MEMORANDUM
Date:
August
27,
2003
To:
Docket
ID
No.
A­
2000­
56
(
Iron
and
Steel
Foundries)

From:
Kevin
Cavender,
EPA
Subject:
Determination
of
the
MACT
Floor
Metal
HAP
Emission
Limits
for
Iron
and
Steel
Foundries
Purpose
The
purpose
of
this
memorandum
is
to
document
the
procedures
used
to
determine
the
MACT
floor
metal
HAP
emission
limits
for
iron
and
steel
foundries
and
to
summarize
the
results.

Conclusion

PM
is
a
reasonable
surrogate
for
metal
HAP
emissions
for
melting
and
pouring
emission
sources.


The
MACT
floor
PM
emission
limits
for
melting
furnaces
and
pouring
stations
at
existing
iron
and
steel
foundries
are
as
follows:

S
cupolas
=
0.006
grains
per
dry
standard
cubic
feet
(
gr/
dscf),

S
electric
induction
furnaces
(
EIFs)
and
scrap
preheaters
(
SPHs)
=
0.005
gr/
dscf,

S
electric
arc
furnaces
(
EAFs)]
=
0.005
gr/
dscf,
and
S
pouring
stations
=
0.010
gr/
dscf.


The
MACT
floor
PM
emission
limits
for
melting
furnaces
and
pouring
operations
at
a
new
iron
and
steel
foundry
are
as
follows:

S
cupolas
=
0.002
gr/
dscf,

S
EIFs
and
scrap
preheaters
=
0.001
gr/
dscf,

S
EAFs
=
0.002
gr/
dscf,
and
S
pouring
stations
and
pouring
areas
=
0.002
gr/
dscf.


An
equivalent
alternative
MACT
floor
metal
HAP
emission
limit
can
be
calculated
from
the
MACT
floor
PM
emission
limits
using
8%
total
metal
HAP.
2
General
Approach
A
substantial
body
of
information
is
available
on
the
types,
configurations,
and
operating
conditions
of
the
operational
units
and
air
pollution
control
devices
applied
across
the
iron
and
steel
foundry
source
category.
This
information
was
collected
through
our
comprehensive
survey
of
known
iron
and
steel
foundries
conducted
in
1998.
From
this
survey,
detailed
data
are
available
for
595
iron
and
steel
foundries
which
provided
survey
responses
(
U.
S.
EPA,
1998).
The
detailed
survey
responses
contained
information
regarding
the
type
of
process
unit,
the
air
pollution
control
device
used,
design
information
about
the
control
device,
and,
in
some
cases,
actual
performance
test
data.

Based
on
the
performance
test
data
available
for
the
melting
furnaces,
baghouses
consistently
performed
much
better
than
wet
scrubbers.
In
terms
of
both
outlet
concentrations
and
mass
of
emissions
per
ton
of
metal
melted,
the
baghouse
emission
control
systems
have
lower
PM
emissions
than
other
control
systems
(
wet
scrubber,
electrostatic
precipitators,
or
uncontrolled
units).
For
this
category
of
sources,
there
is
very
little,
if
anything,
that
can
be
done
to
meet
HAP
emission
limits
aside
from
installing
and
operating
an
effective
emission
control
system.
A
scrap
selection
and
inspection
program
can
reduce,
to
some
extent,
the
amount
of
metal
HAP
entering
a
furnace,
however,
certain
metal
HAP
are
necessary
constituents
in
the
cast
product.
The
proportion
of
the
metal
HAP
constituents
that
can
(
and
generally
are)
removed
by
a
scrap
selection
and
inspection
programs
are
a
small
fraction
of
the
total
HAP
contained
in
the
scrap
material.
As
such
the
primary
means
of
reducing
HAP
emissions
from
this
category
of
sources
is
through
effective
PM
emission
control
systems.
As
baghouses
were
used
in
well
over
12
percent
of
the
melting
furnaces,
we
concluded
that
the
wet
scrubbers
and
uncontrolled
units
were
not
among
the
best
performing
12
percent
of
existing
sources.
Therefore,
only
data
for
melting
furnaces
equipped
with
baghouses
needed
to
be
evaluated
in
detail
in
order
to
identify
the
level
of
performance
associated
with
the
MACT
floor.
Once
we
identified
the
control
systems
that
were
among
the
top­
performing
12
percent
of
units,
the
performance
of
these
systems
were
evaluated
and
ranked
individually.

We
have
interpreted
the
MACT
floor
for
existing
sources
(
i.
e.,
the
average
emission
limitation
achieved
by
the
best
performing
twelve
percent
of
existing
sources)
to
be
the
performance
achieved
by
the
median
source
of
the
top
12
percent
best
performing
sources,
or
the
6th
percentile
unit.
We
use
the
6th
percentile
unit
as
the
most
representative
estimate
of
the
average
emission
limitation
achieved
by
the
best
performing
12
percent
of
existing
sources
because
the
6th
percentile
points
to
the
performance
of
a
specific
unit
with
a
combination
of
actual
emissions
control
measures.
The
6th
percentile
unit
is
identified
by
multiplying
the
total
number
of
units
for
which
we
have
emissions
information
by
0.06.
The
resulting
unit
that
represents
the
6th
percentile
unit
is
then
identified
in
the
ranking
of
control
systems,
and
the
performance
of
this
unit
is
considered
the
MACT
floor
performance
limit
for
existing
sources.

For
new
sources,
the
MACT
floor
is
the
emission
control
that
is
achieved
in
practice
by
the
best­
controlled
similar
source.
Therefore,
the
new
source
MACT
floor
is
simply
the
performance
of
the
top
ranked
control
system.
3
Information
Available
to
Estimate
of
Control
Device
Performance
Three
types
of
information
were
generally
available
to
assess
an
individual
unit's
level
of
performance.
These
were:
actual
performance
test
data;
control
device
outlet
concentration
design
values;
and
control
device
percent
emission
reduction
design
values.
The
design
values
(
either
outlet
concentrations
or
percent
removal)
represent
the
specifications
the
control
device
vendor
used
in
designing
the
emission
control
system.
Generally,
the
foundry
operator
desires
to
achieve
a
given
emissions
limit
or
emissions
reductions.
He
relays
this
information
to
the
control
device
vendor,
and
the
control
device
vendor
designs
the
control
device
to
meet,
at
a
minimum,
the
target
emissions
performance.
These
design
values
are
often
guaranteed
performance
limits
for
which
the
control
device
vendor
may
be
liable.
If
the
performance
of
the
unit
does
not
meet
the
design
or
guaranteed
performance
requirements,
the
control
device
vendor
may
be
responsible
to
upgrade
or
replace
the
control
system
at
their
own
cost.
As
such,
these
design
values
represent
the
minimum
performance
that
these
units
will
consistently
meet.
Therefore,
we
believe
that
these
values
provide
a
reasonable
estimate
of
a
given
unit's
performance
under
the
most
adverse
circumstance
which
can
reasonably
be
expected
to
recur.

The
outlet
concentration
design
values
are
outlet
PM
concentration
values
(
either
in
gr/
acfm
or
gr/
dscf)
that
the
baghouse
vendor
gaurantees
the
system
can
meet.
As
such,
the
outlet
concentration
design
values
directly
represent
a
PM
concentration
emission
limit
that
the
control
system
could
meet
on
a
continuous
basis.
Actual
performance
test
data,
on
the
other
hand,
provides
a
measure
of
a
unit's
performance
for
a
specific
(
short)
period
of
time,
but
may
not
represent
the
actual
range
of
performance
that
the
unit
experiences
over
the
course
of
typical
operating
variations
(
e.
g.,
variations
in
production
rates
or
in
seasonal
temperature,
etc.).
Therefore,
we
applied
statistical
analyses
(
which
are
described
in
more
detail
presently)
to
estimate
the
actual
performance
limits
(
in
terms
of
outlet
PM
concentration
limits)
that
the
control
system
can
consistently
achieve.
For
units
for
which
both
types
of
emissions
information
were
available,
the
outlet
concentration
design
value
was
compared
to
the
statistically
derived
performance
limits
based
on
actual
source
test
data.
Both
types
of
information
generally
lead
to
the
same
performance
limit,
indicating
that
both
types
of
information
are
reasonable
estimates
of
actual
performance
the
units.

The
percent
emission
reduction
design
values
had
to
be
converted
to
an
outlet
PM
concentration
performance
limit
so
that
the
performance
of
the
control
devices
could
be
compared
and
ranked
on
a
consistent
basis.
We
determined
the
outlet
concentrations
associated
with
each
percent
emission
reduction
design
value
by
looking
at
the
available
test
data
for
systems
that
also
reported
a
percent
emission
reduction
design
value
and
by
calculating
the
outlet
PM
concentration
from
the
anticipated
inlet
(
uncontrolled)
PM
concentration
for
that
emission
source.
By
comparing
the
calculated
outlet
PM
concentrations
with
the
performance
limits
derived
from
actual
test
data,
we
ensured
that
the
outlet
PM
concentrations
performance
limits
calculated
from
the
percent
emission
reduction
design
values
are
reasonable
estimates
of
actual
performance
of
the
units.
4
For
each
unit,
the
(
long­
term)
emission
performance
limit
was
calculated
based
on
the
available
information.
For
some
units,
multiple
types
of
information
were
available
to
assess
the
unit's
performance.
The
following
hierarchy,
in
order
of
preference,
was
therefore
used
to
assess
the
performance
limits
for
any
given
unit.


performance
limits
based
on
actual
performance
test
data;


outlet
concentration
design
values
(
as
directly
reported);
and

performance
limits
based
on
percent
emission
reduction
design
values.

Performance
limits
based
on
actual
test
data
was
given
preference
because
the
test
data
provide
credible
evidence
of
actual
performance.
Outlet
concentration
design
values
were
given
second
priority
because
this
design
value
is
in
the
same
concentration
units
as
the
desired
performance
limits.
Nonetheless,
based
on
a
comparison
of
the
performance
limits
when
multiple
types
of
emissions
information
were
available,
we
believe
that
the
each
of
the
reported
design
values
provide
a
reasonable
estimate
of
actual
system
performance.

Statistical
Approach
for
Estimating
Long­
Term
Performance
from
Test
Data
The
MACT
floor
performance
limit
must
include
a
consideration
for
the
unavoidable
variability
inherent
in
the
process
operations
and
control
device
performance.
The
MACT
floor
must
reflect
a
level
of
performance
that
the
source
upon
which
the
floor
is
based
meets
under
the
most
adverse
circumstance
which
can
reasonably
be
expected.
Therefore,
we
must
account
for
unavoidable
variability.
However,
the
"
achievability"
test
is
not
achievability
by
anyone,
but
achievability
by
the
MACT
floor
unit.

Although
an
emissions
source
test
gives
a
good
indication
of
the
level
of
control
achieved
by
a
unit
during
the
time
of
the
emissions
test,
we
do
not
believe
a
single
emissions
source
test,
or
even
a
handful
of
tests,
can
be
used
as
an
estimate
of
the
long­
term
emission
limitation
achieved
for
that
source
due
to
normal
variations
in
process
and
control
device
performance
and
other
factors,
such
as
the
inherent
imprecision
of
sampling
and
analysis,
which
can
not
be
controlled.
Therefore,
where
emissions
source
test
data
were
available,
we
used
a
statistical
method
to
estimate
the
emission
limitation
that
a
unit
could
achieve
consistently
over
time.

For
each
furnace
where
emissions
source
test
data
were
available,
the
emission
limitation
achieved
for
that
furnace
was
estimated
at
the
99th
percentile
outlet
PM
concentration
using
a
one­
sided
z­
statistic
test
(
i.
e.,
the
emission
limitation
which
the
furnace
is
estimated
to
be
able
to
achieve
99
percent
of
the
time).
We
evaluated
several
options
to
estimate
the
standard
deviation
that
is
needed
to
perform
the
z­
statistic
test.
Ultimately,
we
estimated
an
average
relative
standard
deviation
(
RSD)
value
of
0.4
based
on
a
pooling
of
all
of
the
available
emissions
source
test
data
for
all
metal
HAP
emissions
controlled
by
baghouses.
5
Detailed
Assessment
of
the
MACT
Floor
Performance
Limits
The
available
emissions
data
are
provided
in
the
background
information
document
(
BID)
for
the
proposed
rule
(
U.
S.
EPA,
2002).
These
data
were
used
to
analyze
the
relative
standard
deviation
associated
with
each
individual
source
test.
These
data,
in
combination
with
other
emissions
information
received
in
response
to
the
detailed
industry
survey
were
used
to
estimate
the
performance
of
the
control
systems
for
a
given
emission
source.

Use
of
PM
as
a
Surrogate
for
Metal
HAP
Emissions
There
are
very
limited
data
available
to
directly
characterize
the
metal
HAP
emissions
from
foundries.
These
data
are
inadequate
to
directly
develop
emission
limits
for
specific
metal
HAP.
We
do
have,
however,
a
reasonable
number
of
PM
emissions
source
test
data
for
each
of
the
metal
HAP
emission
sources.
For
the
metal
HAP
emission
sources,
we
believe
that
PM
is
a
reasonable
surrogate
for
metal
HAP
because,
at
the
temperatures
at
which
the
metal
HAP
emission
control
devices
operate,
the
metal
HAP
are
particulate
in
nature.
Therefore,
we
believe
that
effective
control
of
PM
emissions
also
provides
effective
and
roughly
equivalent
control
of
metal
HAP
emissions.

We
conducted
source
tests
on
two
cupolas
(
one
equipped
with
a
baghouse
and
another
equipped
with
a
venturi
scrubber)
and
measured
the
PM
and
the
metal
HAP
emissions
at
the
inlet
and
at
the
outlet
of
each
control
device
(
U.
S.
EPA,
1999a
and
1999b).
The
overall
metal
HAP
removal
efficiency
measured
for
these
control
devices
was
essentially
equivalent
to
the
PM
removal
efficiency
of
the
device
tested.
The
wet
scrubber
achieved
an
overall
metal
HAP
removal
efficiency
of
94.2
percent
while
averaging
a
95.7
percent
PM
removal
efficiency.
The
baghouse
achieved
an
overall
metal
HAP
removal
efficiency
of
99.84
percent
while
averaging
a
99.88
percent
PM
removal
efficiency.
Additionally,
both
of
the
primary
metal
HAP
constituents
in
these
cupola
exhaust
streams
(
manganese
and
lead)
each
exhibited
constituent­
specific
emission
reductions
that
were
essentially
equivalent
to
the
PM
removal
efficiency
of
the
control
device.
These
data
support
our
hypothesis
that
PM
is
a
reasonable
surrogate
for
metal
HAP
emissions
and
that
effective
control
of
PM
emissions
also
provides
effective
control
of
metal
HAP
emissions.

Although
we
have
comparative
PM
to
metal
HAP
emission
reduction
data
only
for
cupola
emission
control
system,
we
have
no
reason
to
believe
that
there
is
any
different
relationship
between
metal
HAP
amd
PM
emissions
from
any
other
emission
point
in
the
foundry.
Thus,
our
data
also
suggests
that
PM
is
a
reasonable
surrogate
for
metal
HAP
emissions
for
all
of
the
metal
HAP
emission
sources.
First,
all
of
the
metal
HAP
emission
sources
(
cupola,
EAF,
EIF
and
pouring
operations)
are
sources
of
metal
fume
emissions
with
similar
composition
based
on
a
comparison
of
outlet
PM
and
metal
HAP
data
for
these
various
sources.
Second,
the
percent
of
metal
HAP
emissions
that
is
not
particulate
in
nature
(
and
therefore
may
not
be
well
correlated
to
PM
emissions)
is
expected
to
be
the
highest
in
control
systems
with
higher
operating
temperatures.
As
cupola
control
systems
generally
operate
at
much
higher
temperatures
than
control
systems
used
for
EAFs,
EIFs,
and
pouring
stations/
areas,
the
cupola
emissions
data
would
6
be
the
most
likely
to
exhibit
poor
metal
HAP
control
for
volatile
metals
(
such
as
lead)
while
achieving
substantial
PM
emission
control.
Since
the
emissions
data
suggest
that
the
cupola
control
systems
effectively
control
these
metal
HAP
emissions
(
commensurate
with
the
PM
control
efficiency
suggesting
the
metal
HAP
is
associated
with
the
PM),
it
is
reasonable
to
conclude
that
the
metal
HAP
emissions
within
the
control
systems
used
for
all
of
the
metal
HAP
emission
sources
are
also
particulate
in
nature
and
that
effective
control
of
PM
emissions
by
these
systems
also
provides
effective
control
of
metal
HAP
emissions.

Pooled
RSD
Value
for
Metal
HAP
Emission
Sources
We
evaluated
several
options
to
estimate
the
standard
deviation
that
is
needed
to
perform
the
z­
statistic
test.
We
decided
not
to
estimate
the
standard
deviation
for
each
furnace
separately
based
on
the
available
emissions
data
for
just
that
furnace
since
most
furnaces
only
have
three
data
points
to
use
in
estimating
the
standard
deviation,
one
data
point
for
each
run
in
a
three
run
emissions
source
test.
Instead,
we
calculated
the
relative
standard
deviation
(
RSD)
for
each
source
test,
which
is
the
standard
deviation
of
the
individual
source
test
runs
divided
by
the
average
of
the
individual
source
test
runs.
This
statistic
appeared
to
be
relatively
consistent
across
units
that
had
widely
different
average
outlet
PM
concentrations.
There
was
no
pattern
suggesting
that
the
best
performing
units
had
less
variability
(
in
terms
of
their
RSD
values)
than
other
baghouse
controlled
units.

Table
1
presents
the
relative
standard
deviation
(
RSD)
values
calculated
for
baghouse
controlled
units
for
which
emission
source
test
data
were
available.
The
RSD
values
were
averaged
for
each
type
of
metal
HAP
emission
source
and
for
the
pooled
set
of
all
RSD
values
for
all
emission
sources.
An
analysis
of
variance
(
ANOVA)
was
performed
on
the
data
and
there
was
no
statistically
significant
difference
in
the
standard
deviation
estimates
for
the
different
metal
HAP
emission
sources.
There
are
three
general
sources
of
variability
during
an
emission
source
test.
These
are:
1)
variability
in
the
process
operations,
such
as
melt
rate,
that
might
affect
the
emissions
rate
from
the
process;
2)
variability
in
control
device
performance;
and
3)
variability/
uncertainty
associated
with
the
sampling
and
analysis
method.
Since
these
test
data
are
generated
for
the
same
basic
type
of
control
system
(
fabric
filters/
baghouse
systems)
using
the
same
test
methods,
the
latter
two
sources
of
variability
are
similar
for
the
different
emission
sources.
It
appears
that
either
these
latter
two
sources
of
variability
overwhelm
the
process
variability
or
the
process
variability
for
these
different
sources
are
roughly
equivalent.
In
any
case,
the
relative
standard
deviation
appears
to
be
consistent
for
all
MACT
control
systems
regardless
of
the
type
melting
furnace
generating
the
emissions.
There
is
limited
data
available
for
pouring
stations,
which
makes
the
average
RSD
value
for
this
emission
source
more
uncertain
than
the
RSD
values
for
the
melting
furnaces.
The
individual
RSD
values
for
the
pouring
stations
fall
within
the
same
range
as
the
RSD
values
for
the
melting
furnaces
and
the
median
value
is
similar
to
the
average
and
median
values
for
the
other
emission
sources.

After
evaluating
these
data,
we
decided
not
to
estimate
the
standard
deviation
for
a
specific
metal
HAP
emission
source
based
on
just
the
data
available
for
that
source
because
we
have
very
limited
information
on
some
sources
(
EAFs
and
pouring
stations)
and
because
the
standard
deviation
estimates
for
the
different
emission
sources
were
not
statistically
different.
7
Additionally,
there
are
many
common
sources
of
variability
in
the
emission
test
results
for
these
different
emission
sources.
We
elected,
therefore,
to
use
the
pooled
RSD
value
for
all
metal
HAP
emission
sources.
Also,
when
considering
the
uncertainty
associated
with
the
average
RSD
value,
we
decided
to
round
the
average
RSD
value
to
one
significant
digit.
Consequently,
an
average
RSD
value
of
0.4
was
used
to
calculate
the
99th
percentile
performance
limits
(
using
the
zstatistic
for
all
metal
HAP
emission
sources.

Table
1.
Summary
of
Relative
Standard
Deviations
Calculated
from
Source
Test
Data
for
Metal
HAP
Emission
Sources.

Unit
Number
Relative
Standard
Deviation
(
RSD)
Value
for
Unit
Cupola
Baghouse
EIF
Baghouse
EAF
Baghouse
Pouring
Baghouse
1
0.43
0.36
0.34
0.37
2
0.18
0.14
0.09
0.29
3
1.56
0.45
0.49
0.18
4
0.37
0.33
0.51
1.13
5
0.74
0.37
0.20
0.52
6
0.60
0.64
1.08
7
0.36
0.58
0.35
8
0.11
0.23
0.06
9
0.32
0.35
0.11
10
0.58
0.45
0.13
11
0.23
0.81
0.17
12
0.34
0.07
0.28
13
0.27
0.00
14
0.01
0.08
15
0.24
0.33
16
0.17
0.58
17
0.21
0.41
Average
by
Emission
Source
0.39
0.36
0.32
0.50
Pooled
Average
for
all
Emission
Sources
0.38
EIF
=
electric
induction
furnace
EAF
=
electric
arc
furnace
8
By
using
the
pooled
RSD
value
rounded
to
one
significant
digit,
the
RSD
value
used
becomes
insensitive
to
any
single
test
result.
Additionally,
by
pooling
the
data,
there
are
an
adequate
number
of
data
points
to
warrant
the
use
of
the
z­
statistic
in
calculating
the
performance
limits.
Without
using
the
pooled
data
set,
the
calculated
RSD
values
can
be
very
sensitive
to
a
single
test
value,
especially
for
small
data
groups.
Furthermore,
the
statistic
used
to
calculate
the
confidence
interval
for
these
small
data
groups
(
the
t­
statistic)
is
a
function
of
the
number
of
values
within
the
group.
Consequently,
without
pooling
the
data,
the
99th
percentile
performance
limits
(
and
subsequently
the
MACT
floor
performance
limits)
could
be
very
sensitive
to
a
single
performance
test
value
and
the
MACT
floor
determinations
would
have
to
be
reassessed
each
time
new
data
become
available.
Thus,
there
are
both
statistical
and
practical
reasons
to
use
the
pooled
data
set
in
developing
the
performance
limits
based
on
the
99th
percentile
performance.

The
following
equation
was
used
to
calculate
the
performance
limit
from
the
average
PM
concentration
from
a
single
emissions
source
test:

Performance
Limit
=
Average
PM
Concentration
×
(
1
+
RSD
pooled
×
z­
statistic).

The
value
of
the
z­
statistic
for
a
one­
sided
99
percent
confidence
level
is
2.326.
Given
the
pooled
RSD
value
of
0.4,
the
performance
limit
is
simply:

Performance
Limit
=
Average
PM
Concentration
×
1.93.

MACT
Floor
for
Cupolas
Emissions
information
was
available
for
143
cupolas.
Based
on
industry
survey
data
and
information
regarding
new
control
device,
there
are
62
cupolas
(
43
percent)
controlled
by
baghouses,
71
(
50
percent)
controlled
by
venturi
scrubbers,
1
(
1
percent)
controlled
by
an
ESP,
and
8
(
6
percent)
that
are
uncontrolled
for
metal
HAP.
We
also
have
emissions
data
for
PM
from
source
tests
conducted
on
36
cupolas:
12
controlled
by
baghouses;
23
controlled
by
wet
scrubbers;
and
1
controlled
by
an
electrostatic
precipitator
(
ESP).

We
compared
the
available
PM
emissions
data
for
cupolas
using
baghouses,
wet
scrubbers,
and
the
ESP
in
terms
of
control
efficiency,
outlet
concentration,
and
mass
emissions
per
ton
of
metal
melted.
Based
on
all
of
these
indicators
of
performance,
we
determined
that
cupolas
with
baghouses
perform
substantially
better
than
cupolas
with
scrubbers
or
ESPs
in
controlling
PM
emissions.
For
example,
the
worst
performing
baghouse
system
had
an
average
outlet
PM
concentration
that
was
half
of
the
value
of
the
best
performing
wet
scrubber.
Similarly,
on
the
basis
of
pounds
of
PM
emitted
per
ton
of
metal
melted,
all
of
the
baghouse
controlled
units
performed
as
well
or
better
than
the
best
performing
wet
scrubber.
Based
on
the
available
emissions
data,
baghouses
are
clearly
the
better
performers
of
the
three
control
technologies
used
to
control
PM
and
HAP
metal
emissions
from
cupolas.
Consequently,
units
with
venturi
scrubbers
and
ESPs
were
not
included
in
the
detailed
ranking
of
the
best
performing
units
because
the
available
emissions
source
test
data
clearly
demonstrated
that
the
furnaces
controlled
with
these
devices
were
not
among
the
best
performing
12
percent
of
sources.
9
We
therefore
ranked
the
available
emissions
information
for
units
controlled
with
baghouses.
The
performance
limits
for
cupola
units
that
only
reported
percent
removal
efficiencies
for
their
control
systems
were
assigned
as
follows:


systems
reporting
removal
efficiencies
of
99
%
or
greater
were
assigned
a
performance
limit
of
0.007
gr/
dscf;


systems
reporting
removal
efficiencies
less
than
99
%
were
assigned
a
performance
limit
of
0.015
gr/
dscf.

These
performance
limits
were
based
on
both
an
examination
of
the
available
performance
data
for
cupola
units
that
also
reported
a
design
percent
removal
efficiency
and
on
the
calculation
of
performance
limits
using
the
reported
design
percent
removal
efficiency
for
the
unit's
control
system
and
the
uncontrolled
emission
factor
for
cupolas.
By
comparing
the
calculated
outlet
PM
concentrations
with
the
performance
limits
derived
from
actual
test
data,
we
ensured
that
the
outlet
PM
concentrations
performance
limits
calculated
from
the
percent
emission
reduction
design
values
are
reasonable
estimates
of
actual
unit's
performance.

Table
2
presents
the
ranking
of
the
top­
performing
cupola
units.
The
6th
percentile
unit,
which
characterizes
the
MACT
floor
for
existing
sources,
is
the
9th
ranked
cupola
(
143
×
0.06
=
9).
The
performance
limit
achieved
by
this
unit
is
an
outlet
PM
concentration
of
0.006
gr/
dscf.
The
top­
ranked
cupola
achieved
a
performance
limit
of
0.002
gr/
dscf,
which
is
the
MACT
floor
for
new
sources.

Table
2.
Ranking
of
Best
Performing
Cupola
Control
Systems
Cupola
Rank
Number
Performance
Limit
(
gr/
dscf)
Comment
1
0.002
New
source
MACT
Floor;
"
horizontal"
baghouse
2
0.002
"
horizontal"
baghouse
3
0.003
4
0.004
5
0.004
6
0.004
7
0.006
8
0.006
9
0.006
Existing
source
MACT
Floor
10
0.006
11
0.007
12
0.007
10
The
two
cupolas
that
achieved
99th
percentile
outlet
PM
concentrations
of
0.002
gr/
dscf
both
employed
a
novel
pulse­
jet
baghouse
with
horizontally
supported
bags
rather
than
the
traditionally
designed
vertically
hanging
bags.
According
to
an
operator
of
one
of
these
novel
baghouses
(
Cavender,
2002),
a
lighter
weight
fabric
can
be
used
when
the
bags
are
horizontally
supported.
When
bags
hang
vertically
(
as
in
traditional
baghouses),
the
tops
of
the
bags
must
be
strong
enough
to
hold
up
the
weight
of
the
entire
bag
(
generally
2
or
3
ft
long)
and
the
entire
filter
cake
on
that
bag.
A
light­
weight
bag
would
not
be
able
to
support
the
weight,
and
would
tear.
By
having
the
bags
supported
horizontally,
they
are
able
to
reduce
the
weight
the
bag
material
supports
to
only
the
small
amount
under
the
horizontal
support
(
typical
bags
are
4
to
6
inches
in
diameter).
The
light­
weight
bag
is
easier
to
clean
and
is
more
permeable,
which
allows
for
a
more
even
distribution
of
the
air
flow.
Heavier
weight
bags
tend
to
get
more
material
caught
in
the
bag
material,
and
as
a
result
need
to
be
cleaned
more
frequently
and
more
vigorously.
The
contact
indicated
that,
"
since
80%
of
emissions
are
associated
with
cleaning,"
by
lowering
the
cleaning
frequency,
the
baghouse
emissions
are
lowered.
The
light­
weight
bag
is
also
more
permeable,
so
that
pressure
drop
is
reduced,
and
air
flow
is
more
evenly
distributed.
This,
along
with
the
low
A/
C
ratio
for
these
baghouses,
allows
more
of
the
PM
material
to
be
collected
on
the
bag
surface,
rather
than
becoming
impregnated
into
the
fabric,
making
it
easier
to
clean
the
bags.

MACT
Floor
for
Electric
Induction
Furnaces
(
EIFs)
and
Scrap
Preheaters
(
SPHs)

There
are
1,394
EIFs
at
iron
and
steel
foundries
that
provided
survey
responses
to
the
comprehensive
survey
of
iron
and
steel
foundries.
There
are
also
177
SPHs
at
iron
and
steel
foundries,
and
all
these
SPHs
are
used
specifically
with
EIF
melting
operations.
Baghouses
and
cartridge
filters
(
or
fabric
filters)
are
used
for
controlling
melting
operations
for
388
EIFs
(
28
percent);
wet
scrubbers
are
used
for
21
EIFs
(
1.5
percent);
and
cyclones
are
used
for
2
EIFs
(
0.1
percent).
Of
the
177
SPHs
used
at
iron
and
steel
foundries,
64
are
controlled
by
baghouses,
11
are
controlled
by
cyclones,
and
2
are
controlled
by
wet
scrubbers.
Of
the
64
SPHs
that
are
controlled
by
baghouses,
59
are
employed
in
conjunction
with
EIFs
that
are
also
equipped
with
baghouses.
Of
those
59
SPHs,
43
are
controlled
by
the
same
baghouses
as
their
associated
EIF.
Because
SPHs
are
used
in
conjunction
with
EIFs
and
because
PM
emissions
from
SPHs
are
typically
controlled
with
the
same
control
device
used
to
control
the
PM
emissions
from
their
associated
EIF,
we
elected
to
establish
a
single
MACT
limit
for
both
EIFs
and
SPHs.

We
have
credible
emissions
source
test
data
for
57
EIFs
(
and
15
SPHs)
controlled
by
19
fabric
filters
(
17
baghouses
and
2
cartridge
filter),
2
EIFs
controlled
by
venturi
scrubbers,
2
EIFs
controlled
by
cyclones,
and
7
uncontrolled
EIFs.
We
ranked
these
data
in
terms
of
outlet
PM
concentration
achieved.
Again,
the
performance
of
units
controlled
by
a
baghouse
was
significantly
better
than
the
performance
of
units
that
controlled
by
other
control
systems
or
units
that
had
no
emissions
control.
Given
the
predominant
use
of
baghouses
and
the
relative
performance
of
the
baghouse
controlled
units
compared
to
other
SPH/
EIF
units,
we
only
needed
to
consider
units
controlled
with
fabric
filters
to
identifying
the
top­
performing
units.
11
We
calculated
the
performance
limit
for
each
unit
using
the
available
performance
data
and
reported
control
device
design
values.
The
performance
limits
for
EIF
and
SPH/
EIF
units
that
only
reported
percent
removal
efficiencies
for
their
control
systems
were
assigned
as
follows:


systems
reporting
removal
efficiencies
greater
than
99.95
%
were
assigned
a
performance
limit
of
0.002
gr/
dscf;


systems
reporting
removal
efficiency
between
99.5
%
and
99.95
%
(
inclusive)
were
assigned
a
performance
limit
of
0.005
gr/
dscf;


systems
reporting
removal
efficiencies
less
than
99.5
%
were
assigned
a
performance
limit
of
0.015
gr/
dscf.

These
performance
limits
were
based
on
both
an
examination
of
the
available
performance
data
for
EIF
units
that
also
reported
a
design
percent
removal
efficiency
and
on
the
calculation
of
performance
limits
using
the
reported
design
percent
removal
efficiency
for
the
unit's
control
system
and
the
uncontrolled
emission
factor
for
EIFs.
By
comparing
the
calculated
outlet
PM
concentrations
with
the
performance
limits
derived
from
actual
test
data,
we
ensured
that
the
outlet
PM
concentrations
performance
limits
calculated
from
the
percent
emission
reduction
design
values
are
reasonable
estimates
of
actual
unit's
performance.

Table
3
presents
the
ranking
of
the
top­
performing
EIFs.
Based
on
1,394
EIF
and
SPH/
EIF
emission
sources,
the
6th
percentile
would
be
represented
by
the
84th
best
performing
unit
(
1,394
×
0.06
=
84).
Based
on
our
ranking
of
the
performance
limits
achieved
by
the
existing
EIFs
and
SPH/
EIFs,
we
determined
that
the
MACT
floor
for
metal
HAP
control
at
existing
sources
is
a
PM
emission
concentration
of
0.005
gr/
dscf.
The
new
source
MACT
floor
is
represented
by
the
top­
ranked
EIF
or
SPH/
EIF
unit,
which
is
a
performance
limit
of
0.001
gr/
dscf.
Again,
the
control
system
operated
by
the
best
performing
source
is
technologically
different
than
the
traditionally
operated
baghouse
control
system.
In
this
case,
a
series
of
traditional
fabric
filter
control
technologies
(
a
baghouse
and
a
cartridge
filter)
is
used
with
a
high­
efficiency
particulate
arresting
(
HEPA)
filter
for
the
top­
performing
EIF
emission
source.

[
Note:
if
SPHs
are
treated
as
separate
sources,
the
MACT
floor
for
both
new
and
existing
sources
would
be
the
same
as
the
combined
source.
Also,
if
we
considered
that
there
are
1,574
SPH/
EIF
sources
(
1,394+
177),
the
existing
source
MACT
floor,
which
would
be
the
94th
topranked
source,
would
still
be
0.005
gr/
dscf].

MACT
Floor
for
Electric
Arc
Furnaces
(
EAFs)

Based
on
the
information
collected
through
our
comprehensive
survey
of
iron
and
steel
foundries,
there
are
83
iron
and
steel
foundries
(
out
of
595
respondents)
that
reported
using
a
total
of
163
melting
EAFs.
Of
these
163
EAFs,
baghouses
are
used
to
control
melting
emissions
for
160
EAFs
(
98
percent),
with
only
3
EAFs
uncontrolled
for
melting.
12
Table
3.
Ranking
of
Best
Performing
EIF
and
SPH/
EIF
Control
Systems
EIF
or
SPH/
EIF
Rank
Number
Performance
Limit
(
gr/
dscf)
Comment
1
thru
8
0.001
New
source
MACT
Floor;
control
system
of
a
baghouse,
cartridge
filter,
and
HEPA
filter
operated
in
series.

9
thru
27
0.001
28
thru
48
0.002
49
ane
50
0.003
51
thru
67
0.004
68
thru
147
0.005
Existing
source
MACT
Floor
Outlet
PM
concentration
data
are
available
for
10
baghouses
that
are
used
to
control
the
emissions
from
23
EAFs
operated
by
iron
and
steel
foundries.
No
emissions
data
were
available
for
uncontrolled
units.
However,
it
is
reasonable
to
assume
that
the
controlled
units
perform
better
than
uncontrolled
units.
Furthermore,
as
98
percent
of
the
units
are
controlled
by
a
baghouse
system,
it
is
reasonable
to
expect
that
the
3
uncontrolled
EAFs
were
not
among
the
best
performing
units.

We
calculated
the
performance
limit
for
each
EAF
using
the
available
performance
data
and
reported
control
device
design
values.
The
performance
limits
for
EAFs
that
only
reported
percent
removal
efficiencies
for
their
control
systems
were
assigned
as
follows:


systems
reporting
removal
efficiency
of
99.5
%
or
greater
were
assigned
a
performance
limit
of
0.005
gr/
dscf;


systems
reporting
removal
efficiencies
less
than
99.5
%
were
assigned
a
performance
limit
of
0.015
gr/
dscf.

These
performance
limits
were
based
on
both
an
examination
of
the
available
performance
data
for
EAF
units
that
also
reported
a
design
percent
removal
efficiency
and
on
the
calculation
of
performance
limits
using
the
reported
design
percent
removal
efficiency
for
the
unit's
control
system
and
the
uncontrolled
emission
factor
for
EAFs.
By
comparing
the
calculated
outlet
PM
concentrations
with
the
performance
limits
derived
from
actual
test
data,
we
ensured
that
the
outlet
PM
concentrations
performance
limits
calculated
from
the
percent
emission
reduction
design
values
are
reasonable
estimates
of
actual
unit's
performance.

Table
4
presents
the
ranking
of
the
top­
performing
EAFs.
As
there
is
emissions
information
on
163
EAF
sources,
the
6th
percentile
would
be
represented
by
the
10th
best
performing
unit
(
163
×
0.06
=
10).
Based
on
our
ranking
of
the
emission
limitation
achieved
by
the
existing
EAFs,
we
determined
that
the
MACT
floor
for
metal
HAP
control
at
existing
EAF
sources
is
a
PM
emission
concentration
of
0.005
gr/
dscf.
The
new
source
MACT
floor
is
13
represented
by
the
top­
ranked
EAF,
which
is
a
performance
limit
of
0.002
gr/
dscf.
Unlike
the
new
source
MACT
floor
control
systems
for
cupola
and
EIF,
there
is
no
apparent
technological
difference
in
the
control
systems
used
for
the
highest­
ranked
EAF
and
the
other
top­
ranked
EAF
control
systems
(
all
are
traditionally
designed
baghouses).
It
is
clear,
however,
that
new
EAF
units
could
meet
this
limit
with
either
traditional
baghouse
control
systems
or
with
the
advanced
systems
that
are
currently
used
for
some
cupolas
and
EIFs.

Table
4.
Ranking
of
Best
Performing
EAF
Control
Systems
EAF
Rank
Number
Performance
Limit
(
gr/
dscf)
Comment
1
0.002
New
source
MACT
Floor
2
0.003
3
0.003
4
0.003
5
0.003
6
0.004
7
0.004
8
0.004
9
0.004
10
0.005
Existing
source
MACT
Floor
11
0.005
12
0.005
MACT
Floor
for
Pouring
Stations
We
have
information
on
1,317
pouring
stations
from
our
survey
of
the
industry.
Baghouses
are
used
to
control
178
(
or
13.5
percent)
of
these
pouring
stations
and
wet
scrubbers
are
used
to
control
35
(
or
2.7
percent)
of
the
pouring
stations.
The
majority
of
pouring
stations
(
1,104
pouring
stations
or
83.8
percent)
do
not
control
PM
(
or
metal
HAP)
emissions.

Outlet
PM
concentration
data
are
available
for
8
baghouses
used
to
control
16
pouring
station
emissions
and
2
wet
scrubbers
used
to
control
5
pouring
station
emissions.
Because
there
was
some
overlap
in
the
performance
of
the
baghouse
and
wet
scrubber
control
systems
when
applied
to
pouring
station
emission
control,
and
because
of
the
limited
number
of
controlled
units,
we
included
the
performance
data
for
both
baghouses
and
wet
scrubbers
when
ranking
the
performance
of
pouring
stations.
We
calculated
the
performance
limit
for
each
pouring
station
using
the
available
performance
data
and
the
reported
control
device
design
values.
The
performance
limits
for
pouring
stations
that
only
reported
percent
removal
efficiencies
for
their
control
systems
were
assigned
as
follows:
14

systems
reporting
removal
efficiency
of
99.8
%
or
greater
were
assigned
a
performance
limit
of
0.002
gr/
dscf;


systems
reporting
removal
efficiency
of
99.5
%
or
greater
but
less
than
99.8%
were
assigned
a
performance
limit
of
0.005
gr/
dscf;


systems
reporting
removal
efficiency
of
99
%
or
greater
but
less
than
99.5%
were
assigned
a
performance
limit
of
0.010
gr/
dscf;


systems
reporting
removal
efficiencies
less
than
99
%
were
assigned
a
performance
limit
of
0.020
gr/
dscf.

These
performance
limits
were
based
on
both
an
examination
of
the
available
performance
data
for
pouring
stations
that
also
reported
a
design
percent
removal
efficiency
and
on
the
calculation
of
performance
limits
using
the
reported
design
percent
removal
efficiency
for
the
unit's
control
system
and
the
uncontrolled
emission
factor
for
pouring
stations.
By
comparing
the
calculated
outlet
PM
concentrations
with
the
performance
limits
derived
from
actual
test
data,
we
ensured
that
the
outlet
PM
concentrations
performance
limits
calculated
from
the
percent
emission
reduction
design
values
are
reasonable
estimates
of
actual
unit's
performance.

Table
5
presents
the
ranking
of
the
top­
performing
pouring
stations.
The
6th
percentile
of
1,317
sources
is
the
performance
of
the
79th
best
performing
unit.
Based
on
our
ranking
of
the
emission
limitation
achieved
by
these
pouring
stations,
we
determined
that
the
MACT
floor
for
metal
HAP
control
at
existing
sources
is
a
PM
emission
concentration
of
0.010
gr/
dscf.
The
new
source
MACT
floor
is
represented
by
the
top­
ranked
pouring
station
control
system,
which
is
a
performance
limit
of
0.002
gr/
dscf.

Table
5.
Ranking
of
Best
Performing
Pouring
Station
Control
Systems
Pouring
Station
Rank
Number
Performance
Limit
(
gr/
dscf)
Comment
1
thru
18
0.002
New
source
MACT
Floor;
all
of
these
pouring
stations
use
baghouses.
19
and
20
0.003
Baghouses
21
and
22
0.004
Baghouses
23
thru
37
0.005
Baghouse
used
for
13
pouring
stations;
wet
scrubber
used
for
2
pouring
stations.
38
and
39
0.006
Baghouses
40
thru
43
0.007
Baghouses
40
thru
43
0.008
Baghouse
used
for
3
pouring
stations;
wet
scrubber
used
for
1
pouring
station.
49
thru
53
0.009
Baghouses
54
thru
146
0.010
Existing
source
MACT
Floor;
baghouse
used
for
78
pouring
stations;
wet
scrubber
used
for
15
pouring
stations.
147
thru
213
0.020
15
MACT
Floor
for
Pouring
Areas
We
have
information
on
435
pouring
areas
from
the
foundry
industry
survey.
Baghouses
are
used
to
control
20
(
or
4.6
percent)
of
these
pouring
areas
and
wet
scrubbers
are
used
to
control
2
(
or
0.5
percent)
of
the
pouring
areas.
A
total
of
413
(
or
95
percent)
of
the
435
pouring
areas
do
not
control
pouring
emissions.

Outlet
PM
concentration
data
are
available
for
only
one
pouring
area.
Due
to
the
lack
of
data
for
pouring
areas,
we
used
the
same
algorithm
developed
for
pouring
stations
to
estimate
the
performance
limits
for
pouring
areas
that
only
reported
percent
removal
efficiencies
for
their
control
systems.
We
believe
that
this
is
reasonable
because
of
the
similarities
in
the
emissions
source
and
the
control
systems,
when
used,
for
pouring
stations
and
pouring
areas.
Specifically,


systems
reporting
removal
efficiency
of
99.8
%
or
greater
were
assigned
a
performance
limit
of
0.002
gr/
dscf;


systems
reporting
removal
efficiency
of
99.5
%
or
greater
but
less
than
99.8%
were
assigned
a
performance
limit
of
0.005
gr/
dscf;


systems
reporting
removal
efficiency
of
99
%
or
greater
but
less
than
99.5%
were
assigned
a
performance
limit
of
0.010
gr/
dscf;


systems
reporting
removal
efficiencies
less
than
99
%
were
assigned
a
performance
limit
of
0.020
gr/
dscf.

We
calculated
the
performance
limit
for
each
pouring
area
with
a
control
system.
As
with
pouring
stations,
data
for
both
baghouse
and
wet
scrubber
controlled
pouring
areas
were
included
in
the
ranking.
Table
6
presents
the
ranking
of
the
top­
performing
pouring
areas.
The
6th
percentile
of
413
sources
is
the
performance
of
the
25th
best
performing
unit.
As
only
22
pouring
areas
have
control
systems,
the
pouring
area
that
reflects
the
MACT
floor
does
not
control
emissions.
Therefore,
the
MACT
floor
for
pouring
areas
at
existing
sources
is
no
reduction
in
metal
HAP
(
or
PM)
emissions.
The
new
source
MACT
floor
is
represented
by
the
top­
ranked
pouring
area
control
system,
which
is
a
performance
limit
of
0.002
gr/
dscf.

Table
6.
Ranking
of
Best
Performing
Pouring
Area
Control
Systems
Pouring
Area
Rank
Number
Performance
Limit
(
gr/
dscf)
Comment
1
and
2
0.002
New
source
MACT
Floor;
both
of
these
pouring
areas
use
baghouses.
3
0.005
Wet
scrubber
4
0.007
Baghouse
5
thru
13
0.010
Baghouses
14
thru
22
0.020
Baghouse
used
for
8
pouring
areas;
wet
scrubber
used
for
1
pouring
area.
23
thru
413
No
emission
reduction
Existing
source
MACT
Floor;
No
controls
used
for
these
pouring
areas.
16
Determination
of
Alternative
MACT
Floor
Performance
Limits
Although
we
believe
that
PM
is
a
good
surrogate
for
metal
HAP
emission
sources
at
iron
and
steel
foundries,
there
are
certain
circumstances
in
which
high
PM
emissions
may
not
necessarily
translate
into
high
metal
HAP
emissions.
For
example,
a
foundry
may
use
dry
scrubbing
techniques
to
control
sulfur
dioxide
(
SO
2)
emissions
that
can
result
in
an
increase
in
PM
emissions
without
a
corresponding
increase
in
metal
HAP
emissions.
The
PM
emission
limits
for
the
iron
and
steel
foundries
may
have
the
unintended
consequence
of
discouraging
dry
scrubbing
techniques
for
reducing
SO
2
emissions,
which
may
be
particularly
important
for
cupola
melting
furnaces.
As
we
would
like
to
encourage
these
emission
reduction
measures,
we
sought
to
develop
a
total
metal
HAP
emission
alternative
to
the
PM
emission
limits.

As
stated
previously,
we
have
insufficient
direct
metal
HAP
emissions
data
to
identify
and
characterize
the
top
performing
12
percent
of
sources
with
respect
to
metal
HAP
emissions.
Therefore,
we
cannot
directly
establish
MACT
floor
performance
limits
for
specific
metal
HAP
or
for
total
metal
HAP.
However,
we
do
have
some
information
on
the
metal
HAP
emissions
from
various
metal
HAP
sources
at
iron
and
steel
foundries.
Although
some
of
these
data
are
for
units
that
do
not
meet
the
MACT
floor
PM
emission
limits,
the
metal
HAP
to
PM
emission
ratio
for
units
controlled
with
baghouses
that
meet
the
MACT
floor
PM
emission
limits
are
not
statistically
different
than
the
metal
HAP
to
PM
emission
ratio
for
units
controlled
with
wet
scrubbers
that
do
not
meet
the
MACT
floor
PM
emission
limits.
Additionally,
for
two
units
for
which
metal
HAP
emission
reduction
data
are
available
(
one
unit
controlled
with
a
baghouse
and
one
controlled
with
a
wet
scrubber),
the
metal
HAP
emission
reduction
efficiency
was
equivalent
to
the
PM
emission
reduction
efficiency.
As
such,
the
relative
fraction
of
metal
HAP
emitted
as
a
percent
of
PM
emissions
does
not
appear
to
be
a
function
of
the
emission
control
system.
By
normalizing
the
HAP
emission
data
by
the
PM
emissions,
a
larger
pool
of
data
is
available
to
assess
the
expected
metal
HAP
emissions
performance
and
unavoidable
variability
of
units
that
meet
the
MACT
floor
PM
emissions
limit.
That
is,
by
normalizing
the
HAP
emission
data
by
the
PM
emissions
and
aggregating
these
data
for
the
various
emission
sources
at
iron
and
steel
foundries,
we
can
develop
a
reasonable
estimate
of
the
magnitude
and
variability
of
the
HAP
content
of
the
PM
emitted
from
these
sources.
We
can
then
develop
a
total
metal
HAP
emission
limit
that
is
equivalent
to
the
PM
MACT
floor
emission
limit
by
applying
this
information
to
the
specific
system
that
established
the
PM
MACT
floor
emissions
limit.

The
basis
of
this
alternative
emission
limit
is
still
the
PM
MACT
floor
determination
as
described
in
the
previous
section.
Therefore,
we
consider
this
total
metal
HAP
emission
limit
to
be
an
alternative
that
demonstrates
compliance
with
the
level
of
control
reflected
by
the
PM
emission
limit,
but
that
translates
the
PM
limit
into
a
total
metal
HAP
format
to
allow
for
demonstration
of
compliance
by
sources
that
have
artificially
high
PM
emissions.
This
metal
HAP
limit
is
not
a
separate
and
distinct
MACT
floor
requirement.

The
available
data
regarding
the
metal
HAP
emissions
as
a
percent
of
PM
are
summarized
in
Table
7.
Separate
statistical
distributions
were
generated
for
lead
and
manganese
content,
the
two
HAP
that
were
generally
present
at
the
highest
concentrations,
and
a
third
distribution
was
17
generated
for
the
sum
of
all
other
metal
HAP.
[
Note:
the
"
other"
metal
HAP
category
generally
included
data
for
cadmium,
chromium,
nickel
and
mercury.
The
"
other"
metal
HAP
category
may
underestimate
the
total
concentration
of
other
metal
HAP
because
several
of
the
tests
appeared
to
measure
only
select
metal
HAP
and
did
not
attempt
to
measure
or
report
all
the
metal
HAP
that
would
be
measured
using
EPA
Method
29.]
These
three
components
of
the
total
metal
HAP
concentration
were
assumed
to
be
independent
distributions,
each
normally
distributed
around
its
average.
The
minimum
concentration
for
each
component
was
used
as
the
lower
limit
of
the
distribution
for
that
component,
ensuring
no
negative
concentrations
were
projected.
Five
thousand
independent
random
values
were
generated
for
each
HAP
component
(
given
the
minimum
concentration
limits),
and
the
total
metal
HAP
content
was
calculated
as
the
sum
of
the
lead,
manganese
and
other
metal
HAP
for
a
given
randomization.
This
procedure
generated
5,000
random
estimates
of
the
total
metal
HAP
content
as
a
percent
of
the
PM
emissions.

Table
7.
Metal
HAP
Emissions
Data
as
a
Percent
of
PM
Emissions
Test/
Facility
Docket
Item
Furnace
Type
%
Pb
%
Mn
%
other
metal
HAP
American
Brass
and
Iron
II­
I­
27
Cupola
­
BH
1.99%
1.73%
0.12%
Pacific
Steel
II­
I­
20
EAF
8.64%
0.69%
Charlotte
Pipe
Cupola
BH
0.26%
1.08%
0.07%
U.
S.
Pipe
­
NJ
II­
I­
49
Cupola
BH
2.04%
0.77%
U.
S.
Pipe
­
NJ
II­
I­
24
Cupola
BH
2.35%
Euvrard&
Jackson
II­
I­
30
Cupola
0.12%
6.20%
0.06%
Northern
Steel
II­
D­
114
EAF
0.22%
0.07%
Atchison
Casting
II­
D­
100
Cupola
Gray
Iron
0.36%
0.85%
Atchison
Casting
II­
D­
100
Cupola
Nodular
0.19%
0.49%
Great
Lakes
II­
D­
50
Cupola
0.21%
1.76%
0.01%
GM­
SMCO
II­
A­
32
Cupola
WS
1.10%
7.28%
0.14%
Waupaca­
Tell
City
II­
A­
30
Cupola
BH
1.03%
0.59%
1.88%
Griffin
Pipe
­
NJ
II­
D­
60
Cupola
3.98%
0.22%
Briggs
&
Stratton
II­
D­
80
Cupola
1991
1.28%
0.03%
Briggs
&
Stratton
II­
D­
80
Cupola
1994
0.06%
2.08%
0.05%
Griffin
Wheel
AL­
11
EAF
0.11%
2.68%
1.21%
CMI
MI­
13
Cupola
WS
0.29%
2.18%
Auburn
IN­
12
Cupola
WS
1.62%
3.33%
0.98%
Auburn
IN­
12
Preheater
cyclone
0.70%
1.35%
0.71%
Auburn
IN­
12
EIF
0.25%
0.74%
Blanchester
OH­
11
Cupola
WS
2.20%
CERP
Mexico
II­
I­
63
Pouring
0.42%
2.01%
2.08%
Average
1.01%
2.70%
0.58%
Std
Deviation
1.08%
2.49%
0.64%
Minimum
0.060%
0.225%
0.006%
18
For
each
of
the
emissions
source
with
an
existing
source
PM
emissions
limit
(
i.
e.,
for
cupolas,
EAFs,
EIFs
and
scrap
preheaters,
and
pouring
stations),
the
average
PM
emissions
concentration
for
the
source
that
was
identified
as
the
6th
percentile
unit
(
i.
e.,
the
source
that
set
the
MACT
floor
performance
limit)
was
used
as
the
average
PM
emissions
concentration.
The
average
RSD
value
of
0.4
was
used
to
calculate
the
relative
standard
deviation
associated
with
the
average
PM
emissions
concentration
for
each
source.
Five
thousand
independent
random
PM
concentration
values
were
generated
for
each
PM
emissions
source.
The
5,000
PM
emissions
concentrations
for
each
emissions
source
were
randomly
paired
with
the
5,000
estimates
of
total
metal
HAP
content
(
as
a
percent
of
the
PM
emissions)
to
project
the
distribution
of
metal
HAP
emissions
concentrations
for
each
emissions
source.
The
99th
percentile
total
metal
HAP
concentration
was
then
selected
from
the
projected
distributions.
Using
this
approach,
we
account
for
the
co­
variability
in
the
metal
HAP
concentrations
and
in
the
PM
emissions.
We
believe
that
this
approach
provides
an
emissions
limit
that
is
equivalent
to
the
MACT
floor
PM
emissions
limit
because
the
metal
HAP
emission
limits
were
based
on
the
average
performance
of
the
6th
percentile
emissions
source
(
as
a
measure
of
the
average
performance
of
the
top
12
percent
of
sources).
The
99th
percentile
metal
HAP
concentrations
determined
from
these
distributions
are
equivalent
to
7.5
percent
of
the
99th
percentile
PM
emissions
limit
(
i.
e.,
the
MACT
floor
PM
emissions
limit)
for
each
of
the
emission
sources.
As
the
"
other"
metal
HAP
content
may
be
slightly
biased
low
and
because
of
the
relative
accuracy
of
the
assessment
(
generally
one
significant
figure),
it
is
appropriate
to
round
the
metal
HAP
to
PM
ratio
to
one
significant
digit
or
8
percent.
Therefore,
a
total
metal
HAP
emissions
limit
that
is
equivalent
to
the
PM
MACT
floor
emissions
limit
can
be
calculated
as
8
percent
of
the
PM
emissions
limit
(
i.
e.,
0.08
times
the
PM
emissions
limit).
As
the
identification
of
the
unit
that
represents
the
MACT
floor
is
based
on
the
PM
emissions
performance,
we
do
not
believe
that
the
metal
HAP
emission
limits,
as
derived
here,
represent
a
separate
MACT
floor
that
must
be
met
at
all
sources,
but
rather
an
alternative
emissions
limit
that
is
expected
to
be
equivalent
to
the
MACT
floor
PM
emissions
limit
that
is
offered
to
provide
regulatory
flexibility.

References
Cavender
K.,
2002.
Personal
communication
with
Tom
McManamy,
as
recorded
in
"
Grede
baghouse
information"
e­
mail
from
K.
Cavender,
EPA/
OAQPS/
ESD/
Metals
Group
to
A.
Vervaert,
EPA/
OAQPS/
ESD/
Metals
Group
and
J.
Coburn,
RTI.
April
25,
2002.

U.
S.
Environmental
Protection
Agency,
1998.
Compilation
of
Information
from
Questionnaire
Forms
Submitted
by
Iron
and
Steel
Foundries
to
the
U.
S.
EPA
Office
of
Air
Quality
Planning
and
Standards.
Office
of
Air
Quality
Planning
and
Standards,
Research
Triangle
Park,
NC.

U.
S.
Environmental
Protection
Agency,
1999a.
Iron
and
Steel
Foundries
Manual
Emissions
Testing
of
Cupola
Baghouse
at
Waupaca
Foundry
in
Tell
City,
Indiana.
Office
of
Air
Quality
Planning
and
Standards,
Research
Triangle
Park,
NC.
EPA­
454/
R­
99­
017A
and
EPA­
454/
R­
99­
017B.
June.
19
U.
S.
Environmental
Protection
Agency,
1999b.
Iron
and
Steel
Foundries
Manual
Emissions
Testing
of
Cupola
Wet
Scrubber
at
General
Motors
Corp.,
Saginaw,
Michigan.
Office
of
Air
Quality
Planning
and
Standards,
Research
Triangle
Park,
NC.
EPA­
454/
R­
99­
025A
and
EPA­
454/
R­
99­
025B.
July.

U.
S.
Environmental
Protection
Agency,
2002.
National
Emission
Standards
for
Hazardous
Air
Pollutants
(
NESHAP)
for
Iron
and
Steel
Foundries
­­
Background
Information
for
Proposed
Standards.
Air
Quality
Planning
and
Standards,
Research
Triangle
Park,
NC.
EPA­
453/
R­
02­
013.
December.
