
Cl­/
106BTU
­
0.0047
lb
Cl­/
106BTU)
=
1.05
<
1.94
(
0.73
x
0.003
lb
Cl­/
106BTU)

Since
1.05
<
1.94
then
the
EFCI
for
Hog
Fuel
has
been
successfully
verified;
EFCI
remains
the
same
at
0.0047.

4B.
FCI
Verification
(
Case
2):

Coal
Established
Fuel
Chloride
Input(
EFCI)
=
0.20
lb
Cl­/
106BTU
Verification
Test
Average
(
VTA)
=
0.52
lb
Cl­/
106BTU
Standard
Deviation
for
EFCI
(
SEFCI)
=
0.10
lb
Cl­/
106BTU
Eq.
1
(
VTA
­
EFCI)
/
(
0.73
SEFCI)
<
1.94
(
0.52
lb
Cl­/
106BTU
­
0.20
lb
Cl­/
106BTU)
=
4.38
>
1.94
(
0.73
x
0.10
lb
Cl­/
106BTU)

Since
4.38
>
1.94
then
a
new
FCI
must
be
established
and
evaluated.

5.
Threshold
Evaluation
of
New
FCI
Using
the
heat
input
from
the
IPT
and
the
new
Coal
FCI,
a
comparison
is
made
to
the
threshold
value.
Assume
that
the
new
VTA
is
used
for
the
new
FCI.

Coal
FCI
=
0.52
lb
Cl­/
106BTU
Hog
Fuel
FCI
=
0.0047
lb
Cl­/
106BTU
Coal
­
(
260x106BTU/
hr)(
0.52
lb
Cl­/
106BTU)
=
135.20
lb
Cl­/
hr
8
HF
­
(
80x106BTU/
hr)(
0.0047
lb
Cl­/
106BTU)=
0.38
lb
Cl­/
hr
Max.
Fuel
Cl­
Input
=
135.58
lb
Cl­/
hr
Max.
Heat
Input
­
340x106BTU/
hr
Max.
Fuel
Chloride
Input
=
135.58
lb
HCl/
hr
=
0.398
lb
Cl­/
106BTU
340x106BTU/
hr
Comparing
this
value
to
the
Threshold
in
Item
3
above,
no
new
IPT
is
needed.

0.398
lb
Cl­/
106BTU
<
0.99
lb
Cl­/
106BTU
