Improvements for Heavy-Duty Engine
and Vehicle Test Procedures
Response to Comments
£% United States
Environmental Protect
Agency
-------
Improvements for Heavy-Duty Engine
and Vehicle Test Procedures
Response to Comments
Assessment and Standards Division
Office of Transportation and Air Quality
U.S. Environmental Protection Agency
United States
Environmental Protection
^1 Agency
EPA-420-R-22-014
July 2022
-------
Introduction 2
List of Commenters 2
List of Acronyms 2
1.1 General Support or Opposition 3
1.2 Use of GEM 3.5.1 for Model Year 2021 3
1.3 Concerns with GEM 3.5.1 and GEM 3.8 for Custom Chassis Applications 4
1.4 Concerns with GEM 3.8 Adjustment Factors 5
1.5 Improving the accuracy of GEM 3.8 6
1.6 Adjustment Factors 7
1.7 Features of GEM 3.8 9
1.8 Powertrain Test 11
1
-------
Introduction
The Environmental Protection Agency ("EPA") published a supplemental Notice of Proposed Rulemaking
("NPRM") on June 29, 2021 (86 FR 34189) to propose changes to the test procedures for heavy-duty
engines and vehicles to improve accuracy through additional amendments for certain aspects of the
modeling parameters in the Greenhouse gas Emissions Model ("GEM").
This action amends the regulations that implement our air pollutant emission standards for heavy-duty
engines and vehicles. The amendments in this final rule include corrections, clarifications, additional
flexibilities, and adjustment factors to the GEM compliance tool for heavy-duty vehicles. These
amendments modify the existing test procedures for heavy-duty highway engines and vehicles and
apply to the measurement of C02 emissions.
This Response to Comments contains a detailed summary of the comments we received on the
supplemental NPRM as well as our analysis and response to the comments. The supplemental Final
Rulemaking ("FRM") published in the Federal Register includes the final regulations resulting from this
rulemaking, along with further description and rationale for our conclusions.
List of Commenters
Commenter
Docket ID
Anonymous public comment
EPA-HQ-OAR-2019-0307-0101
REV Group, Inc.
EPA-HQ-OAR-2019-0307-0102
Allison Transmission, Inc.
EPA-HQ-OAR-2019-0307-0103
Truck and Engine Manufacturers Association ("EMA")
EPA-HQ-OAR-2019-0307-0104
List of Acronyms
Acronym
Definition
CFR
Code of Federal Regulations
C02
Carbon Dioxide
EPA
Environmental Protection Agency
FEL
Family Emission Limit
FR
Federal Register
FRM
Final Rulemaking
GEM
Greenhouse Gas Emissions Model
MY
Model Year
MPH
Miles-per-hour
NPRM
Notice of Proposed Rulemaking
V
Version (e.g., "V3.8" means GEM version 3.8)
2
-------
1.1 General Support or Opposition
Comment
Response
From: Anonymous public comment
The improvements to the greenhouse gas emissions model
(GEM) seems to be the right direction to go when assessing
the C02 standards for heavy-duty vehicles. By taking the
existing test procedures and making modifications to the way
C02 measurements are measured, the model will better be
able to capture how efficiently fuel is being used. The GEM
model is a tool that will help determine fuel efficiency and the
changes in the 3.7 model will make all the difference. In the
3.8 model the changes to the torque inputs and the
adjustments to the idle fueling and the input value for neutral
coasting are both that will be necessary to upgrade the model
to where it needs to be.
The comment generally affirms the
proposed rule.
1.2 Use of GEM 3.5.1 for Model Year 2021
Comment
Response
From: EMA
GHG Phase 2 rule became effective with model year ("MY")
2021 and, since that MY has begun, it may not be feasible
for manufacturers to utilize GEM V3.8 for certification and
compliance to the rule. We therefore support the proposal
in the SNPRM to allow manufacturers the option of using
the previous GEM version, V3.5.1, for demonstrating
compliance to the GHG Phase 2 standards for MY 2021
vehicles.
We have finalized provisions
intended to provide appropriate
flexibility to transition to a revised
version of GEM, see Section III.B. of
the preamble "Allowable version of
GEM for certification and
compliance" for more details.
3
-------
1.3 Concerns with GEM 3.5.1 and GEM 3.8 for Custom Chassis Applications
Comment
Response
From: REV Group, Inc.
The subject NPRM proposes to make corrections and add
adjustment factors to the EPA
Greenhouse Gas Emissions Model (GEM) to improve this
compliance tool for heavy-duty
vehicles while more closely matching the outputs produced by
the original GEM version 3.0 that was used to establish the
C02 standards for model years 2021 and later in the 2016
Heavy-duty Phase 2 Final Rule.
REV Group business units utilized the provisions in 40 CFR
1037.105(h) for custom chassis
certification in the other bus and emergency vehicle categories
and prepared MY21certification applications using GEM V3.0.
However, when these business units subsequently ran the
same configurations using GEM 3.5.1, FEL results were an
average of 3.6% higherthan GEM 3.0. When the business units
ran the same configurations using GEM 3.8, FEL results were
an average of 3.8% higherthan GEM 3.0.
REV Group requests that EPA staff look into this issue and add
adjustment factors for the
custom chassis other bus and emergency vehicle categories to
GEM 3.8 that would result in achieving the stated intent of this
NPRM
We agree with the comment that
GEM 3.5.1 gives different results for
custom chassis vehicles than with
GEM 3.0, and we have made
multiple changes to GEM that are
included in GEM 4.0 to align the
custom chassis results with the
results from GEM 3.0. The first of
these changes was to align the
default engine fuel maps used for
custom chassis vehicles between the
two version of GEM. The second
change that was made was including
adjustment factors in GEM 4.0 to
more closely match the outputs
produced by the same vehicle
configurations when using the
original GEM version 3.0 (GEM 3.0
was used in the 2016 Heavy-duty
Phase 2 Final Rule to establish the
C02 standards for model years 2021
and later.)
From: Allison Transmission Inc.
Regarding EPA's improvement to correct how GEM adjusts the
idle fueling of the transient cycle, Allison is concerned that EPA
did not achieve the outcome that they expected for the
Custom Chassis category. For this improvement, EPA used
correction factors to adjust the FEL scores in GEM 3.8. Allison's
understanding is that the goal is to adjust FEL scores to be
closer to GEM 3.0 values. Allison staff analyzed the results
using GEM 3.8 with the correction factors and found that the
results in Custom Chassis category were unfavorable. The
unfavorable outcome was especially noticeable in the Refuse
and Other Bus applications and even resulted in an adverse
impact in the School Bus application. The Refuse and Other
Bus FEL values were only slightly improved. The School Bus FEL
values were worse. Please refer to plots in Appendix A.
4
-------
1.4 Concerns with GEM 3.8 Adjustment Factors
Comment
Response
From: Allison Transmission Inc.
As Allison has stated in earlier comments of October 1,
2015, transit buses are disadvantaged in GEM because
transit buses operate in urban areas and have a duty cycle
of frequent starts and stops, with no 65 mph operation and
very limited 55 mph operation so the current vocational
test cycle does not represent how they operate. And, we
are learning from transit bus OEM customers that it is
challenging to determine a compliance path. The Custom
Chassis option was supposed to offer flexibility for specialty
OEMs. Now, further compounding the challenge is the
unexpected outcome of the proposed correction factors.
Allison recommends that EPA re-evaluate the GEM 3.8 fix
using correction factors, especially for the Custom Chassis
category. Additionally, Allison recommends EPA consider a
special category in GEM for transit buses.
We recognize that GEM utilizes a
fixed set of drive cycles and limited
number of duty cycle weights. The
Urban weighting of 90% ARB
Transient and 10% 55 mph cruise is
not too far off the operation
described. The custom chassis
"other bus" category uses the same
weighting factors that are used in
the Urban category. The included
Adjustment Factors in GEM 4.0, that
are described in Section III of the
Preamble, were determined so that
the outputs of GEM 4.0 more closely
match the outputs produced by the
original GEM version 3.0 (GEM 3.0
was used to establish the vocational
and custom chassis C02 standards
for model years 2021 and later in
the 2016 Fleavy-duty Phase 2 Final
Rule.) Therefore, the results from
GEM 4.0 for transit buses should be
comparable to the same vehicles
run in GEM 3.0.
5
-------
1.5 Improving the accuracy of GEM 3.8
Comment
Response
From: EMA
GEM is required to model CO2 emissions to show
compliance with the vehicle standards in the Greenhouse
Gas ("GHG") Phase 2 rule, and the SNPRM proposes GEM
improvements and adjustment factors for GEM results. The
SNPRM proposes to incorporate the changes in a new
version of GEM, V3.8. We appreciate the collaborative
approach the Agency has taken toward improving GEM and
offer these comments in that same constructive spirit.
GEM is a sophisticated compliance tool that models the CO2
emissions and fuel consumption of a vehicle over the
speeds and loads in prescribed duty cycles. GEM also must
accurately simulate real-world operation of a vehicle and
accurately assess the impacts of different engine and
vehicle fuel-saving technologies. Since trucking fleets place
a high value on fuel consumption performance when
purchasing a new commercial vehicle, the technologies that
manufacturers develop must perform effectively in-use and
show an equivalent benefit in GEM. Otherwise, there would
be a disconnect between the real-world performance of
fuel-saving technologies and technologies that show a CO2
and fuel consumption benefit in GEM. Accordingly, we
support EPA's efforts to refine GEM, and we believe that
incorporating the following proposed changes in GEM V3.8
will enhance the accuracy of the model
• Changed limits on engine input to allow small
negative torque inputs.
• Corrected how GEM adjusts the idle fueling of the
transient cycle by using the same idle duration time
both for subtracting the idle fuel rate from the
transient cycle average engine fuel map and for
adding back in the simulated idle fuel rate.
• Added an option for vocational vehicles to input a
value for neutral coasting in GEM and amend the
related test procedure in 40 CFR 1037.520(j)(l).
• Corrected manual and automated manual
transmissions to perform clutched upshifts for
Heavy HDV.
We agree with all of the changes to
GEM that EMA suggested in their
comments to improve the accuracy
of the model. These changes are
incorporated in GEM 4.0 and are
described in Section III.A. of the
Preamble.
6
-------
1.6 Adjustment Factors
Comment
Response
From: EMA
Adjustment Factor Calculations
The SNPRM proposes adjustment factors to the GEM V3.8
results to "ensure that these changes to GEM do not
change the effective stringency of the GHG Phase 2 CO2
standards." See, id. at 34,192. ThSe SNPRM proposes that
manufacturers apply the adjustment factors to their
unrounded GEM V3.8 outputs. However, the process EPA
used to determine the proposed adjustment factors
included a minor mathematical flaw that negatively
affected the results of the calculations. Even though the
SNPRM proposes applying the adjustment factors to
unrounded GEM V3.8 outputs, and the adjustment factors
are carried out to four decimal places, the Agency
calculated the adjustment factors using GEM results
rounded to the nearest whole number for vocational
vehicles and the nearest tenth for tractors. We believe that
calculating the adjustment factors using unrounded GEM
results would yield more accurate adjustment factors.
Accordingly, EPA should recalculate all the proposed
adjustment factors using unrounded GEM results and
include those more accurate adjustment factors in Table 10
of 40 C.F.R. § 1037.520.
We agree with the comment to
calculate adjustment factors from
unrounded GEM results. The
adjustment factors included in GEM
4.0 have been determined from
unrounded results.
7
-------
Comment
Response
From: EMA
Applying the Adjustment Factors
The SNPRM proposes that manufacturers first produce GEM
outputs using V3.8 and then modify those results by
applying the adjustment factors. Such a two-step process
would be resource intensive and wasteful. Moreover, the
SNPRM proposes that each manufacturer apply one of the
33 different adjustment factors to each of the multitude of
GEM outputs it produces, with the 33 adjustment factors
changing with each of the three stringency steps in the rule.
Such an after-the-fact application of the myriad and
changing adjustment factors is sure to lead to errors.
Instead of that burdensome and error-prone approach, EPA
should incorporate the adjustment factors into the GEM
programming so the computer model will apply the proper
adjustment factor before producing a result.
We agree with the comment to
include adjustment factors in GEM.
GEM 4.0 includes the adjustment
factors and applies them to the
unrounded composite GEM result
before outputting the "Default FEL
C02 Emissions".
From: EMA
Adjustment Factors for Tractors with Automatic
Transmissions
As stated before, we endorse the proposal to correct GEM
V3.8 programing for clutched upshifts for tractors with
manual and automated manual transmissions. The
performance of tractors with those transmissions in GEM
V3.8 will more closely match real-world operation, with the
adjustment factors appropriately increasing the GEM
output to align with the GHG Phase 2 stringency values.
However, the GEM modification for clutched upshifts does
not affect a small but significant number of tractors that are
built with automatic transmissions, because those
transmissions do not benefit from the clutched upshift
change due to their unique design. Thus, the adjustment
factors inappropriately increase the GEM V3.8 outputs for
those tractors. To correct that error, EPA should set the
adjustment factors to zero for tractors with automatic
transmissions.
We agree with the comment that
the adjustment factor for tractors
with automatic transmissions should
be set to zero and have included this
in GEM 4.0.
8
-------
1.7 Features of GEM 3.8
Comment
Response
From: EMA
Input Files Generated with Prior GEM Versions
GEM V3.8 will not run properly with a fuel map, a
transmission power loss map, or a rear axle power loss map
that was generated using a prior version of GEM.
Reproducing all component input maps using GEM V3.8
would consume tremendous resources and produce nearly
identical maps. To avoid unnecessarily wasting time and
resources, EPA should modify the GEM V3.8 programing to
fully accept component input maps produced with prior
versions of GEM. Additionally, EPA must only audit
component input maps using the same version of GEM that
the manufacturer used to produce the map.
We agree with the comment that
GEM should allow input files from
previous versions of GEM as long as
the inputs are the same. GEM 4.0
allows inputs that were created for
prior versions of GEM as long as the
input file includes all the inputs
needed for GEM 4.0.
We agree that EPA will only audit
component inputs for fuel maps
using the same version of GEM that
the manufacturer used to create the
input.
From: EMA
Steady-State Fuel Map Inconsistency
GEM V3.8 correctly removes from the default steady-state
fuel maps points that are below 105% of the maximum
speed and 120% of the maximum torque. However, the
regulatory text in 40 C.F.R. § 1036.535(d)(2) calls for
removing points that are below 115% of the maximum
speed and 115% of the maximum torque. EPA should
correct the regulatory text.
We agree with the comment that
GEM correctly removes from default
steady-state fuel map points and as
discussed in Section IV of the
Preamble we have removed this
requirement from 40 CFR 1036.535.
We have finalized the removal of
the text from 40 CFR 1036.535
because the default fuel maps are
now included in GEM starting with
version 3.5.1, so the only reason for
keeping the regulatory text would
be to note what GEM does.
From: EMA
Transmission Cost Map Error
GEM V3.8 includes a programming error that affects the
generation of transmission cost maps. To correct the error,
EPA should revise GEM V3.8 to use between 6 and 70 points
to determine whether a default steady-state fuel map is
needed for the transmission cost map.
We agree with the comment that
GEM can incorrectly generate a
transmission cost map in
circumstances where only the idle
portion of the fuel map is input into
GEM and have changed how GEM
determines what data is used to
determine the transmission cost
map in GEM 4.0 to address this
issue. If 25 or fewer points are input,
a default map will be used to
determine the transmission cost
map, otherwise the cost map will be
constructed from the provided
steady state points. We chose 25 or
fewer points because 25 is above
the upper limit of points needed to
define the engine fuel map at idle,
9
-------
but below the number of points
needed for the complete engine fuel
map.
Comment
Response
From: EMA
Drive Idle Fueling Interpolation Error
The interpolation of drive- idle fueling in GEM V3.8 is
inappropriately impacted by the surrounding default
steady-state fuel map points when the option of using three
cycle-average fuel maps is utilized. To correct the error, EPA
should modify GEM V3.8 to only use measured drive-idle
fuel map points for the interpolation of drive- idle fuel
consumption. That is, GEM should not merge measured
drive-idle fuel map points with default steady-state fuel
map points.
We agree with the comment, and
have made changes included in GEM
4.0 to not merge measured drive-
idle fuel map points with default
steady-state fuel map points.
When a default map is required, it
is used for simulation and the idle
data is used in the post process
calculations.
From: EMA
Cycle-Average Fuel Map Regression Error
GEM V3.8 uses Regression Method 7 to determine fueling
during the 55 and 65 MPH cruise duty cycles in cycle-
average fuel maps. However, that planar regression is not
accurate. EPA should modify GEM to use Regression
Method 13 instead. Please note that GEM V3.8 should
continue to use Regression Method 7 to determine fueling
during the transient duty cycle in cycle-average fuel maps.
We agree with the comment that
regression Model 7 is not accurate
for all engines for the 55 and 65
mph cruise duty cycles and have
changed the regression model in
GEM 4.0 for the cycle average cruise
cycles as described in Section III of
the Preamble. The regression
model used is not Method 13
because Method 13 requires more
data points to prevent overfitting of
the model. The Method finalized is
more accurate than 7 and 13 and
works when the fuel maps for the
55mph and 65mph cruise cycles are
combined. By combining the 55mph
and 65mph fuel maps, the number
of test points available to fit the
model double without requiring
additional testing.
From: EMA
Engine Speed Tolerance Error
The engine speed tolerance is too sensitive in GEM V3.8.
Specifically, GEM will report an error for engine idle speed
when slight testing variations impact the speed. For
example, if an engine's lowest measured idle speed is
700.08 RPM, GEM will report an error when attempting to
We agree with the comment and
have changed the idle speed
tolerance in GEM 4.0 as described in
Section III of the Preamble.
10
-------
simulate 700 RPM. EPA should modify GEM V3.8 to include
a reasonable tolerance for engine speeds.
1.8 Powertrain Test
Comment
Response
From: EMA
Idle Reduction in Powertrain Test
The benefit of automatic engine shutdown idle reduction
technology is not fully realized during a powertrain test.
Additionally, realizing benefits for neutral idle and stop-
start technologies requires testing with the features
enabled, which may lead to significant file proliferation to
cover multiple combinations of idle reduction technologies.
To ensure the full benefits of idle reduction technologies,
and to reduce the number of powertrain input files, EPA
should modify GEM V3.8 to allow manufacturers the option
of directly identifying those features.
We agree with the comment to
include in GEM the ability to
recognize idle shutdown
technologies when the powertrain
test method is used to generate the
fuel maps. As described in Section
III of the Preamble we have included
in GEM 4.0 changes that will allow
automatic engine shutdown and
stop-start to be selected in the
vehicle input file instead of requiring
these technologies to be captured in
the powertrain fuel maps.
From: Allison Transmission Inc.
As we look ahead to using GEM for Powertrain Certification,
Allison suggests that EPA consider an enhancement of the
GEM Technology Improvement section. Today, GEM's
Technology Improvement section provides a Yes/No entry
for Neutral Idle. For Powertrain Certification, however, an
entry of "Yes" in the Neutral Idle field simply allows a trivial
computation using "Drive Idle". Allison recommends that
for Powertrain Certification, that "Yes" in the Neutral Idle
field runs the same or similar Neutral Idle computation as
the one used for stand-alone components. This
enhancement would provide an option for the certifier to
use a default computation when the collection of
powertrain data is overly costly and burdensome.
We understand the comment as a
request to include in GEM the ability
to recognize if the vehicle includes
Neutral Idle technology with a
Yes/No entry when the powertrain
test method is used to generate the
fuel maps. To allowthis, GEM
would have to make an assumption
about the load from the
transmission on the engine when
neutral idle is enabled and disabled.
To do this in a representative way,
GEM would need additional inputs,
which may require additional test
procedures. For these reasons, with
GEM 4.0 we have not enabled the
ability to select Neutral Idle with a
Yes/No entry at this time.
11
------- |