MEMORANDUM
TO:
304(
m)
Record
(
EPA
Docket
Number
OW­
2004­
0074)

FROM:
Cortney
Itle,
ERG
DATE:
August
23,
2004
SUBJECT:
Comparison
of
Methodologies
for
Estimation
of
Values
Below
the
Detection
Limit
for
the
Steam
Electric
Industry
Facilities
that
have
NPDES
permits
submit
discharge
monitoring
report
data
(
DMR)
to
their
permitting
authority.
The
DMR
data
is
submitted
to
EPA's
Permit
Compliance
System
(
PCS)
database.
For
parameters
that
are
not
detected
during
monitoring,
permittees
have
the
option
of
reporting
the
detection
limit
and
that
the
measured
concentration
fell
below
detection
limit
(
BDL).
BDL
means
that
the
actual
concentration
is
between
zero
and
the
detection
limit.
Therefore,
the
following
options
have
been
used
to
estimate
loadings
from
PCS
data:


BDL
is
set
equal
to
zero;


BDL
is
set
to
the
detection
limit;
and

BDL
is
set
equal
to
half
the
detection
limit.

For
the
PCSLoads2000
database,
ERG
initially
calculated
loadings
with
BDL
set
equal
to
half
the
detection
limit.
This
approach
resulted
in
relatively
high
calculated
mass
loads
for
some
parameters
reported
with
a
wide
range
in
detection
limits
(
e.
g.,
PCBs
and
dioxins).
The
high
mass
loads
reflected
high
reported
detection
limits
for
discharges
in
which
the
parameters
were
never
detected
possibly
as
a
result
of
relatively
insensitive
laboratory
analyses.
However,
by
setting
the
BDL
to
zero,
ERG
believes
the
pollutant
discharges
are
likely
to
be
underestimated
because
facilities
generally
monitor
for
only
those
pollutants
likely
to
be
generated
by
their
industrial
processes.

Therefore,
ERG
developed
a
hybrid
approach
where
BDL
was
set
equal
to
zero
for
parameters
never
detected
by
the
facility
in
2000.
For
parameters
sometimes
detected
and
sometimes
not,
BDL
was
set
equal
to
half
of
the
detection
limit.
ERG
considered
half
the
detection
limit
a
reasonable
estimate
of
the
unknown
concentration
for
an
initial
screening­
level
review
and
prioritization
of
the
toxic­
weighted
pollutant
loadings
discharged
by
industrial
categories.

ERG
performed
an
analysis
of
pollutant
loads
associated
with
the
steam
electric
industry
to
compare
the
resulting
TWPE
when:
°
BDL
values
are
set
equal
to
zero
(
abbreviated
in
the
spreadsheet
as
KGY0);

°
BDL
is
estimated
to
be
equal
to
half
the
detection
limit
(
abbreviated
in
the
spreadsheet
as
KGYE);
and
°
BDL
is
set
equal
to
zero
for
parameters
never
detected
by
the
facility
in
2000,
and
for
parameters
sometimes
detected
and
sometimes
not,
BDL
is
set
equal
to
half
of
the
detection
limit
(
this
is
the
approach
used
to
develop
PCSLoads2000).

Toxic­
weighted
pound
equivalent
(
TWPE)
loads
were
estimated
from
PCS
data
for
all
of
the
circumstances
listed
above.
Note
that
chlorine
loads
were
excluded
from
this
analysis
(
see
Section
5.4.4
of
the
TSD
for
a
discussion
of
chlorine
loads).

The
total
industry
TWPE
estimated
with
BDL
set
to
zero
represents
72
percent
of
the
total
industry
TWPE
estimated
by
ERG
in
PCSLoads2000.
In
other
words,
the
total
TWPE
for
the
steam
electric
industry
in
PCSLoads2000
are
38
percent
higher
than
loads
estimated
with
BDL
set
equal
to
zero.

The
total
industry
TWPE
estimated
with
BDL
set
to
half
the
detection
limit
represents
1,410
percent
of
the
total
industry
TWPE
estimated
by
ERG
in
PCSLoads2000.
In
other
words,
the
total
TWPE
for
the
steam
electric
industry
in
PCSLoads2000
are
93
percent
lower
than
loads
estimated
with
BDL
set
equal
to
half
the
detection
limit.

Please
see
the
attached
excel
workbook
for
the
data
used
to
estimate
the
total
industry
TWPE
for
each
BDL
circumstance.
The
total
load
(
in
pounds
and
TWPE)
is
shown
by
parameter
for
each
BDL
circumstance.
