Summary
of
and
Response
to
Peer
Review
Comments
on
the
Draft
NONROAD2002
Emissions
Inventory
Model
Pertaining
to
the
Nonroad
Diesel
Engine
Final
Rulemaking
1.0
Review
of
Nonroad
Engine
Growth
Estimates
1.1
Comments
from
Andrew
Bollman,
E.
H.
Pechan
1.1.1
Comment:
The
data
and
approach
used
to
develop
growth
factor
estimates
is
reasonable
and
has
advantages
over
other
potential
approaches.
In
addition
to
the
advantages
noted
in
the
report
(
e.
g.,
ability
to
reflect
fuel­
specific
growth
trends),
two
major
advantages
are
that
the
approach
is
non­
data
intensive
and
easy
to
explain.

EPA
Response:
No
response
necessary.

1.1.2
Comment:
When
the
PSR­
based
fuel­
specific
growth
rates
(
from
columns
4­
7
in
Table
1)
are
applied
to
the
1996
population
data,
the
resulting
total
sector
engine
population
growth
rates
will
not
equal
the
total
growth
rates
calculated
from
the
Power
Systems
Research
(
PSR)
historical
data
(
from
column
3
in
Table
1).
It
is
my
understanding
that
EPA
adjusts
the
fuel­
specific
engine
population
projection
so
that
the
resulting
sector­
level
population
growth
rates
equal
the
PSR
sector­
level
population
growth
rates
(
from
column
3
in
Table
1).
If
this
understanding
is
correct,
the
report
should
explain
how
these
adjustments
are
made.
I
suggest
that
an
example
of
how
these
adjustments
are
made
would
help
clarify
the
procedure
(
perhaps
at
the
bottom
of
page
4).
As
an
alternative,
I
suggest
that
EPA
consider
developing
values
for
each
year
that
represent
the
percentage
of
total
equipment
in
each
sector
that
uses
each
of
the
four
fuel
types.
For
example,
the
percentage
of
Construction
sector
equipment
in
1996
that
is
diesel
is
70.92,
while
the
percentage
in
1996
that
is
gasoline
is
29.08.
Based
on
applying
the
growth
rates
for
equipment
in
each
fuel
type
from
Table
1
to
the
1996
equipment
populations
in
Appendix
A,
these
percentages
would
be
71.53
and
28.47
respectively
in
1997.
This
approach
should
simplify
the
computations
and
be
more
straightforward
to
explain
in
the
nonroad
engine
growth
estimate
report.

EPA
Response:
It
is
correct
that
the
growth
rates
were
calculated
with
an
attempt
to
account
for
the
market
shifting
somewhat
from
gasoline
to
diesel
fueled
engines.
The
total
growth
rates
in
column
3
of
Table
1
do
indeed
match
the
weighted
average
of
growth
rates
from
columns
4­
7,
using
the
1996
PSR
fuel­
specific
populations
from
Appendix
A
Table
1
to
do
the
weighting.
The
following
text
can
be
added
to
the
report
to
clarify
how
this
weighting
was
done.
"
The
total
growth
rates
in
column
3
of
Table
1
were
calculated
using
a
weighted
average
of
the
fuel­
specific
growth
rates
from
columns
4­
7
of
that
table.
The
weighting
was
based
on
the
1996
PSR
fuelspecific
populations
from
Appendix
A
Table
1.
For
example,
the
overall
farm
growth
rate
of
2.6%
equals
(
3%
x
0.70341)
+
(
1.8%
x
0.29442)
+
(­
10.2%
x
0.002141),
where
0.70341
=
3,302,604
/
4,695,124,
which
is
the
1996
PSR
diesel
population
divided
by
the
total
1996
PSR
farm
equipment
population,
and
similarly
for
each
of
the
fuel
types."

1.1.3
Comment:
For
aircraft
ground
support
equipment,
there
is
no
discussion
of
how
EPA
plans
to
develop
growth
data
beyond
the
period
available
from
the
FAA
source
(
i.
e.,
beyond
1From
BEA
web­
site
located
at
http://
www.
bea.
doc.
gov/
bea/
dn/
gdplev.
htm.

2
2011).
I
first
note
that
there
is
a
more
up­
to­
date
version
of
the
FAA
forecast
data
source
that
now
goes
through
2012.
Secondly,
to
extrapolate
beyond
2012,
I
suggest
that
EPA
review
the
possibility
of
using
projections
data
from
a
separate
FAA
report
that
provides
forecasts
for
2015,
2020,
and
2025.
This
June
2000
report
is
entitled,
"
FAA
Long­
Range
Aerospace
Forecasts
Fiscal
Years
2015,
2020,
and
2025"
(
see
http://
api.
hq.
faa.
gov/
apo_
pubs.
htm#
ANCHOR00_
10).

EPA
Response:
EPA
will
take
the
suggested
FAA
report
into
consideration
as
it
continues
to
analyze
its
options
for
updating
and
improving
the
NONROAD
model's
growth
rates.

1.1.4
Comment:
I
suggest
clarifying
the
discussion
comparing
PSR
growth
with
BEA
growth
by
identifying
the
specific
BEA
data
that
are
being
used
in
the
comparison.
It
is
not
clear
if
you
are
referring
to
BEA
earnings
data
or
BEA
gross
state
product
data
(
note
that
I
would
expect
the
BEA
earnings
data
would
show
significantly
lower
growth
than
the
BEA
gross
state
product
data
since
the
gross
state
product
data
reflect
increases
in
output
produced
per
employee).

EPA
Response:
EPA
used
the
gross
state
product
data
for
the
comparisons
with
PSR
equipment
population
historic
growth
values.
This
is
being
addressed
in
the
next
update
of
the
technical
report
on
growth
in
the
NONROAD
model,
NR­
008c.

1.1.5
Comment:
Comments
on
Current
Methodology
I
agree
with
the
comment
in
the
report
that
there
is
a
significant
concern
with
the
assumption
that
the
1989­
1996
engine
population
growth
rate
will
equal
the
growth
rate
over
the
1997­
2045
period.
This
approach
implicitly
assumes
that
the
underlying
factors
that
affected
engine
populations
during
this
short
time­
frame
will
continue
to
affect
future
engine
populations
throughout
the
long
forecast
period.
More
importantly,
this
approach
assumes
that
the
trends
in
these
underlying
factors
will
remain
the
same
throughout
the
entire
forecast
period.
One
way
of
reviewing
the
reasonableness
of
these
assumptions
is
to
compare
total
economic
growth
over
the
1989­
1996
period
with
long­
run
average
growth.
Based
on
BEA
data,
constant
dollar
gross
domestic
product
(
GDP)
grew
at
an
annual
average
rate
of
2.1
percent
over
the
1989­
1996
period.
1
In
the
post­
war
era
(
1946­
2000),
constant
dollar
GDP
grew
at
an
annual
rate
of
3.4
percent,
and
constant
dollar
GDP
grew
3.6
percent
per
year
between
1996­
2000.
These
GDP
growth
rates
indicate
that
economic
growth
over
the
1989­
1996
period
is
unrepresentative
of
the
long­
run
average.
Assuming
that
total
economic
growth
is
related
to
nonroad
engine
population
growth,
then
the1989­
1996
growth
in
nonroad
engine
populations
may
not
be
representative
of
long­
term
growth
rates.

EPA
Response:
EPA
has
recently
evaluated
alternative
approaches
to
revising
the
growth
rates
in
the
NONROAD
Model,
as
well
as
the
method
used
to
generate
them,
as
part
of
the
current
nonroad
diesel
engine
rulemaking
process.
This
analysis
in
discussed
in
the
RIA
and
the
summary
and
analysis
of
comments
for
the
Nonroad
Diesel
Engine
Tier
4
Rule.
3
1.1.5.1
Comment:
Publically­
Available
Regional
Forecast
Data
Sources
The
growth
factors
included
in
the
draft
NONROAD
model
are
National
and
assume
that
1989­
1996
engine
population
growth
rates
will
continue
over
the
next
40
plus
years.
The
EPA
report
requests
recommendations
on
how
to
develop
regional
or
State­
level
growth
factors.
The
three
sources
of
publically
available
sub­
National
level
forecast
data
that
I
am
aware
of
are:

1.
The
BEA
forecast
data
described
in
the
EPA
nonroad
engine
growth
estimates
report.
2.
Economic
Growth
Analysis
System
(
EGAS)
Version
4.0
and
underlying
data
sources.
3.
DOE
forecasts
of
energy
consumption
by
sector
and
fuel
type.

The
BEA
data
have
two
main
advantages
over
the
other
two
sources
 
they
are
more
geographic­
specific
and
they
cover
the
entire
forecast
period
of
the
NONROAD
model.
However,
the
BEA
data
are
more
outdated
(
prepared
in
1995)
and
less
source­
specific
(
at
the
2­
digit
SIC
code
level)
than
the
other
two
sources.

The
Regional
Economic
Models,
Inc.
(
REMI)
data
that
underlie
EGAS
4.0
are
generally
specified
at
the
State­
level,
or
for
states
with
ozone
nonattainment
areas,
at
the
ozone
nonattainment
area/"
rest­
of­
state"
level.
Version
4.0
of
EGAS
has
projections
capability
through
the
year
2020,
but
the
REMI
models
that
underlie
EGAS
project
through
2035.
The
National
growth
factor
data
from
EPA's
draft
NONROAD
model
are
incorporated
into
EGAS.
For
nonroad
sectors
not
included
in
the
NONROAD
model
(
e.
g.,
commercial
marine
vessels),
EGAS
uses
State
or
nonattainment
area/
rest­
of­
state
constant
dollar
output
data
by
Bureau
of
Labor
Statistics
sector
(
2­
or
3­
digit
SIC
code),
or
in
some
cases,
a
regression
equation­
based
approach.
The
regression­
based
approach
uses
an
equation
that
was
statistically
identified
from
the
historical
relationship
between
the
emissions
activity
for
the
category
(
e.
g.,
amount
of
diesel
fuel
consumed
by
vessels)
and
a
socioeconomic
indicator
available
from
the
EGAS
REMI
models
(
e.
g.,
constant
dollar
output
in
the
Water
Transportation
sector).
This
regression­
based
approach
is
also
similar
to
that
used
in
EPA's
Nonroad
Engine
and
Vehicle
Emission
Study
(
NEVES)
to
identify
surrogate
nonroad
geographic
allocation
indicators.
However,
the
NEVES
analyses
were
flawed
in
that
they
used
PSR
state­
level
data
as
dependent
variables
in
the
regression
analyses.
The
PSR
state­
level
data
are
not
actual
equipment
population
data,
but
instead
are
summations
of
PSR
county
data.
The
PSR
county
data
were
estimated
by
applying
surrogate
county­
level
allocation
indicators
to
National
equipment
populations.
Therefore,
the
indicators
that
were
identified
in
NEVES
as
statistically
correlated
with
the
PSR
data
actually
correlate
with
the
geographic
allocation
indicators
used
by
PSR,
and
not
with
equipment
population
data.

For
its
"
Annual
Energy
Outlook"
publication,
the
DOE
develops
energy
production
and
consumption
projections
on
a
regional
basis
(
Census
division)
through
2020.
Some
of
the
DOE
data
have
applicability
for
nonroad
sectors
(
e.
g.,
oil
production
forecasts),
but
many
nonroad
sectors
do
not
have
an
acceptable
growth
indicator
available
from
the
DOE
data.
4
Alternative
Methodology
Recommendation
Any
of
the
above
forecast
data
sources
could
assist
in
developing
regional­
level
growth
factors.
My
recommendation
concerning
an
alternative
growth
factor
development
methodology
is
to
implement
the
more
exploratory
regression­
based
approach
used
for
some
source
categories
in
EGAS
4.0
(
and
similar
to
the
approach
used
to
identify
geographic
allocation
indicators
in
the
NEVES).
This
approach
first
entails
compiling
National
time­
series
nonroad
engine/
equipment
data
for
the
complete
historical
period
available
from
PSR
(
i.
e.,
1989­
1998?).
These
data
would
then
be
used
to
identify
equations
that
relate
surrogate
indicators
with
the
engine/
equipment
data.
These
equations
would
be
identified
based
on
regression
analysis,
and
the
surrogate
indicators
that
would
be
tested
for
correlations
with
the
PSR
data
would
come
from
sources
that
have
historical
and
regional
forecast
data
available.
One
potential
source
of
indicators
for
this
effort
is
EGAS
4.0
and
its
associated
underlying
data
sources
(
e.
g.,
REMI
socioeconomic
data).
After
statistically
identifying
relationships
between
the
historical
PSR
and
REMI
data,
the
EPA
could
use
the
REMI
forecast
data
to
develop
National
or
sub­
national
engine
population
growth
factors.
This
approach
is
similar
to
the
method
that
is
used
for
some
EGAS
nonroad
categories
that
are
not
included
in
EPA's
NONROAD
model.
This
method
could
easily
be
incorporated
into
future
versions
of
EGAS.

For
Agricultural
sector
equipment
populations,
for
example,
the
EPA
could
develop
an
equation
relating
equipment
populations
to
surrogate
indicators.
To
identify
this
equation,
EPA
would
regress
the
PSR
National
time­
series
data
against
potential
explanatory
variables
(
e.
g.,
Farm
sector
and
Agricultural
Services
sector
employment
and
output)
that
are
available
from
the
REMI
models
that
underlie
EGAS.
Recent
work
conducted
to
estimate
aircraft
ground
support
equipment
provides
a
similar
example.
Page
16
of
the
latest
NONROAD
geographic
allocation
report
identifies
the
use
of
regression
equations
that
predict
growth
in
aircraft
ground
support
equipment
populations
as
a
function
of
the
number
of
landings
and
take­
offs
(
LTOs)
by
narrow
and
wide­
body
aircraft
(
the
distinction
being
that
more
pieces
of
ground
support
equipment
are
needed
to
service
a
wide­
body
aircraft
than
a
narrow
body
one).
These
regressions
are
from
a
May
1999
report
entitled,
"
Technical
Support
for
the
Development
of
Airport
Ground
Support
Equipment
Emission
Reductions."
Because
LTO
forecast
data
are
available
from
the
FAA,
it
may
be
possible
to
use
these
data
in
conjunction
with
the
regression
equations
to
forecast
ground
support
equipment
(
note
that
it
will
be
necessary
to
determine
whether
the
LTO
forecasts
differentiate
between
narrow
versus
wide­
body
aircraft).

This
approach
does
not
estimate
fuel­
specific
trends
in
equipment
populations.
To
estimate
these
trends,
EPA
could
conduct
time­
series
regression
analyses
based
on
the
historical
percentage
of
equipment
in
each
sector
using
each
type
of
fuel.
In
their
simplest
form,
these
equations
would
simply
assume
that
historical
trends
will
continue
throughout
the
forecast
period.
This
is
similar
to
the
current
EPA
approach.
Unless
there
is
evidence
that
each
sector's
fuelspecific
trends
will
continue
indefinitely
into
the
future,
EPA
should
consider
employing
upper/
lower­
limits
on
the
percentage
of
each
sector's
equipment
powered
by
each
fuel
type.
A
more
exploratory/
robust
approach
would
attempt
to
project
fuel
type
percentages
based
on
5
potential
correlations
with
explanatory
variables
such
as
fuel
prices.
This
second
approach,
however,
is
likely
to
be
hindered
by
the
availability
of
long­
term
forecast
data
for
the
major
variables
that
affect
the
proportion
of
engines
that
use
each
type
of
fuel.

EPA
Response:
EPA
has
recently
evaluated
alternative
approaches
to
revising
the
growth
rates
in
the
NONROAD
Model,
as
well
as
the
method
used
to
generate
them,
as
part
of
the
current
nonroad
diesel
engine
rulemaking
process.
This
analysis
in
discussed
in
the
RIA
and
the
summary
and
analysis
of
comments
for
the
Nonroad
Diesel
Engine
Tier
4
Rule.

At
this
time,
EPA
is
not
actively
considering
the
implementation
of
state
or
regional
growth
rates
in
the
NONROAD
model.
However,
EPA
will
take
this
information
under
consideration
for
the
next
generation
of
mobile
source
emissions
model,
the
Motor
Vehicle
Emission
System
(
MOVES).
MOVES
will
most
likely
include
the
capability
to
model
state
or
regional
growth
rates.

1.1.6
Comment:
Forecast
Data
Alternatives
The
purpose
of
this
comment
is
to
make
EPA
aware
of
alternative
forecast
data
sources;
it
was
beyond
the
scope
of
this
peer
review
to
analyze/
compare
these
sources
against
EPA's
current
PSR­
based
approach.

Freedonia
Group,
Inc.

The
Freedonia
Group,
Inc.
(
Freedonia)
is
a
market
research
firm
that
develops
forecasts
for
many
sectors
of
the
economy
including
engine/
equipment
demand
projections
for
certain
nonroad
applications.
The
Freedonia
forecasts
are
limited
in
geographic
and
temporal
scope
in
that
they
are
National
forecasts
that
are
available
for
5­
year
intervals
over
a
ten­
year
forecast
period.
For
an
Emissions
Inventory
Improvement
Program
(
EIIP)
review
of
projections
data,
EPA
purchased
the
following
nonroad­
related
Freedonia
forecast
data,
which
are
reported
in
constant
dollars:

°
Recreational
vehicle
shipments;
°
Lawn
mowers
sold;
°
Lawn
and
garden
tractor
shipments;
°
Marine
equipment
diesel
engine
demand;
°
Construction
equipment
diesel
engine
demand;
°
Agricultural
equipment
diesel
engine
demand;
and
°
Other
markets
equipment
diesel
engine
demand
(
Includes
railroad
equipment,
nonmarine
military
equipment,
recreational
vehicles,
and
residential
generators.)

It
is
important
to
note
these
data
are
sales
projections,
not
projections
of
total
engine/
equipment
populations.
For
more
information
on
the
Freedonia
data,
please
refer
to
the
EIIP
report
entitled
6
"
Evaluation
of
Emission
Project
Tools
and
Emission
Growth
Surrogate
Data,
Final
Draft,"
November
17,
2000
(
see
http://
www.
epa.
gov/
ttn/
chief/
eiip/
committee/
projections/
projects.
html).

Department
of
Energy
As
an
alternative
to
the
use
of
BEA
data
for
the
oil
field
equipment
sector,
EPA
may
want
to
consider
the
use
of
domestic
oil
production
forecasts
available
from
Department
of
Energy
(
DOE)'
s
"
Annual
Energy
Outlook."
The
advantages
of
these
forecasts
are:
(
1)
they
are
a
direct
measure
of
oil
production
activity
(
barrels
produced),
rather
than
a
measure
of
oil
sector
economic
activity;
(
2)
they
are
much
more
recent
than
the
BEA
forecasts,
which
were
released
in
1995;
and
(
3)
they
explicitly
model
factors
that
affect
domestic
oil
production
such
as
oil
price,
international
oil
supply,
and
domestic
oil
demand.
The
two
major
disadvantages
compared
with
the
BEA
data
are:
(
1)
the
last
forecast
year
is
2020;
and
(
2)
the
forecasts
are
less
geographicspecific
(
i.
e.,
are
regional,
while
the
BEA
data
are
State­
specific).

EPA
Response:
EPA
has
recently
evaluated
alternative
approaches
to
revising
the
growth
rates
in
the
NONROAD
Model,
as
well
as
the
method
used
to
generate
them,
as
part
of
the
current
nonroad
diesel
engine
rulemaking
process.
This
analysis
in
discussed
in
the
RIA
and
the
summary
and
analysis
of
comments
for
the
Nonroad
Diesel
Engine
Tier
4
Rule.

1.2
Comments
from
Rick
Baker,
Eastern
Research
Group
1.2.1
Comment:
In
general
the
report
is
very
clearly
written
and
well
organized.
The
Background
section
is
particularly
effective
in
communicating
the
complexity
of
alternative
approaches.

EPA
Response:
No
response
necessary.

1.2.2
Comment:
In
the
third
full
paragraph
on
page
3,
EPA
discusses
how
oil
field
equipment
growth
rates
were
handled
differently
from
other
equipment
categories
due
to
their
declining
populations.
Is
this
the
only
equipment
type
negative
growth
rates?
If
so,
please
state
as
such.
If
not,
what
other
types
of
equipment
have
declining
populations,
and
how
were
these
handled?

EPA
Response:
Other
equipment
categories
that
have
negative
growth
rates
include
CNG
agricultural
equipment,
gasoline
industrial
equipment,
and
diesel
logging
equipment.
These
growth
rates
were
derived
using
the
PSR
data.
EPA
included
a
separate
discussion
of
oil
field
equipment
because
it
used
BEA
estimates
of
gross
state
product
of
domestic
oil
production
rather
than
PSR
data.
However,
EPA
will
clarify
this
in
the
next
revision
of
the
growth
rate
technical
report.
7
1.2.3
Comment:
In
the
first
paragraph
on
the
second
page,
it
is
acknowledged
that
economic
indicators
and
models
in
recent
years
have
under­
predicted
growth
in
the
national
economy.
In
light
of
the
recent
economic
downturn,
it
may
be
that
econometric
models
do
in
fact
provide
reasonable
long­
term
growth
predictions,
although
their
ability
to
predict
short
and
medium
term
fluctuations
is
admittedly
limited.

EPA
Response:
This
is
a
relevant
point
that
EPA
has
considered
and
should
emphasize
in
the
growth
technical
report.

1.2.4
Comment:
In
the
first
paragraph
on
the
second
page,
it
is
noted
that
"
Economic
indicators
may
not
be
able
to
adequately
predict
the
effects
of
substitution
of
equipment
for
labor
in
the
market."
While
this
is
true,
it
is
not
clear
that
such
substitution
is
a
significant
factor
when
it
comes
to
many
nonroad
equipment
applications.
In
general,
one
might
anticipate
significant
substitution
in
the
farm
sector
as
smaller
farms
continue
to
undergo
consolidation
by
large
agribusiness
interests.
However,
such
a
trend
is
not
self­
evident
for
most
of
the
other
equipment
sectors.
For
example,
substantially
new
"
labor
saving
technologies"
have
not
been
developed
for
many
years
in
most
construction
and
industrial
applications,
and
most
engines
and
machines
continue
to
be
used
in
much
the
same
way
they
have
for
years
(
e.
g.,
welders,
pavers,
backhoes).
And
in
the
case
of
recreational
equipment,
the
"
substitution
factor"
is
not
even
applicable.

EPA
Response:
This
is
a
good
point.
The
NONROAD
model
does
not
have
the
ability
to
account
for
subtle
economic
influences,
even
if
it
could
be
proven
that
they
exist
or
have
a
significant
impact
on
the
equipment
population
of
a
given
equipment
category.
EPA
will
clarify
this
in
the
next
revision
of
the
growth
technical
report.

1.2.5
Comment:
Although
I
do
have
reservations
concerning
the
points
made
in
the
first
paragraph
of
page
2
regarding
the
limitations
of
economic
indicators,
I
am
in
substantial
agreement
with
the
second
paragraph,
concerning
the
ability
to
identify
trends
within
sectors.
For
this
reason
I
concur
with
EPA's
decision
to
use
historical
equipment
populations
as
the
basis
for
determining
growth
rates.

EPA
Response:
No
response
necessary.

1.2.6
Comment:
In
the
third
paragraph
on
page
2,
EPA
notes
that
equipment
population
is
a
reasonable
surrogate
for
activity,
given
the
high
capital
cost
of
equipment
relative
to
operating
costs.
I
would
also
add
that
future
increases
in
equipment
costs
resulting
from
the
adoption
of
increasingly
stringent
emission
standards
will
tend
to
emphasize
this
effect.

EPA
Response:
This
is
a
good
point.
EPA
will
consider
incorporating
this
point
in
the
next
revision
of
this
technical
report.

1.2.7
Comment:
In
the
last
sentence
of
the
third
paragraph
on
page
2,
it
is
not
reasonable
to
assume
that
equipment
will
be
"
fully"
utilized.
This
implies
100%
utilization.
I
suggest
saying
8
"
efficiently
utilized"
instead.

EPA
Response:
This
is
a
good
point.
EPA
will
consider
incorporating
this
point
in
the
next
revision
of
this
technical
report.

1.2.8
Comment:
EPA
proposes
to
use
simple
linear
regressions
of
historical
engine
populations
to
predict
future
populations.
I
believe
this
is
a
reasonable
approach,
as
long
as
annual
average
growth
factors
are
not
substantially
higher
than
overall
population
and
economic
growth.
(
Of
course
exceptions
are
to
be
expected,
such
as
with
the
recent
upsurge
in
ATV
populations,
but
these
should
be
handled
on
a
case­
by­
case
basis.)
In
this
light,
most
of
the
annual
growth
rates
presented
in
Table
1
appear
appropriate
for
linear
projections,
with
the
possible
exceptions
of
light
commercial
(
4.0%),
and
logging
(
4.5%).
Unless
evidence
can
be
provided
to
justify
the
relatively
rapid
growth
in
these
sectors
indefinitely,
I
believe
an
adjustment
should
be
made
to
either
the
annual
growth
rates,
or
the
functional
form
of
the
predictive
relationship,
in
order
to
slow
the
growth
in
the
medium/
long
term.

EPA
Response:
Significant
uncertainty
exists
concerning
the
best
growth
rates
to
use
for
the
NONROAD
model.
The
uncertainty
only
increases
when
trying
to
determine
what
the
rate
of
growth
will
be
for
each
nonroad
category
in
the
long
term.
EPA
will
continue
to
study
this
issue
and
refine
its
approach,
especially
in
the
development
of
the
MOVES
model.

It
should
be
noted
that
EPA
does
not
use
the
total
growth
rates
from
Table
1
in
the
model.
Rather,
the
NONROAD
model
applies
the
separate
growth
rates
for
diesel,
gasoline,
CNG,
and
LPG
engines
to
each
equipment
category
by
fuel
type.
This
will
be
clarified
in
the
next
revision
of
the
growth
rate
technical
support
document.

1.2.9
Comment:
Why
is
the
Airport
Service
equipment
category
excluded
from
Table
1?

EPA
Response:
Table
1
compares
growth
rates
used
in
the
NONROAD
model
to
BEA
growth
rates.
BEA
most
likely
did
not
have
a
category
for
aircraft
ground
support
equipment.
However,
not
including
a
summary
table
of
all
of
the
growth
rates
used
in
NONROAD
was
an
oversight.
In
the
next
revision
of
the
growth
rate
technical
report,
EPA
will
add
another
table
including
all
the
growth
rates
used
in
the
NONROAD
model.

1.2.10
Comment:
On
page
5
EPA
acknowledges
that
there
is
some
"
concern
about
basing
longterm
growth
estimates
for
nonroad
equipment/
emissions
on
the
seven
years
of
data
from
PSR."
In
order
to
help
assess
the
significance
of
this
concern,
EPA
should
provide
some
indication
of
the
year­
to­
year
variation
in
the
data
by
sector.
For
example,
from
Appendix
A
we
see
that
the
yearto
year
population
increases
in
the
construction
and
airport
service
sectors
are
relatively
constant,
varying
by
less
than
2.0%
over
the
seven
year
period.
While
not
definitive,
this
may
imply
that
projections
based
on
seven
years
of
historical
data
may
be
relatively
accurate
for
these
sectors.
On
the
other
hand,
annual
changes
in
recreational
equipment
varied
by
4.5%,
and
rail
equipment
by
6.1%
over
this
same
period,
perhaps
implying
more
volatility
in
sales
and
less
confidence
in
9
future
year
projections.
Therefore
I
recommend
at
least
providing
a
short
discussion
of
year­
toyear
variance
in
the
existing
data,
and
possibly
additional
analysis
of
variance,
standard
deviations,
etc.

EPA
Response:
A
comment
about
this
issue
is
being
added
to
the
technical
report
on
growth
(
NR­
008c).

1.2.11
Comment:
The
annual
growth
rates
for
commercial
and
logging
equipment
seem
inordinately
high.
Although
admittedly
uninformed
on
the
logging
industry,
I
was
under
the
impression
that
logging
activity
had
been
substantially
curtailed
in
certain
areas
of
the
country
such
as
the
Pacific
Northwest.
If
activity
in
the
Southeast
has
increased
dramatically
to
more
than
offset
this
downturn
elsewhere,
please
note.

EPA
Response:
Various
sources
of
information
yield
widely
varying
estimates
of
nonroad
engine/
equipment
growth.
In
regard
to
logging
equipment,
EPA
uses
a
negative
one
percent
growth
rate
for
diesel
logging
equipment.
Positive
growth
in
this
sector
comes
from
sparkignition
logging
equipment,
specifically
industrial
chainsaws
over
six
horsepower.

1.2.12
Comment:
An
annual
recreational
equipment
growth
rate
of
0.9%
appears
low,
much
lower
than
the
demographic
growth
rate
(
which
one
would
expect
recreational
equipment
to
follow.)
And
given
that
recreational
equipment
purchases
are
likely
tied
to
increased
levels
of
disposable
income,
I
would
expect
such
purchases
to
have
increased
substantially
during
the
mid
1990s.

EPA
Response:
The
growth
rate
of
0.9
percent
from
Table
1
is
a
BEA
estimate.
EPA
did
not
use
the
BEA
estimates
in
the
NONROAD2002
model.
In
addition,
EPA
revised
the
recreational
growth
rates
for
modeling
done
to
support
the
Recreational
and
Large
Spark­
Ignition
Engine
Rule.
EPA
developed
separate
growth
rates
for
ATVs,
snowmobiles,
and
off­
highway
motorcycles
based
on
data
provided
by
the
Motorcycle
Industry
Council
and
the
International
Snowmobile
Manufacturers
Association.
This
is
discussed
in
the
August
7,
2002
memorandum,
"
Updated
Population
Growth
Projections
for
Snowmobiles,
ATVs,
and
Off­
Highway
Motorcycles.
This
memorandum
was
submitted
to
the
peer
reviewer.
Recreational
marine
equipment,
including
boats
using
diesel
engines,
continued
to
use
the
PSR­
based
growth
estimate
of
3.3
percent.
EPA
is
incorporating
the
changes
made
to
the
recreational
growth
rates
in
the
next
revision
to
the
technical
report
addressing
growth
estimates
used
in
the
model
(
NR­
008c)
and
will
attempt
to
present
the
information
in
a
clearer
manner.

1.2.13
Comment:
The
current
projection
methodology
based
on
national
level
population
data
can
be
adjusted
to
reflect
locality­
specific
factors
at
the
county
level.
In
order
to
account
for
growth
at
the
county
level,
readily
available
surrogates
must
first
be
determined
for
each
sector.
For
example,
county­
level
population
projections
can
be
obtained
based
on
US
Census
data,
which
can
be
used
to
adjust
national
level
projections
for
the
construction,
industrial,
lawn
and
garden,
recreational,
and
possibly
light
commercial
sectors.
(
Logging,
railway,
and
farm
10
equipment
populations
may
or
may
not
lend
themselves
to
this
sort
of
adjustment,
depending
on
the
availability
of
other
appropriate
surrogates
 
e.
g.,
acreage
in
crop
production
by
county/
state
and
year
for
the
farm
sector.)

Under
this
proposal,
demographic
growth
rates
at
the
county
level
could
be
normalized
relative
to
national
demographic
growth
rates,
with
the
sum
of
future
county
equipment
population
totals
constrained
to
equal
the
national
equipment
total
projection.
This
approach
admittedly
cannot
account
for
regional
differences
in
per
capita
equipment
ownership,
but
should
still
provide
an
improvement
over
the
current
scheme.

EPA
Response:
EPA
will
be
considering
regional,
state,
and
county­
specific
growth
rates
for
the
MOVES
model,
which
will
be
the
next
generation
model
for
mobile
source
emissions
inventory
modeling.

2.0
Review
of
CI
Emission,
Deterioration,
and
Transient
Adjustment
Factors
[
Penny]

2.1
Comments
from
Dr.
Nigel
Clark,
University
of
West
Virginia
2.1.1
Comment:
Consider
the
language
concerning
"
Percentage
Phase­
in
Allowance,"
on
page
3,
as
well
as
Table
1
and
Appendix
A
Table
A1.
The
"
cumulative
total
of
eighty
percent"
language
does
not
clearly
convey
the
0.2
 
0.1
 
0.1
(
for
2
years
each)
sequence
in
Table
A1.
If
it
is
the
intent
to
inform
the
reader
fully,
some
extra
words
may
be
needed
on
page
3.
If
it
is
the
intent
just
to
refer
the
user
to
Table
A1
as
the
primary
source,
the
page
3
language
is
fine
as
it
stands.

EPA
Response:
EPA
will
seek
to
clarify
this
in
the
next
version
of
the
diesel
emission
factor
technical
report
(
NR­
009b).

2.1.2
Comment:
The
issue
of
crankcase
emissions
(
p.
5)
is
very
interesting.
Presumably
the
report
uses
the
2%
number
across
the
board
because
so
little
is
known
about
these
emissions,
but
it
would
be
good
to
cite
the
source
on
p.
5
(
the
source
is
referenced
only
toward
the
end
of
the
report).
Crankcase
emissions
are
equivalent
in
volume
to
the
ring
and
valve
stem
blowby,
and
are
probably
around
one
or
two
percent
of
exhaust
flow
for
most
engines.
Since
combustion
is
incomplete
during
some
of
the
blowby,
the
hydrocarbon
concentration
in
the
crankcase
gases
is
likely
to
be
higher
than
in
the
exhaust
gases.
The
"
2%"
value
may
be
reasonable
in
some
cases,
but
more
likely
underestimates
the
crankcase
emissions.
However,
crankcase
emissions
are
clearly
small
from
an
overall
inventory
perspective,
so
that
no
recommendation
to
change
the
present
document
is
made,
other
than
to
add
the
reference
on
p.
5.
In
the
longer
term,
crankcase
emissions
should
receive
some
experimental
attention.
Even
though
these
emissions
will
be
controlled
by
requiring
positive
crankcase
regulation
in
the
future,
off­
road
engines
enjoy
a
long
life,
and
open
crankcase
emissions
will
remain
in
the
inventory
for
decades
to
come.

EPA
Response:
EPA
will
add
a
citation
of
the
source
on
page
5
during
the
next
revision
of
the
11
technical
report
(
NR­
009b).

2.1.3
Comment:
On
p.
5,
perhaps
state
why
the
tier
2
emission
factors
have
the
0.2%
sulfur
default
level.

EPA
Response:
This
will
be
added
during
the
next
revision
of
the
emission
and
deterioration
factor
technical
report.

2.1.4
Comment:
The
correction
of
NMHC
to
THC
using
1.02
in
Table
2
appears
without
reference
or
comment,
though
I
doubt
it
is
of
much
concern!

EPA
Response:
The
references
for
the
conversion
from
NMHC
to
THC
and
vice­
versa
can
be
found
in
NR­
002a,
"
Conversion
for
Hydrocarbon
Emission
Factor
Components".
It
should
be
noted
that
the
conversion
table
in
NR­
002a
contains
different
values
because
the
conversion
is
from
THC
to
NMHC
(
0.984
for
diesel),
as
well
as
other
hydrocarbon
types,
while
the
value
in
the
footnote
of
Table
2
in
NR­
009b
(
1.02)
is
to
convert
from
diesel
NMHC
to
THC
(
1/
0.984).
EPA
will
be
clarify
this
in
the
next
revision
of
NR­
009b.

2.1.5
Comment:
Consider
Table
D5.
There
is
an
entry
above
the
Tier
1
standard
in
the
50­
100
hp
category
for
oxides
of
nitrogen
(
80hp,
line
4,
p.
D23,
7.61
grams).
How
can
this
be?
Page
3
language
suggests
that
it
cannot
be
a
"
Percentage
Phase­
in
Allowance,"
since
it
would
be
allowed
only
for
engines
25
to
50hp
in
1999.

Consider
Table
D5,
1999
standards
for
25
to
50hp
engines
(
pp.
D21­
22).
I
examined
oxides
of
nitrogen
values.
Is
this
table
meant
to
contain
values
for
engines
that
should
have
been
Tier
1,
but
that
had
higher
emissions
due
to
the
"
Percentage
Phase­
in
Allowance?"
If
so,
none
appears,
and
this
does
not
jibe
with
the
p.
3
statement,
that
manufacturers
took
full
advantage
of
the
"
Percentage
Phase­
in
Allowance."
I
assume
that
"
Phase­
in"
engines
do
not
appear
on
this
list,
because
they
were
Tier
0
and
hence
uncertified,
and
no
data
were
provided.

EPA
Response:
Some
engine
families
are
certified
above
the
standard
if
they
are
part
of
the
Average
Banking
and
Trading
(
ABT)
program.
In
other
words,
the
manufacturer
can
use
ABT
credits
to
certify
an
engine
family.
The
entry
referred
to
by
the
reviewer
is
part
of
the
ABT
program.
The
NOx
Family
Emission
Limit
(
FEL)
for
that
engine
family
is
8.3
g/
bhp­
hr.
"
Phasein
engines
are
not
included
in
the
tables
in
Appendix
D
of
the
report.

2.1.6
Comment:
On
p.
11,
the
text
states
that
NONROAD's
Tier
1
emission
factors
are
based
on
EPA
certification
data,
and
refers
the
reader
to
Appendix
D.
However,
Appendix
D
Table
D1
states
that
only
boldface
numbers
were
used.
Why
do
small
differences
between
cert.
and
actual
NONROAD
values
differ
in
some
cases?
(
Example:
Tier
1
NOx
Tables
4
and
A2
are
4.7279
for
25
to
50hp,
but
cert.
average
in
Table
D1
is
4.640553.
If
explanation
appears
in
the
text,
I
apologize
for
overlooking
it.
12
EPA
Response:
For
categories
less
than
50
hp,
the
Tier
1
standard
for
HC
and
NOx
is
expressed
as
a
combination
of
HC
and
NOx.
As
a
result,
most
of
the
certification
data
for
these
engines
is
provided
as
the
sum
of
HC
and
NOx,
although
there
are
some
data
reported
for
HC
and
NOx
separately.
Using
the
example
for
25
to
50
hp
engines
above,
72
engines
reported
NOx
separately,
while
270
engines
reported
results
as
a
combination
of
HC
and
NOx.
The
cert
average
in
Table
D1
of
4.640553
is
based
on
the
data
reporting
NOx
separately;
however,
due
to
the
smaller
sample
size,
this
value
was
not
used.
Instead,
HC
fractions
of
HC+
NOx
emissions
were
calculated
for
those
tests
which
reported
both
HC
and
HC+
NOx
emission
factors.
The
average
HC
fraction
was
then
calculated
and
multiplied
by
the
sales­
weighted
average
HC+
NOx
emission
factor
to
obtain
an
HC
emission
factor
(
referred
to
as
HC
calc
in
the
tables).
The
remaining
fraction
of
the
HC+
NOx
emission
factor
was
assigned
as
the
NOx
emission
factor
(
referred
to
as
NOx
calc
in
the
tables).
For
25
to
50
hp
engines,
the
NOx
calc
value
of
4.7279
in
Table
A2
was
used.
This
approach
is
discussed
in
Appendix
D
of
the
report.

2.1.7
Comment:
On
p.
11,
Option
(
1)
seems
to
assume
implicitly
that
timing
changes
(
and
maybe
increased
internal
EGR
due
to
valve
timing
changes)
are
the
only
tools
to
be
used,
hence
yielding
a
NOx
 
PM
tradeoff.
This
is
probably
reasonable,
although
continuous
full
load
operation
at
about
4.5
g/
bhp­
hr
NOx
may
thermally
challenge
an
engine
manufactured
with
conventional
materials.
In
the
case
of
highway
engines
with
conventional
materials,
thermal
relief
was
provided
by
"
off­
cycle"
timing
operation
that
persisted
through
the
90'
s.
None
of
the
options
is
flawed
in
its
arguments.
However,
why
have
the
seemingly
arbitrary
value
of
75%
in
Option
(
4)
when
the
highest
value
in
appendix
E
is
27%?
Also,
Option
(
4)
is
worded
generically,
whereas
data
are
presented
only
for
NOx
and
PM
in
Appendix
E.
I
do
understand
that
these
margins
may
be
large
for
CO
or
HC.

EPA
Response:
Appendix
E
only
presents
results
for
NOx
and
PM
because
Option
(
4)
was
only
applied
for
these
pollutants.
For
CO
and
HC,
the
margins
were
greater
than
80%,
which
is
the
basis
for
the
upper
limit
of
75%.

2.1.8
Comment:
On
p.
14,
is
it
fair
to
say
that
high
HC
compliance
margins
are
"
unrealistically
large?"
The
issue
is
that
in
the
case
of
highway
engines,
the
HC
standards
have
not
driven
technology
development
at
all,
and
hence
the
standards
for
HC
are
unrelated
to
cert.
values.
These
compliance
margins
will
map
from
one
application
to
another
only
if
the
standards
are
technology
drivers.
Upon
re­
reading
the
report
language,
I
do
not
think
that
I
disagree
with
the
report's
intent,
but
a
wording
change
might
be
good.

EPA
Response:
This
clarification
will
be
added
to
the
next
version
of
the
CI
emission
factor
technical
report.

2.1.9
Comment:
The
choices
of
options
on
p.
14
seem
reasonable.
I
am
curious
as
to
whether
an
overall
comparison
was
made
(
in
a
working
table)
to
see
how
much
the
options
differed.
For
NOx
and
PM
(
where
all
options
were
credible),
it
would
be
comforting
to
know
that
the
choice
did
not
matter
substantially.
For
HC
and
CO,
clearly
some
options
were
unsuited
and
comparison
13
would
not
be
useful.

EPA
Response:
Such
an
overall
comparison
was
not
made,
but
will
be
considered.

2.1.10
Comment:
On
p.
16,
the
deterioration
factor
approach
is
reasonable,
but
could
be
improved
in
future
models.
The
use
of
a
linear
deterioration
rate
is
acceptable
over
the
engine
life,
but
clearly
engines
operate
beyond
expected
life
and
continue
to
decline
in
emissions
control
(
and
ultimately
in
fuel
efficiency),
particularly
as
they
are
sold
down
the
food
chain.
Unfortunately,
it
is
hard
to
extrapolate
this
deterioration.
Also,
the
Age
Factor,
being
based
on
cumulative
work
performed
by
the
engine
over
its
life,
is
quite
simple.
In
real
use,
I
would
expect
that
an
engine
running
at
high
speed
and
low
load
(
e.
g.
An
oversized
constant
speed
generator
set)
may
suffer
substantial
wear
relative
to
the
useful
cumulative
work
done.
The
safety
net
is
that
deterioration
is
averaged
in
the
inventory
over
many
applications,
but
this
might
not
apply
if
a
specific
engine
application
prevailed
in
a
region.
The
report
does
observe
that
there
is
a
lack
of
deterioration
data
for
nonroad
engines,
and
perhaps
this
could
be
an
area
for
future
study.

EPA
Response:
Agreed.
No
response
necessary.

2.1.11
Comment:
On
p.
19,
why
is
HC
subtracted
in
arriving
at
carbon
dioxide
emissions,
while
CO
is
not?
They
are
both
consuming
fuel
carbon
sources.
However,
they
are
also
both
small,
so
predictions
will
not
be
affected
much.

EPA
Response:
HC
is
used
as
the
correction
for
unburned
fuel
in
this
equation.
We
agree
that
this
correction
is
insubstantial
for
diesel
emissions.

2.1.12
Comment:
On
p.
19,
0.87
is
"
hard
coded"
as
the
carbon
mass
fraction
in
diesel.
Allowing
this
to
be
a
variable
would
adjust
for
runs
with
less
aromatic
fuels,
but
it
will
not
enhance
the
model's
utility
appreciably.
In
fact,
it
is
probably
lost
in
the
relative
inaccuracy
of
BSFC
data.

EPA
Response:
Agreed.
No
response
necessary.

2.2
Comments
from
Michael
Hutcheson,
URS
Corp.

2.2.1
Comment:
The
reviewer
would
like
to
state
at
the
outset
that
the
description
of
brake
specific
fuel
consumption
(
BSFC)
as
a
pollutant
or
emission
is
an
incorrect
characterization.
BSFC
should
be
referred
to
and
described
in
a
purer
context
as
a
measure
of
the
amount
or
rate
of
fuel
consumed
per
horsepower
of
work
output.
BSFC
is
a
function
of
engine
speed
and
therefore
varies
with
engine
load
conditions.
Charts
of
BSFC
versus
engine
speed
vary
from
engine
to
engine.
Correspondingly,
BSFC
is
dependent
on
the
equipment
design
and
equipment
use
but
should
not
be
confused
with
an
emission.

EPA
Response:
This
is
a
good
point.
EPA
will
revise
this
in
the
next
version
of
NR­
009b.
14
2.2.2
Comment:
On
page
6,
when
discussing
the
use
of
Tier
0
BSFC
for
all
engine
model
years
(
presumably
by
hp/
category),
it
should
be
clarified
that
this
has
an
affect
on
both
CO
2
and
SO
2
emission
factors.
The
presumption
being
that
since
BSFC
is
not
assumed
to
change
then
the
related
emissions
of
CO
2
and
SO
2
do
not
change
(
for
a
given
fuel)
even
though
new
emission
reduction
technologies
are
being
implemented.
This
correlation
should
be
identified
in
order
to
clarify
the
affect
of
this
assumption.

EPA
Response:
The
correlation
between
BSFC
and
CO
2
and
SO
2
will
be
stated
more
clearly
in
a
future
version
of
the
CI
emission
factor
technical
report.
CO
2
and
SO
2
will
change
slightly
as
HC
emissions
are
changed,
since
the
carbon
and
sulfur
that
goes
to
exhaust
HC
(
unburned
fuel)
is
subtracted
in
the
CO
2
and
SO
2
equations.
For
diesel
emissions,
this
is
insubstantial.

2.2.3
Comment:
In
the
description
of
the
sources
of
data
used
for
generating
emission
factors
for
certain
engine
horsepower
classifications
the
report
states
that
no
certification
data
is
available
for
tier
2
engines
<
300
hp
and
>
600
hp.
Since
Tier
2
for
the
600
hp
to
750
hp
category
started
in
2002,
data
for
Tier
2
for
this
horsepower
class
should
be
available.
In
addition,
Tier
2
data
for
engines
from
100
hp
to
300
hp
should
be
available
now
(
and
at
the
time
of
writing
the
report)
since
the
standard
applies
in
2003.
It
is
understandable
that
the
100
hp
to
300
hp
data
would
not
have
been
ready
before
the
Nonroad2002
revision
but
the
600
hp
to
750
hp
Tier
2
certification
data
should
have
been
ready
and
used.

EPA
Response:
This
is
a
good
point.
Unfortunately,
EPA
was
not
able
to
incorporate
the
Tier
2
certification
data
for
600
to
750
horsepower
engines
before
the
model
needed
to
be
locked
down
for
use
in
supporting
the
proposed
Nonroad
Diesel
Engine
Rule.
EPA
will
address
this
during
the
next
revision
of
NONROAD's
diesel
emission
factors.

2.2.4
Comment:
When
describing
the
methods
used
to
determine
emission
factors
for
Tier
2
and
Tier
3
engines
in
each
horsepower
category,
the
report
states
that
a
different
method
of
determining
the
emission
factor
was
used
for
HC
and
for
NOx
for
the
600
hp
and
above
hp
categories.
For
HC,
the
certification
values
for
Tier
2
standards
in
the
300
to
600
hp
category
were
used.
For
NOx,
a
default
compliance
margin
of
10
%
was
used.
This
does
not
make
intuitive
sense.
It
is
recognized
that
the
Tier
2
certification
values
for
NOx
in
the
300
hp
to
600
hp
engine
category
did
not
meet
the
10
%
compliance
margin
EPA
assumes
manufacturers
will
try
to
meet.
The
available
certification
evidence
indicates
that
manufacturers
are
not
beholden
to
a
10
%
compliance
margin.
Since
the
standard
for
all
these
hp
categories
are
the
same
and
the
report
states
that
"
the
use
of
certification
data
is
preferable
to
applying
a
compliance
margin"
the
reviewer
believes
the
NOx
certification
data
for
300
hp
to
600
hp
category
should
be
used
for
the
horsepower
categories
above
600
hp.

EPA
Response:
The
approach
described
by
the
reviewer
could
also
have
been
used.
Since
the
two
approaches
yield
emission
factors
that
only
differ
by
5
percent,
the
current
approach
was
retained.
15
3.0
Review
of
Geographic
and
Temporal
Allocation
Factors
3.1
Comments
from
Kirsten
Thesing,
E.
H.
Pechan
3.1.1
Seasonal/
Monthly
Allocation
Factors
3.1.1.1
Comment:
The
discussion
of
the
calculation
of
national
average
seasonal
and
monthly
activity
allocations
was
somewhat
unclear
(
as
described
in
the
March
1999
report).

°
For
Step
1,
it
states
that
emission
estimates
are
generated
for
each
of
50
States.
Are
the
emissions
then
summed
to
the
national
level
by
accounting
for
each
State?

°
In
looking
briefly
at
the
season.
dat
file,
the
number
of
different
equipment
types
for
each
region
is
65,
and
the
number
of
different
equipment
types
for
the
total
US
is
111.
Additional
national
allocations
are
provided
for
commercial
vs.
residential
lawn
and
garden,
golf
carts,
oil
field
equipment,
and
other
underground
mining
equipment.
It
is
not
clear
how
national
average
allocations
are
calculated
for
these
additional
equipment
categories.

°
Also,
in
doing
quality
assurance
of
a
1996
county­
level
inventory
developed
using
NONROAD
(
in
conjunction
with
EPA's
Emission
Factor
and
Inventory
Group),
I
found
that
the
results
from
a
specific
single
State
run
did
not
equal
results
for
the
same
State
when
doing
a
50­
State
run.
This
was
discovered
during
May
1999,
and
Craig
Harvey
documented
this
glitch
in
a
May
11,
1999
email.
Do
the
average
national
seasonal
and
monthly
activity
allocations
reflect
any
corrections
to
fix
this
known
error
(
since
the
latest
version
of
the
"
season.
dat"
file
is
dated
3/
22/
99).

EPA
Response:

In
order
to
calculate
national
average
seasonal
allocations,
EPA
did
12
runs
of
the
model
to
generate
outputs
for
all
50
states
in
one
run
for
each
month.
The
emissions
for
each
of
the
50
states
were
summed
to
the
national
level
to
account
for
each
state.

The
monthly
allocations
at
the
state/
regional
level
are
given
by
aggregate
SCC's
wherever
possible,
such
as
using
the
entire
Rec
Equipment
or
Lawn
&
Garden
segment,
or
even
an
overall
average
(
2270000000
all
diesel
equipment
not
otherwise
specified).
But
to
properly
weight
together
the
allocations
for
all
the
states,
taking
into
account
the
different
populations
and
horsepower
distributions
within
each
state,
it
needs
to
be
done
at
the
level
of
individual
SCC's.
Reaggregating
the
individual
SCC
monthly
allocations
to
the
higher
levels
would
actually
decrease
the
precision
of
the
results.

The
discrepancy
between
the
results
of
a
single
state
run
and
the
same
state
in
a
50
state
run
has
been
fixed.
16
3.1.2
Weekday/
Weekend
Day
Allocation
3.1.2.1
Comment:
I
found
this
report
to
clearly
explain
the
methods
and
data
behind
the
weekday
and
weekend
day
fractions.

EPA
Response:
No
Response
needed.

3.1.2.2
Comment:
There
was
no
discussion
of
how
the
weekday
and
weekend
day
fractions
were
developed
for
railway
maintenance
category.
This
category
showed
the
highest
total
weekday
use
(
90%),
and
the
lowest
weekend
day
use
(
10%).
Were
these
values
based
on
a
survey,
or
expert
judgement?

EPA
Response:
Lacking
any
definitive
data
for
railway
maintenance
equipment
weekly
usage
patterns,
the
week
day
and
weekend
day
allocations
were
based
on
expert
judgement.

3.1.2.3
Comment:
You
assigned
the
same
weekday/
weekend
temporal
profile
for
the
agricultural
category
that
is
assigned
to
other
commercial­
related
categories
which
tend
to
operate
equipment
more
during
the
week
than
the
weekend.
Depending
on
the
season,
though,
and
especially
during
harvest,
it
would
seem
that
agricultural
equipment
may
need
to
be
operated
throughout
the
week,
including
weekends.
In
the
absence
of
actual
survey
data,
I
understand
the
assignment
you
made,
but
I
would
think
that
farm
equipment
has
the
potential
to
be
used
as
much
on
the
weekends
as
the
weekdays.

EPA
Response:
EPA
is
aware
that
agricultural
equipment
could
be
operated
on
the
weekends
more
than
the
current
allocation
allows.
EPA
would
like
to
find
documented
information
which
would
indicate
whether
a
change
to
the
allocation
is
warranted.

3.1.3
Geographic
Allocation
3.1.3.1
Comment:
Commercial
and
Industrial
Equipment:
As
a
result
of
my
company's
research
and
investigations
into
economic
output
data
for
other
projects
(
which
may
concern
developing
emissions
projections
of
data,
as
well
as
geographical
allocations),
we
have
not
found
that
any
SIC­
specific
output
data
are
available
at
the
county
level
for
the
whole
U.
S.

The
surrogate
used
for
industrial
equipment,
manufacturing
employees,
would
appear
to
be
a
reasonable
choice.
The
surrogate
for
light
commercial,
(
establishment
for
SIC
50:
wholesale
trade
­
durable
goods)
is
consistent
with
what
was
used
in
the
1991
NEVES,
as
determined
by
Energy
and
Environmental
Analysis.
However,
there
is
a
limitation
in
how
the
surrogates
were
selected,
as
explained
in
the
following
write­
up
from
Pechan's
September
1997
report,
Evaluation
of
PSR
Nonroad
Population
Data
Base.

"
As
part
of
the
NEVES,
Energy
and
Environmental
Analysis
(
EEA)
estimated
county
populations,
and
ultimately,
nonattainment
area
populations,
from
national
data
(
EEA,
1991).
In
17
developing
its
methodology,
EEA
estimated
regression
equations
based
on
PSR
State­
level
data.
With
some
notable
exceptions,
these
regression
equations
were
typically
developed
using
number
of
employees
in
related
industries
(
e.
g.,
State­
level
logging
equipment
was
regressed
versus
the
number
of
employees
in
SIC
241
 
Logging).
This
approach
uses
indicators
that
are
statistically
correlated
with
PSR
State
distributions.
The
major
concern
with
the
EEA
approach
is
that
it
is
based
on
relationships
with
PSR's
State
estimates,
which
PSR
has
indicated
are
simply
summations
of
its
county
estimates."

In
addition
to
this
limitation,
SIC
50
may
not
be
capturing
nonroad
engine
activity
related
to
all
commercial
activities.
In
developing
forecasts
for
Economic
Growth
Analysis
System,
version
4.0
(
E­
GAS
4.0),
for
commercial
sector
categories,
we
have
in
some
cases
chosen
to
use
the
number
of
establishments
for
SICs
50
through
89,
which
encompasses
the
SICs
for
wholesale
trade,
retail
trade,
finance,
insurance
and
real
estate,
and
services.
The
finance,
insurance
and
real
estate
SICs
may
not
be
appropriate
to
include
for
nonroad
equipment,
but
the
other
three
would
intuitively
seem
to
correlate
better
than
wholesale
trade
alone.
Ideally,
one
would
want
to
perform
regressions
to
test
how
this
group
of
SICs
correlates
to
light
commercial
activity/
populations.
Based
on
the
types
of
equipment
that
are
classified
as
light
commercial
(
compressors,
generators,
welders),
including
establishments
for
retail
trade
and
services
in
the
allocation
factor,
as
well
as
wholesale
trade,
may
better
reflect
activity.

EPA
Response:
To
the
best
of
EPA's
knowledge,
the
geographic
allocation
factors
in
the
NONROAD
2002
model
used
for
the
proposed
Nonroad
Diesel
Rule
were
the
best
available
at
the
time.
EPA
appreciates
receiving
comments
indicating
better
methods
or
data.
EPA
will
consider
improvements
to
the
NONROAD
2002
model
for
the
final
Nonroad
Diesel
Rule
analysis.

3.1.3.2
Comment:
Logging
Equipment:
You
requested
comments
on
the
availability
of
data
on
the
number
of
forest
acres
harvested
per
county.
In
trying
to
develop
estimates
for
slash
burning
for
EPA's
Emission
Factor
and
Inventory
Group,
we
looked
into
obtaining
these
data
at
a
county
or
State
level,
since
the
number
of
acres
logged
would
provide
an
initial
indicator
of
the
amount
of
acres
that
are
ultimately
slash
burned.
The
U.
S.
Forest
Service
does
track
the
number
of
acres
harvested,
but
they
acknowledge
that
the
uncertainty
associated
with
these
data
at
a
county
level
is
very
high.
The
reliability
increases
when
counties
are
grouped
into
multi­
county
regions,
or
at
the
State
level.

The
U.
S.
Forest
Service
does
maintain
an
on­
line
timber
product
output
(
TPO)
data
base
retrieval
system
which
provides
county
level
estimates
of
the
total
timber
removal,
with
a
subset
for
roundwood
product
removal
(
available
at
http://
srsfia.
usfs.
msstate.
edu/
rpa/
tpo/.
Logging
equipment
activity
would
be
expected
to
correlate
with
the
volume
of
timber
removed.
The
data
are
reported
for
all
land
ownerships,
including
National
Forest,
Other
Public,
Forest
Industry
and
Other
Private.
Through
the
current
on­
line
system,
you
can
select
all
counties
to
obtaining
data
at
the
county
level
for
the
whole
U.
S.,
but
this
may
be
a
bit
cumbersome.
In
speaking
to
contacts
at
the
Forest
Service
that
work
with
the
TPO
data
base,
data
could
also
be
compiled
by
them
for
18
little
to
no
fee.

The
reliability
of
the
timber
volume
data
at
the
county
level
is
higher
than
the
acres
harvested
data
at
the
county
level.
However,
as
explained
in
the
documentation
for
the
TPO
database,
there
are
limitations
associated
with
the
data
(
e.
g.,
some
counties
have
been
combined
due
to
data
paucity
or
sensitivity,
and
the
data
contained
is
subject
to
being
updated
or
revised
based
on
reviewer
and
user
comments).
The
employment
data
from
County
Business
Patterns
(
CBP)
is
likely
to
be
a
more
consistent
database,
but
the
volume
of
timber
removed
may
more
accurately
reflect
logging
equipment
use.
As
this
database
is
reviewed
and
refined
further,
this
may
be
a
potential
allocation
source
for
logging
activity
in
the
future.

EPA
Response:
This
appears
to
be
a
promising
source
of
data.
EPA
will
take
this
information
into
consideration
during
development
of
the
nonroad
portion
of
the
MOVES
model.

3.1.3.3
Comment:
Railroad
Maintenance
and
AC/
Refrigeration
Units:
You
requested
comment
on
the
availability
of
railroad
track
mileage
by
county.
In
doing
work
for
specific
States
(
e.
g.,
Pennsylvania)
railroad
track
mileage
by
county
has
been
obtained
and
used
to
apportion
State­
level
rail
activity
and
emissions
to
counties.
I
am
not
aware
of
one
national
source
that
would
provide
these
data
for
each
State
by
county.
It
is
likely
that
each
State's
Department
of
Transportation
or
equivalent
would
need
to
be
contacted
to
obtain
these
data.

For
AC/
Refrigeration
Units,
since
this
activity
is
tied
to
vehicular
truck
travel,
you
may
want
to
consider
allocating
to
States
based
on
vehicle
miles
traveled
(
VMT)
for
light­
duty
gas
trucks
(
LDGTs),
light­
duty
diesel
trucks
(
LDDTs),
heavy­
duty
gasoline
vehicles
(
HDGVs),
and
heavy­
duty
diesel
vehicles
(
HDDVs).
Consistent
with
how
we
allocate
VMT
to
the
counties
for
Trends,
though,
county
level
VMT
would
need
to
be
estimated
based
on
population.
So,
this
method
may
improve
upon
the
National
to
State
allocation,
but
the
county
allocation
is
still
based
on
population.
To
the
extent
that
AC/
Refrigeration
units
may
also
be
located
on
railroad
boxcars,
an
approach
based
solely
on
VMT
would
be
insufficient.

EPA
Response:
These
might
be
useful
sources
of
allocation
data.
The
reviewer
highlighted
a
challenge
that
EPA
faces
with
allocation
factors,
namely
finding
county­
level
data
with
nationwide
coverage.

3.2
Comments
from
Sam
Wells,
Consultant
3.2.1
Comment:
The
three
reports
regarding
geographic
and
temporal
allocation
for
NONROAD2002
are
well
written.
The
discussion
of
activity
versus
equipment
population
seems
a
trifle
awkward
at
times
and
could
use
some
editing
 
the
main
concept
should
be
that
we
are
using
secondary
surrogates
that
would
best
describe
activity
and
population,
taken
together,
given
that
accurate
ground
count
data
is
not
available.
In
fact,
use
of
surrogates
is
preferable
to
using
ground
data
in
some
cases
because
of
the
mobile
nature
of
the
sources;
the
exception
would
be
large
construction
projects
that
would
be
measured
in
terms
of
fractions
of
billions
of
dollars.
19
EPA
Response:
EPA
will
try
to
clarify
the
discussion
of
activity
versus
equipment
population.

3.2.2
Geographic
Allocation
by
Equipment
Category
3.2.2.1
Comment:
Construction
Discussion
of
the
development
of
adjustment
factors
from
previous
surveys
done
in
Houston,
Texas
is
a
little
confusing
because
there
have
been
at
least
five
major
efforts
dating
back
to
the
before
the
1991
NEVES.
The
proposed
factors
(
for
residential,
highway,
and
other
economic
sectors)
are
not
well
documented,
although
it
is
obvious
that
construction
activity
is
more
intensive
in
highway
and
utility
applications
rather
than
others
such
as
residential.
One
recommendation
would
be
to
use
the
existing
factors
and
then
further
study
the
issue
with
special
attention
to
regional
construction
practices.
For
example,
the
cost
of
labor
and
fuel
can
substantially
affect
how
the
"
dollars
of
construction
value"
actually
translates
into
construction
activity.
As
an
example,
it
was
found
that
the
cost
of
labor
in
Dallas
was
approximately
6
percent
higher
than
in
Houston.
Therefore,
one
could
make
a
further
adjustment
by
dividing
construction
revenue
statistics
by
1.06.
This
approach
has
not
been
verified
by
comparing
the
method
to
hard
data
but
could
be
something
to
consider.

Also,
one
might
consider
the
difference
between
construction
machinery
used
in
relatively
soft
material
(
e.
g.,
sandy
clay)
versus
hard
material
(
e.
g.,
limestone
and
granite).
Cranes
and
excavators
are
more
likely
to
be
used
in
soft
media;
rock
drills,
rock
saws,
boring
machines,
and
trenchers
are
more
likely
to
be
used
in
harsher
media.
I
am
not
exactly
sure
how
to
adjust
the
kinds
of
equipment
from
region
to
region,
although
when
the
Austin
construction
inventory
was
compared
to
one
done
for
Houston,
we
noted
that
Houston
had
zero
trenching
equipment
 
and
46
were
found
just
in
the
Austin
area
(
some
were
up
to
1,000
HP,
and
had
a
significant
contribution
to
the
emission
inventory).
Perhaps
use
of
USGS­
GIS
maps
to
determine
alluvial
sands
versus
rocky
soils
would
be
something
to
explore.

As
a
word
of
caution,
the
equipment
rental
sector
is
very
difficult
to
ascertain,
and
is
a
"
wild
card"
in
any
scheme
that
uses
secondary
surrogates.
For
example,
in
most
cases
cranes
are
obtained
from
rigging
companies
that
specialize
in
bridges
and
elevated
highways,
commercial
high­
rise
buildings,
or
even
industrial
uses
(
e.
g.,
refinery
turnaround
construction).
To
my
recollection,
none
of
the
studies
conducted
in
Houston
were
able
to
get
a
handle
on
the
crane
market
sector,
so
extrapolations
to
other
areas
in
the
U.
S.
would
probably
have
to
be
done
in
a
top­
down
manner.
The
ENVIRON
memorandum
notes
that
port
and
industrial
activity
in
Houston
might
not
be
characteristic
of
other
urban
cities,
but
an
unintended
result
is
that
perhaps
other
areas
of
application
would
suffer
from
potential
misprediction
of
specific
types
of
equipment
such
as
heavy­
duty
cranes
(
150­
750
HP,
excluding
the
larger
ship
cranes
up
to
2,500
HP).
Again,
these
kinds
of
refinements
can
be
incorporated
over
the
next
few
years
but
are
worthy
of
further
attention.

With
this
said,
I
applaud
the
EPA
for
considering
several
metrics
instead
of
one
"
total
dollar
of
construction
revenue"
figure
per
county
for
the
purposes
of
spatial
allocation.
20
EPA
Response:
EPA
is
aware
of
differences
in
the
cost
of
construction
by
region
and
the
impact
this
has
on
under
or
over­
predicting
construction
equipment
emissions.
EPA
plans
to
address
this
in
future
revisions
to
the
model.
The
rental
sector
is
another
area
in
which
EPA
will
need
to
do
further
investigation
in
the
future.

EPA
believes
that
the
adjustments
made
to
the
dollar
amount
of
construction
allocation
factors
are
generally
representative
of
construction
equipment
usage
on
infrastructure
projects
versus
usage
on
less
intense
building­
related
construction
projects.
Since
it
is
a
national
model,
EPA's
NONROAD
model
does
not
have
the
capability
to
account
for
all
local
variations
in
construction
equipment
usage.

3.2.2.2
Comment:
Agricultural
Use
of
crop
land
acreage
is
the
best
method
to
allocate
agricultural
activity.
However,
the
USA
Counties
data
set
is
very
crude,
at
best,
and
one
could
attempt
to
utilize
other
more
robust
data
sources
such
as
from
the
U.
S.
Natural
Resources
Conservation
Service
(
NRCS):

http://
www.
nrcs.
usda.
gov/
technical/
NRI/
1997/
summary_
report/
index.
html
Note
that
total
crop
land
includes
cultivated
and
non­
cultivated
crop
land,
so
variations
in
noncultivated
crop
land
statistics
would
skew
the
allocation
of
agricultural
activity.
On
average
one
sees
approximately
10
percent
of
the
total
acreage
is
not
cultivated
(
meaning
little
tractor
and
no
combine
activity),
but
in
non­
cultivated
crop
land
in
California
is
approximately
30
percent
and
in
Connecticut
it
is
larger
than
50
percent
of
the
total.
Presumably,
non­
cultivated
crop
land
would
be
land
that
would
be
cultivated
but
is
left
fallow
due
to
federal
incentives
and
other
factors.
I
strongly
recommend
the
use
of
this
publicly­
available
NRCS
database
because
it
also
includes
irrigated
versus
non­
irrigated
crop
land,
crop
type,
and
historical
trends.
I
am
not
sure
if
the
data
includes
orchards
as
being
"
crop
land,"
but
it
would
be
obvious
that
orchard
spraying
might
need
to
be
considered
for
the
purposes
of
allocating
sprayers
(
especially
SCC
2270005035).

EPA
Response:
It
should
be
clarified
that
EPA
used
the
number
of
harvested
acres
of
crop
land
in
the
NONROAD
model,
so
that
fallow
crop
land
is
not
counted.
The
EPA
will
examine
the
U.
S.
Natural
Resources
Conservation
Service
(
NRCS)
data
during
development
of
the
nonroad
portion
of
the
MOVES
model.

3.2.2.3
Comment:
Airport
Ground
Support
Equipment
(
GSE)
Suggested
is
that
passenger
jetliner
landing
and
take­
off
operations
(
LTO)
data
from
the
Federal
Aviation
Administration
(
U.
S.
DOT)
be
used
instead
of
employment
to
allocate
GSE.
Only
certified
commercial
carriers
are
included.
The
amount
of
GSE
activity
ought
to
be
directly
proportional
to
the
number
of
jet
airliner
LTO.
I
have
not
found
an
internet
site
for
the
data,
since
we
always
used
paper
copies
in
the
past.

It
should
be
noted
that
while
GSE
is
oriented
towards
passenger
operations,
there
is
a
significant
commercial
cargo
component
(
e.
g.,
Airborne,
Emory,
FEDEX,
UPS)
involving
a
unique
piece
of
21
equipment
I
have
not
seen
in
the
NONROAD
model:
is
it
a
curious
cross
between
a
forklift
and
an
aerial
lift
that
perhaps
the
EPA
should
consider
them
in
the
future.
Again,
allocation
should
be
commercial
cargo
LTO
for
this
category.

Finally,
many
airports
have
special
GSE
agreements
to
reduce
emissions,
which
should
be
reflected
in
the
improved
model,
and
many
GSE
surveys
now
include
support
vehicles
as
being
part
of
the
mix
(
security,
food
&
beverage,
refueling
trucks,
etc.).
For
this
reason
it
might
be
prudent
to
"
hard
code"
some
of
the
larger
international
airports
because
each
has
a
unique
blend
of
LTO
by
airline
and
equipment
usage.

EPA
Response:
EPA
is
aware
of
the
FAA's
landing
and
take­
off
operations
data,
but
gaps
and
suspected
errors
in
the
data
set
prevented
EPA
from
utilizing
it
in
the
model.

3.2.2.4
Comment:
Light
Commercial
and
Industrial
The
proposed
indicators
appear
to
be
rational
although
I
have
concerns
about
(
1)
the
rental
market,
especially
for
aerial
lifts,
and
(
2)
industrial
forklift
activity.
Aerial
lifts
are,
at
least
from
my
experience,
mostly
rented
and
therefore
the
using
number
of
equipment
rental
companies
might
be
beneficial.
The
situation
with
industrial
forklifts
is
far
more
complicated
because
many
are
used
in
wholesale
and
retail
distribution
centers,
not
just
at
manufacturing
establishments.

EPA
Response:
This
is
good
point.
EPA
will
take
this
into
consideration
during
the
development
of
the
nonroad
portion
of
the
MOVES
model.

3.2.2.5
Comment:
Logging
Equipment
The
USDA
Forestry
Service
has
its
Forest
Inventory
and
Analysis
(
FIA)
unit
that
maintains
the
Timber
Product
Output
(
TPO)
database.
Annual
round­
wood
harvest
statistics
can
be
obtained
at
state
and
county
levels
of
detail.
Please
see:
http://
fia.
fs.
fed.
us/

Use
of
employment
data
(
SIC
2410),
while
easy
to
obtain,
can
lead
to
misleading
results
because
the
logging
industry
consists
of
felling
(
the
main
purpose
for
using
non­
road
equipment),
transportation,
and
processing
(
e.
g.,
de­
limbering,
de­
barking,
etc.)
for
delivery
to
a
mill.
The
TPO
output
for
round­
wood
is
also
useful
because
it
contains
data
fields
for
firewood
(
an
area
source,
when
burned).
Units
are
usually
expressed
in
terms
of
millions
of
cubic
feet
(
MCF).

EPA
Response:
EPA
will
take
this
under
consideration
during
the
development
of
the
nonroad
portion
of
the
MOVES
model.
However,
another
peer
reviewer
stated
that
the
TPO
database
may
have
significant
limitations.
22
3.2.2.6
Comment:
Oil
Field
Equipment
I
have
found
that
using
SIC
codes
1381,
1382,
and
1389
(
what
you
call
SIC
1300)
is
highly
misleading,
as
it
usually
identified
the
company
headquarters,
which
in
Texas
is
predominantly
based
in
Houston.
However,
little
oil
and
gas
extraction
actually
occurs
in
Houston
or
Harris
County.
Baker­
Hughes,
a
leading
information
center
for
the
industry,
has
a
database
of
oil
and
gas
drilling
sites
that
would
be
much
more
useful
in
allocating
this
kind
of
source.
Please
see:
http://
www.
bakerhughes.
com/
investor/
rig/
index.
htm
Note
that
these
statistics
are
available
only
for
entire
states
or
districts
within
states
(
as
is
the
case
with
Texas).
Therefore,
the
EPA
would
probably
have
to
purchase
special
county­
based
data
from
Baker
Hughes,
similar
to
working
with
the
Dodge
Reports
for
allocating
construction
equipment.
One
benefit
of
using
"
real"
rig
count
data
is
that
it
includes
large
rotary
rigs
and
smaller
work­
over
rigs,
which
might
be
helpful
in
allocating
drill
rig
HP
categories.

Equipment
other
than
drill
rigs
are
also
used
at
oil
and
gas
extraction
sites,
and
this
should
be
addressed
because
there
could
be
substantial
non­
road
emissions
other
than
just
the
platform,
tower
crane,
and
winch
engine
combination.
Mud
pumping,
well
fracturing,
and
service
pumps
seem
to
be
completely
ignored.

EPA
Response:
EPA
will
take
this
under
consideration
during
the
development
of
the
nonroad
portion
of
the
MOVES
model.

3.2.2.7
Comment:
Railway
Maintenance
I
have
not
been
able
to
find
a
source
of
data
preferable
to
use
of
human
census
population,
but
I
am
sure
that
the
Federal
DOT
and
the
American
Association
of
Railways
(
AAR)
know
track
mileage
and
gross
ton­
miles
by
state
and
county.
The
purpose
of
such
a
refinement
would
be
to
not
allocate
railway
maintenance
activity
to
counties
that
do
not
have
any
significant
track
mileage
or
locomotive
traffic.
For
example,
approximately
a
third
of
the
254
Texas
counties
do
not
have
any
active
track
mileage.
My
suspicion
is
that
railway
maintenance
would
be
proportional
to
the
amount
of
gross
ton­
miles
on
a
given
route
and
county
(
in
Texas,
it
is
though
that
only
about
40%
of
the
track
mileage
is
actually
used
on
a
daily
basis).
Since
all
Class
I
railways
must
report
their
GTM
data,
it
should
not
be
hard
to
allocate
railway
maintenance
in
a
more
logical
manner.

EPA
Response:
EPA
will
take
this
under
consideration
for
the
development
of
the
nonroad
portion
of
the
MOVES
model.
23
3.2.2.8
Comment:
A/
C
Reefers
Like
railway
maintenance,
this
category
was
proposed
to
be
allocated
by
human
census
population.
However,
there
is
significant
truck
traffic
(
trucks
having
refrigeration
units)
even
in
rural
counties,
such
as
along
major
U.
S.
highways.
Therefore,
the
Highway
Performance
Monitoring
System
(
HPMS)
would
be
a
preferable
candidate
for
allocating
this
source
type,
using
the
HDDV
vehicle
miles
of
travel
category
as
the
main
surrogate.

While
use
of
the
HPMS
would
be
a
definite
improvement,
there
still
could
be
some
refinements
such
as
by
using
employment
in
certain
food
services
(
cold
warehouses,
slaughterhouses,
grocery
distribution
centers,
import
export
clearing
houses,
etc.);
this
would
probably
require
a
hybrid
method
of
some
kind.

EPA
response:
EPA
will
take
this
under
consideration
for
the
development
of
the
nonroad
portion
of
the
MOVES
model.

3.2.3
Seasonal
Allocation
3.2.3.1
Comment:
The
method
to
develop
monthly
temporal
factors
appears
to
be
logical,
although
perhaps
it
would
help
to
explain
why
the
SEASON.
DAT
file
for
use
in
nationwide
runs
requires
annual
and
monthly
NONROAD
runs
for
each
of
the
65
types
of
equipment,
as
the
rationale
for
doing
so
is
not
obvious
in
the
text.

EPA
Response:
This
approach
is
necessary
due
to
the
different
seasonal
activity
allocations
and
different
geographic
allocations
of
the
various
types
of
equipment.
Any
set
of
equipment
that
has
the
same
temporal
and
geographic
allocations
can
be
handled
together,
which
is
why
there
are
only
65
different
types
of
equipment
for
this
purpose.
EPA
is
clarifying
this
in
the
next
revision
of
the
seasonal
allocation
technical
report.

3.2.4
Weekday/
Weekend
Day
Allocation
3.2.4.1
Comment:
This
paper
is
well­
researched,
and
the
effort
to
include
local
data
from
the
ARB
and
NCTCOG
and
others
is
truly
appreciated
(
recreational
marine,
residential
lawn
and
garden).
One
thing
that
was
obvious
in
the
California
studies
was
the
indication
of
a
potential
"
Friday
Effect"
that
could
be
explored
in
the
future.
In
many
of
the
travel
demand
models
for
highway
transportation,
Monday
through
Thursday
is
treated
as
an
average
weekday,
while
Friday
has
the
highest
peaks
in
vehicle
miles
of
travel.
Therefore,
for
certain
types
of
equipment,
especially
recreational
vehicles
and
vessels,
Friday
could
conceivably
be
treated
as
a
"
ramp­
up"
day
that
could
have
a
higher
allocation
than
the
weekdays.

A
final
recommendation
that
could
affect
both
seasonal
and
weekday
documentation
packages
would
be
special
events
such
as
Memorial
Day,
the
4th
of
July,
and
Labor
Day
vacation
periods,
which
probably
have
higher
residential
and
recreational
activities
than
the
average
weekday
would
indicate
in
the
model.
This
could
be
an
important
consideration,
even
though
it
24
would
be
difficult
to
implement,
because
many
of
the
ozone
and
particulate
matter
episodes
being
modeled
occur
on
these
kinds
of
vacation
periods.
For
example,
the
Dallas
attainment
SIP
was
based
on
the
4th
of
July;
the
Houston
attainment
SIP
has
a
September
episode
that
follows
Labor
Day.

EPA
Response:
EPA
is
aware
of
the
"
Friday
Effect"
and
that
holidays
falling
on
weekdays
other
than
Friday
could
be
more
similar
to
weekend
days.
EPA
is
currently
considering
how
to
handle
these
issues.

4.0
Review
of
Activity,
Load
Factors,
and
Median
Life
Values
4.1
Comment
from
Sam
Wells,
Consultant
4.1.1
Comment:
Several
years
ago
I
expressed
some
displeasure
with
the
logistic
curve
associated
with
the
median
useful
life
calculations
and
desired
that
EPA
would
allow
users
to
input
stated
by­
model­
year
distributions
 
if
local
data
were
available.
The
problem
is
that
if
one
modifies
the
median
useful
life
values,
one
would
corrupt
several
other
parts
of
the
model,
thereby
causing
a
"
ripple
effect"
on
engine
useful
life
and
other
calculations
in
the
model.

The
issue
is
related
to
load
factors
because
load
factor
is
a
key
ingredient
in
estimating
"
in­
use"
useful
life
 
after
the
value
has
been
entered
into
the
model.
Basically,
what
the
model
assumes
is
that
median
age
is
a
function
of
100
percent
load,
but
load
factor
fractions
are
applied
to
extend
the
"
in­
use"
useful
life
in
a
very
theoretical
and
largely
undocumented
manner.
As
a
crude
example,
an
engine
might
last
100
hours
at
full
load
but
because
it
operated
at
33
percent
load
factor,
it
would
last
three
times
longer
(
300
hours).
Perhaps
the
reader
could
benefit
from
a
discussion
of
how
the
following
information
sources
were
combined
to
obtain
average
load
factor
ratings:

Laboratory
testing
using
fully
loaded
versus
"
average"
loads
(
SwRI
method)

External
data
based
upon
maximum
fuel
consumption
and
average
fuel
consumption
(
PSR
method)

Manufacturer
load
factors
that
may
have
to
do
with
de­
rating
the
engines,
as
opposed
to
true
load
factor
statistics
EPA
Response:
Currently,
for
diesel
engines,
EPA
only
uses
laboratory
testing
data
to
calculate
load
factors.
PSR
data
is
not
longer
used,
and
manufacturer­
supplied
load
factors
have
never
been
used.

4.2
Comments
from
Rick
Baker,
Eastern
Research
Group
4.2.1
Comment:
What
procedure
did
EPA
use
to
determine
which
test
cycle
(
High,
Low,
25
Steady­
State)
"
most
closely
represents"
each
NONROAD
CI
equipment
category?
Were
any
inuse
data
used?
Was
it
based
on
engineering
judgement?
Were
any
third
party
experts
consulted?
On
a
related
issue,
it
is
not
clear
why
using
three
highly
aggregated
load
factor
categories
provides
an
improvement
in
accuracy
and/
or
precision
over
the
previous
data
derived
from
PSR.
Please
justify.

EPA
Response:
First,
equipment
types
were
designated
as
tending
more
toward
transient
versus
steady­
state
operation.
Then,
those
that
were
considered
transient
were
further
examined
and
assigned
to
one
of
the
seven
transient
test
cycles.
These
assignments
were
made
based
on
engineering
judgment
and
our
understanding
of
the
type
of
equipment
involved.
These
assignments
were
checked
by
an
independent
non­
EPA
peer
reviewer,
and
minor
adjustments
were
made.
We
believe
using
engine
test
results
based
on
various
operating
cycles
developed
specifically
for
nonroad
engines
is
preferable
to
PSR's
survey­
based
approach
for
estimating
load
factors.

4.2.2
Comment:
If
the
Regulatory
Useful
Life
estimates
reflect
the
effect
of
in­
use
loads,
and
the
New
NONROAD
Median
Life
estimates
represent
hours
at
full
load,
then
shouldn't
the
ratio
of
the
two
be
determined
by
the
model's
diesel
load
factors?
This
does
not
appear
to
be
the
case.
As
shown
in
the
footnote,
load
factors
ranging
from
0.21
to
0.59
would
require
either
much
lower
Median
Life
estimates
or
much
higher
Regulatory
Useful
Life
estimates.

EPA
Response:
These
two
measures
of
engine
life
serve
different
purposes
and
are
not
directly
comparable.
The
NONROAD
median
life
values
are
intended
to
represent
an
overall
average
expected
life
for
the
purpose
of
estimating
engine
replacement
timing
and
thus
in­
use
populations
from
initial
sales
estimates.
The
regulatory
useful
life
is
the
time
during
which
the
engine
must
meet
the
applicable
standard,
as
determined
in
the
rulemaking
notice
and
comment
process.

4.2.3
Comment:
Appendix
A
­­
There
are
several
SCCs
listed
with
0
hrs/
year,
but
non­
zero
load
factors.
These
include:

°
Concrete
pavers
(
all
fuel
types)
°
Logging
equipment,
Fellers/
Bunchers
(
all
fuel
types)
°
Diesel
ATVs
and
MCs
(
unused)

If
these
SCCs
are
no
longer
used,
they
should
either
be
removed
from
the
table
or
footnoted.

EPA
Response:
These
equipment
types
are
not
presently
used
but
have
been
left
in
the
activity
input
file
in
case
they
are
needed
again
in
the
future.

4.2.4
Comment:
An
extensive
off­
road
engine
cycle
development
effort
has
just
been
completed
for
telescoping
boom
excavators,
wheeled
loaders,
small
utility
engines,
and
single
and
tandem
axle
dump
trucks
for
the
Texas
Department
of
Transportation.
The
analysis
utilized
in­
26
use
data
loggers
for
engine
speed
and
load
estimates.
I
recommend
reviewing
this
report
as
an
independent
assessment
of
the
representativeness
of
the
cycle
categories
and
load
factors
used
for
CI
engines.
The
final
report
will
be
available
March
15,
2003.
Contact
Don
Lewis,
TXDOT,
(
512)
416­
2085.

Similarly,
for
an
independent
source
of
activity
data
for
CI
engines,
please
refer
to
Eastern
Research
Group,
"
Development
of
a
Revised
Emissions
Inventory
for
Construction
Equipment
in
the
Houston­
Galveston
Ozone
Non­
Attainment
Area",
April
20,
2000,
prepared
for
The
Houston­
Galveston
Area
Council
and
the
TNRCC
Area
and
Mobile
Source
Emissions
Assessment
Section.
Estimates
for
annual
hours
are
developed
from
high­
response
rate
surveys
of
diesel
construction
equipment
owners/
operators
in
the
Houston
area.

EPA
Response:
EPA
will
try
to
obtain
these
reports.
They
may
indeed
be
useful
as
a
check
on
the
load
factors,
activity,
and
cycles
used
in
the
model.
EPA
plans
to
incorporate
local/
regional
data
into
nonroad
emissions
modeling,
most
likely
as
part
of
the
development
of
the
MOVES
model,
EPA's
next
generation
emissions
modeling
system
for
mobile
sources.

4.3
Comments
from
Michael
Hutcheson,
URS
Corp.

4.3.1
Comment:
The
determination
of
median
life
for
diesel
engines
appears
to
be
flawed.
It
does
not
appear
that
the
EEA
analysis
of
PSR
data
was
used
in
a
meaningful
manner.
Instead
it
appears
that
the
existing
PSR
data
was
ignored
in
favor
of
a
more
intuitive
approach
which
lacks
hard
data.

A
review
of
the
EEA
analysis
in
Appendix
B
of
the
report
clearly
shows
that
PSR
data
indicates
that
the
average
life
for
all
diesel
engines
(
from
the
50
to
500
hp
range)
is
around
10,000
to
11,000
hours
at
rated
loads.
The
value
for
median
life
used
for
NONROAD
is
7,000
hours
for
engines
above
300
hp
and
4,667
hours
for
engines
from
50
to
300
hp.
If
the
PSR
data
is
to
be
ignored,
a
better
reasoning
needs
to
be
applied
than
EEA's
subjective
assertion
that
the
PSR
average
life
values
"
appeared
incorrect".

EEA
also
gives
no
real
factual
basis
or
reference
for
the
reduction
in
the
median
life
estimate
for
medium
duty
engines
and
this
reduction
has
been
carried
through
in
NONROAD.
This
is
particularly
troubling
since
this
hp
range
has
a
very
large
population
profile.

Additionally,
the
EEA
analysis
appears
to
place
a
large
value
on
the
difference
between
PSR
population
estimates
and
other
estimates
of
populations.
The
reviewer
does
not
place
great
confidence
in
any
estimate
of
population
short
of
extensive
regional
polling.
As
a
result,
the
reviewer
believes
EEA's
assertion
that
the
PSR
data
is
potentially
in
serious
error
is
presumptive
and
not
necessarily
based
on
fact.

EPA
Response:
PSR
determines
engine
life
at
rated
load
using
a
proprietary
formula
that
is
undocumented.
EEA
attempted
to
validate
the
PSR
life
estimates
by
comparing
the
PSR
27
population
estimates
with
other
population
estimates.
We
acknowledge
the
reviewer's
concern
regarding
the
confidence
in
existing
population
estimates,
but
the
comparison
with
a
variety
of
other
sources
does
suggest
that
PSR's
diesel
engine
life
estimates
generally
appear
to
be
too
high,
even
if
this
cannot
be
quantified
reliably.

Given
the
limitations
with
the
PSR
data,
EEA
developed
independent
estimates
based
on
a
combination
of
manufacturer
comments
and
available
data
from
on­
highway
engines.
As
such,
we
disagree
with
the
statement
that
the
EEA
approach
lacks
hard
data.

With
regard
to
the
reduction
in
the
median
life
estimate
for
medium
duty
engines,
manufacturers
generally
agreed
that
EPA's
original
durability
finding
that
medium
heavy­
duty
engines
have
twothirds
the
life
of
heavy­
heavy
duty
engines
was
a
reasonable
estimate.
We
continue
to
believe
that
the
useful
life
estimates
in
NONROAD
are
based
on
the
best
available
information.

4.3.1.
a
Comment:
EEA
asserts
that
the
problems
with
the
PSR
data
result
from
errors
in
engine
life
data
particularly
for
smaller
engines.
It
occurs
to
the
reader
that
the
flaw
in
both
analyses
of
the
PSR
data
results
from
an
incorrect
characterization
of
scrappage
instead
of
an
incorrect
characterization
of
median
life.
The
reviewer
believes
that
scrappage
occurs
not
only
as
a
result
of
a
given
engine
reaching
the
end
of
its
useful
life,
but
as
a
result
of
damage
to
equipment,
owner
movement
to
newer
models,
and
outright
abandonment.
If
median
life
for
scrappage
and
median
life
for
emissions
degradation
are
to
be
equated
as
is
done
in
NONROAD,
then
the
other
factors
affecting
scrappage
must
be
taken
into
account.

Adding
in
the
other
factors
affecting
scrappage
to
useful
life
inputs
increases
scrappage
rates
resulting
in
a
lower
median
life.
This
is
proper
when
calculating
growth
estimates
but
is
not
proper
when
using
median
life
to
estimate
emissions
degradation.
The
PSR
data
may
be
a
better
estimate
of
median
life
as
it
relates
to
emissions
than
it
does
growth.

As
a
result,
the
reviewer
believes
that
the
median
life
estimates
are
low
for
diesel
engines
especially
in
the
50
to
300
hp
range.
Even
though
the
regulatory
useful
life
values
represent
inuse
loads,
these
should
be
lower
bounds
for
median
life
because
regulatory
useful
life
should
guarantee
that
100
%
of
the
engines
maintain
the
emissions
standard.
Therefore,
median
life
is
logically
higher
than
regulatory
useful
life
in
emission
degradation
estimates.

EPA
Response:
We
concur
that
part
of
the
reason
for
PSR's
relatively
long
useful
life
estimates
for
smaller
engines
is
likely
due
to
considering
only
the
engine
life
without
considering
equipment
scrappage
due
to
non­
engine
causes.
Additionally,
given
the
proprietary
nature
of
PSR's
life
estimation
method,
there
may
be
other
issues
regarding
use
of
the
PSR
life
estimates
in
the
NONROAD
model.
Thus,
as
we
acquire
documented
and
reasonable
alternatives
to
the
PSR
values
we
reconsider
the
values
used
in
NONROAD.
The
concept
of
using
different
life
estimates
for
purposes
of
scrappage
versus
deterioration
may
have
some
technical
validity,
but
it
is
beyond
the
current
capabilities
of
the
model,
and
since
it
applies
mainly
to
the
smallest
diesel
engines,
its
effect
on
emissions
inventory
estimates
is
considered
too
small
to
justify
a
significant
redesign
of
28
the
model.

4.3.2
Comment:
The
reviewer
believes
EPA
has
appropriately
used
available
data
for
determination
of
load
factor
as
well
as
the
averaging
method
used.
The
use
of
load
factor
categories
also
appears
to
be
a
good
use
of
resources,
however,
the
application
of
those
categories
should
be
based
on
a
few
definitive
rules.
For
instance,
for
SCC
categories
which
include
an
all
or
general
category,
use
of
the
average
7­
cycle
load
factor
should
prevail.
A
thorough
review
of
Table
10
of
the
report
was
made
by
the
reviewer
and
the
following
changes
to
load
factor
assignments
are
suggested.

SCC
Current
Load
Factor
Proposed
Load
Factor
Reason
2270001000
Lo
lf
Avg.
7­
cycle
All
category
2270002027
Avg
7­
cycle
Lo­
lf
Low
power
requirements
similar
to
Arc
Welder
2270002033
Avg
7­
cycle
Lo
LF
Drill
rigs
have
substantial
down­
time
and
are
over
powered
for
difficult
applications
2270002054
Avg
7­
cycle
Hi
LF
High
Power
requirements
at
steady
state
2270002081
Hi
LF
Avg.
7­
cycle
General
category
2270004060
Avg
7­
cycle
Lo
LF
Substantial
down
time.

2270005035
hi
lf
avg.
7­
cycle
same
as
2270005050,
steady
state
use.

2270005055
hi
LF
Avg.
7­
cycle
General
category
Since
the
errors
in
Table
9
have
been
fixed,
no
other
comments
are
included.

EPA
Response:
The
EPA
response
for
each
of
these
equipment
types
is
listed
below.
In
considering
these
assignments
it
should
be
kept
in
mind
that
the
load
factor
category
assignment
(
high,
low,
or
steady­
state)
is
also
used
in
application
of
the
in­
use
transient
adjustment
factors
(
TAFs)
to
the
g/
hp­
hr
emission
factors.
2270001000
Recreational
Equipment:
The
only
diesel
equipment
in
this
category
is
specialty
vehicles/
carts.
Thus,
even
though
the
SCC
is
general,
the
actual
application
of
the
load
factor
is
more
specific.
2270002027
Signal
Boards:
These
tend
to
have
much
more
steady­
state
operation
than
welders,
which
have
very
transient
load
characteristics.
2270002054
Crushing/
Processing
Equipment:
Although
these
would
certainly
have
high
power
requirements,
it
does
not
change
the
more
steady­
state
operating
characteristics
expected
of
this
type
of
equipment.
2270002081
Other
Construction
Equipment:
Although
it
is
a
general
category,
the
types
of
equipment
in
this
category
tend
to
be
used
in
transient
cycle
operations.
And
since
most
other
29
specific
construction
equipment
types
have
a
high
LF
assignment,
it
is
considered
appropriate
to
also
apply
the
high
LF
to
this
category.
2270004060
Wood
Splitters:
No
longer
used
as
an
individual
equipment
type
in
the
model
(
see
Population
input
file).
Regarding
the
other
equipment
types
mentioned
in
the
comment:
2270002033
Bore/
Drill
Rigs:
We
concur
with
the
recommendation.
This
could
be
considered
similar
to
the
transient
load
operating
characteristics
of
welding
equipment.
2270005035
Sprayers:
We
concur
with
the
recommendation.
2270005055
Other
Agricultural
Equipment:
We
concur
with
the
recommendation.
EPA
will
address
these
modifications
in
future
revisions
of
the
technical
reports
dealing
with
load
factors
and
diesel
transient
adjustment
factors.

4.4
Comments
from
Dr.
Nigel
Clark,
University
of
West
Virginia
4.4.1
Comment:
The
emissions
equation
presented
on
p.
1
represents
a
well­
established
approach.
Below
the
equation
"
Average
Power"
should
be
changed
to
"
Average
Rated
Power"
to
avoid
confusion.
The
authors
may
wish
to
consider
adding
a
single
sentence
to
describe
what
is
meant
by
"
average
rated
power"
to
avoid
confusion,
because
equation
1
may
be
mis­
applied
if
taken
out
of
context.

EPA
Response:
We
will
make
this
change
during
the
next
revision
of
NR­
005b.

4.4.2
Comment:
On
p.
2,
the
equation
to
determine
median
lifetime
is
reasonable
provided
that
the
equipment
is
used
in
the
original
intended
fashion.
The
following
comments
are
an
observation,
and
should
not
be
considered
a
recommended
revision
to
this
version
of
the
modeling
approach.
One
assumes
that
engines
are
not
generally
oversized
for
the
equipment
that
they
power,
and
that
load
factors
will
therefore
be
a
reasonably
large
fraction.
Difficulties
arise
when
an
engine
is
used
at
far
below
its
rated
capacity,
because
the
load
factor
may
be
very
low,
and
the
median
lifetime
implied
by
the
p.
2
equation
may
therefore
be
high.
As
an
example,
a
generator
set
that
sees
high
occasional
peak
loads
(
and
is
therefore
specified
to
be
large
in
power
rating)
but
usually
sees
very
light
loads,
would
have
a
very
low
load
factor.
However,
the
engine
would
suffer
wear
that
is
disproportionately
high
in
terms
of
the
load­
based
median
life
equation,
because
wear
still
occurs
at
light
load.
I
recognize
that
the
median
life
equation
is
not
applied
to
individual
pieces
of
equipment,
but
it
can
be
argued
that
some
types
of
equipment
that
run
for
long
hours
at
light
loads
may
have
a
shorter
actual
life
than
the
median
lifetime
equation
implies.
This
is
because
wear
is
partly
a
product
of
"
cumulative
revolutions
turned"
(
which
may
be
represented
reasonably
by
cumulative
hours
of
operation)
and
partly
a
product
of
"
cumulative
energy
produced"
(
which
is
the
concept
behind
the
p.
2
equation).
The
EEA
report
(
attached
to
my
copy
of
NR005b)
agrees
with
this
concept
when
it
reports
"[
The
manufacturers]
agreed
that
it
(
the
load
factor)
would
overestimate
life
at
very
light
load
factors
and
underestimate
life
as
the
load
factor
approached
one."
I
have
discussed
possible
future
remedies
for
this
issue
towards
the
end
of
this
report.
30
EPA
Response:
EPA
will
take
this
into
consideration
for
the
development
of
the
nonroad
portion
of
the
MOVES
model.

4.4.3
Comment:
On
p.
2,
the
report
states
that
"
larger
engines
last
longer
than
smaller
ones."
The
reader
will
not
know
what
"
larger"
means
in
this
case.
The
text
should
be
more
specific
and
use
a
tangible
term,
such
as
"
rated
power,"
"
displacement,"
or
"
displacement
per
cylinder."
Perhaps
the
strongest
indicator
of
engine
design
longevity
is
the
rated
speed
(
which
is
closely
related
to
displacement
per
cylinder.)
I
do
believe
that
engines
of
larger
displacement
per
cylinder
last
longer,
but
this
is
not
synonymous
with
the
statement
in
the
report.
On
a
"
grand
scale"
displacement
per
cylinder
rises
in
sympathy
with
engine
power
rating,
but
these
two
measures
are
not
uniquely
related.

°
Consider
that
a
6
cylinder
DDC
6V92
may
be
rated
at
around
300hp,
and
a
12
cylinder
12V92
may
be
rated
at
twice
that
value.
There
is
no
reason
to
believe
that
their
lifetimes
would
differ,
although
the
statement
would
imply
that
the
12V92
might
last
longer.

°
In
the
same
vein,
a
"
highly
boosted"
1800
rpm
two­
liter­
per­
cylinder
engine
may
be
rated
the
same
as
a
conservatively
turbocharged
1200
rpm
four­
liter­
per­
cylinder
engine.
I
believe
the
larger
displacement
engine
would
prevail
in
lifetime.
There
may
even
be
some
concern
that
a
high
power
version
of
an
engine
with
the
same
block
and
hardware
may
have
a
slightly
shorter
life
(
an
issue
discussed
in
the
EEA
attachment
to
NR006b.)

°
To
some
extent
the
concept
that
more
powerful
engines
last
longer
flies
in
the
face
of
the
median
lifetime
equation.
Consider
two
engines
with
identical
hardware
(
including
pistons,
liners,
valves
and
bearings),
but
different
fueling
calibrations,
so
that
one
engine
is
rated
at
300hp
and
the
other
is
rated
at
400hp:
this
is
not
uncommon.
If
they
are
both
used
in
a
100
hp
(
average
load)
application,
the
load
factor
for
one
will
be
0.33
and
for
the
other
will
be
0.25.
The
median
life
equation
would
suggest
that
the
400hp
version
would
last
longer,
which
clearly
it
would
not.

There
is
no
immediate
modeling
remedy
for
this
dilemma,
but
in
the
report
the
"
Engine
life
varies ."
sentence
(
p.
2)
might
be
altered
or
softened.
Perhaps
the
authors
could
add
that
it
is
recognized
that
"
within
a
power
class
there
can
still
be
variation
in
lifetime,
although
this
level
of
detail
cannot
be
captured
without
substantial
additional
data."
It
is
evident
that
the
PSR
lifetimepower
estimates
(
as
reported
by
EEA)
were
philosophically
directly
opposed
to
the
lifetimepower
relationship
expressed
in
NR
005b,
and
the
confusing
nature
of
the
issues
that
I
raise
above
simply
highlight
the
disagreement.
Of
course,
of
the
two,
the
NR005b
approach
(
that
lifetime
increases
with
power
rating)
is
far
more
correct,
although
it
is
approximate.

EPA
Response:
EPA
appreciates
the
peer
reviewer
raising
this
issue
and
will
clarify
the
discussion
of
engine
life
in
the
technical
report.

4.4.4
Comment:
On
p.
2,
it
seems
reasonable
to
adjust
the
ARB
horsepower
ranges
slightly
to
31
suit
NONROAD.
Since
the
cutpoints
in
this
case
are
being
used
for
inventory
rather
than
enforcement,
no
entity
is
immediately
injured
by
this
adjustment.
Moreover,
two
of
these
changes
for
diesel
produce
no
change
in
median
lifetime,
they
are
a
moot
point
for
two­
stroke,
and
they
all
cause
no
change
at
all
for
four­
stroke
and
CNG/
LPG
lifetime.

EPA
Response:
No
response
is
necessary.

4.4.5
Comment:
An
on­
highway
tractor­
trailer
engine
might
have
a
life
of
750,000
miles
before
a
rebuild,
and
could
accumulate
these
miles
at
60mph
cruise,
which
would
represent
about
50%
of
rated
power
(
e.
g.
200hp.
at
60mph
on
flat
land
out
of
400hp.
rated.)
This
would
be
12,500
hours
at
50%
load,
implying
6,250
hours
of
life
at
full
load.
This
quick
estimation
supports
the
7,000
hour
full­
load
lifetime
adopted
in
this
EPA
report.

EPA
Response:
No
response
is
necessary.

4.4.6
Comment:
On
p.
4,
it
is
mentioned
that
off­
highway
engines
are
generally
derated
from
their
on­
road
counterparts.
They
may
be
derated
with
respect
to
power
for
the
same
displacement
in
comparison
to
an
on­
road
engine.
However,
in
many
cases
on­
road
engines
"
lead"
in
the
horsepower
race
due
to
the
need
for
light
engines
at
high
power
ratings
in
truck
applications.
The
off­
highway
version
may
have
older
technology
and
be
reflective
of
the
on­
road
engines
of
a
few
model
years
earlier.
In
this
way
some
engines
may
not
actually
be
derated,
but
have
less
sophisticated
hardware
components
and
hence
a
lower
rating.
This
observation
does
not
warrant
any
change
to
the
report.

An
exception
to
the
"
derating"
of
off­
highway
engines
is
in
marine
applications,
where
power
ratings
are
very
high.
Assured
cooling
allows
these
ratings.

EPA
Response:
No
response
is
necessary.

4.4.7
Comment:
On
p.
7
it
should
be
made
clear
how
engine
idle
is
included
in
the
load
factor
(
and
in
the
activity).
Since
the
p.
2
median
life
equation
is
based
solely
on
engine
energy,
it
does
not
really
matter
if
(
a)
activity
means
"
key
on",
and
load
factors
are
low
due
to
long
idling
times,
or
(
b)
whether
activity
does
not
include
long
idling
periods
and
load
factors
are
higher.
However,
there
is
real
concern
that
if
a
load
factor
and
an
activity
number
(
hours/
year)
are
acquired
from
two
different
sources,
and
are
inconsistent
in
definition
with
regard
to
idle,
that
the
results
will
be
in
error.
This
is
more
a
concern
for
a
user
choosing
to
use
other
data
than
for
the
default
values
in
NR005b,
because
both
values
in
most
cases
came
from
the
same
source.

EPA
Response:
This
is
correct.
The
load
factors
are
intended
to
take
into
account
the
virtually
zero­
load
operation
when
an
engine
is
idling.
And
"
activity"
refers
to
all
times
that
the
engine
is
running
at
any
load,
as
it
would
show
up
on
an
engine
hour
meter.
Thus
it
would
be
important
for
any
user­
substituted
model
inputs
to
be
consistent
with
this
approach.
32
4.4.8
Comment:
On
pp.
7
and
8
the
use
of
fuel
consumption
data
to
measure
load
factors
is
mentioned.
No
change
to
the
report
is
recommended,
although
the
following
observations
are
made.
Although
no
detail
is
given
in
the
report,
the
use
of
fuel
consumption
to
determine
the
load
factor
may
prove
approximate,
particularly
for
throttled
engines.
If
an
engine
is
rated
at,
say,
2,400rpm,
but
is
used
at
full
torque
at
1,600
rpm,
it
will
use
less
fuel
than
an
engine
at
2,400rpm
at
part
load.
The
Pechan
report
(
ref
.8)
cites
Duleep
in
observing
that
the
ratio
of
fuel
consumed
may
overestimate
the
load
factor.
This
is
true.
There
are
other
confounding
factors,
such
as
full
load
enrichment
and
enrichment
during
transients
for
gasoline
engines,
and
the
considerable
disparity
in
fuel
consumption
at
part
load
between
gasoline
and
diesel
engines.
However,
it
may
be
more
accurate
to
estimate
engine
life
based
on
the
total
fuel
consumed
by
an
engine
over
its
life,
so
that
this
concern
may
prove
to
be
a
blessing!

EPA
Response:
Agreed.
No
response
is
necessary.

4.4.9
Comment:
The
CI
equipment
load
factors
represent
the
most
mature
and
reliable
data
in
the
report.
Clearly
there
is
still
a
concern
that
only
seven
representative
cycles
(
for
seven
equipment
types)
exist,
and
that
other
equipment
must
be
assigned
to
a
"
similar"
cycle,
but
this
is
still
an
advance.

EPA
Response:
Agreed.
No
response
is
necessary.

4.4.10
Comment:
It
is
important
that
each
equipment
type
assigned
to
"
low"
or
"
high"
falls
into
the
same
"
bin"
that
it
would
for
NONROAD
transient
adjustment
factors
for
purposes
of
internal
consistency,
and
I
assume
that
Table
10
has
the
same
assignments
for
load
factors
as
for
transient
adjustment
factors.
The
present
associated
match
of
equipment
to
cycle
for
transient
adjustment
factors
does
show
very
reasonable
assignments.
Table
10
is
a
showpiece
for
demonstrating
the
need
for
more
usage
data
for
off­
road
equipment:
the
division
of
CI
equipment
into
three
load
factor
assignments
is
far
too
approximate,
but
this
cannot
be
improved
without
greater
knowledge.
It
would
have
been
possible
to
use
alternative
approaches,
namely
either
°
to
assign
at
least
the
seven
frequently
encountered
pieces
of
equipment
for
which
there
were
original
SwRI
transient
cycles
the
exact
load
factor
implied
in
the
associated
cycle,
or
°
to
assign
each
piece
of
equipment
the
same
load
factor
as
for
the
cycle
chosen
to
represent
its
transient
behavior
best.

However,
Occam's
razor
("
one
should
not
increase,
beyond
what
is
necessary,
the
number
of
entities
required
to
explain
anything")
is
probably
the
greatest
friend
of
this
005b
report
and
suggests
that
the
meager
information
available
should
not
be
"
over­
editorialized."
The
three
bins
should
stand
for
now,
but
should
be
reconsidered
as
more
data
become
available.

It
interesting
to
note
inconsistencies
that
arise
due
to
changing
methodologies.
CI
leafblowers
(
infrequently
encountered!)
have
a
load
factor
of
0.43
(
from
Table
10)
whereas
the
SI
version
33
would
have
a
load
factor
of
0.94
(
Table
6).
This
is
true
for
several
classes
in
Table
10,
but
it
cannot
be
helped
until
a
more
consistent
field
data
set
is
available.
Of
course,
some
of
these
classes
in
Table
10
do
not
exist
in
practice
anyway.

EPA
Response:
Agreed.
No
response
is
necessary.

4.4.11
Comment:
There
is
one
entry
in
Table
10
that
may
raise
concern,
but
this
paragraph
presents
an
observation
for
future
consideration
rather
than
a
recommended
correction.
Sailboat
auxiliary
CI
engines
(
a
common
class)
have
been
assigned
a
load
factor
from
the
7­
cycle
composite
because
there
is
no
other
provision
in
the
methodology
that
is
reasonable.
However,
the
marine
engine
assignments
for
SI
were
0.21.
This
inconsistency
could
be
remedied
only
by
relaxing
the
CI
methodology.
In
reality,
sailboat
inboard
engines
are
often
modest
in
output
and
are
probably
run
at
fairly
high
load
factors,
but
the
same
issue
also
arises
for
recreational
pleasure
craft.
Larger
recreational
vessels
may
be
diesel­
powered
and
are
not
likely
to
deviate
much
in
load
factor
from
SI
counterparts.

EPA
Response:
Agreed.
EPA
will
keep
this
in
mind
for
the
development
of
the
nonroad
portion
of
the
MOVES
model.

4.4.12
Comment:
The
load
factors
in
Table
10
will
be
used
to
determine
engine
life.
In
this
case
they
are
not
fuel
consumption
based,
but
are
assignments
to
load
factors
derived
from
real
equipment
operation
(
even
if
coarsely
binned.)
The
concern
that
engine
life
is
a
product
of
both
operating
hours
and
cumulative
energy
output
arises
again.
A
detailed
model
of
engine
life
would
be
a
substantial
undertaking,
but
some
simple
remedies
are
possible.
One
possibility
is
to
propose
a
"
life
factor"
for
purposes
of
calculating
engine
life.
This
factor
could,
for
example,
have
the
simple
linear
form:

Life
factor
=
a*(
load
factor)
+
b
Where
an
assignment
of
about
0.1
for
"
b"
and
0.8
for
"
a"
could
be
used.
In
this
way
engines
at
full
load
would
live
a
little
longer
than
is
now
projected,
and
engines
at
very
light
loads
would
live
a
little
shorter
than
now
projected.
In
fact,
"
a"
and
"
b"
could
be
selected,
with
justification,
such
that
over
the
population
of
engines
the
grand
average
of
engine
age
remained
the
same.
This
approach
is
justifiable
in
that
an
idling
engine
(
or
worse,
a
high­
speed
idling
engine,
as
in
a
generator
application)
is
performing
indicated
work
at
the
pistons
and
is
suffering
wear
from
inertial
forces.
Indicated
power
at
idle
is
typically
five
to
ten
percent
of
indicated
rated
power,
which
leads
to
a
suggestion
that
"
b"
might
be
in
the
region
of
0.1.

When
the
load
factor
is
calculated
from
fuel
consumption,
it
is
probably
philosophically
quite
close
to
the
"
life
factor"
suggested
above.

EPA
Response:
EPA
will
consider
this
approach
for
the
development
of
the
nonroad
portion
of
the
MOVES
model.
34
4.4.13
Comment:
In
the
future,
it
will
be
important
to
watch
technology
whereby
generator
sets
need
not
be
constant
speed
devices,
but
may
be
operated
at
variable
speed
and
high
torque,
with
the
electrical
output
waveform
crafted
by
high
current
semiconductor
devices.
Generators
of
this
kind
will
differ
in
efficiency
and
operation
from
the
generators
in
use
today.

EPA
Response:
EPA
appreciates
the
reviewer
identifying
this
as
an
issue
to
monitor
in
the
future.

5.0
Equipment
Population
5.1
Comment
from
Sam
Wells,
Consultant
5.1.1
Comment:
It
is
not
clear
why
the
1998
PSR
data
was
not
used;
instead,
the
1996
data
was
used
in
conjunction
with
the
scrappage/
median
life
function
"
in
order
to
maintain
consistency."
This
would
indicate
that
the
1996
and
1998
PSR
estimates
for
CI
engines
were
not
consistent,
and
that
the
EPA
was
attempting
to
smooth
data
that
did
not
fit
their
mathematical
curves
(
e.
g.,
the
"
dog
leg"
in
MOBILE5
that
did
not
fit
a
linear
or
nonlinear
function).
Conversely,
one
could
also
surmise
that
the
1998
PSR
data
was
somehow
faulty.
In
my
opinion,
new
data
should
be
used
if
the
survey
design
is
valid,
since
as
a
contractor
I
have
had
to
revise
many
of
the
CI
population
estimates,
in
some
cases
being
twice
or
half
what
we
found
(
that
is,
in
a
given
urban
area,
not
on
a
statewide
basis).
Secondly,
the
on­
road
MOBILE
models
are
usually
revised
every
three
years
or
so,
resulting
in
emission
rate
changes
of
up
to
50
percent,
so
I
do
not
understand
the
"
consistency"
issue.

EPA
Response:
EPA
has
updated
the
population
input
data
for
CI
equipment
in
NONROAD2004
using
PSR's
data
for
2000,
the
most
recent
available.

5.2
Comment
from
Rick
Baker,
Eastern
Research
Group
5.2.1
Comment:
The
USDA
Agricultural
Census
may
be
a
better
source
of
data
for
certain
agricultural
equipment,
including
agricultural
tractors
and
combines.
The
data
is
available
for
1997
at
the
county
level,
and
is
based
on
comprehensive
bottom­
up
survey
methods.
While
not
fully
disaggregated,
information
is
available
on
age
and
hp
bins.
At
the
least,
this
information
should
be
used
as
a
validation
of
the
current
PSR
data.
(
Note
that
a
preliminary
comparison
of
national
totals
from
Appendix
A
and
the
USDA
Census
indicate
large
discrepancies
(
e.
g.,
NONROAD
combines
at
206,517
versus
USDA
census
at
460,606
 
although
admittedly
the
base
year
is
off
by
one
year.)
If
this
information
was
intentionally
not
used,
please
indicate
why.

EPA
Response:
In
1999,
as
part
of
an
effort
to
reconcile
NONROAD's
diesel
fuel
consumption
estimates
to
those
provided
by
the
Department
of
Energy's
(
DOE)
Energy
Information
Administration
(
EIA),
EPA
compared
PSR's
1997
agricultural
equipment
populations
to
those
found
in
the
1997
USDA
Census
of
Agriculture.

As
the
reviewer
mentions,
the
USDA
National
Agricultural
Statistics
Service
(
NASS)
conducts
35
periodic
surveys
of
farm
equipment
and
agricultural
fuel
consumption.
The
Census
of
Agriculture
is
conducted
every
five
years.
Results
from
the
1997
Census
are
available,
and
the
2002
Census
is
underway.
The
Census
requests
selected
respondents
to
report
information
on
"
Equipment
on
Place"
and
expenditures
for
diesel
fuel
"
purchased
for
the
farm
business."
The
equipment
population
data
is
of
good
quality,
but
limited
in
that
results
do
not
distinguish
between
gasoline
and
diesel
equipment.
Equipment
is
distinguished
by
size,
but
the
size
classes
are
very
coarse
relative
to
those
used
by
EPA
to
define
diesel
emission
standards.
Equipment
age
is
only
addressed
by
dividing
it
into
two
ranges:
0­
4
years
and
greater
than
4
years
old.
These
three
limitations
preclude
the
construction
of
age
distributions
and
apportionment
of
diesel
equipment
by
regulatory
tier,
which
is
necessary
in
the
estimation
and
projection
of
equipment
inventories
at
the
level
of
detail
required
in
the
rulemaking
analyses.

Differences
between
equipment
populations
as
reported
in
the
Census
of
Agriculture
and
as
estimated
using
data
provided
by
PSR
are
difficult
to
resolve
directly,
and
are
further
complicated
by
questions
of
scope.
The
Census
covers
farm
equipment
used
in
within
the
agricultural
sector
only;
it
does
not
cover
"
farm"
equipment
(
such
as
tractors),
purchased
by
organizations
in
other
sectors,
nor
does
it
cover
equipment
of
other
"
non­
farm"
types,
purchased
and
used
in
agriculture.
In
contrast,
the
populations
used
in
NONROAD
represent
"
entire"
populations,
across
all
economic
sectors.

Thus,
the
fact
that
Census
populations
exceed
NONROAD's
is
difficult
to
explain.
One
possibility
is
that
high
proportions
of
equipment
in
the
populations
reported
by
the
Census
represent
equipment
that
has
very
low
usage
rates,
whereas
the
activity
rates
used
in
NONROAD
might
be
more
representative
of
more
"
active"
proportions
of
equipment
populations.
In
any
case,
the
tenor
of
most
comments
concerning
NONROAD's
fuel
consumption
and
emissions
estimates
is
that
they
appear
to
be
overestimates
relative
to
corresponding
estimates
derived
from
data
published
by
the
EIA
for
all
sectors,
including
agriculture.
In
addition,
an
estimate
of
diesel
fuel
consumption
in
agriculture
derivable
from
NASS's
estimates
of
fuel
expenditures
and
prices
is
in
agreement
with
a
corresponding
estimate
derived
from
DOE
sources.
Thus,
resolving
the
differences
between
the
Census
of
Agriculture
and
NONROAD
appears
to
require
improved
understanding
of
the
appropriate
relationships
between
equipment
populations
and
corresponding
usage
rates
in
the
agricultural
sector.
At
the
present
time
however,
combining
equipment
populations
from
the
Census
of
Agriculture
with
usage
rates
estimated
by
PSR
(
NASS
does
not
report
usage
rates),
would
increase
the
difference
between
NONROAD's
fuel
consumption
and
emissions
estimates
and
the
information
from
the
EIA
that
we
used
as
a
benchmark
in
our
analysis.

5.3
Comments
from
Dr.
Nigel
Clark,
University
of
West
Virginia
5.3.1
Comment:
The
palette­
mounted
engine
is
discussed
as
an
example
in
the
"
stationary
vs.
mobile"
debate.
Conversely
it
would
also
be
wise
to
comment
illustratively
on
a
piece
of
equipment,
such
as
a
front­
end
loader
or
forklift,
which
moves,
but
is
confined
for
many
years
to
a
very
small
radius
of
operation
in
a
yard
or
building:
it
is
still
mobile.
36
The
stationary/
mobile
split
is
not
critical,
so
long
as
the
stationary
and
mobile
inventories
do
not
double
count
or
miss
equipment.
Is
there
any
additional
guarantee
that
this
will
not
happen?

EPA
Response:
EPA
is
aware
of
the
potential
for
double­
counting
mobile
source
emissions.
A
risk
does
exist
that
nonroad
equipment
may
be
double­
counted
if
a
stationary
source/
facility
submits
an
emissions
inventory
to
their
state
that
includes
nonroad
equipment
used
there.
However,
EPA
assigns
all
emission
sources
a
unique
source
category
code
(
SCC).
Given
that
a
stationary
source/
facility
inventory
would
detail
all
of
the
SCCs
included
in
its
total
inventory,
a
state
or
other
downstream
user
of
the
data
would
most
likely
be
able
to
detect
and
remove
any
nonroad
equipment
emissions
that
were
included
in
a
stationary
source
total
if
the
potential
to
double­
count
existed.
There
may
also
be
purposes,
such
as
for
an
environmental
impact
statement,
where
it
might
be
necessary
to
include
emissions
for
nonroad
equipment
(
as
well
as
for
on­
highway
vehicles)
used
on
the
sight
with
the
stationary
source's
emissions.

5.3.2
Comment:
On
p.
8
the
forklift
population
is
discussed,
and
this
raises
a
specific
issue.
Equipment
such
as
forklifts
and
backhoes
is
assigned,
elsewhere
in
the
model
(
see
NR005b),
an
annual
activity
(
as
hours/
year
of
operation),
and
this
activity
does
not
vary
with
equipment
age.
Scrappage
models
usually
employ
a
retirement
curve
that
suggests
that
most
equipment
is
retired
close
to
its
average
engine
life
(
based
on
total
life
hours
and
load
factor),
and
that
all
of
the
equipment
is
scrapped
by
the
time
twice
the
average
life
is
reached.
In
reality,
many
pieces
of
equipment
will
linger
on
in
a
reduced
activity
mode
as
they
become
older,
so
that
they
last
many
more
years
(
but
not
more
total
life
hours
of
operation)
than
a
scrappage
model
might
suggest
[
Ref.
2
of
NR006b
,
the
Pechan
1998
report,
supports
this
argument].
Even
though
the
scrappage
model,
in
using
a
fixed
number
of
hours
for
annual
activity,
will
deny
the
existence
of
very
old
equipment,
that
equipment
is
usually
not
very
active
in
reality,
and
the
ultimate
emissions
inventory
may
not
be
in
great
error.
Alternatively,
it
would
be
reasonable
to
consider
an
actual
audited
population
of
(
say)
forklifts,
but
then
older
forklifts
in
this
population
would
need
to
be
associated
with
reduced
activity
in
computing
emissions
production.
A
danger
arises
if
the
actual
population
is
used,
and
is
then
combined
with
annual
activity
(
hours
per
year)
that
is
not
indicative
of
the
operation
of
older
equipment.
In
this
way,
the
inventory
could
be
overestimated.
Nonroad
models
would
be
most
effective
if
the
scrappage
model
considered
declining
annual
activity.
In
this
way
the
scrappage
model
would
more
closely
reflect
the
true
population
curve
(
with
a
"
long
tail"
to
the
curve
for
old
age
equipment)
which
could
then
be
combined
with
activity
that
reduces
with
age
in
reaching
the
inventory.
The
purpose
of
this
discussion
is
twofold.
First,
I
suggest
that
in
future
the
use
of
activity
that
declines
with
age
(
to
feed
scrappage
models
and
compute
the
inventory)
might
be
employed.
Second,
I
observe
that
one
needs
to
be
cautious
about
taking
the
product
of
activity
and
population
if
the
two
factors
are
found
in
different
ways.
It
would
be
valuable,
in
this
NR
006b
report,
to
illustrate
to
the
reader
how
much
the
population
of
forklifts
would
change
if
a
scrappage
model
were
used.
The
difference
between
populations
determined
using
different
(
but
credible)
methods
illustrates
to
some
extent
the
uncertainty
in
the
model.

EPA
Response:
EPA
would
like
to
incorporate
the
relationship
between
equipment
age
and
activity.
EPA
has
looked
for
data
reflecting
this
relationship.
Anecdotal
information
suggests
37
that
activity
does
decline
as
equipment
ages.
However,
EPA
has
only
been
able
to
find
data
on
a
handful
of
equipment
types,
which
may
not
extrapolate
well
to
other
types
of
equipment
(
e.
g.
construction
versus
recreational
equipment).
Although
this
comment
was
brought
up
in
reference
to
the
population
technical
report,
EPA
will
consider
incorporating
a
discussion
of
the
effect
of
equipment
age
on
activity
in
the
NR­
005b
(
activity,
median
life,
and
load
factor)
technical
report,
where
it
is
more
appropriate.

5.3.3
Comment:
On
p.
8,
there
are
491,321
forklifts.
The
report
explains
that
approximately
80%
of
these
are
propane,
so
that
approximately
10%
are
diesel
and
approximately
10%
are
gasoline.
So
the
LNG
percentage
of
SI
engines
should
reflect
the
fraction
80/(
80
+
10),
which
is
89%.
Table
4
shows
95%.
I
assume
that
this
disagreement
is
not
due
to
any
CNG
forklift
assignment.

EPA
Response:
Actually,
as
shown
in
the
population
input
files,
some
of
the
SI
forklifts
are
assigned
to
CNG.
According
to
the
PSR
estimates,
roughly
ten
percent
of
all
SI
forklifts
were
CNG
fueled,
and
they
were
all
in
the
40­
50
hp
range.
EPA
is
clarifying
this
in
the
next
revision
of
the
equipment
population
technical
report.

5.3.4
Comment:
I
follow
the
logic
of
preparing
Table
4,
but
it
is
clear
that
more
field
or
sales
data
are
really
needed
to
be
equitable.
One
example
is
the
assignment
of
100%
LPG/
CNG
to
generators
but
only
50%
to
compressors,
welders
and
crushing/
processing
equipment.
(
If
generators
are
fueled
by
pipeline
natural
gas,
then
they
must
surely
be
stationary,
so
I
assume
that
most
of
these
LPG/
CNG
generators
are
intended
to
be
LPG.)
One
assumes
that
there
are
some
SI
gasoline
portable
generators
in
use,
so
that
the
100%
assignment
to
LPG/
CNG
is
probably
high.
Also,
there
is
some
similarity
between
generators
and
compressors
(
as
power
sources)
in
site
use,
so
that
I
am
concerned
about
the
strong
contrast
between
the
100%
and
50%
LPG/
CNG
assignments
for
those
two
equipment
types.
However,
this
cannot
be
argued
further
without
new
data
in
the
future,
and
the
006b
report
should
not
be
altered
at
present.

EPA
Response:
We
agree
with
these
observations
and
the
desirability
of
more
data.
Although
it
was
not
explicitly
stated
in
the
population
technical
report
or
in
the
rulemaking
document
from
which
this
information
was
taken
(
Final
Regulatory
Support
Document
for
the
Final
Rule
published
November
8,
2002,
EPA420­
R­
02­
022),
the
use
of
100%
LPG/
CNG
for
large
SI
generators
was
based
on
recent
sales
and
other
market
data.
Due
to
the
lack
of
such
data
for
compressors
and
other
large
SI
equipment,
the
50%
default
value
was
used
for
these
equipment
types.

5.3.5
Comment:
On
p.
11
the
language
suggests
that
there
is
population
dating
back
to
1948.
Surely
the
scrappage
model
would
exclude
such
equipment,
unless
the
load
factors
or
annual
activity
was
extremely
low.
If
there
is
no
population
that
is
55
years
old
using
this
scrappage
method,
then
the
report
should
give
an
indication
that
these
older
years
have
no
population.
I
realize
that
quirks
of
marine
engine
use
did
cause
old
population
overestimation
in
the
marine
case,
but
I
do
not
believe
that
this
was
the
case
in
construction.
38
EPA
Response:
As
suggested
in
the
comment,
there
are
certain
equipment
types
for
which
the
combination
of
load
factor
and
annual
activity
are
very
low,
which
results
in
the
calculation
of
a
small
number
of
them
still
in
use
up
to
50
years,
which
is
the
maximum
age
that
the
model
can
handle.
The
report
language
will
be
updated
to
reflect
the
use
of
year
2000
instead
of
1998
as
the
point
from
which
50
years
are
subtracted.

5.3.6
Comment:
Assignments
for
sailboat
engines
are
reasonable.
I
would
reword
the
sentence
on
p.
12
to
emphasize
that
the
sailboat
engines
resemble
outboard
engines
only
in
horsepower
ratings:
the
present
wording
could
be
misinterpreted.

EPA
Response:
EPA
will
clarify
this.

5.3.7
Comment:
This
report
does
not
consider
the
rebuilding
of
engines
as
a
mechanism
for
preserving
equipment
life.
In
real
use,
some
engines
(
e.
g.,
lawnmower
engines)
are
abandoned
when
the
equipment
is
scrapped,
while
some
engines
enjoy
a
longevity
similar
to
the
longevity
of
the
technology
that
they
power,
and
some
are
outlasted
by
the
equipment.
In
on­
road
use
the
truck
population
is
swelled
by
vehicles
with
rebuilt
engines,
and
in
off­
road
use
the
expense
of
replacing
large
pieces
of
equipment
will
cause
engines
to
be
rebuilt.
Therefore,
in
the
higher
horsepower
categories,
population
estimation
from
scrappage
might
consider
rebuilds
in
the
future.
This
could
be
achieved
simply
by
estimating
a
rebuild
fraction,
and
assigning
a
second
median
life
(
somewhat
less
than
for
a
new
engine)
to
the
rebuilds.
I
do
realize
that
little
information
is
available
to
account
for
rebuilds
and
that
the
term
"
rebuild"
is
poorly
defined.
Engines
may
be
partly
rebuilt
(
just
bearings,
or
just
heads)
to
extend
their
lives
and
the
terms
"
rebuild",
"
repair"
and
"
maintenance"
may
merge
seamlessly.
Ultimately,
the
rebuild
and
repair
activity
may
just
serve
to
cause
an
extended
effective
average
engine
life,
applicable
mainly
to
large
bore,
expensive
engines.
No
improvement
to
the
NR
006b
approach
is
possible
without
acquiring
more
real
world
data.

EPA
Response:
This
is
a
good
suggestion.
EPA
will
address
this
during
the
development
of
the
nonroad
portion
of
the
next
generation
MOVES
model.

5.4
Comments
from
Michael
Hutcheson,
URS
Corp.

5.4.1
Comment:
It
is
not
clear
why
NONROAD
needs
to
break
down
the
power
levels
above
750
hp
into
5
categories
for
populations.
Emission
factors
and
median
life
estimates
only
require
one
category
above
750
and
the
need
for
extra
categories
for
populations
should
be
explained.

EPA
Response:
EPA
included
five
power
level
categories
above
750
horsepower
for
flexibility
in
case
more
detailed
data
became
available
in
the
future
or
future
regulatory
efforts
needed
that
level
of
detail.

5.4.2
Comment:
The
reviewer
agrees
with
the
changes
in
SCC
codes
to
accommodate
fuel
types
and
new
equipment
classes
and
power
ranges.
As
in
comments
on
other
NONROAD
39
reports,
the
reviewer
would
like
to
see
better
descriptions
of
the
SCC
classifications
for
nonroad
equipment.

EPA
Response:
A
complete
description
of
the
equipment
and
corresponding
source
category
codes
(
SCCs)
modeled
by
NONROAD2002
is
included
in
Appendix
B
of
the
user's
guide.
EPA
will
include
a
reference
to
this
in
the
next
revision
of
the
Nonroad
Engine
Population
Estimates
report,
NR­
006b.

5.4.3
Comment:
The
reviewer
agrees
in
principle
with
the
use
of
the
B.
A.
H.
data
on
mobile
vs.
stationary
engine
breakdowns
for
appropriate
source
categories.
The
report
should
clarify
the
exact
SCC
categories
to
which
these
data
apply
and
how
the
data
is
applied
to
the
population
estimates.
For
instance,
if
the
percent
mobile
equipment
fractions
are
applied
to
populations
internally
in
NONROAD
then
the
user
should
have
this
information.
If
the
fractions
are
applied
prior
to
inputting
the
estimates
into
NONROAD,
then
users
who
make
their
own
estimates
need
to
know
to
apply
mobile
equipment
percentages
to
their
data
if
appropriate.

EPA
Response:
Table
2
in
the
Nonroad
Engine
Population
Estimates
report,
NR­
006b,
provides
the
source
category
codes
(
SCCs)
of
the
equipment
types
that
include
both
mobile
and
stationary
applications
in
the
heading.
EPA
is
clarifying
how
the
mobile/
stationary
fractions
are
applied
to
the
engine
populations
in
the
next
revision
of
this
report
(
NR­
006c).

5.4.4
Comment:
The
reviewer
disagrees
with
the
use
of
the
median
life
values
developed
for
NONROAD
for
use
in
population
estimates.
As
detailed
in
comments
submitted
previously
on
NR­
005b,
median
life
values
and
scrappage
estimates
do
not
take
into
account
all
factors
affecting
scrappage
and
therefore
are
inherently
inaccurate
when
estimating
growth.

EPA
Response:
Please
see
the
EPA
responses
to
comments
4.3.1
and
4.3.1a.
